IMPACT OF MICRO HEALTH INSURANCE-

CASE STUDY OF SAMPOORNA SURAKSHA

PROGRAMME IN KARNATAKA
THESIS

Submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

By

SAVITHA

(Reg.No.: 081043HM08P03)

DEPARTMENT OF HUMANITIES, SOCIAL SCIENCES AND

MANAGEMENT
NATIONAL INSTITUTE OF TECHNOLOGY KARNATAKA,
SURATHKAL, MANGALORE -575025

NOVEMBER 2012
 DECLARATION

I hereby declare that the Research Thesis entitled ‘IMPACT OF MICRO HEALTH
INSURANCE- CASE STUDY OF SAMPOORNA SURAKSHA PROGRAMME IN
KARNATAKA’ which is being submitted to the National Institute of Technology
Karnataka, Surathkal in partial fulfillment of the requirements for the award of the Degree
of Doctor of Philosophy in Humanities, Social Sciences and Management is a bonafide
report of the research work carried out by me. The material contained in this Research
Thesis has not been submitted to any University or Institution for the award of any degree.

(Signature of the Research Scholar)

Name : Savitha

Reg.No. : 081043HM08P03

Department : Humanities, Social Sciences and Management

Place : NITK, Surathkal

Date :
 CERTIFICATE

This is to certify that the Research Thesis entitled ‘IMPACT OF MICRO HEALTH
INSURANCE- CASE STUDY OF SAMPOORNA SURAKSHA PROGRAMME IN
KARNATAKA’ submitted by Savitha (Register number: 081043HM08P03) as the record
of the research work carried out by her, is accepted as the Research Thesis submission in
partial fulfillment of the requirements for the award of degree of Doctor of Philosophy.

Dr.K.B.Kiran

Research Guide

(Signature with Date and Seal)

Chairman-DRPC

(Signature with Date and Seal)

 ACKNOWLEDGEMENT
When I embarked on my research work, I thought it to be a difficult lonely
journey of many years. When I look back, I realize that many people gave their time and
energy to make my journey memorable. It would not have been possible to write this
doctoral thesis without the help and support of the kind people around me, to only some
of whom it is possible to give particular mention here. Firstly, I would like to
acknowledge my deep sense of gratitude to Prof. Dr. K.B. Kiran, my research guide, who
has been invaluable on both an academic and a personal level, for his constant motivation
and encouragement throughout my research work. His support throughout my research
period was unimaginable. I appreciate all his contribution to make my PhD experience
productive and stimulating, at the same time giving me confidence to pursue the
extensive fieldwork with perseverance and enthusiasm.
The panel members of Research Proposal Appraisal Committee, National Institute
of Technology Karnataka have contributed immensely by their valuable comments and
constructive suggestions. I take this opportunity to express my gratitude to Prof. Aloysius
H.Sequeira, Dr. Shashikantha Koudur, Dr. A. Kandasamy and Dr. Vidya Shetty K. I
would like to acknowledge the friendship and academic association of Dr. Naveen Kumar
Shetty, National Institute of Bank Management, Pune. I use this opportunity to thank Dr.
Ankur Sarin, Indian Institute of Management Ahmadabad for stimulating comments and
valuable advice on research methods.
I sincerely thank Dr. Veerendra Heggade, Dharmadhikari, Shri Kshetra
Dharmasthala, for giving me an opportunity to research on Sampoorna Suraksha
Programme. I am particularly grateful to Dr.L.H. Manjunath, Executive Director and Mr.
K.V.Bhat, Director, Sampoorna Suraksha Programme for their enthusiasm and co-
operation during data collection. I would like to thank Mr. Sampat Kumar, Mr.Satish
Shetty, Ms.Krishnaveni, Mr.Umanath, Mr.Umesh, Mr.Mahabala, Ms.Kumuda,
Mr.Yogish, Mr. Ganesh, Mr.Gopal, Mr.Jagadeesh, Mr. Keshava, Ms.Nethravathi,
Ms.Latha and many other supervisors and field staffs for their unfailing support during
my ‘on-foot’ survey and data collection. Collecting data is invariably a trying experience
and many people helped with this, for which I would like to thank them wholeheartedly.
Thousands of people in the districts of Dakshina Kannada, Gadag and Uttara Kannada
have given their valuable time to not only let me pilot my questionnaire without which
the data would have been far less useful; but also to administer the survey questionnaires
and focus group discussions. Without their generosity, there would be nothing to work
with. I thank them all for their support.
I would also like to acknowledge Dr. Devaraj K., Director, S.D.M. Post Graduate
Centre for Management Studies and Research, Mangalore for his support and
encouragement. I will forever be thankful to Fr. Denzil Lobo S.J., Director, Aloysius
Institute of Management and Information Technology, Mangalore for encouraging me to
take initial steps in research.
Lastly, no words can adequately express my debt of gratitude to my late father,
Shri Chandrashekar Basri and my mother Smt. Jayalaxmi Basri for generating in me a
perennial interest in higher studies and motivating me to achieve the apex of education.
Today I bow down to them for everything, without which I would be nowhere. I would
like to thank my brothers Mr.Sudhakar Basri, Mr.Suresh Basri and sister Ms. Sumithra
for their support. I will be failing in my duty if I do not mention here the tremendous
support, motivation, moral and emotional support I received from my husband Prof.Dr.
B.P.Shelley. He introduced the research world to me, motivated me in difficult times and
was instrumental in instilling confidence. He had faith in my intellect and unconditionally
supported me during my good and bad times. I deeply appreciate the tolerance,
encouragement and understanding of my loving son Adithya. No words can express my
appreciation for their continuous love and support.

Savitha
 ABSTRACT

Micro health insurance (MHI) is a pivotal and innovative health financing

mechanism that mitigates iatrogenic poverty thereby providing financial protection to the
informal sector. However, the limited evidence from India does raise research questions
regarding the effectiveness of MHI schemes in achieving these objectives. Thus, this
thesis focuses on this area of research gap in MHI to evaluate its impact on (i) financial
protection, (ii) social inclusion, (iii) resource mobilization, and (iv) the role of scheme
characteristics on its performance. Sampoorna Suraksha Programme (SSP), a MHI
scheme in Karnataka was chosen to study the research gap on impact and performance
using a descriptive survey research methodology using self-administered validated
questionnaire. Multi-stage, clustering design with random selection procedures was
adopted to collect quantitative data from 416 insured, 366 newly insured and 364
uninsured self-help group (SHG) households of Dakshina Kannada, Uttara Kannada and
Gadag districts in Karnataka State. Qualitative data was collected using in-depth
interviews with network hospital staff, field staff and administrators of SSP, and focus
group discussion with the members of SHG households.
The results on financial protection indicated a lower out of pocket expenditure
and catastrophic health expenditure, higher utilisation of inpatient services, desired
health-seeking behaviour and decreased borrowing for the insured households. There was
no impact on access to care and social inclusion in enrolment. On the contrary, the
inadequate resource mobilization evident from the study resulting from high claims ratio
and lower premium collection would pose a threat to the long-term financial
sustainability of SSP. The study indentified certain design features that influenced the
outcome of SSP. The findings of this study provide adequate evidence to substantiate the
effectiveness and positive impact of SSP on financial protection and MHI certainly is
advocated as a financing alternative to mitigate iatrogenic poverty.
Key words: Micro health insurance, impact, financial protection, health financing,
catastrophic.
 CONTENTS

Declaration
Certificate
Acknowledgement
Abstract
Table of contents
Abbreviations
Page No.
CHAPTER 1 INTRODUCTION
1.1 Background 1
1.2 Micro Health Insurance 5
1.3 Statement of the problem and Research Questions 9
1.4 Research Objectives 11
1.5 Research Hypothesis 12
1.6 Scope of the study 14
1.7 Organization of thesis 15
CHAPTER 2 REVIEW OF LITERATURE
2.1 Introduction 17
2.2 Health system goals and the role of micro health insurance 17
2.3 Basic concepts and principles of micro health insurance 20
2.4 Financial protection 20
2.4.1 Access to health care services 20
2.4.2 Health seeking behavior 22
2.4.3 Utilisation of health services 24
2.4.4 Out-of –pocket and catastrophic expenditure 26
2.5 Ex-post risk coping strategies 29
2.6 Social inclusion 31
2.6.1 Demand for health insurance 32
2.6.2 Adverse selection and moral hazard 34
2.7 Resource mobilization 36
2.7.1 Quality of care 36
2.8 Association between characteristics of the scheme and outcome 37
2.8.1 Technical characteristics 38
2.8.2 Management characteristics 39
2.8.3 Organizational characteristics 40
2.8.4 Institutional characteristics 41
2.9 Summary 42
CHAPTER 3 RESEARCH METHODOLOGY
3.1 Introduction 49
3.2 Research Approach 49
3.3 Data collection methods and data sources 52
3.4 Sampling design 54
3.4.1 Sampling procedure 55
3.5 Study settings 56
3.5.1 Profile of Gadag 58
3.5.2 Profile of Dakshina Kannada 58
3.5.3 Profile of Uttara Kannada 59
3.6 Shri Kshetra Dharmasthala Rural Development Project 59
3.6.1Sampoorna Suraksha Programme 60
3.6.2 Key features of Sampoorna Suraksha Programme 60
CHAPTER 4 IMPACT OF SAMPOORNA SURAKSHA
PROGRAMME ON FINANCIAL PROTECTION
4.1 Description of surveyed households 71
4.2 Impact of Sampoorna Suraksha Programme on Access to
Care
4.2.1 Introduction 77
4.2.2 Incidence of illness in the sampled households 78
4.2.3 Access to health care 79
4.2.4 Barriers to access care 82
4.2.5 Summary 83
4.3 Impact of Sampoorna Suraksha Programme on Health
Seeking Behaviour
4.3.1 Introduction 85
4.3.2 HSB by insurance status 86
4.3.3 Sequence of health seeking behavior 89
4.3.4 Discriminant analysis of health seeking behaviour of 90
insured and uninsured
4.3.5 Discriminant analysis of underlying reasons for health 92
seeking behaviour
4.3.6 Econometric estimation of HSB 95
4.3.6.1 Income related equity in health seeking behaviour 96
4.3.6.2 Gender related equity in health seeking behaviour 97
4.3.6.3 Area related equity in health seeking behaviour 97
4.3.6.4 Results of econometric estimation 98
4.3.7 Summary 100
4.4 Impact of Sampoorna Suraksha Programme on Utilisation
of Health Services
4.4.1 Introduction 103
4.4.2 Utilisation of health services and moral hazard behaviour 104
4.4.3 Patient perceived result of the treatment 105
4.4.4 Econometric estimation on the probability of 105
hospitalisation
4.4.4.1 Need factor determining utilisation 106
4.4.4.2 Predisposing factors determining utilisation of 106
health services
4.4.4.3 Enabling factors determining utilisation 107
4.4.4.4 Results of econometric estimation 110
4.4.5 Summary 116
4.5 Impact of Sampoorna Suraksha Programme on Out of
Pocket Expenditure
4.5.1 Introduction 119
4.5.2 Out of pocket expenditure incurred for health care 120
services
4.5.3 Out of pocket expenses as percentage of annual 121
consumption expenditure
4.5.4 Econometric estimation on the probability of out of 123
pocket expenses
4.5.4.1 Characteristics of individuals 123
4.5.4.2 Characteristics of households 127
4.5.4.3 Characteristics of the community 129
4.5.4.4 Results of econometric estimation 130
4.5.5 Analysis of Sampoorna Suraksha claims 134
4.5.6 Summary 134
4.6 Impact of Sampoorna Suraksha Programme on
Catastrophic Health Expenditure
4.6.1 Introduction 137
4.6.2 Catastrophic expenditure among insured and uninsured 138
households
4.6.3 Number of visits to health care facilities and CHE 140
4.6.4 Econometric estimation on the probability of CHE 140
4.6.4.1 Characteristics of individuals 141
4.6.4.2 Characteristics of households 145
4.6.4.3 Characteristics of community 149
4.6.4.4 Results of econometric estimation 151
4.6.5 Summary 161
4.7 Impact of Sampoorna Suraksha Programme on Risk
Coping Strategies
4.7.1 Introduction 165
4.7.2 Access to self finance during health crisis 167
4.7.3 Risk coping strategies during health crisis 168
4.7.4 Relationship between borrowing and health insurance 170
status
4.7.4.1 Determinants of borrowing for treatment 171
4.7.4.1a Characteristics of individuals 171
4.7.4.1b Characteristics of households 171
4.7.4.1c Characteristics of the community 174
4.7.4.2 Econometric estimation on incidence of 176
borrowing
4.7.4.3 Econometric estimation of amount of borrowing 185
4.7.5 Relationship between health insurance status and use of 188
savings
4.7.5.1 Determinants of the use of savings 188
4.7.5.1a Characteristics of individuals 188
4.7.5.1b Characteristics of households 189
4.7.5.1c Characteristics of the community 190
4.7.5.2 Econometric estimation on the probability of 191
use of savings
4.7.5.3 Econometric estimation of the amount of 194
savings used for treatment
4.7.6 Sale of assets to pay for medical expenses 196
4.7.7 Summary 196
CHAPTER 5 ENROLMENT IN SAMPOORNA SURAKSHA
PROGRAMME
5.1 Introduction 201
5.2 Incidence of illness among insured and uninsured households 202
5.3 Determinants of enrolment in the surveyed households 202
5.3.1 Enabling factors 203
5.3.2 Predisposing factors 204
5.3.3 Adverse selection 206
5.3.4 Econometric estimation of the determinants of 208
enrolment in SSP
5.4 Reasons for enrolment in SSP 212
5.5 Non enrolment in SSP 213
5.5.1 Demand side factors 214
5.5.2 Scheme related factors 216
5.6 Summary 217
CHAPTER 6 RESOURCE MOBILIZATION AND EFFECT
OF FEATURES OF SSP ON THE OUTCOME
6.1 Introduction 221
6.2 Resource mobilization 223
6.2.1 Revenue mobilization 223
6.2.1.1 Premium structure of SSP 223
6.2.1.2 Enrolment and premium collection 224
6.2.1.3 Allocation of premium to insurance company 226
and SSP
6.2.1.4 Benefits provided by SSP since inception 227
6.2.2 Financial sustainability of SSP 229
6.2.3 Perceived quality of care of hospitals 231
6.2.4 Summary 233
6.3 Effect of features of SSP on Financial Protection, 237
Enrolment and Resource Mobilization
6.3.1 Technical design characteristics 237
6.3.1.1 Revenue collection 237
6.3.1.2 Risk pooling 242
6.3.1.3 Strategic purchasing 243
6.3.2 Management characteristics 245
6.3.2.1 Staff 245
6.3.2.2 Culture 245
6.3.2.3 Access to information 246
6.3.3 Organizational characteristics 246
6.3.3.1 Forms of organization 246
6.3.3.2 Incentive regime 247
6.3.3.3 Linkages 247
6.3.4 Institutional characteristics 248
6.3.4.1 Stewardship 248
6.3.4.2 Governance 249
6.3.4.3 Insurance markets 249
6.3.4.4 Factor and product markets 249
6.3.5 Summary 253
CHAPTER 7 SUMMARY OF FINDINGS, CONCLUSION
AND SUGGESTIONS
7.1 Introduction 259
7.2 Summary of findings 259
7.2.1 Findings on the impact of SSP on financial protection 260
7.2.2 Findings on the impact of SSP on social inclusion 262
7.2.3 Findings on the resource mobilization of SSP 262
7.2.4 Findings on characteristics of SSP and its effect on 263
enrolment, financial protection and resource mobilization
7.3 Main findings and conclusion 264
7.4 Managerial implications 266
7.5 Policy implications 268
7.6 Limitations 270
7.7 Suggestions for future research 271
APPENDICES
I English household questionnaire used for the survey 273
II English qualitative interview schedules used for the survey 280
III Pilot study report 286
IV Surveyed Districts,Taluks, Valayas and Karyakshetras 287
V Profile of Karnataka and sampled districts 288
REFERENCES 289
RESEARCH PUBLICATIONS 305
BIODATA 307
 LIST OF TABLES
Table No. Title Page No.
1.1 Comparisons of micro health insurance schemes in India 8
2.1 Literature review on financial protection 44
2.2 Literature on the determinants of enrolment 46
2.3 Definition of the concepts used in the study 48
3.1 Data collected using questionnaire 53
3.2 Key features of Sampoorna Suraksha Programme 62
4.1 Socio-economic characteristics of surveyed households 73
4.2 Basic socio-economic characteristics of head of the 74
households
4.3 Assets ownership of surveyed households 75
4.4 Demographic and health related characteristics of ill 79
persons in the sample
4.5 Socio-economic characteristics and access to care 81
4.6 Health seeking behaviour in the first visit: Comparison by 88
insurance status
4.7 Health seeking behaviour in the second visit: comparison by 88
insurance status
4.8 Discriminant analysis of health seeking behaviour 92
4.9 Health seeking behaviour– Reasons given by the surveyed 93
individuals
4.10 Discriminant analysis of the factors determining the choice 94
of health care facility
4.11 Income related equity in health seeking behaviour 96
4.12 Gender of the ill person and HSB in private and public 97
hospitals
4.13 Health seeking behaviour of surveyed individuals in the 98
first episode of illness
4.14 Definition and measurement of variables 99
4.15 Probability of hospitalisation in private facilities (Model 1) 100
4.16 Effect of the treatment perceived by the individuals 105
4.17 Gender of the ill persons and type of treatment 107
4.18 Description of predisposing, enabling and need factors 109
4.19 Definition and measurement of variables 111
4.20 Probability of hospitalisation: Results of Model 2a 113
4.21 Probability of hospitalisation: Results of Model 2b 114
4.22 Probability of hospitalisation: Results of Model 2c 115
4.23 Probability of hospitalisation: Results of Model 2d 116
4.24 Cost of medical care 121
4.25 Out of pocket expenses as a percentage of the annual 123
consumption expenditure
4.26 Total expenses incurred for different illness 124
4.27 Gender of ill persons and total out of pocket expenditure 125
4.28 Average out of pocket expenses incurred by women 126
4.29 Average out of pocket expenses incurred for outpatient and 127
inpatient treatment
4.30 Total out of pocket expenses by income quintile 128
4.31 Area of residence and total out of pocket expenses 130
4.32 Estimation of Model 3a: Health insurance and OOP 132
expenditure
4.33 Estimation of Model 3b: Health insurance and OOP 133
expenditure
4.34 Catastrophic health expenditure by insurance status 139
4.35 Catastrophic health expenses and the number of visits to 140
health facility
4.36 Catastrophic health expenditure and gender of ill person 142
4.37 Catastrophic health expenditure and type of illness 143
4.38 Catastrophic health expenditure and treatment 144
4.39 Association between job status and CHE 145
4.40 Head of the household as unskilled labourer and CHE 146
4.41 Catastrophic health expenditure: Intra-income class 147
comparison
4.42 Catastrophic health expenditure: Inter-income class 148
comparison
4.43 Catastrophic health expenditure and area of residence 149
4.44 Independent variables included in CHE binary logistic 150
regression model
4.45 Measurement of independent variables: CHE models 152
4.46 Probability of Catastrophic health expenditure: Estimated 155
results of Model 4a
4.47 Estimated results of Model 4b: Probability of Catastrophic 156
health expenditure
4.48 Estimated results of Model 4c: Probability of Catastrophic 158
health expenditure
4.49 Estimated results of Model 4d: Probability of Catastrophic 159
health expenditure
4.50 Estimated results of Model 4e: Probability of Catastrophic 161
health expenditure
4.51 Availability of money to pay the medical expenses 168
4.52 Source of financial resources during crisis- Ex ante 169
strategies
4.53 Amount of money mobilized to pay for health care 170
4.54 Borrowing and job status of head of the household 173
4.55 Borrowing in income class: Comparison by health 174
insurance status
4.56 Borrowing and area of residence 175
4.57 Borrowing and district of residence 175
4.58 Description of independent variables of borrowing model 176
4.59 Measurement of independent variables included in the 178
regression analysis
4.60 Probability of Borrowing: Estimated results of Model 5a 181
4.61 Probability of Borrowing: Estimated results of Model 5b 182
4.62 Probability of Borrowing: Estimation of Model 5c 183
4.63 Probability of Borrowing: Estimation of Model 5d 185
4.64 Estimation of Model 5e: Health insurance and amount of 187
borrowing
4.65 Description of independent variables included in the 191
savings model
4.66 Probability of the use of savings: Estimated results of 193
Model 5f
4.67 Estimated results of Model 5g: Health insurance and 195
amount of savings
5.1 Incidence of illness and enrolment in SSP 202
5.2 Intra-income comparison of health risk 207
5.3 Basic characteristics of independent variables of Enrolment 208
model
5.4 Measurement and coding of independent variables 209
5.5 Probability of enrolment: Results of logistic model 6a 210
5.6 Probability of enrolment: Estimated results of Model 6b 211
5.7 Rotated Component Matrix: Factors underlying enrolment 213
in SSP
6.1 Description of the premium and eligible limit for cashless 224
treatment
6.2 Premium collected and coverage of families under SSP 225
6.3 Premium shared with insurance companies 227
6.4 Benefits given under the scheme since inception 228
6.5 Incurred claims ratio, incurred expense ratio and combined 231
ratio of SSP
6.6 Quality of care at hospitals: Comparison of insured and 232
uninsured groups
6.7 Effect of characteristics of SSP on the outcome of the 251
programme
 LIST OF FIGURES

Figure No. Title Page No.

2.1 Basic conceptual framework linking MHI characteristics, 19
performance and health system goals
2.2 Research framework on the impact of micro health 47
insurance
3.1 Organization structure of SSP 63
3.2 Client servicing and claim management process 68
4.1 Overview of health care seeking behaviour of individuals 87
4.2 Health seeking behaviour of surveyed individuals 89
4.3 Sequence of health seeking behaviour during illness 90
4.4 Direct out of pocket expenses as percentage of the annual 121
consumption expenditure incurred by insured members
4.5 Direct out of pocket expenses as percentage of the annual 122
consumption expenditure
4.6 Risk coping strategies of sampled individuals - Ex post 169
strategies
5.1 Reasons for enrolment in SSP 212
5.2 Reasons for enrolment: Comparison of renewed insured 212
and newly insured households
6.1 Growth rate of enrolment over a period of time 226
6.2 Amount sanctioned by SSP per claim 229
 ABBREVIATIONS

BPL Below poverty line

CHE Catastrophic health expenditure
FGD Focus group discussions
FP Financial Protection
FS Financial sustainability
HI Health insurance
HSB Health seeking behavior
ILO International Labour Organization
IRDA Insurance Regulatory and Development Authority
MDGs Millennium Development Goals
MFI Micro finance institution
MHI Micro health insurance
MHO Mutual health organizations
NCMS New Cooperative Medical Scheme
NGO Non-government organization
NSSO National Sample Survey Organization
NCMH National Commission on Macroeconomics and Health
OOPE Out of pocket expenditure
RM Resource mobilization
RMHC Rural Mutual Health Care
RSBY Rashtriya Swasthya Bhima Yojana
SEWA Self- Employed Women's Association
SI Social inclusion
SHG Self help group
SKDRDP Shri Kshetra Dharmasthala Rural Development Project
SSP Sampoorna Suraksha Programme
WHO World Health Organization
 CHAPTER 1

INTRODUCTION
1.1 Background

The poor face a wide variety of health and non-health related risks that include
death, unemployment, natural disasters, fire outbreak and death of livestock. Illness is
the second most frequent risk after crop failure in rural areas and the most common
shock faced by poor in urban areas that jeopardizes normal life of people with long-
term negative effect (Dercon 2004). Ill health causes poverty through loss of wages,
catastrophic expenses and repeated medical treatment (World Bank 2004). Spiraling
health care expenses often lead to impoverishment of poor households who have to
borrow money, mortgage or sell assets to pay for healthcare expenses, or just forgo
treatment (Wang et al. 2005). On the other side, poverty is associated with ill health
due to low income, high debt and social expenses (Krishna 2005). There is an intricate
connection between poverty and ill health resulting in indebtedness and
impoverishment. Hence, health has highest priority in international development goals
as an issue of economic growth and not just a medical issue (WHO 2000). Realising
fundamental association between health and development, Millennium Development
Goals (MDGs) considered the achievement of an equitable provision of health care as
one of the priorities for all the nations, especially the developing countries. Health
security should be an integral part of any poverty reduction programme. This is more
important in developing countries since majority of the population lives in rural areas
with scanty health infrastructure or work in informal sector. These countries account
for 84 percent of the world’s population, 90 percent of the worldwide burden of
disease, 20 percent of the global gross domestic product and 12 percent of the global
health spending (Gottret and Schieber 2006). International Labour Organisation
(2005) highlights the gloomy picture of 1.3 billion people lacking access to affordable
and effective health care facilities and 44 million households facing financial
difficulties due to high medical expenditure. Many nations promised to adopt Alma
Ata declaration ‘Health-for-all’ of 1978 that urged countries all over the world to
provide universal access to quality health care to their population by the year 2000.
Such an intervention can reduce medical illness induced poverty, known as iatrogenic
poverty by curtailing negative impact of ill health on the life of people.

1
 The performance of the Government of India in the health sector is
unsatisfactory. India has 16 percent of the world's population, 18 percent of the
world's mortality, 20 percent of the world's morbidity but its healthcare expenditure is
a miniscule one percent of global health expenditure (WHO 2004). The health care
expenditure is 3.6 percent of GDP in 2006; almost 75 percent of total health spending
in India is private expenditure, and 25 percent is public expenditure (WHO 2009). Of
the total private expenditure, 91.4 percent is the out-of-pocket expenses (OOPE) and
point-of-service expenditure. Global comparison reveals a dismal picture in terms of
central government outlay as a percent of total outlay in the social sector. In 2003,
India’s outlay was 1.63 percent whereas Sri Lanka (5.1%), Nepal (5.44%), Tunisia
(5.7%) and other comparable underdeveloped countries had higher outlay for health
(NCMH 2005). At about 1.36 percent of the GDP in 2008, India’s public health
spending appears even poorer in comparison with China, Sri Lanka and Thailand
(1.95%, 1.8% and 3.06% respectively) (WHO 2009).
Insufficient funding for health care by governments, inadequate health
financing mechanisms, poor delivery of health care especially in public facilities
(Patel 2010) and excessive reliance on unregulated high cost private providers has
resulted in massive OOPE and consequent impoverishment of the poor. Poor families
have to resort to desperate measures such as borrowing, sale of assets or postpone
care when sick. In fact, 40 percent of the families hospitalised borrowed money or
sold assets, which establishes the inimical position of the poor due to lack of effective
health insurance system (Peters et al. 2002). Poorest were 2.6 times more likely to
forgo the health services than the richest and one quarter of Indians seeking care
plummets below the poverty line (Peters et al. 2002). In 1995-96, 15 percent of rural
ailments were untreated that increased to 28 percent in 2004. The story is no different
in urban areas. The percent of ailments untreated due to the financial barriers was 10
percent in 1995-96 and 20 percent in 2004 (NSSO 2004). In addition, indirect cost of
care is high in rural areas due to travelling to nearby providers in city or towns and
loss of wages, as most of them are daily labourers (Sodani 1999). Thus, high medical
cost and OOPE has given rise to iatrogenic poverty (Messen 2003). Managing the
health risks has been a challenging task for the poor households in India.

2
 There are different ways of reducing the OOPE related to use of health
services namely government provision of health services, social insurance, private
health insurance and micro health insurance (MHI) schemes. In India, successive
governments gave least importance to health expenditure in their budgets. The
government spending on health as a proportion of GDP was 1.17 percent in 2009,
which is very low compared to other countries (WHO 2009). India’s Five-Year Plan
Programme targets to achieve a high rate of growth in all sectors. However, the
government has decreased its plan outlay on health rather than increasing it (Planning
commission of India 2010). Public sector cannot mobilise the required resources to
provide free health care due to large informal sector consisting of low-income
population. In addition, rich have squandered government health subsidy aimed at the
poor. World Health Organisation’s report observed only one tenth of it going to the
poorest while the richest reap one third of the subsidy (Ramachandran and
Rajalakshmi 2009). Hence, the performance of the government in the provision of
health services is inadequate.
Another health financing mechanism is Social Health Insurance (SHI). Main
reasons for choosing SHI as a method of health care financing are that SHI can
provide a stable source of revenue, a visible flow of funds into the health sector and a
combination of risk pooling with mutual support. Nevertheless, it is not a solution to
plethora of health financing problems in India due to a large share of informal sector
in the economy. In fact, organised sector employs only 9.4 percent of the total
workforce in India and the rest of working population is in unorganised sector (Datt
1997) that does not have social security benefits including Employees’ State
Insurance and Central Government Health Scheme. Therefore, scaling up of SHI in
India is limited. Hence, a large proportion of total health expenditure is private health
expenditure. It was 78.43 percent in 1975-76, 77.8 percent in 1998 and 75 percent in
2009 (WHO 2009). The OOPE exposes poorer section of the society to
impoverishment and low quality of life. Thus, private or micro health insurance as an
important mechanism to reduce household catastrophic payments has emerged (Xu et
al 2003; WHO 2000).

3
 Health insurance provides coverage against unexpected events that causes
financial loss. Based on the principle of risk pooling, it compensates economic loss
such as medical charges and income loss of daily labour due to illness for insured
individuals. Since mid-80s when health insurance got the recognition as a separate
industry, it became an important mechanism to pool risks faced by the people. In
addition, socio-economic changes such as increased awareness, higher literacy rates
and brand development by insurance companies contributed to the growth of the
industry. TPAs (Third Party Administrators) have revolutionised the administration of
policies, settlement of claims, servicing of policyholders, technical support and
customer services.
In 1987, private health insurance (PHI) in India took birth with Mediclaim
policy. Despite 25 years of its existence, the coverage of PHI is limited to less than 5
percent of population (Data monitor 2005). It is the fastest growing segment of the
non-life industry in India. It is almost one fifth of the total non-life insurance market
and is the second biggest component of the total non-life premium in the country
(Mayur, 2009). However, it is underdeveloped and lacks deep penetration especially
in rural market. The main reasons for the slow development of rural insurance market
were i) high administration costs, ii) lack of regulation and control on provider
behaviour, iii) unaffordable premiums, iv) high claim ratios, v) exclusion of many
diseases from the coverage and vi) co-variate risks (NCMH 2005). Private insurance,
being expensive and urban-centric is unaffordable by majority of population working
in informal sector. Thus, inadequate government spending on health services,
ineligibility to avail social health insurance and exorbitant private health insurance
narrow the options available to the poor in informal sector to either MHI or OOPE.
The World Bank and other multilateral and bilateral agencies have stressed the need
of PHI for better-off section of the society and MHI for those below the poverty line.
Thus, micro health insurance has emerged as a viable option to protect the poor from
iatrogenic poverty, improve access to health care and better health status.

4
1.2 Micro Health Insurance
Micro health insurance is a type of micro insurance that finances health care
expenses through the principle of risk pooling. MHI is different from the PHI, i)
individuals can not choose a coverage level at a given price (usually low premium), ii)
premium is based on community rating and iii) group contract distributed through
nodal agency such as non-government organisation (NGO) or micro finance
institution (MFI). MHI, community health funds (CHF), mutual health organisations
(MHO), community based health insurance (CBHI), rural health insurance, revolving
drugs funds, and community involvement in user-fee management have been referred
as community-based financing (Preker et al. 2002). MHI is any not-for-profit
insurance scheme aimed primarily at the informal sector and formed based on a
collective pooling of health risks, in which the members participate in its management
(Musau 1999). MHI broadly covers financing schemes that have three key features;
community control, voluntary membership, and prepayment for health care by
community members (Hsiao 2004). These schemes target low-income households
living in the same district or the members of MFIs. The membership is usually
voluntary unlike SHI.
As a health financing mechanism, MHI aims to provide adequate financial
resources to ensure timely access to health care services and help individuals escape
from the poverty trap caused by illness. MHI schemes require small contributions
from the community members to provide the pooling benefits according to the local
needs of the people. These tailor-made products cannot offer generous benefit
package due to the resource constraints owing to low income of the target population.
However, MHI schemes intend to provide financial protection to poor families and
safeguard them from falling into indebtedness or impoverishment. Additional benefits
are low transaction costs and better health behaviour through the health education.
MHI has the advantage of scientific organisation of the private insurance and
advantages of local knowledge and trust enjoyed by NGOs at the grass-root level.
Hence, the design and implementation depends on the local context. MHI brings
down the burden of health care expenditure on poor, improves the health status,
increases utilisation of services and reduces the financial barriers to access health care

5
while balancing the local requirements and affordability (Preker et al. 2002). Due to
considerable flexibility in the contract with the insurance companies and the hospitals,
scaling up of MHI is easier.
MHI deals with the information asymmetry problems with efficacy through
certain design features and implementation mechanisms. Due to the constant flow of
information among the people in rural communities, information asymmetry will be
less prevalent and much lesser the possibility of adverse selection. Over-utilisation by
some members results in higher premiums and cost shifting to other members who
would disallow moral hazard practices. Generally, NGO initiated MHIs connect the
community and formal insurance companies and hence, improve the participation and
efficiency. Despite these advantages, limitation of small pool due to modest size of
membership, inadequate benefit packages, lack of external subsidies, non-financial
barriers to access health care, limited management capacity and lack of awareness
inhibit the successful working of such schemes (Ranson 2003).
The evolution of MHI began in Africa with Bamako initiative in 1987,
followed by Germany and Japan in which MHI preceded the establishment of SHI.
The poorer countries in the world are still experimenting with this mechanism. In
1952, MHI activities started in India with Student’s Health Home scheme in West
Bengal. Since then, a few micro insurance schemes were designed and implemented
by NGOs or MFIs adopting different models of MHI as an extension of the existing
micro-credit activities. Insurance Regulatory and Development Authority (Micro-
Insurance Regulations 2005) require the private/public health insurance companies to
develop and distribute micro insurance products to rural areas. These insurance
companies tied up with the NGOs to meet the regulatory requirements in order to
reduce the transaction costs and overcome informational disadvantage. This promoted
NGO mediated health insurance schemes for the low-income people in India.
Sampoorna Suraksha Programme (SSP), VHS (Voluntary Health Services), BAIF
(Bharat Agro Industries Foundation), DHAN (Development of Human Action),
RAHA (Rajgarh Ambikapur Health Association), SEWA (Self Employed Women’s
Association), ACCORD (Action for Community Organisation, Rehabilitation and
Development), Karuna Trust, Yeshasvini Trust, and Navsarjan are some of the

6
successful MHI schemes. A comparison of SSP with other MHI reveals that except
Karuna, Yeshasvini and SSP, most of the MHI schemes require co-payments from
insured individuals (Table 1.1). Moreover, southern states of India dominate in terms
of the number of operating schemes in India because of superior social organisation.
Most of the MHI schemes exclude certain diseases from risk coverage except VHS
and Karuna Trust.
There are three models of MHI namely the provider model, insurer model and
linked (partner agent) model. In the provider model (Types I), the hospital provides
insurance facility along with the delivery of health services. In the insurer model
(Types II), voluntary organisation or NGO offers the insurance and purchases health
services from the hospitals. In the partner agent model (Types III), NGO or voluntary
organisation purchases the insurance from the insurance companies and health
services from the providers. Partner-agent model appears to be the dominant
institutional arrangement for the delivery of the MHI in India. This model enables
access to existing target market, educate and encourage preventive measures,
collection of premium, disbursement of the claim amount, use of existing distribution
channels and combines credit/ savings activities with insurance to realise economies
of scale and scope. Economies of scale mean decline in the ratio of expenses to
premium as the volume of premium increases due to increase in branches and
informational advantages (Hensely 1962). Property and liability insurance companies
were found to have economies of scale until an optimum size of premium volume due
to the ability to attract better management talent and better use of resources
(Hammond et al. 1971). Economies of scope refer to reduction in the average cost
(sharing of inputs, brand names, managerial expertise, shared marketing) due to
diversity of products offered (Cummins et al. 2007).

7
 Taable 1.1 Comparisons of Micro Health
H Insurancee Schemes in Ind
dia
Name, accronym, Taarget population Types of Unit of
U Preemium Coveerage ng on
Ceilin Provideer Benefit
location, year MHI e
enrolment perr (% of target beneffit ( ) paymennt package
perrson ( ) popuulation)
Student Health
H Fuull time students Provider School or 4 per
p 5 lak
kh Unlim
mited Third paarty Inpatient
Home, West
W Bengal c
college stu
udent studeents
(1952)
VHS, Tam mil Nadu Poopulation of Provider F
Family Sliiding 12% Unlim
mited Third paarty Inpatient
(1972) caatchment area scaale 80
RAHA, Pooor living in Insurer I
Individual 20 58% 1200 Third paarty Inpatient
Chhattisg
garh (1980) caatchment area
ACCORD D, Tamil Sccheduled tribe Provider I
Individual 22 36% 1500 Third paarty Inpatient
Nadu (19992)
SEWA, Gujarat
G SEEWA Union Linked I
Individual 100 10% 2000 Insurancce Inpatient
(1992) meembers companny
DHAN fo oundation, M
Members of Insurer I
Individual 100 40% 10000
0 KKVS Inpatient
Tamil Naadu (2000) coommunity reimburrses
baanking scheme patientss
BAIF, Maharashtra M
Members of Linked I
Individual 225 58% 5000 Insurancce Inpatient
(2001) coommunity companny
baanking scheme reimburrses
Karuna Trust,
T Sccheduled tribes Linked I
Individual 30 31% 2500 Third paarty Inpatient
Karnataka (2002) annd scheduled paymennt
caaste
Yeshasvin
ni, M
Members of co- Insurer I
Individual 120 48% in 2009 20000
00 Cashlesss Surgery
Karnataka (2003) opperative societiess treatmennt Outpatiennt
SSP, Karn
nataka SHHG members of Linked F
Family 350 32% in 2011 5000 Cashlesss Inpatient
(2004) SKKDRDP treatmennt
Source: Devaadasan et al (2006), Ownn compilation

8
1.3 Statement of the Problem and Research Questions
Health care has been a problem area for India, a nation with a large population
that has a substantial portion living below the poverty line. Consequently, health care
access and equity have become the major thrust areas. Owing to inadequate approach
of successive governments, public sector continues to face the problems of poorly
motivated work force, inadequate funding and other issues, especially in rural and
remote areas. These factors force the poor people to rely on expensive private sector
health care providers. Moreover, PHI is underdeveloped in India, the world’s 5th
largest economy. Hence, an effective financing policies are crucial to mitigate
iatrogenic poverty caused by high OOPE and it is highly imperative to undertake
studies to evaluate its effectiveness.
MHI is a poverty reduction strategy in developing countries but the empirical
evidence on the effect of such schemes on the household strategies to finance medical
expenditures is limited. While there is a reason to believe that households in different
contexts cope with health shocks differently, determining the pattern across countries
is conceivably of great interest. It becomes important to understand the risk coping
mechanism employed by the people in the face of major health adversity in India, as
socio-economic factors are different from other countries.
In India, the success of microcredit operations motivated NGO initiated MFIs
to diversify the product portfolio into the micro insurance sector. Promulgation of
such schemes on a large scale necessitates constant evaluation of existing
programmes. However, far too little attention was given on the effectiveness of the
MHI schemes in providing financial protection and reducing impoverishment.
Advocates of MHI highlight its potential of increasing access and utilisation of care,
reducing OOPE, catastrophic health expenditure (CHE) and lesser reliance on ex-post
risk coping strategies (termed as financial protection). MHI is expected to include the
poorest as members (social inclusion) and be financially sustainable (depends on the
resource mobilisation). Strategic purchasing, technical design features, management
and organisational characteristics of MHI schemes determine the performance in
terms of financial protection, social inclusion and resource mobilisation. Little
research has been carried out to test these propositions in the Indian context. The

9
efficacy of MHI has to be established before promoting them on a pan-India basis.
Sustainability of MHI schemes increases if the policymakers extend financial
support in the form of subsidies, technical assistance and links to more formal
financing arrangements. This is possible only if they are convinced of the benefits and
problems faced by MHI schemes. Because of paucity of evidence on the impact of
MHI; existing schemes, policy makers and regulators cannot push MHI as a viable
mechanism to achieve health system goals.
There is an increasing concern that enrolment in MHI schemes remains low.
The low level of enrolment in the MHI schemes may be due to the absence of
evidence on the effectiveness of MHI in reducing iatrogenic poverty. Certain scheme
characteristics may negatively affect enrolment, financial protection, social inclusion
and mobilisation of resources. An understanding and assessment of the contribution of
various characteristics of MHI schemes on its performance facilitates the definition of
critical success factors and the need to consider certain characteristics as constraints
while designing community-financing schemes. Identification and modification of
such characteristics is required to keep MHI as a sustainable and viable health
financing mechanism. This would help existing MHI schemes and newer schemes to
design and modify the benefit package for better impact.
In the international literature, majority of studies were on the schemes that
were supported by the governments, large or international organisations and not NGO
initiated MHIs. Moreover, the available literature on financial protection is mainly
from Africa and recently from China. The application of the findings of these studies
to the Indian context is undesirable since the context and the environment in which
MHI operates differs all across the world. Moreover, the schemes in Africa (known as
MHOs) are different from the Indian schemes. NGOs in India initiated many schemes
along with a broader development programme that leverages the trust, a crucial
element for the success of the MHIs. Most of the MHI models are linked models, not
found in Sub-Saharan Africa and other countries. In addition, the community does
not participate in overall decision-making process; instead, the professionals perform
technical functions. MHI is the most promising health care financing alternative and it
is highly relevant to assess its impact on members and the effect of characteristics on

10
the outcome achieved by the schemes. To study the impact of MHI, we have chosen a
case of Sampoorna Suraksha Programme, a MHI programme in Karnataka and
identified the following questions.
1. What is the impact of SSP on financial protection of members?
2. What is the impact of SSP on risk coping strategies?
3. What is the effect of SSP on social inclusion of the poor?
4. What is the impact of SSP on resource mobilisation?
5. Do SSP characteristics influence its outcome?
1.4 Research Objectives
The purpose of this research is to assess the impact of MHI on the members
and add to the existing knowledge that would help policymakers and scheme
administrators to bring about desirable changes in the scheme to realise better
outcome. The research questions are addressed by specific research objectives.
1. To assess effect of SSP on financial protection.
1.1 To learn the impact on access to health care.
1.2 To study the impact on health seeking behaviour of members.
1.3 To understand the effect on health care utilisation.
1.4 To assess the impact on out of pocket expenses.
1.5 To know the effect on catastrophic health expenditure.
2. To evaluate the impact of SSP on risk coping strategies of households.
2.1To understand the impact of SSP on coping strategies used by the households to
meet health expenditure.
2.2 To assess the effect of SSP on medical cost induced borrowing.
2.3 To know the effect of SSP on the use of savings to meet medical expenses.
2.4 To learn the impact of SSP on the sale of assets to pay for medical expenses.
3. To study the impact on social inclusion of the poor.
3.1 To look into the determinants of enrolment in SSP.
3.2 To understand the inclusion of the poor in SSP.
3.3 To explore the reasons for joining SSP.
3.4 To study adverse selection in SSP.

11
4. To analyse the impact of SSP on resource mobilisation.
4.1 To identify the amount of resource mobilised by SSP.
4.2 To assess financial sustainability of SSP.
4.3 To explore the impact of SSP on patient perceived quality of care.
5. To explore the effect of characteristics of SSP on financial protection (FP),
enrolment and resource mobilisation (RM).
5.1 To study the role of technical characteristics and the performance of SSP in
terms of FP, enrolment and RM.
5.2 To explore the management related factors and its influence on enrolment and
RM.
5.3 To learn about the relationship between the organisational characteristics and
financial sustainability.
5.4 To understand the role of institutional characteristics on the viability of SSP.
1.5 Research Hypothesis
Based on the extensive literature review pertaining to the current field of
investigation, the study hypothesises the following for further investigation.
1. H1: SSP increases access to care for insured individuals compared to uninsured
and newly insured individuals.
Insured need not incur high OOPE for treatment that reduces the financial
barriers to access care. Income acts as a major barrier to access care that prevents
low-income people from seeking care when they fall sick (Gotsadze 2005). Since SSP
compensates low income by the insurance coverage, insured individuals can access
timely care.
2.H2: Insured members seek care from formal private sector providers than other
providers (including public facilities or informal care) compared to newly insured
members and uninsured individuals.
SSP insured individuals would seek care at the private facilities due to
superior quality of the network hospitals and higher level of awareness owing to
frequent health education programmes conducted by SSP. Network hospitals are
expected to provide quality care at agreed price to insured members as per the

12
contractual agreement between the hospitals and SSP. Treatment in these hospitals
would be less expensive that reduces financial barrier to access care.
3. H3: SSP increases hospitalisation among insured members of SSP compared to
uninsured and newly insured individuals.
SSP covers hospitalisation expenses, insured need not incur higher expenses
compared to uninsured and newly insured individuals. Hence, SSP removes the
financial barriers to utilisation resulting in higher hospitalisation.
4. H4: SSP reduces OOPE associated with illness for insured members due to claim
benefits.
By providing financial assistance during hospitalisation, SSP reduces OOPE.
5. H5: SSP reduces CHE for insured individuals compared to newly insured and
uninsured individuals.
SSP reduces the direct cost of treatment; there would be lower incidence of
CHE for insured individuals.
6. H6: SSP reduces the reliance on other strategies with negative consequences
(borrowing, use of savings and sale of assets) for insured individuals compared to
newly insured and uninsured individuals.
SSP meets the major part of the total medical cost and stabilizes the
expenditure that fluctuates due to illness. Hence, the need for additional finance was
less for SSP members.
7. H7a: Incidence of borrowing would be less for SSP insured compared to uninsured
and newly insured individuals.
SSP provides financial benefits to insured; hence, the need to borrow would be
less for insured individuals compared to uninsured and newly insured individuals.
8. H7b: Insured individuals compared to uninsured and newly insured individuals
would borrow lower amount.
Since SSP covers hospitalisation expenses, the amount of borrowing would be
lower for insured compared to uninsured and newly insured individuals.

13
9. H8a: Incidence of use of savings will be less for insured compared to uninsured
and newly insured individuals.
SSP covers most of the direct expenses of hospitalisation; hence, insured use
savings less compared to uninsured and newly insured individuals.
10. H8b: Insured use lesser amount of savings compared to uninsured and newly
insured individuals.
Due to financial claim from SSP, members have to spend small amount to
meet indirect expenses or outpatient expenses.
11. H9: SSP insured sell fewer assets compared to uninsured and newly insured
individuals.
Owing to SSP, the need to sell assets for insured was not as much as that for
uninsured or newly insured individuals since SSP covers most of hospitalisation
expenses.
12. H10: SSP includes the poorest as members
Social inclusion is one of the main objectives of any health care financing
mechanism including MHI. SKDRDP (Shri Kshetra Dharmasthala Rural
Development Project), a socio-economic development programme launched SSP
targeted at poor households in the informal sector. Hence, larger percent of poorest
would be SSP members.
13. H11: SSP does have adverse selection
SSP enrols entire household as the unit of enrolment. However, lack of
medical examination of prospective members and waiting period exposes SSP to
adverse selection. Moreover, the upper age limit for enrolment is 80 years that
encourages older high-risk individuals to enrol in SSP.
1.6 Scope of the Study
Recent developments in health financing have heightened the need for MHI to
achieve universal health coverage in India. We do not have adequate empirical
evidence to support such schemes, both the impact on members and wider
implications for the society in India. Hence, this study focuses on the impact of SSP
on financial protection and social inclusion. It also aims to know demand and supply
factors that determine enrolment. Other aspects looked into were resource

14
mobilisation by SSP and the influence of technical, management, organisational and
institutional factors on financial protection, social inclusion and enrolment. The
questionnaire was designed to collect information on the basic socio-economic
characteristics, access and utilization of health services, health-seeking behavior, cost
of treatment, quality of care and risk coping methods. Qualitative data includes
barriers to access health care and enrolment and participation in scheme management.
Premium and claims data was obtained from annual reports and SSP head office.
Cross-sectional survey was carried out in Karnataka to collect qualitative and
quantitative data from 416 insured households, 366 newly insured household and 364
uninsured households of SKDRDP in the first half of the year 2011. The outcome of
the present study would guide the policymakers and the scheme administrators to
provide more impetus to expand and scale up MHI schemes, especially when
government of India is focusing on universal access to health care by 2015.
1.7 Organisation of Thesis
The thesis is structured into 7 chapters. Chapter 1 explicates background
information, context and relevance, research problem, objectives and hypothesis.
Chapter 2 reviews the literature on financial protection, social inclusion, resource
mobilisation and design characteristics influencing outcome. Chapter 3 explains the
research methodology, study settings and Sampoorna Suraksha Programme. Chapter 4
describes the socio-economic characteristics of sample households and explains the
impact of SSP on access to care, health seeking behaviour, utilisation of care, OOPE,
CHE, and risk coping strategies (Objective 1 and 2) . Chapter 5 investigates the social
inclusion, determinants of enrolment and analyses adverse selection in SSP (Objective
3). Resource mobilisation and the association of features of SSP and its performance
is the theme of chapter 6 (Objective 4 and 5). The last chapter summarises the thesis
findings and provides policy implications.

15
 CHAPTER 2

REVIEW OF LITERATURE
2.1 Introduction
This chapter explains the concepts used in the study and reviews the relevant
literature with a view to derive conceptual map and design research methodology.
Firstly, a description of the role of MHI in achieving health system goals highlights
the relevance of the study. Next section discusses the review of literature on selected
themes such as financial protection (including risk coping strategies), social inclusion,
adverse selection, resource mobilisation and determinants of performance of MHI.
2.2 Health System Goals and the Role of Micro Health Insurance
Universal coverage of health services is the main agenda of many nations’
development programmes including India. It requires access to an affordable health
care to all without regard to one’s ability to pay. This idea of equity in access and
financing stipulates health care system in any country to achieve better health status
and health equality, to be responsive to people’s non-medical expectations and to
ensure fairness in financial contribution (WHO 2000). This broad objective can be
broken down to equity in utilisation, financial protection and sustainability. Equity is
interpreted in relation to both income and gender equality of access to health care.
Health system performs four main functions namely i) provision of health services, ii)
resource generation (investment and training), iii) health financing (risk pooling) and
iv) government stewardship (governance and oversight) to achieve these objectives
(WHO 2000).
Among these main functions, health-financing sub function is to be prioritised
in India due to its impoverishing effects on the poor. MHI is one of the health-
financing mechanisms that involve the provision of adequate financial resources to
ensure timely access to public and private health care services. The functions of MHI
include revenue collection, risk pooling and strategic purchasing (WHO 2000). In the
revenue collection function, determination and mobilisation of the financial resources
from the households, enterprises and other organisations takes place that in turn
depends on enrolment and ratio of prepayment. Enrolment depends on affordability of
premium, unit of membership, timing of the collection of premium, quality of care
offered, and geographical location of the household (Carrin et al. 2005). The pooling
function allows the sharing of financial resources between healthy and sick that

17
involves accumulation and management of contributions of members to spread the
risk of illness among the members. Strategic purchasing happens when a continuous
search to buy the best health services, contract with best providers, use best payment
methods and contracting arrangements (WHO 2000) exists.
By performing these functions, MHI aims to achieve three independent goals
namely mobilisation of resources, protecting the households from financial
consequences of illness and the inclusion of the poorest by making them active
participants in health care system which ultimately contributes to the objectives of the
health system. Resource mobilisation denotes cost recovery ratio, amount of resources
raised through community-financing arrangements as a share of the country’s total
health revenues and indirectly by efficiency and quality impact on health care and
moral hazard effects (Ekman 2004). Financial protection is the reduction in annual
health expenditure as a percent of total annual household income. It denotes reduction
in OOPE, access to health care and utilisation of health care. The size of poorest
members in a scheme measures the social inclusion (Jakab and Krishnan 2001). Also
demand side factors (income, size of family, education and gender of head of the
household) and supply side factors (scheme design and implementation) determine
enrolment and social inclusion.
The performance of functions of MHI to achieve its objectives depends upon
the design of the schemes in terms of technical, management, organisational and
institutional characteristics (Preker et al. 2002). Technical characteristics namely
benefit packages, structure of premium, purchasing of health services and allocation
mechanisms determine revenue collection, risk pooling and enrolment. The level of
pre-payment, types of contribution (compulsory or voluntary), degree to which
contributions is progressive, tools to address adverse selection, flexibility in the
payment of premium and provision of subsidies affect the revenue collection (Preker
et al. 2004). Size of the insurance scheme, trust and confidence in the management of
MHI and moral hazard control mechanisms affect the risk pooling. Provider payment
mechanisms, referral systems, waiting period provisions, contents of benefit package,
and contract specifications in health services are factors that determine the extent of
strategic purchasing (Carrin et al. 2005). Management characteristics include staff

18
(leadership, extent of capacity building), culture (management style, structure), and
access to information (financial, health information, resources, and behaviour).
Organisational characteristics include organisational forms, incentive regime (degree
of autonomy, accountability, financial responsibility), and linkages with health care
providers. Institutional characteristics are stewardship (government and donor
support), governance, insurance markets, and factor and product markets (Preker et al.
2002). Figure 2.1 depicts the broad conceptual framework of the study.

Universal coverage to Adequate access at

tackle iatrogenic poverty Affordable cost

Health system goals Health system Health system

intermediate goals ultimate goals

Health system functions

Equitable utilisation Health status & Health equality

Reduce impoverishment Responsiveness to non-medical expectations
Health financing sub-function Sustainability of resource Fairness in financial contribution
mobilisation

Micro Health Insurance model

Micro Health
Determinants Insurance impact

Technical characteristics
Performance Objectives
Revenue collection
Risk pooling Resource- mobilisation
Strategic purchasing Financial protection
Social inclusion

Management characteristics
Organisational Characteristics
Institutional Characteristics

Figure 2.1 Basic Conceptual Framework Linking MHI Characteristics,

Performance and Health System Goals
A systematic assessment of any program intervention requires reliable and
generally agreed upon performance indicators that ensures good quality of body of
evidence. A set of variables as proposed by Ekman (2004), Jakab and Krishnan
(2001) and Preker (2002) are used to evaluate the impact of MHI on members that
ultimately contributes to the achievement of health system goals. The following
section reviews the available literature on determinants of performance and impact of
MHI as given in the basic conceptual framework (Figure 2.1)

19
2.3 Review of Literature on the Impact of Micro Health Insurance
A review of 45 published and unpublished reports and conference proceedings
by Preker (2002) advocated an important role of community financing as it provides
financial protection against cost of illness, improves access to care by the poor and
promotes efficient use of scarce health care resources. The review of available
literature on financial protection, social inclusion, resource mobilisation and effect of
characteristics on the performance of the scheme highlights the research gap.
2.4 Financial Protection
Resource mobilisation, reduction in impoverishment (financial protection) and
equitable utilisation (social inclusion) measures the performance of MHI in realising
universal coverage. Fair (financial) contribution denotes distribution of cost of illness
based on the ability to pay. In the literature, due to the absence of any relevant
validated instrument, financial protection acts as a proxy. Reduction in OOPE and
CHE, access and utilisation of health care measure financial protection (Ekman 2004).
In addition to these measures, risk coping strategies represent a comprehensive
measure of financial protection.
2.4.1. Access to Health Care Services
Access to health services refers to the entry into the health care system
determined by the need to improve current health status and the capacity to benefit
from health care. Individual’s perception of the need depends on the knowledge of
health care. It also depends on the perception of what is ‘normal’ with regard to their
health. Sometimes there is a need but no demand, which denotes unmet need.
Information deficiencies, supply factor (lack of health services) and demand factors
(income and prices of health care) are the causes of unmet need (Morris et al. 2007).
In addition, financial constraints or non-financial constraints limit one’s ability to
obtain health services when needed (Liu et al. 2002). Health provider’s diagnosis is
termed as evaluated need (Aday and Awe 1997). Perceived need depends on health
status, illness symptoms and days of disability whereas evaluated need relies on
diagnosis, surgery or urgency of presenting conditions to a physician. These needs
differ depending on the diverse factors that influence the entry to health system and
organisation of the system to give care (Aday and Andersen 1974).

20
 Through time and age, health depreciates that can be improved by seeking
care and investing time, effort and money. Hence, demand for health services is a
derived demand to improve health (Morris et al. 2007). The most important factors
determining the demand for health care are price, income, price of
substitutes/complementary goods, tastes and trends. Consumer choice theory predicts
higher demand when the price falls, ceteris paribus. Nevertheless, demand for the
health services by an individual is a special case that would not follow this established
relationship between price and demand. Even if the price falls, a person may not take
treatment. However, a cumulative addition of individual demand results in the market
demand that would be downward sloping with respect to price (Morris et al. 2007).
When the price reduces, there will be a movement along the demand curve. Thus,
demand for health care depends on ability and willingness to pay for the care. In this
context, affordable health care is defined as expenses that do not reduce the
consumption or investment in essential commodities below levels that may affect
either future health, earning capacity or future expenditures (Russell, 1996). Since
health insurance reduces the price of care, insured individuals move downwards along
the demand curve and use more services.
Access to care or propensity to use care is to be distinguished from utilisation
of care or volume of utilisation once in health care system (Newbold et al. 1995).
Individual characteristics and demand-side factors determine access to care whereas
patient role and health care system or provider behaviour decide utilisation (Alberts et
al. 1997). Access is one’s ability to obtain health services when needed or the
likelihood of visiting a health provider that is contingent upon two conditions:
financial constraints and non-financial constraints (Liu et al. 2002).
Yip (2007) studied the success of Rural Mutual Health Care (RMHC in
China), a MHI scheme, in achieving the objective of improving access to health care
while Msuya and others (2004) made an observation of higher access to curative
health care in Tanzania as the members of the community health funds (CHF) were
financially better protected against health shocks. Devadasan (2005) from his study on
ACCORD demonstrated a positive impact of MHI as insured access health care more
than the uninsured and highlighted the importance of trust in health insurance

21
enrolment. Gotsadze (2005) demonstrated that people with less financial resources
postpone seeking care or spend a higher proportion of monthly expenditure on health
due to financial barriers in Georgia. These barriers hindered access to medicines and
fostered inequalities in access to basic care in Tajikistan (Tediosi et al. 2008). Non-
financial barriers include area of residence (Auchincloss et al. 2001), mismanagement
of the scheme and contract with ineffective health care providers (Jutting 2003; Dror
et al. 2005). Furthermore, income, education, position at work and access to
outpatient care/ diagnostic technology determines access to care (Liu et al. 2002).
2.4.2 Health Seeking Behaviour
The pattern of actual utilisation is the end process of access to care that differs
in terms of types, site, purpose and time interval (Andersen and Newman 1973).
Types of service can be hospital, physician (clinics) and pharmacist. Site refers to a
place of care namely outpatient departments, clinics, casualty or emergency room and
hospital wards. The purpose of care can be preventive or curative. This in turn would
determine treatment-seeking behaviour. Time interval means whether or not visit to
health care facility takes place in time of illness, frequency of visits and the process of
receiving care.
Considering population at risk as unit of analysis, Aday and Anderson (1974)
label predisposing factors (demographics), enabling factors (income, area of
residence, distance to hospitals and health insurance) and need factors (either
perceived by individuals or evaluated by health delivery system) as determinants of
access and utilisation of care. Research on utilisation of health services differentiates
policy variables from control variables. Health insurance, income, source of care, ease
of getting care, general health care attitudes and knowledge of health care are
amenable to change. Age, sex, marital status, education, religion, size of family and
area of residence are control variables (Aday and Andersen 1974). Thus, MHI is an
enabling variable viable to alteration by a suitable health policy to affect access and
utilisation of care by poor population.
The types of service availed or health-seeking behavior is one of the
characteristics of utilization behaviour. It is an activity undertaken by individuals with
a health problem to find an appropriate remedy (Ward et al. 1997). This is shaped by a

22
number of factors including historical pattern of use, illness types and severity, pre-
existing lay beliefs about illness causation, range and accessibility of therapeutic
options and their perceived efficacy, convenience, opportunity costs, quality of
service, staff attitudes as well as age, gender and social circumstances of the sick
individual (Tipping and Segall 1995). Considering direct and indirect cost involved in
seeking care, the effective decision for the very poor may be not to seek care at all or
to go to traditional healers or resort to partial treatment (Russell 1996). Efficiency of
healthcare system would be negatively affected when people resort to self- treatment
including self-prescription when drugs are freely available in the market and seeking
care directly from specialists’ while by–passing primary care providers (Gotsadze et
al. 2005). Self-medication has been associated with lack of access to professional
healthcare, lack of government-sponsored health insurance coverage and socio-
economic status related to lower education, living in rural areas, lower income, and
fewer assets (Pag´an et al. 2006). Ahmed (2003) carried out a study on gender related
changes in health seeking behaviour and report formal care to depend on gender (men
more than women), geographic location, greater socio-economic status and serious
illness of long duration. A study from Vietnam found low income people substituting
drug vendors for formal care in health facilities in order to save time and money
(Deolalikar 2002). This study observed a large proportion of public subsidies being
captured by elite class. Falkingham (2004) documented the tendency of the poor to
use home remedies, primary care facilities and providers, whereas the better off
approached high cost polyclinic and hospital care in Tajikistan.
Health insurance has an incentive effect as insured tend to use inpatient
facilities and public providers more than uninsured in Vietnam (Jowett 2004). As
health insurance removes any financial barriers to access care in good quality
hospitals, insured would use inpatient facilities and private providers more than
uninsured that may reduce the demand for self-medication and change the perceptions
about the benefits of modern medicine (Pagan et al. 2006). Mutual health
organisations (MHO) members were more likely to seek formal health care in Ghana
and Mali, although this result was not confirmed in Senegal (Chankova et al. 2008). It
has been found in Tanzania that members of a community health fund were more

23
likely to seek formal medical care when ill than non-members (Msuya et al. 2004).
Schneider (2001) documented a shift of demand for care from traditional to modern
health sector by MHO members. Ahmed (2000) showed women from BRAC member
households to use qualified allopathic care than women from poor non-member
households. Contrary to these findings, impact on health seeking behaviour has been
negligible in Senegal (Jutting and Tine 2000). Hence, evidence base provides
inconclusive evidence on the impact of MHI, albeit majority of studies document
positive impact. Only a few studies were household data based that used statistical
analysis whereas the rest were descriptive based on facility data.
2.4.3 Utilisation of Health Services
Utilisation of health services can be described in terms of site or place where
actual care is received, which can be outpatient departments, clinics, emergency
room, and inpatient treatment in hospitals. Utilisation refers to guaranteeing an
effective and needed health services for the promotion of health, prevention and
treatment of illnesses and rehabilitation of good health. Need, enabling and
predisposing factors determine utilisation of health services. The use of health care
facilities or the length of stay in hospitals can also measure utilisation of health
services during a period of illness.
Demand for health services is price inelastic (McPake 1993), however, poorer
display more elastic demand than rich people (McPake et al 2002). RAND Health
Insurance experiment in US and in other settings estimated the elasticity of demand
for health services to lie between 0.1 and 0.7 (Morris et al. 2007). The elasticity of
demand for health care is income elastic (more than 1), hence any increase in income
(notional) due to ‘income effect’ of decrease in price of care would enhance the use
despite health services being a necessity, especially by low-income households. Thus,
any rise in income results in more than proportionate increase in the use of health
services (Morris et al. 2007). In addition, non-price access costs (transport and time)
determine demand for health services.
A review study concluded that a minority of MHI schemes paid explicit
attention to utilisation. Out of the 258 schemes reviewed, 14 studies, of which only
one study with internal validity, found a positive impact of MHI on utilisation of

24
health care services (Baeza et al.2002). In our review, twenty studies on utilisation
impact of MHI from different parts of the world reported positive impact and five
documented negative impact. Studies from China (Bogg et al.1996; Wagstaff et al.
2008a), Congo (Criel and Kegels 1997), Ghana (Atim 1999), Kenya (Musau 1999),
Tanzania (Msuya et al. 2004) and Senegal (Ju¨tting 2003) revealed increased
utilisation of health services in those schemes that cover hospital inpatient care.
However, moral hazard and cost escalation have been observed in schemes that cover
inpatient care (Bennett et al. 1998). In micro insurance units in Philippines, higher
rates of professionally-attended deliveries, lower rates of delivery at home, a higher
frequency of primary-care physician encounters as well as diagnosed chronic
diseases, and better drug compliance among chronically ill was observed (Dror et al.
2005). Studies from Senegal, West, Central and East Africa confirm the positive
effect on hospital utilisation due to MHO coverage (Atim 1999; Ju¨tting 2003). Rao
(2009) found positive impact of community health funds in Afghanistan on the
utilisation of health services, due to reduction in financial barrier. A study on MHOs
in Mali by Franco (2008) found positive effect on utilisation of priority health
services, although it could not achieve complete coverage of the poorest.
There are reports in literature which document the absence of impact on
utilisation from Ghana (Chankova et al. 2008), India (Ranson 2001), Jordan (Ekman
2007a) and China (Yip et al. 2007). Chankova and colleagues (2008) found
inconclusive evidence on the positive impact of MHO membership on utilisation in
Africa. Higher utilisation was observed in Mali and Senegal but not in Ghana. A study
from India confirms these findings as SEWA (Self- Employed Women's Association,
India) members’ utilisation of health care services did not increase compared to
uninsured (Ranson 2001; Gumber 2001). Another study carried out in Jordan found
no significant impact on the probability of utilising health care (Ekman 2007a). A
study on RMHC in China found an increase in utilisation of village clinic but no
impact on hospitalisation or outpatient utilisation (Liu et al. 2002).
Soucat and others (1997) have reported increased utilisation of health services
after the introduction of Bamako Initiative which is attributed to the availability of
drugs and improved quality of services brought about by the community involvement.

25
This finding is supported by other researchers (Schneider and Diop 2001; Jutting,
2003, Dror et al. 2005). Moreover, location of the residence, distance to healthcare
provider, occupation. (Liu et. al. 2009), income and education, inconvenience caused
to the family, domestic responsibilities borne by women, lack of awareness of benefits
of insurance (Sinha et al. 2005) also influence utilisation of health care services.
Studies from China found rich members benefiting more than poor members
(Wang et al. 2005). In contrast, a study from Philippines found that insured
households across all income groups use hospitalisation and consultations equitably
(Dror et al. 2005). In spite of the growing literature, the evidence is inconclusive and
the question whether members of MHI are financially better protected than uninsured
still remains. Lack of such evidence stems from the paucity of large studies based on
household/individual level data, and only few studies utilised rigorous statistical
methodology such as quantitative techniques, in particular regression analysis.
2.4.4 Out-of –Pocket and Catastrophic Expenditure
Financial protection means reduction in the proportion of income spent as
health expenditure due to health insurance. It means the household is not required to
contribute directly or indirectly more than acceptable proportion of its total income in
order to gain access to adequate health services (Baeza et al. 2002). In addition,
absence of financial protection exists when excessive health expenditure reduces
households’ consumption to below the poverty line. There are different methods of
defining financial protection. One preliminary method is to use specific or arbitrary
limits on health expenditure for the lowest income quintiles. Usually, this method sets
excessive expenditure at a level of certain proportion of total household income
equivalent to the cost of a standardized package of services. Another approach defines
a limit on health expenditure as a proportion of disposable income available to
household after deducting the expenditure for the consumption of other goods, and
services (Baeza et al. 2002). OOPE were used as a direct measure of financial
protection in earlier studies, which has been disapproved by Baeza et al (2002) who
suggested the use of catastrophic health expenditure (CHE) as validated direct impact
indicator of financial protection. In this context, World Health Organisation proposed
health expenditure (non-food expenditure) to be considered catastrophic when it is

26
above 40 percent of the capacity to pay (Kawabata et al. 2002, Xu et al. 2003).
Another definition of CHE is a health expenditure that is more than 10 percent of total
household income (Pradhan and Prescott 2002; Ranson 2002).
The value of health insurance depends on the value of expensive health care
that becomes affordable. Health insurance, thus, is expected to reduce the burden of
cost of care. Litvack and Bodart (1992) postulate the beneficial effect of community
based health insurance that facilitate access to care for low-income populations who
otherwise have no financial protection against the cost of illness. There are reports of
improved financial protection in terms of reduction in OOPE (Jütting and Tine 2000;
Schneider and Diop 2001). In addition, households with access to micro-finance loans
reported lower OOPE per reported illness (Dekker and Wilms 2009). A recent study
from India found substantial financial protection provided to the members by
reducing the need to borrow money or sell assets to meet medical expenses (Aggarwal
2010). However, there are studies which had documented ‘marginal’ or ‘limited’
impact (Jütting 2001; Carrin et al. 1999; Wagstaff et al. 2008b; Yip et al.2007; Rao et
al.2009). The study on the impact of SEWA Scheme in India found the burden of
seeking care on the household budget to be higher among SEWA members than
among those insured by other mechanisms (Gumber 2001) and uninsured members
(Ranson 2002). A study on RMHC found no impact on OOPE for outpatient treatment
or hospitalisation (Yip et al. 2007). MHO membership was observed to provide
protection against OOPE related to hospitalisation in West Africa (Chankova et al.
2008) but it did not have a positive effect on curative outpatient care. Study on
NCMS, a public health insurance scheme by Lie and Lin (2009) found that insurance
scheme membership neither decreases OOPE nor increases utilisation of formal
medical service or improves health status (as measured by self-reported health status)
but changes the health seeking behaviour from traditional Chinese folk doctors to
formal preventive care.
The literature on the impact on CHE provides conflicting evidence. Ranson
(2002) and Devadasan (2007) using the facility data showed positive yet partial effect
of MHI on CHE. A study on Universal Insurance in Mexico found evidence of
reduction in probability of CHE and a reduction of expenditure on medicines and

27
outpatient care among insured families (Galarraga et al. 2010). Pradhan and Prescott
(2002) indicate the absence of positive impact on CHE for community-financing
members in Indonesia due to low benefit coverage. NCMS in China could only
provide partial protection because of high medical costs, low effective reimbursement
levels, inadequate benefit package, policies on co-payment, ceilings and deductibles
and complex reimbursement procedure (Zhang et al. 2009).
Ekman’s (2007b) investigation provides contrary evidence of lack of impact in
Zambia. In his study, health insurance was found to increase risk of catastrophic
payments due to expensive medical care as insured accessed care at later stages of
illness. This study recommends the consideration of health care needs, in addition to
health care utilisation patterns and expenditures when analysing the effects of health
insurance. Wagstaff and colleagues (2008b) confirm these findings from his study on
NCMS in China. He argues that health insurance increases the risk of high and
catastrophic spending as it encourages people to seek care from higher-level
providers. Another study by the same author on Health Care Fund for the Poor (HCF)
suggests that the scheme substantially increased inpatient service utilisation and
reduced the risk of catastrophic spending. Nevertheless, it was not successful in
reducing out-of-pocket spending, and had negligible impact on utilisation among the
poorest deciles (Wagstaff 2007). The conflicting evidence reflects the diverse socio-
economic, political and cultural settings of these studies. Many of these studies were
descriptive based on facility data without rigorous statistical analysis (Table 2.1).
While designing MHI scheme, ignorance of various factors that contribute to
CHE and high OOPE would result in partial effect. These factors are household size,
incidence of illness, presence of 'smokers or drinkers' in the household (Kawabata et
al. 2002, Arhin-Tenkorang 2001), high medical costs and low effective
reimbursement levels (Zhang et al. 2010). Effective financial protection depends on
the reasonable balance of funds maintained by the scheme, which is critical to ensure
that the schemes are sustainable and effective in offering financial protection to
members.

28
2.5 Ex-Post Risk Coping Strategies
The most common risk faced by poor is health shock which is defined as
unpredictable illnesses that diminish health status (Leive and Xu 2008). Risk
management strategies would ensure a steady income that mitigates health shocks.
Negative effect of health shock may be transient if the affected household has certain
ex-ante and ex-post measures to tackle health risks. Ex-ante strategies are
diversification or entry into low risk- low return activities and reduced average
consumption spending due to precautionary savings (Dercon 2004). Coping strategies
used after the health shock (ex-post) can be divided into, (i) behaviour-based
strategies (less consumption or increase labour supply) (ii) asset-based strategies (use
savings, assets, borrowing money), (iii) assistance from informal or public sources
(Heltberg and Lund 2009; Dercon 2004) and iv) self-insurance. These strategies rely
on formal or informal coping instruments that could have harmful consequences for
the households who already consume less, have low savings and face barriers to non-
exploitative credit (Heltberg and Lund 2009). Self-insurance can occur in two ways; i)
use of savings or building up suitable liquid assets in good years that can be depleted
during a bad year ii) informal risk sharing arrangements, based on reciprocal gifts or
contingent credit within family, friends or neighbours for consumption smoothing
during the episode of illness (Dercon 2004). In addition, survival strategies such as
sacrifice of human capital (sending additional household member for work), sale of
productive assets, borrowing from banks and charity were usually used in times of
health crisis (Dercon 2002). However, some of these strategies have adverse impact
on future household consumption as they would have less income due to sale of
productive assets and repayment of loan.
The strategies with negative consequences such as use of savings, sale of
assets, borrowing and reduction in consumption may lead to iatrogenic poverty and
worsening of health status (Gotsadze et al. 2005; Msuya et al.2004). These strategies
increase the vulnerability to future health shocks, reduces asset base for future wealth
creation and adversely affects nutrition and human capital (Dercon 2002). Some
strategies such as engagement in activities other than normal work or selling labour

29
(Sauerborn, et al. 1996) especially sending school going children for work
(Mutangadura et al. 1999) may have adverse consequences.
Of the available resources to deal with risks, cash at hand, access to a loan
from either a MFI or community and savings deposits are used by the households first
and the more productive and protective assets are used as a last resort (Sebstad and
Cohen 2002). Borrowing ability and financial capacity to repay a loan, existence of
social capital and availability of assets determine coping strategies of the households.
Wilms (2006) observed the size and degree of uncertainty of loss to be the
determinants of the use of savings and credit. Smaller the size and degree of
uncertainty, savings may be more appropriate. Although borrowing is a preferred
strategy to cope with financial shocks, it has negative consequences that vary directly
with income of the household (Wilms 2006). Number of shocks experienced and the
resultant health expenditure of the household, cost of the shock, household size and
area of residence determine the likelihood of selling assets (Wagstaff 2006). Leive
and Xu (2008) documented asset based strategies used by African households to cope
with inpatient medical bills. In the same study, current income and savings financed
outpatient spending. A study on informal risk sharing arrangements in a rural area of
North-Western Burkina Faso found asset sale to be an important health financing
strategy and relatives were the first resort in terms of financial arrangements
(Sommerfield et al. 2002).
Poor households had to sell land and other assets, exchange food or labour for
cash, take loans or use common property to deal with health shocks (Russell 1996). In
Burkina Faso, selling livestock, grain and borrowing was the common strategy used
by households (Sauerborn et al. 1996). In a study of coping strategies in Uganda,
Leliveld (2006) observed that households sold land, cattle, or goats or used their
savings to respond to illness. This may jeopardize the future economic status of
household through indebtedness and deprive the future income generated by the use
of the productive assets (Scheil-Adlung et al. 2006). A study on adaptive behaviour of
people in Tbilisi in response to high medical bills found that borrowing money and
selling assets were frequently resorted that resulted in impoverishment and worsening
health status (Gotsadze et al. 2005). Heltberg and Lund (2009) found economic

30
shocks financed by savings and natural or agricultural shocks were coped with
borrowing in Pakistan.
Given the inadequacy of ex-post measures in fully protecting the households,
ex-ante measures especially health insurance need to be a part of a comprehensive
system of protection against risk (Dercon et al. 2004). The difference between actual
loss after insurance indemnification and what would have been lost without it
measures the impact of health insurance. Health insurance enhances the welfare of the
household by providing financial protection, shortening the duration of illness and
improving health (Young et al. 2006). It reduces the use of impoverishing risk coping
strategies (Dekker and Wilms 2009) and makes the poor less vulnerable to poverty
induced by health shocks (Wilms 2006). One way to achieve this is to link credit and
health insurance, which not only helps in building assets but also increases the ability
to cope with health shocks (Dercon 2002). Another way is to provide health insurance
at subsidised rates to poor population (Dercon et al. 2004).
Aggarwal (2010) in his recent study from India on Yeshasvini programme,
show that insured borrowed less or sold fewer assets compared to uninsured to meet
surgical expenses, thereby insured experienced substantial financial protection. In
case of hospitalisation, there was no effect on borrowing or sale of assets. Dekker and
Wilms (2009) found MHO members to rely less on risk coping strategies in Uganda.
Insured households were less likely to sell assets to finance health expenditure and the
value of sold assets was lower. Insured households did borrow or sold assets but
lesser amount per illness episode compared uninsured (Dekker and Wilms 2009).
There is dearth of literature on the impact of MHI on risk coping strategies used by
the poor, especially from India.
2.6 Social Inclusion
Equity has been considered as a major objective of health care policy in
international community. Social exclusion refers to inadequate or unequal
participation in social life or exclusion from a place in the consumer society, often
linked to social role of employment or work (Duffy 1995). Moreover, income and
self-rated health is linked and the very poor are most likely to report bad health
compared to high income earners (Subramanian et al. 2003). Considering this, health

31
financing interventions have been advocated to reduce socio-economic inequalities in
health to alleviate poverty among individuals (Weich et al. 2002). Thus, social
inclusion is one of the objectives of health financing mechanisms including MHI. In
this regard, MHI schemes aim to include the poorest as members to lower health
expenditure (Jowett 2002).
Some researchers suggest that the poorest of the poor and socially
disadvantageous groups were excluded in community-based initiatives for financing
of health care (Jutting 2003). Payment of premium can be a significant barrier to
social inclusion. Jutting and Tine (2000) highlight the problem of social exclusion in
which the community’s poorest members had no opportunity to participate due to lack
of resources to pay the required premium. His finding was supported by other studies
(Sinha et al. 2005; Msuya et al. 2004; Schneider and Diop 2001) carried out in
different settings. An extensive review of literature on the impact of community
health insurance found strong indications that these schemes exclude the poorest and
have little effect on access to care (Ekman 2004). Exclusion of poor in the MHI
scheme is a major limitation that affects equitable access to health care system. Any
health financing mechanism that aims to include the poor has to get external funds
especially when the internal funds are inadequate. However, MHI schemes nested
within a larger organisation (like MFIs) that address other needs of the poor and
charges affordable premium will be able to meet the goal of social inclusion (Ranson
2002).
2.6.1 Demand for Health Insurance
Seminal paper by Arrow (1963) highlighted the role of risk aversion and
uncertainty of future health as motivators for the purchase of health insurance (HI).
On the other hand, Pauly (1968) put forward a proposition that consumers will be
worse off with HI that had deductibles and coinsurance rates. A recent theory by
Nyman (2003) refutes utility function of Neumann and Morgenstern and contradicts
the theory of Pauly. According to him, consumer compares expected utility lost from
the payment of premium and expected utility gained from insurance claim if ill. The
consumer demands HI in order to obtain a transfer of income from the healthy if she
were to become ill. Gaining access to unaffordable health care services during illness

32
is highly valuable to the consumer that motivates the purchase of HI. This theory is
more applicable in India since unaffordable medical care restricts access to care to the
poor. Hence, HI can be advocated as a mechanism to provide health security and
better access to care.
Demand for HI or enrolment in a MHI scheme determines not only social
inclusion but also resource mobilisation and hence, sustainability of the scheme in the
long run. Renewal and enrolment of members has been identified as one of the
challenges faced by MHIs in most of the countries (Ahmed et al. 2005) that reflects
member satisfaction, and trust in the programme (Supakankunti 2004). Enrolment in a
scheme depends on household income that can be paid as premium (Msuya et al.
2004; Jutting and Tine 2000) and inability to pay premium, even a small amount, acts
as a major barrier to enrolment in MHOs in West Africa (Chankova et al. 2008). In
their review of 83 HI schemes for the informal sector, Bennett and others (1998)
found that very few schemes adopted sliding scales or exemptions for poor despite
being aware of the problem of affordability. Most schemes relied on flat-rate
premiums and several schemes charged unaffordable premiums which acted as a
major deterrent to participation.
A study from Rwanda found household characteristics such as the district of
residence, education level of household head, family size, distance to the health
facility, trust and radio ownership influence enrolment in a scheme but did not find
evidence on the role of health and economic indicators in enrolment decisions
(Schneider and Diop 2001). Research work in West Africa on the impact of MHOs
found that gender and education of the household head and economic status of the
household to be positively associated with MHO membership (Chankova et al. 2008).
Lack of involvement in the management of the scheme, difficulties to get specified
families enrolled as per scheme guidelines, long distance from the provider’s
facilities, unattractive benefit package were the reasons for low enrolment in Uganda
(Basaza et al. 2008) (Table 2.2).
Sinha and others (2007) propose demand-side factors (characteristics of
individuals, households or groups in the target population) and supply-side factors
(characteristics of the MHI scheme) that determine enrolment in a scheme. On the

33
demand-side are the factors that underlie the likelihood of benefits perceived by the
members. These include age, education, health status, trust in the scheme, previous
claim experience and participation in the scheme by friends and neighbours. On the
supply-side, factors such as opportunity to enrol, knowledge of the scheme, additional
services offered by the organisation, frequency of contact with the members,
education provided regarding insurance and opportunity provided by the scheme to
renew like accurate and up-to-date records of members for determine enrolment
(Sinha et al. 2007).
2.6.2 Adverse Selection and Moral Hazard
Selection bias or adverse selection is the phenomenon in which people who
anticipate high medical care costs would purchase health insurance (Rothschild and
Stiglitz 1976). Adverse selection leads to financial un-sustainability as the premium
set will be lower compared to the average risk of the population covered (Cutler and
Zeckhauser1999; Pauly and Nicholson 1999). It also denotes inadequate pooling
(Cutler and Zeckhauser 1999) because healthier may not be interested to enrol and
less healthy may be interested to sign up resulting in higher healthcare costs and
financial loss to the scheme. Due to asymmetric information, buyers of insurance
know their own risk levels but sellers are unable to distinguish between risks
(Rothschild and Stiglitz 1976). Thus, heterogeneity in health risk faced by individuals
gives rise to selection bias. Many insurance companies adopt strict selection criterion
to screen applicants who are suspected to use expensive medical care which includes
refusal to issue or renew a policy and exclusion of pre-existing illness from the
coverage, waiting period, mandatory reference system and family enrolment (Dercon
et al, 2004). Also, collective membership, social cohesion and high penetration of
target group can eliminate adverse selection (Atim 1998).
Evidence of adverse selection in insurance market in developed countries is
ample (Cutler and Zeckhauser 1999; Savage and Wright 2003). But, the evidence on
adverse selection in MHI shows mixed results (Atim 1998; Jakab and Krishnan 2001).
A study on SEWA in India reports a positive association between older age and
higher frequency of illness and membership in SEWA’s insurance scheme (Ranson
2001). In their review of 82 schemes, Bennett and colleagues (1998) observed the

34
prevalence of adverse selection despite having waiting period and exclusion of pre-
existing or chronic diseases from coverage which was found to be encouraged by
scheme functionaries by allowing the households to insure the member who is most
likely to fall ill. Kutzin and Barnum (1992) examined the impact of Bwamanda
Hospital Community Financing Health on efficiency and equity in health sector and
concluded the presence of moral hazard and adverse selection. RMHC, a social health
insurance scheme in China reported individuals with worse health status to enrol more
than individuals with better health status, especially from partially enrolled
households. Although there was adverse selection, due to high enrolment rate it was
not a threat to financial sustainability of the scheme (Wang et al. 2006). Absence of
selection bias was demonstrated in MHI Units in Philippines (Dror et al. 2005) and in
Senegal (Jutting and Tine 2000).
Moral hazard refers to the tendency for insured individuals to increase their
consumption of health services (Nyman, 2003). Two behavioural changes due to
insurance are ex- ante moral hazard which involves reduction in the use of preventive
care (that increases the chance of falling ill) and ex-post moral hazard which means
increase in the use of health care (especially expensive services once the person is ill)
(Jowett 2004). Moreover, an expected future consumption of health services
determines both health insurance choice and use of health care (Cameron et al. 1988).
Moral hazard has been a problem for the MHI schemes that include inpatient
care in the benefit packages (Bennett et al. 1998). Abuse of the scheme has been
reported in Ghana and Senegal as the identification of beneficiaries was not checked
by the scheme managers, instead other agencies or hospital staff was entrusted with
the job. Even the restriction on benefit package did not curtail moral hazard due to
lack of reference system and provision of full coverage without co-payments
stipulations. Cameron (1988) found higher utilisation of services because of both
adverse selection and moral hazard in Australia. Savage and Wright (2003) support
this finding as insured had longer duration of hospital stay by a factor of up to 3 due
to private health insurance.
Empirical observations in Hong Kong found no evidence of moral hazard
although insured had a higher probability of visiting a doctor or being admitted in the

35
hospital but did not incur more bed days which reflect that realised access was due to
genuine health need than over-utilisation of services (Wong et al. 2006). A good
practice is reported from South Borgou MHOs (Benin) and two CPH schemes in
Nigeria, in which the manager was the first line in the reference system, and
beneficiaries had to get slip from her before going to the facility (Atim 1998). It
should be kept in mind that insurance purchase is motivated by the need to access
necessary unaffordable care. Hence increased use of services cannot be considered as
undesirable in developing countries (Nyman 2003). It may be welfare-promoting as it
removes financial barriers to access care and results in higher utilisation which
otherwise may not be possible for poorer people.
2.7 Resource Mobilisation
Resource mobilisation is directly measured by cost recovery ratio; amount of
resources raised as a share of the country’s total health revenues and indirectly by
quality impact on health care and moral hazard effects (Ekman 2004). Ratio of
prepayment to total healthcare costs indicates degree of financial protection and
access to health services during the need (Carrin et al. 2005). Ekman (2004) found
MHI to mobilise insufficient amounts of resources which was confirmed by Preker
(2002) that MHI could not raise sufficient financial resources from the target
population. Financial sustainability in MHI schemes is difficult due to low penetration
of target populations and insufficient premium collection rates and low income of
target population (Atim 1998; Bennet et al. 1998; Hsiao 2001; Jutting 2001) However,
MHI could mobilise some resources which would have been not possible in its
absence (Diop et al. 1995; Soucat et al. 1997). Among three models of MHI, provider-
based schemes made modest contribution to resource mobilisation (Atim 1998) which
stresses the need for external assistance for scheme sustainability.
2.7.1 Quality of Care
Quality of care denotes delivery of care that achieves favourable balance of
medical risks and health benefits, performing interventions that are safe according to
accepted standards of practice. Haddad (1998) identified four factors to assess quality
of care as perceived by patients: healthcare delivery, health facility, interpersonal
aspects of care and access to services. MHI can improve the quality of the services by

36
acting as a "strategic purchaser" of health care services (WHO 2000) and by
stipulating the quality of prescriptions and treatment given to members through its
empowerment of members and their discussion with health centre managers
(Schneider and Diop 2001). A study on Latino population in USA found insured
individuals perceiving better quality of care than uninsured individuals (Perez et al.
2006). However, despite the potential of influencing quality of care, MHOs neither
engage in strategic purchasing nor address quality issues and pricing of the care (Atim
1998; and Ranson 2003). Lack of functional information systems seriously constrains
the ability of purchasers to influence performance (Waters et al. 2004). Poor quality
of care was the single most important contributor to low level of enrolment in
Maliando scheme (Criel and Waelkens 2003).
2.8 Association between Characteristics of the Scheme and Outcome
The design of the schemes in terms of technical, management, organisational
and institutional characteristics determines the performance of MHI in realizing the
objectives of financial protection, resource mobilisation and social inclusion (Preker
et al. 2002). Successful implementation and achievement of goals of MHI depends on
effective design and management (Jakab and Krishnan 2001; Ahmed et al. 2005) that
improves participation, higher cost recovery rates, and social inclusion of poorest
members of the society (Wiesmann and Jutting 2001). In fact, Bennett and colleagues
(1998) link the limited membership of CHF (Community health financing) with
inadequate financial protection to the poor design of the schemes.
Factors that determine success are the mechanisms incorporated in the scheme
to deal with adverse selection, accommodation of non-cash stream of income of
members, ownership of the community, trained and competent management (Preker et
al. 2002). Success of scheme also depends on the organisational linkages between the
scheme and providers, donor support and government funding (Jakab and Krishnan
2001). Moreover, provider-based schemes have moderate positive effect on resource
mobilisation and a limited positive effect on financial protection (Ekman 2004).
Partner agent model is the best method of providing insurance to the poor (Dercon et
al. 2004). Designing a scheme requires the consideration of benefit package,
premium, information asymmetry problems in insurance market, accounting and

37
management and participation of members (Wiesmann and Jutting 2001). Survival of
the scheme depends on the extent of risk-pooling and resource mobilisation it
achieves (De Allegri et al. 2006) and mechanisms to control the problems of
information asymmetry (Wiesmann and Jutting 2001). Literature on technical,
management, organisational and institutional characteristics and their role in scheme
shows the importance of scheme characteristics in shaping the performance of MHI.
2.8.1 Technical Characteristics
Technical expertise in the management of the scheme in the form of design of
benefit packages, revenue collection, pooling and health care purchase mechanism is
essential to improve the efficiency of MHIs (Preker et al. 2004). It also depends on
the adequacy of the benefit package, policies on co-payment, ceilings, deductibles,
and reimbursement procedure adopted by the scheme (Zhang et al. 2010). Revenue
collection appears to be more successful when the contribution scheme takes into
account the nature of the target population’s income (Jakab and Krishnan 2001).
Annual contributions, collected at the time of harvest of cash crops, seem to be
prevalent among schemes in rural areas (Bennett et al. 1998). Flexibility in the
payment of premium in terms of amount or kind and the time of payment would
contribute to better scheme performance (Wiesmann and Jutting 2001). Certain
technical design features such as affordability of premiums, unit of enrolment, timing
of collection of premium and quality of care offered by the providers influence the
enrolment in a scheme (Carrin et al.2005).
Ratio of prepaid contributions to healthcare costs determines revenue
collection and thereby resource mobilisation (Carrin et al. 2005). While calculating
prepaid contributions, all stakeholders that contribute including central and local
governments, corporation and donors are to be included (Carrin et al. 2005). A review
study carried out by Baeza and colleagues (2002) found that most of the schemes did
not bear the bulk of financial risk. In most of the schemes, central and local
government covered the larger part of the cost of health services.
The degree of financial protection provided by an individual MHI scheme
depends upon the extent to which the benefit package offered covers a comprehensive
package of services particularly high cost services and co-payment (Bennett 2004).

38
Moreover, family enrolment as a unit of membership and waiting period provisions
can curtail adverse selection. Referral system is another component of strategic
purchasing which can curtail moral hazard and improves efficiency (Carrin et al,
2005). Practice of strategic purchasing can improve the quality of the services (WHO,
2000) through negotiation with providers, checking the prescriptions and quality of
care provided to their members before effecting payment, and helping to set up
revolving drug funds. A review on the MHI impact concluded that out of 62 schemes
for which information was available, ten schemes had some form of strategic
purchasing (Baeza et al. 2002). Atim (1998) observed lack of experience and
managerial skills and low levels of negotiating power of MHOs in relation to health
care providers that resulted in ineffective purchasing of health care services. MHOs
do not negotiate with providers or check their prescriptions owing to lack of required
medical and pharmaceutical skills, but it represents an important shortcoming.
Payment and reimbursement methods for hospitals are a part of strategic
purchasing. The most common method of payment is line item and global budgets in
low and middle-income countries (Wouters 1999). Paying claims directly to the
providers increases efficiency (improving the administrative cost ratio) and is far
superior for clients than any method of reimbursement (McCord and Osinde2005).
Fee-for-service payment is another method, which is retrospective, and provides
strong incentive for quality in the sense that they encourage the production of
additional services but it may lead to the overproduction of services (Alvarez et al.
2000). Payment systems influence quality of care. Retrospective rather than
prospective and variable rather than fixed payment method allows for the greatest
flexibility for purchasers to incorporate quality standards in purchasing arrangements
(Waters et al. 2004).
2.8.2 Management Characteristics
Second important characteristics is the management of schemes that include
staff (leadership, extent of capacity building), culture (management style, structure),
and access to information (financial, health information, resources, and behaviour)
(Preker et al. 2004). Strong management of the scheme is necessary due to the
possibility of misuse or overuse of insurance claim by members (Jakab and Krishnan

39
2001; Ahmed et al. 2005). Sinha and colleagues (2005) linked member orientation
and strong community networks, good management practices, systems of planning
and implementation, and the commitment of the management to the success of MHI
performance. In addition, local management, accountability and monitoring are
crucial in implementing equitable and accountable CHF schemes (Polonsky et al.
2008).
Top-down interference with the design and management of the schemes has
negative effect on their function and sustainability (Preker et al. 2002). The bias or
priority of management and the board determines effective management and
development of an insurance product. Management capacity is another important
factor that helps in running the scheme effectively and making necessary adjustments
(Musau 1999). Major hindrances to success of the scheme found by Atim (1998) were
lack of skills in setting premium rates, determining benefits packages, marketing and
communication, contracting with providers, accounting, monitoring and evaluation,
and collecting dues.
Community involvement in scheme management leads to improvements in
revenue collection, cost containment, membership and quality of services (Hsiao
2004) and the absence of community involvement in management may lead to
provider capture and monopoly pricing (Jakab and Krishnan 2001). Schemes
providing better information would improve subscribers’ confidence and enrolment
rates and involvement in decision-making has a significant impact on subscribers’
values (Ouimet et al. 2007). Hence, members should participate in decision making
for better performance of the scheme (Wiesmann and Jutting 2001).
2.8.3 Organisational Characteristics
Organisational characteristics include linkages in the form of vertical and
horizontal integration, strategic alliances, administrative capacity and enlarged risk
pools. In addition, organisational forms, incentive regime (degree of autonomy,
accountability, financial responsibility), and linkages with providers determine
success of the scheme (Preker et al. 2004).
Vertical integration depends on the stipulations regarding the nature and scope
of the products supplied by the health care providers (Zweifel 2004). Organisational

40
linkages such as those between schemes and providers and between schemes
themselves (including national government health system and/or social security
system) are a critical determinant of performance of MHIs (Jakab and Krishnan
2001). However, vertical links with NGOs may increase dependence of the scheme on
external party (Mladovsky and Mossialos 2007) that endanger its sustainability in the
end.
2.8.4 Institutional Characteristics
The key institutional characteristics namely the degree of congruence between
the scheme’s operating rules and participating population’s normal behaviour patterns
and health care providers’ past experience with third-party payments has a significant
influence on the nature and extent of community participation in any given scheme, as
well as the quality of its management (Preker et al. 2004). Additional institutional
characteristics include stewardship (government and donor support), governance,
insurance markets, and factor and product markets (Preker et al. 2002). Regarding
governance structures, two key issues for consideration are the strength and the
quality of these overseeing structures. A strong management board with
knowledgeable people and balance of priorities is essential for the long-term
sustainability of MHI schemes (McCord and Osinde 2005).
Community-financing schemes compete in the factor markets with other
organisations involved in financing and providing health care. Negotiation skills to
conclude the contract with providers and other market players determine the
performance (Jakab and Krishnan2001). In any health market, government plays
stewardship role by creating an enabling legal environment, transferring resources in
the form of subsidies to the poor members of the scheme (Bennett et al. 2004) and
regulating and monitoring MHI schemes. However, minimal government regulation
of MHI has been advocated sighting adverse effect of government subsidies in the
form of cream skimming and adverse selection (Pauly et al. 2006). Public subsidies
work best when administrative structures in MHI intersect with local political
structures to facilitate bureaucrats’ loyalty and enthusiasm to become ‘‘embedded’’ in
schemes and put their energy into making them work (Mladovsky and Mossialos
2008).

41
 Indian studies on the factors that determine success or failure of MHI schemes
stresses economic condition of the society (Dave Sen 1997); income adjusted fee
schedule and waiting period to avail benefits (Dave 1993); subsidy in premium
payment (Prasad 1998); strong and dynamic leadership (Dave Sen 1997) and trust in
the management of the scheme. The theoretical framework for research methodology
was derived from the literature review (Figure 2.2). When individuals have illness,
they either seek care or postpone treatment due to financial and non-financial barriers.
If individuals decide to access care, there are different facilities or place where the
health services are provided namely public and private facilities (hospitals, nursing
homes and clinics), traditional treatment (ayurvedic and homeopathic) and informal
providers (quacks, pharmacists and home medicine). Treatment at these facilities can
take place as either outpatient or inpatient that leads to OOPE and CHE. Individuals
adopt various risk coping strategies such as borrowing, savings, sale of assets,
substitution of labour and reduction in consumption to meet the cost of illness.
Undesirable and unforeseen consequences of OOPE, CHE and risk coping methods
can be reduced or eliminated by enrolling in a MHI scheme. However, enrolment in
MHI depends on various household socio-economic and demographic factors that
including adverse selection factors. Enrolment affects volume of premium collection
and cost recovery and indirectly determines resource mobilisation of the scheme.
Resource mobilisation also depends on the quality of care as perceived by MHI
members. Higher resource mobilization enables a scheme to protect members from
negative financial consequences of OOPE and CHE and ensures financial
sustainability. However, financial protection (as measured by access to care,
utilization of care, cost of care and financial consequences), social inclusion (as
measured by enrolment) and resource mobilization (revenue collection, cost recovery
and quality of care) is influenced by various technical, management, organizational
and institutional characteristics of the MHI scheme. These factors act as catalysts or
inhibitors to achieve the objectives of MHI scheme.
2.9 Summary
The existing literature focuses more on economic outcome of MHI schemes
than social and scheme characteristics. Review of literature points at a number of

42
research gaps in the knowledge base on the impact of MHI in India. Although there
are few studies on financial protection of MHI, the findings are inconclusive. Two
studies on SEWA of Gujarat (Ranson 2001, 2002; Gumber 2001) focussed mainly on
financial protection in terms of OOPE as well as utilisation and Yeshasvini of
Karnataka (Aggarwal 2010) on risk coping behaviour. Hence, the data currently
available in the literature on the impact of existing MHI schemes in India and the
factors that determine the success of a scheme are limited. The literature on the
impact of MHI on health seeking behaviour, access to care, catastrophic health
expenditure and adverse selection is also scanty in India.

43
 Table 2.1 Literature on Financial Protection

Author Scheme Findings

Soucat et al. 1997 Bamako Initiative Increased utilization

Programme, Nigeria
Criel and Kegels 1997 Bwamanda Increased hospital utilization
Criel1998 Rwanda,Congo Increased utilisation of health services
Musau 1999 Kenya Increased hospital utilization
Jütting and Tine 2000 ‘Les mutuelles de sante’’, Decrease in OOPE,higher utilisation
Thiès, Sénégal
Gumber 2001 SEWA, India Marginal effect on FP
Schneider and Diop 2001 PPS, Rwanda Impact on OOPE, utilisation was positive
Yip and Burman 2001 Egypt Middle-class children benefited more
Jowett 2002 VHS, Vietnam Reduction in health expenditures more for
the poor
Ranson 2001,2002 SEWA, India No effect on utilisation of care and OOPE
Liu et al 2002 RMHC, China Diverted health care resources from
expensive IP care to OP care
Deolalikar 2002 VLSS, Vietnam Low income people substituting drug
vendors for formal care
Jutting 2003 ‘Les mutuelles de santé’’, Decrease in OOPE, higher utilisation
Thiès,Sénégal
Msuya et al. 2004 CHF, Tanzania Improved access to care; higher utilisation;
exclude the poor
Dror et al. 2005 MIUs, Philippines Improve access; higher utilisation
Wang et al. 2005 RCMS, China Rich members benefiting more than poor
members
Devadasan et al. 2007 ACCORD, India Partial protection against CHE
Yip et al. 2007 RMHC, China No impact on OOPE for OP or IP treatment
44
Ekman 2007a PPS, Zambia Health insurance fails to protect the member
Yip et al. 2007 RMHC, China Substituted self-medication for formal health
care services
Chankova et al. 2008 MHOs in Ghana, Protection against CHE for IP, no positive
Senegal, Mali effect on OOPE
Dror et al.2009 BAIF, Up Lift, Higher & equal utilisation among insured
Nidan, India
Wagstaff et al. 2008b NCMS, China Health insurance increases the risk of high
catastrophic spending
Franco et al. 2008 MHOs in Mali Increased hospital utilisation of priority
health services
Chankova et al. 2008 MHOs in Ghana, Inconclusive evidence on utilisation; poorest
Senegal, Mali included
Polonsky et al. 2008 CHI, Armenia Poorest included
Lie &Lin 2009 NCMS, China No impact on OOPE or utilisation of formal
medical service
Rao et al. 2009 CHF, Afghanistan No evidence of reduced OOPE; but higher
utilization
Low enrolment due to high premium, low
quality of health care
Zhang et al. 2010 NCMS, China Inclusion of the poor

45
 Table 2.2 Literature on Determinants of Enrolment
Author Country Factors
Jutting,J and Tine J Senegal, HH survey Benefit package, contracts with providers, availability of quality health
2000 care provider ; active engagement of local people; prevalence of trust and
solidarity
Schneider and Diop Rwanda, HH survey Education level of household head, family size, distance to the health
2001 facility, trust and sentiments of ownership and radio ownership
Criel and Waelkens West Africa, MHO Decrease in enrolment due to low quality of care offered
2003
Msuya et al. 2004 Tanzania, HH survey Household income was a found to be a significant
determinant
Schneider 2005 Rwanda, HH survey Benefit coverage and the availability of medicines influences quality
perceptions
Basaza et al. 2008 Uganda, HH survey Ability to pay the premium, quality of health service,
distance
Chankova et al. 2008 West Africa, HH Gender and education of the head of household and economic status
survey Inability to pay premium, even when small, acts as a major barrier to
enrolment
Franco et al.2008 Mali, HH survey Enrolment not significantly associated with socio-economic status
MHO Mutual Health Organisation PPS Prepayment scheme VHS Voluntary Health Scheme
RMHC Rural Mutual Health Care VLSS Vietnam Living Standards Survey CHF Community Health Fund
CHI Community Health Insurance MIUs Micro Insurance Units RCMS- Rural Cooperative Medical System
NCMS New Cooperative Medical Scheme IP Inpatient OP Oupatient HH Household

46
 Technical- Revenue collection, risk pooling, strategic purchase

Management- Staff, culture, access to information

Characteristics RQ5 Organisational – Forms, linkages, incentive regime

Institutional- Stewardship, governance, insurance markets, factor and product markets

Financial Protection RQ1

Incidence of Health seeking behaviour Utilisation of care Cost of care Financial consequences
illness
RQ2
Yes Private providers Outpatient Out of pocket
Access care expenditure Borrowing
Public providers Inpatient Savings
No Catastrophic health
Informal providers expenditure Sale of assets
Substitution of labour
Reduce consumption

Enrolment RQ3 Financial sustainability

Resource mobilisation RQ4

Reasons

Socio-economic determinants Cost recovery

Adverse selection Quality of care

Figure 2.2 Research Framework on the Impact of Micro Health Insurance

47
 Table 2.3 Definition of the Concepts Used in the Study
Concept Variable Working definition

Access to health Number of visits to health provider One’s ability to obtain health services
care when needed. Likelihood of visiting a
(Liu et al. 2002) health provider.

Out of Pocket Direct - fees, medicines, diagnostic Ratio of total medical expenses to
expenses tests, surgery, bed charges; Other aggregate household annual
(Gumber 2001) costs- transportation, food expenditure
expenses; Indirect costs-
income/wage loss of the patient,
interest payments on medical
borrowing

Catastrophic Expenses incurred on Ratio of total illness related expenses

Health Expenditure hospitalisation and outpatient to annual per capita household
(Ranson 2002b) treatment income; Catastrophic if household
expenditure for treatment exceeds
10% of the total annual per capita
household income.

Utilisation of Inpatient care Measured by the use of a inpatient

health services health care facilities
(Andersen &
Newman 1973)

Health Seeking Types of providers – formal or Activity undertaken by individuals

Behaviour informal who perceive to have a health
(Ward et al. 1997) problem or to be ill, for the purpose
of finding an appropriate remedy

Horizontal equity Inter-group income class and Effect on low income class/women in
(Liu et al. 2002) gender equity insured group more than similar class
in uninsured group; distribution of
benefits across groups of people of
similar socio-economic status

Vertical equity Intra-group income and gender People with the greatest need be
(Liu et al. 2002) equity given the most care; distribution of
benefits across groups of people
differing in socio-economic status

Moral hazard Duration of stay at hospital Longer stay at hospital than expected
(Wong et al. 2006)

48
 CHAPTER 3

RESEARCH METHODOLOGY
3.1 Introduction
This chapter describes the research methodology of the study. The first
section explains research methods and second section deals with data collection and
sources of data. Third section describes sampling design and fourth section briefly
explains the study settings and Sampoorna Suraksha Programme. The literature
review provided theoretical basis for designing the study and collection of data. The
nature of research problem led to the choice of case study method and use of
quantitative and qualitative methodology.
3.2 Research Approach
Case study method is suitable when complex issue requires in-depth analysis,
especially the effectiveness of a programme intended for the socio-economic
development of the community. This method emphasises detailed contextual analysis
of quantitative and qualitative data to explain both the process and outcome of a
programme, a phenomenon or an entity (Tellis 1997). It is an empirical inquiry into a
phenomenon within its real-life context in which multiple sources of evidence is used
(Yin 1984). The researchers mistake case study to be a qualitative research. However,
it is suitable to collect quantitative evidence especially numerical and categorical
responses of subjects of the study (Block 1986; Yin 1984). “How” and “why” related
to a phenomenon are explored through this method. Researchers have used the case
study method to investigate the effectiveness of CHI schemes in different settings
(Ranson 2001; Jutting 2003; Ekman 2007a; Chankova et al. 2008; Polonsky et al.
2008; Wagstaff 2008a; Zhang 2010; Aggarwal 2010). Hence, single case study
approach was the suitable research method, keeping in mind potential audience for the
final report and research questions.
To answer broad research questions, deductive or realist approach with precise
objectives was the choice. In addition, inductive or constructive approach to answer
refined questions that on scheme design features and its impact of the MHI is
required. The study adopted a combination of both the approaches, starting with
deductive and gradually moving to inductive approach. The design of the study was
based on the research purpose and research approach. Objectives 1 to 4 were
addressed through quantitative methodology whereas objective 5 required qualitative

49
methodology. However, certain aspects of objective one (access to care) and objective
three (barriers to enrolment) used qualitative approach. Thus, triangulation reveals
multiple aspects of a single empirical entity and provides knowledge that is more
comprehensive. Both qualitative and quantitative data from multiple sources (persons
and places) and methods (surveys, interviews and focus group discussions,
documents) were gathered. The cross-sectional household survey data was the basis
for assessing the objectives 1, 2 and 3. Focus group discussions (FGD) and interviews
provided the data to study objective 1 through 5, but emphasis was on objective 5.
Annual reports and interviews with administrators correspond to the objective 4 and
5.
The study adopted a descriptive research to describe the impact of health
insurance. A mix of structured and unstructured approach facilitated a comprehensive
understanding of the research problem. The structured approach pre-determined the
objectives, sample design, tools of data collection and survey instruments. In addition,
unstructured approach helped to understand the problems faced by the people while
accessing health care, barriers to enrolment and association between design
characteristics and outcome of the scheme.
Any impact study suffers from methodological problem of self-selection bias.
It becomes difficult to attribute the positive findings to the programme impact alone
when people self-select to be members. Voluntary membership thus poses a challenge
in that those who enrol may have time invariant unobservable characteristics
(endogenous variables) which influence the outcome. This problem either exaggerates
the significance of findings or undermines the programme impact. If the enrollees
have hidden health risks, they would join and use the health services leading to higher
utilisation, wrongly construed as positive impact of the programme. In addition, due
to high utilisation, out of pocket expenses would be high leading us to conclude that
the programme does not decrease out of pocket expenses.
The present study tackles the problem of endogeneity (self-selection bias) in
three ways. Firstly, a comparison of the newly enrolled Sampoorna Suraksha
Programme members (SSP) (considered uninsured) and the renewed members (taken
as insured) of the programme on various measures of outcome limited the bias. The

50
logic was that both members
m seelf-select in
nto the proogramme soo the unobservable
characteristtics would be
b homogenneous. In ad
ddition, bothh the groupps were mem
mbers of
self-help groups;
g hencce, observaable charactteristics woould be sim
milar. The renewed
r
members were
w renew
wing their enrolment
e in
i the prevvious one oor more yeears. A
comparisonn of the renewed and
a newly insured members
m w
would conttrol the
unobservabble characteeristics thaat induce people
p to get
g insurannce. In add
dition, a
comparisonn of uninsurred self-hellp group (SHG) membbers with thhe insured members
m
corroboratees the findinngs. Insured group con
nsists of rennewed mem
mbers and un
ninsured
group incluudes newly enrolled meembers of SSP
S and uniinsured selff-help group
p (SHG)
members of
o SKDRD
DP (Shri Kshetra
K Dhaarmasthala Rural Devvelopment Project).
P
Minimizingg the contam
mination off the results due to the location off residence required
r
the selectioon of newlly insured, insured an
nd uninsureed househollds from th
he same
location. Seecondly, thee study of adverse
a seleection as a research prroblem wou
uld show
the magnittude of sellf-selection bias. Third
dly, use off Durbin-W
Wu-Haussmaann test
controlled the
t self-seleection bias statistically.
s .
Ressearch questtion on the inclusion
i off the pooresst in MHI scchemes requ
uires the
definition of
o poorest inncome class, measured
d by annuall per capita household income.
In this stuudy, social inclusion means
m the inclusion of
o the poorrest househ
holds as
members of
o Sampoornna Surakshaa Programm
me. The cutooff point off income to classify
householdss to differennt income classes
c is difficult
d duee to lack off consensuss on the
definition of
o poverty. The official line of pov
verty was 368 per peerson per month
m for
rural areas and 559 per personn per montth for urbann areas in 2010-11 (P
Planning
Commissioon of India)). This doess not repressent the reaal picture oof poverty in
i India.
Hence, thiss study adoopts a simplle classificaation of houuseholds baased on ann
nual per
capita income. The claassification relates to sample
s houuseholds onlly and not to
t entire
target popuulation.
Thee classificattion of housseholds into
o five incom
me groups considered
d the per
capita annuual income of the entiire sample. The per caapita annuaal income data
d was
divided into five equaal parts, afteer arranging
g them in an
a ascending order, as quintile
1(first 20%
%), quintile 2 (next 200%), quintile 3 (next 20%), quinntile 4 (sub
bsequent
20%) and quintile 5 (last 20%).. Thus, we defined five dummy variables, one for

51
wealthy (quintile 5), non-poor (quintile 4), vulnerable non-poor (quintile 3), moderate
poor (quintile 2) and extremely poor (quintile 1).
Analysis on the access and utilisation of care, health seeking behaviour, out of
pocket expenditure, catastrophic health expenditure (objective 1) and risk coping
strategies (objective 2) considered cases of households that reported illness in the
previous year of the study. Analysis considered individual cases since more than one
member in a family can fall sick. Analysis of the factors influencing utilisation of
care, health-seeking behaviour, OOPE, CHE and risk coping strategies used
logistic/multiple linear regression models. Enrolment in MHI depends on both supply
and demand factors. Objective three dealt with demand side and investigated the
influence of head of the household, household and community characteristics on
enrolment. Objective five deals with factors (supply side) related to enrolment, design
and implementation of the scheme.
3.3 Data Collection Methods and Data Sources
A small-scale pilot study using Kannada translated questionnaire in December
2010 gave the information on the relevance of questions, ease of administration and
time required to fill the questionnaire. It facilitated the measurement of the validity
and reliability of questionnaire. The sample size was 30 and the respondents were
selected using convenience sampling method. After 15 days, retest on the 15
respondents using the same questionnaire confirmed the reliability. Subject experts
scrutinised the content validity of the questionnaire. Kappa coefficient, Cronbach’s
alpha and intra-class correlation coefficient measured the reliability of various items
on the questionnaire. It was re-drafted after making changes to wordings of the
sentence, order of questions, range of answers on multiple-choice questions and
removal of some questions that was unnecessary or ambiguous.
The questionnaires and interviews form the basis of data collection. The
quantitative methods include questionnaire survey and data on financial performance
of the scheme (objectives 1, 2, 3 and 4) (Table 3.1). The study collected data using
questionnaire from households in Dakshina Kannada, Uttara Kannada and Gadag
districts in Karnataka in the first half of the year 2011. The qualitative instruments
included focus group discussions (FGDs) with insured and uninsured members and

52
 interviews with health care providers and scheme administrators. FGD with members
and non-members addressed the issues related to barriers to access care (financial and
non-financial; objective 1), participation in management and non-enrolment
(Objective 3). Interviews with the health providers provide the data on the strategic
purchasing and the problems faced during administration of the care to members.
Interviews with the managers of insurance administration department in the hospitals
or doctors triangulated the data gathered from field staffs and scheme members
regarding claim settlement and contribution of the scheme to the hospital revenue.
The interview data collected from the field staffs helped to know the moral hazard,
adverse selection, and claim settlement, quality of hospitals and members
participation in management of the scheme. Interview with scheme administrator
helped to know the objective of starting the programme, initial and current difficulties
faced during implementation of the scheme, management and administration of the
scheme, financing arrangements and performance, human resource policies, criterion
for membership, rapport with providers and strategic purchasing, benefit package
decisions and agreement with insurance companies (objective 5).
Table 3.1 Data Collected Using Questionnaire
Demographic characteristics Age, gender, years of education and occupation.
Socio-economic characteristics Amount and sources of income, assets and
monthly expenditure of the household,
household size- gender wise, caste and religion,
place of residence.
Cost of illness, health care access and Episodes of illness, types of care sought, reasons
utilization for choosing providers, result of treatment,
number of days of illness, direct and indirect
cost of illness.
Risk coping methods Risk coping strategy of household for illness,
amount of money borrowed/savings used/asset
sold.
Membership details Years of membership, claim data, reasons for
enrolment, mode of payment of premium.
Quality of hospitals Cleanliness, expertise of doctors, care of nurses,
treatment by other hospital staff, time taken for
examination by doctors, availability of
diagnostic facilities/medicines

53
 This study chose the case of SSP initiated and implemented by SKDRDP in
Karnataka. Information gathered from the members of the programme using
questionnaires and focus group discussions forms the primary data of the study. The
study collected data related to health behaviour and expenses in the previous year of
the study, February 2010 to March 2011 (one-year recall period for inpatient care and
three months for outpatient care). Secondary data sources were books on health
insurance, periodicals, journals that helped in conceptual mapping and preparation of
questionnaire. Annual reports, brochures, information pamphlets and list of hospitals
in the network provided scheme related information.
The hypothesis driven analysis of the data used SPSS version 17.0 and applied
logistic and multiple linear regression models to test the hypothesis. A p-value of 5
percent was the criterion for significant association. The FGDs were (Kannada
language) videotaped, transcribed and translated into English.
3.4 Sampling Design
In 2010-11, nine districts had SSP operations that varied substantially in terms
of income, education, geography, natural resources, disease pattern, sex ratio,
economic development and health indicators. The data set consisted of three districts
and three to five taluks in each district, the sampling included 10 taluks overall.
The population of study is the SHG members of SKDRDP who were newly
insured, insured or uninsured. Districts, taluks, valayas1and karyakeshtras2 were the
clusters and each successive stage selected these clusters randomly. The list of
member households in each karyakshetra formed the sample frame. Households
formed the unit of the study. Head of the households or spouses were the respondents
to provide general demographic, socio-economic data, illness related health seeking
behaviour information.
While calculating sample size, level of precision, level of confidence and
degree of variability in attributes are important considerations (Israel 1992). As the
target population size was 8,92,740 households in 2011-12 (SSP households were
420302 that included insured and newly insured), 385 was considered a desirable
______________________________________________
1
valaya in Kannada means region2karyakshetra in Kannada means division of SSP

54
sample size per group given the confidence interval of 5 percent and confidence level
of 95 percent. As the study intends to compare the performance of these three main
groups across various dimensions, 385 members from each category was the desired
sample size. Replacement of respondents unavailable for administering the
questionnaire was not done.
3.4.1 Sampling Procedure
The sample was drawn using five-stage cluster design with random selection
procedures. In the first stage, nine SSP districts in Karnataka wee listed and re-
arranged in an ascending order based on human development index (HDI). A random
selection of three districts namely Dakshina Kananda, Utatra Kannada and Gadag was
done. In the next stage, taluks in these districts formed clusters. The list of taluks
provided by the district websites formed the basis of selection of taluks. Using the list
of taluks listed according to the literacy index, we selected 10 taluks randomly based
on the probability proportion to population size sampling method (number of taluks
selected depends on the total number of taluks in each district). In the third stage, list
of valayas (obtained from the taluk SSP office) in the selected taluks was used to
randomly select valayas. These taluks had 97 valayas and we chose twenty percent of
valayas for the study (18). One or two valayas from each taluk were selected
depending on the number of valayas in each taluk. In the fourth stage, from the list of
karyakshetras, four to five karyakshetras were selected from each valaya using the
probability proportional to the number of karyakshetra in each valaya. Thus, eleven
percent of the total karyakshetras (84) formed another cluster.
In the next stage, using the list of households (insured, newly insured and
uninsured) in each karyakshetra, desired number of sample was selected using
systematic sampling method. Third and fourth stage relied on the probability
proportional to population size. Fifth stage used systematic sampling method to select
households (10-15) in each karyakshetra. Total sample size included additional five
percent to deal with the problem of non-response or partly filled questionnaire.
Therefore, 18 valayas, 84 karyakshetras were selected and 1260 sample size was
determined taking into consideration the potential problem of non-response. Hence, a
sample of 420 for each category of insured, newly insured and uninsured group was

55
considered. However, due to non-response and incomplete or wrongly filled
questionnaire, data of 416 renewed insured, 366 newly insured and 364 uninsured
households were used for further analysis.
SSP members who have been renewing their SSP status in the previous one
or more years were classified as insured members. Those SHG members enrolled in
2011-12 were newly insured members and SHG members who did not buy health
insurance formed uninsured group. Ten FGDs, one in each taluk comprising fifteen
members from insured and uninsured/newly insured group collected qualitative data
and each FGD lasted for 30 minutes. The group included both men and women to get
an insight into the various issues related to SSP and health care seeking behaviour.
Six in-depth interviews with providers in six taluks and interviews with
administrators/office staff and field staff provided in-depth information on SSP
operations. Purposeful sampling strategy was the basis for the selection of scheme
administrators, health care providers for interviews and members for FGDs.
3.5 Study Setting
India is the second fastest growing major economy and is the tenth largest
economy in the world by nominal GDP and fourth largest in purchasing power parity
in 2011 (IMF 2010). There are 640 districts within 28 states and 7 union territories. In
India, there are 7,000 towns and 6 lakh villages. Population and number of districts
were highest in Uttara Pradesh and lowest in Sikkim and Daman and Diu. India has
17.5 percent of world population. Literacy rate for female is 65.46 percent and for
male is 82.14 percent (Census of India 2011).
Economic reforms have been instrumental in accelerating the growth.
However, India has failed to bring in policies to remove the obstacles in social
development, especially in health sector. To provide quality health care facilities to
people, especially to those below the poverty line, Government of India has initiated
several health programmes. National Health Policy of India (2002) aims to achieve
health system goals such as improvement in the health status of the population and
health standards. National Rural Health Mission in 2005 has a similar objective of
enhancing the availability and access to health care.

56
 Karrnataka, as a state in inddependent India
I coverss an area off 191,976 sq
q. km. or
5.83 percennt of the total geographhical area off India. It raanks eighth largest Indiian state
by area, thee ninth by population
p a seventh
and h in terms of
o Net State Domestic Product.
P
The Gross State Dom
mestic Produuct (GSDP)) of the staate in 20100-11 was 271,956
crore (basee year 1999--2000) (Dirrectorate of Economicss and Statisttics, GoK). The per
capita GSD
DP at currennt prices wass 50,974 in 2009-10 and
a 518588 in 2010-11
1. It has
the sixth hiighest per-ccapita GDP of all statees (Econom
mic survey oof Karnatakaa, 2010-
11).The staate is the maanufacturingg hub for so
ome of the largest
l publlic sector in
ndustries
and premieer science and
a technoloogy researcch centres inn India. It hhas emergeed as the
pan-India leeader in thee field of infformation teechnology.
Thee state receeives exterrnal assistaance to health sectorr from the central
governmennt (15%) andd the remaiining investtment moneey comes frrom the statte funds.
However, the
t GSDP spent
s on heaalth was 0.7
7 percent inn 2004 and less than 4 percent
of total buddget expendditure went to health seector. The state
s failed tto meet the targeted
expendituree of 7 perceent of total budget adv
vocated in National
N Heealth Policy
y (2002).
In additionn, primary services reeceived 50 percent of
o allocatedd funds, seecondary
services goot 13 percennt and tertiarry services received 344 percent off funds. Th
he health
infrastructuure in the state has skewed
s dev
velopment as
a few citiies like Baangalore,
Mangalore and Manippal have exccellent facillities whereeas more than three fiffth cities
and towns lack
l basic health
h care infrastructu
i ure (Econom
mic Survey oof Karnatak
ka 2010-
11). Theree is a largge disparityy in inter-d
district performance iin the health care
infrastructuure and indiicators. Thee poor, espeecially in north
n Karnaataka has to rely on
public healtth care systtem, as privaate hospitalls are expennsive and noot easily avaailable.
SSP
P was active in nine districts
d nam
mely Dakshiina Kannadda, Udupi, Kodagu,
K
Uttara kannada, Chikkmagalur, Shimoga,
S Gadag,
G Haveeri and Dhharwad in the
t year
2011-12. This
T study seelected Dakkshina Kann
nada, Uttaraa Kannada aand Gadag districts
using randoom sampling method.

57
3.5.1 Prrofile of Gaadag
G
Gadag distrrict is locateed in the western
w partt of northerrn Karnatakka. It has an
n
area off 4651 sq. km
k with thee density of
o populatio
on of 229 inn 2011 (Diirectorate off
Econom
mics and Statistics,
S G
GoK 2008)). It has 1.74
1 percennt of totall Karnatakaa
populattion. Literaccy rate was 71.4% and number off female perr 1000 malee was 978 in
n
2010-111. The cityy is popularr for printinng press and
d handloom
m. The disttrict had net
district income waas 281,9488 lakh at cuurrent pricees in 2007--08 that connstitutes 1.3
3
percent of GSDP. The districct had per capita
c incom
me of 21,,600 (2007--08). It wass
ranked 15th in the state
s amongg other distrricts in term
ms of per cappita incomee. It ranks at
a
13th plaace in terms of HDI in the
t state off Karnataka (KHDR 2005). It is onne of the top
p
five disstricts in thee gross enroolment of children
c in the school and one off the bottom
m
five disstricts in terrms of life expectancyy at birth (K
KHDR 20055). It has hiigh value of
educatioon index inn the state after
a 5th place in terms of inncome index
Udupi (0.750), 15 x
(0.525) and 22nd rank in heealth index (0.628). Itt has five taluks nam
mely Gadag-
Betagerri (administtrative headdquarters), Shirhatti,
S Ron,
R Mundarragi, and N
Naragund. In
n
the disttrict, 43.3 percent
p of villages
v mber of villlages is 32),
had sub-centres (total num
21.9 peercent had primary
p heaalth centre, 43.8 percen
nt had goveernment heaalth facility
y,
31.3 peercent had doctors,
d 3.1 percent haad ASHA (A
Accredited Social Heaalth activist))
workerss, and all thhe villages had
h anganvaady workerrs (Nationall Family Heealth Survey
y
2005).
3.5.2 Prrofile of Daakshina Kaannada
Dakshinaa kannada, also know
wn as South
h Kanara, is the southhern coastaal
district of Karnaataka. It has
h five taluks nam
mely Manggalore (addministrativee
headquaarters), Banntwal, Puttuur, Sullia annd Belthangaady. It has an area of 44599 sq.km
m.
Literacyy rate was 88.62 perceent, of whicch males litteracy rate was 93.31 percent and
d
that of females was
w 84.04 percent.
p Thee net districct income was 674,3352 lakh in
n
2007-088 and it conntributed 4.6 percent of
o the GSD
DP (Directorrate of Ecoonomics and
d
Statisticcs, GoK). The
T district had per cappita incomee of 33,1554 and rankks second in
n
terms of
o per capitaa income. Itt ranks secoond in termss of HDI inn the state of Karnatakaa
(KHDR
R 2005). It is one of the top fivve districts in the statte in literaccy rate, lifee
expectaancy at birthh and per capita
c incom
me (KHDR
R 2005). It occupies fifth place in
n

58
education index
i (0.7007), 3rd placce in health
h index (0.8823) and 2nd place in income
index (0.6336) in the sttate (KHDR
R2005). In the
t district, 54.8 percent of villag
ges (total
number of villages is 31) had suub-centres, 12.9 percennt had prim
mary health centres,
64.5 percennt had goverrnment heallth facility, 19.4 percennt had doctoors, 3.2 perccent had
ASHA worrkers, and all
a the villagges had ang
ganvady woorkers (Natioonal Family
y Health
Survey 20005).
3.5.3 Profile of Uttaraa Kannadaa
U
Uttara Kannaada, knownn as North Kanara,
K is onne of the biiggest distriicts with
eleven taluuks. It is in the northerrn coastal part
p of Karnnataka. Uttara Kannad
da had a
population of 14, 36,8847 (Censuss of India 20
001). It hass area of 10,291 sq.km
m and the
density of population was 140 peer sq.km in
n 2011. Uttaara kanandaa has eleveen taluks
namely Kaarwar, Kum
mta, Ankolla, Honnav
var, Bhatkaal, Sirsi, S
Siddapur, Yellapur,
Y
Haliyal, Suupa and Mudagod. Thee district had
d a per capiita income oof 12,043 in 2001
and occupied 11th rankk in the statte in terms of per capitta income. IIt ranks at 7th place
in terms off human devvelopment index in th
he state of Karnataka.
K IIt is one off the top
five districtts in literacyy rate. It rannks 19th (0.6
632) in eduucation index,4thin health index
(0.781) andd 11th in inccome index (0.546) am
mong all the districts in Karnataka (KHDR
2005). Karw
war is the district
d admiinistrative headquarters
h s. In the disstrict, 40 peercent of
villages (tootal numberr of villagees is 35) haad sub-centrres, 22.9 peercent had primary
health centrres, 42.9 peercent had government
g t health faciility, 17.1 ppercent had doctors,
11.4 percennt had ASH
HA workerrs, and 91.4
4 percent of
o the villagges had ang
ganvady
workers (N
NFHS 03).
3.6 Shri Ksshetra Dhaarmasthala Rural Dev
velopment Project
P
SKD
DRDP is a novel proggramme inittiated in 19982 in Belthhangady, Dakshina
D
Kannada unnder the vissionary leaddership of Dharmadhik
D kari Sri Veeerendra Heg
ggade of
Sri Kshetrra Dharmaasthala to uplift the poor and transform
m poverty stricken
householdss’ better liviing throughh self-emplo
oyment. It is registereed under Ch
haritable
Trust Act inn 1991.The well-know
wn microfinaance program
mme was sttarted in 1995, with
a shift from
m the conceept of charitty based deevelopment assistance to self-help
p groups
model. Thee main focuus of SKDR
RDP is ruraal developm munity development
ment, comm
and urban community
c y developmeent for whicch it has suuccessfully implementeed many

59
programmes namely livelihood promotion programmes, self-employment training,
‘Pragathibandhu’ (for small, marginal and landless labourers), agriculture
development programmes, irrigation programmes, ‘Siri’ (provides market outlet to
products of members), ‘Sampoorna Suraksha’ (micro-insurance) and ‘Jnana Vikasa’
(social empowerment). It grabbed ‘Microfinance India Award’ by Hong Kong based
HSBC bank and Access Development Services in 2011. ‘Pragathibandhu’ and
Sampoorna Suraksha,’ were selected as one of the final best three models and have
won the ‘Change Makers Award’jointly promoted by US based Ashoka Foundation
and Citi Bank in 2011. It also bagged‘Ashden Golden Award’ for Global Green
energy in 2012.
3.6.1 Sampoorna Suraksha Programme
Sampoorna Suraksha, meaning total security (Kanishta Nirvahane, Garishta
Bhadrate1) was started in 2004 to provide financial risk coverage to the SHG
members of SKDRDP, staff and their families in case of unforeseen consequences of
ill health, natural disasters and death.The programme also provides credit in case of
excessive inpatient medical expenses to insured families. The benefit package
includes cashless treatment for hospitalisation and delivery expenses, death
compensation, and sickness allowances. Enrolment of members takes place through
SHGs and field staff in the month of February of every year. Initially, it was offered
in Uttara Kannada, Dakshina Kannada, Udupi in 2004, later it was extended to
Chikkamagalur, Shimoga, Kodagu, Dharwad, Gadag and Haveri. It was introduced in
Tumkur and Belgaum districts in 2011-12. The total coverage was 8000 villages and
41 towns.
3.6.2 Key Features of Sampoorna Suraksha Programme
As per IRDA (Micro-insurance) Regulations 2005, private insurance
companies should mobilise seven percent of total premium from rural India in the
sixth/seventh year of operations. Owing to high cost of transaction and serving rural
population scattered in more than six lakh villages, these companies have tied up with
micro - finance institutions (MFIs) to meet the statutory requirements of IRDA. Since
2004, well-known private-for-profit insurance companies and public sector companies
1
In Kannada language, it means minimum management and maximum security

60
have offereed group heaalth insurannce policies to SHGs thhrough SSP model. SSP
P acts as
communityy based agggregator andd a TPA. Itt assumes the
t role of agent or in
nsurance
intermediarry as it usess existing innfrastructuree and establlished channnels of micrro-credit
and micro-ssavings to offer
o insurannce productts to the SH
HGs. The obbjective of SSP
S is to
provide finnancial assisstance to meeet unexpeccted medicaal expenses to the stakeeholders
and their family,
f to facilitate
f acccess to thee best hosppitals and tto provide medical
facilities att lower costt (Sampoorrna Suraksh
ha Brochuree, 2011). Organisation map of
SSP depictts hierarchyy structure (Figure 3.1). Followinng paragrapphs briefly describe
d
salient featuures of SSP
P.
i. Eligibilityy
me is voluntaary offered to SHG members andd their family of SKDR
The schem RDP, its
staff and thheir family members
m inn the age grroup of threee months too 80 years. Family
includes seelf, spouse, unmarried daughters, sisters, brotthers, their wives and children
living undeer the same roof,
r and coook in the saame kitchenn.
ii. Premium
m
Thee premium payable
p for the first meember of a family wass 350 in 2011-12.
2
There has been
b a signnificant chaange in the premium contribution
c n and margiinal cost
per additionnal memberr since 20077.
iii. Enrolmeent and prem
mium paym
ment procedu
ure
mium is too be paid yearly in the month of Februaary of everry year.
Prem
Consideringg the seasonality of income off target poopulation, a credit facility is
provided too SHG mem
mbers to pay the prem
mium at a low interesst rate. Fielld staffs
(called in loocal languaage as sevannirathas) haave to create awareness on the ben
nefits of
SSP prior too the enrolm
ment monthh. Their resp
ponsibilitiess include fillling the registration
forms, colllection of the
t premium
m from thee members and issuinng SSP card
d to the
members. Supervisors
S s have to monitor
m thee enrolmentt in their vvillages and
d submit
consolidateed accountss of the subscription
s n and prem
mium to P
Project Off
fficer of
respective valaya,
v whoo in turn would send it
i to SSP heead office. T
The office pays
p the
m 1st March
premium too the insurance compaany for thee term begiinning from h to 28th
February. In
I 2011-12,, maximum number of family mem
mbers per policy was reestricted
to seven.

61
 Tablee 3.2 Key Feeatures of Sampoorna
S Suraksha
S Prrogramme
Ownersship and govvernance Shrree Kshetra Dharmasthala Rural D Developmen nt
Proj
oject (SKDR RDP) Trust and Insurannce
Com mpanies
Micro health
h insuraance Modell Parrtner-agent model
m for hospitalisatio
h on benefit
andd full servicee model forr special bennefit cover
Insurannce companyy Uniited Insuran nce Companny Ltd., Orieental
Insuurance Com mpany Ltd., New India Assurance
Com mpany Ltd., National Insurance Company
Ltdd. in 2011-12 2
Target population
p Selff Help Grou up (SHG) members
m of SKDRDP
andd their famillies
Enrolm
ment 4200,302 households, 1,6600,185 membbers in
20111-12
Eligibillity Agee group bettween 3 moonths to 80 years, only y
for SHGs of SK KDRDP
Benefit package Lifee, Health annd Assets; 5000/ for
n per person in a familyy
hosspitalisation
Networrk hospitals 1100 in 9 districcts of SKDR RDP and Baangalore
andd Hassan in 2011-12
Processs of reimburrsement Casshless; paym ment directlyy to the hosspital
Methodd of reim mbursemennt to Paidd by Real Time
T Gross Settlementt
hospitall
Externaal funding Nonne
Nature of relationsship with thee Conntract basis
provideer
Role off governmennt Nonne
Commuunity involvvement in Feeedback given at the annnual or monnthly
scheme design andd managemeent meeetings is useed to improvve the schem me design
Role off health caree provider Proovision of heealth care seervices, no
invoolvement inn managemeent or desiggning
bennefit packag
ge
Politicaal context No governmen nt involvement
Structurre and perfoormance of Muulti-tier struccture, privatte sector doominates
health care
c system

62
 Presiden
nt

Board of Trrustee

Executive Diirector

Operational Wing Audit Wing Administratoor Federal Win

ng

Director Director Director

Director

Project Officeers Chief Auditorr Senior Managger P

Project level fedeeration

Field supervissors Villaage levellevel

federattion
Group Auditorr Branch manaager Taluk

Field Stafff

A
Assistant managgers/Office staff// Staff, Dataa entry Operatorrs SHGs

Figure 3.1 Organisatio

on Structuree of SSP
iv. Benefits
Thee benefits of Sampoooran Suraksha covereed in 2011-12 were medical
benefits (hhealth treaatment) annd special benefits (delivery allowancess, death
consolationn, domiciliary treatm
ment, rest allowance and consoolation of natural
calamities).
a. Medical benefits
Heaalth benefitss are providded as cash
hless treatment at netw
work hospitaals up to
the sum asssured ( 5,0000/ per inddividual). The
T schemee offers a fa
family floateer cover
and a multtiple of 5,,000 for seven membeers summinng up to 335,000/. Ou
utpatient
treatment is excluded from coverrage. Unlikee a private for-profit ccompany, SSP does
not exclude pre-existiing diseasees. In addition, there is
i no waitiing period and co-
payment orr deductiblees to be paidd by the poliicyholders. The coveraage is provid
ded only
for general ward admiissions in thhe network hospitals, however
h noon-network hospital
admissions are consideered for reim
mbursement in special cases.

63
b. Speciial benefits
The membbers of the programmee can avail special benefits in thhe nature off
consolaatory benefiits to overccome liquiddity constraaints due too the risk oof ill health
h,
natural calamity orr loss of lifee.
Maternity benefits
Maternity benefits
b covver the firsst two deliv
veries. SSP pays an am
mount of
2,500 foor normal deliveries
d annd 5,000 for
fo caesarian
n section deeliveries. Suubmission of
the Forrm C alongg with meddical bills and dischaarge summary of the hospital iss
requiredd to avail thhis benefit.
Rest alllowance
Rest allowaance providdes a maxim
mum of 1,5
500 per weeek and a minnimum of
50 per week
w for a maximum period
p of one month until
u 2010-111. Doctors’’ certificatee,
requisittion letter ciiting the sevverity of illnness, and nu
umber of working
w mem
mbers in thee
family and
a Form C was requirred to avail the benefit..
Personaal accident cover
A compenssation of 12,500
1 to a primary
p meember in thee event of loosing one off
the major organs inn an accidennt is given. In case of permanent
p d
disability, 25,000/ can
n
be availled.
Death consolation
c
The benefitt package includes a death
d conso
olation coveer of 2,0000 on naturaal
death of any registtered membber and 5,0000 in case of primary member off the family
y.
25,0000 is paid onn the accideental death of the prim
mary membeer on the suubmission off
first infformation report,
r inveestigation reeport, autop
psy report and statem
ments by thee
witnesss and familly members along wiith the certtificate of cause of ddeath and a
requisittion letter.
v. Accesssing mediccal care
Insured meembers cann get meddical treatm
ment in anyy of the 1110-network
k
hospitall, even wiithout the referral lettter from any
a doctor.. Within 224 hours of
admissiion, insuredd has to prooduce SSP card to the hospital reegistration ssection. Thee
hospitall has to sennd the pre-auuthorisationn request to SSP head office.
o Meddical officerr
will verrify line off treatment planned innvestigation
n and total costs. The accounting
g

64
section will check the unclaimed balance and the previous claim record of the
member. If approved, the office will send an online authorisation letter with the
sanction limit to the hospital. Sampoorna Suraksha staff visits the hospital to verify
the admission of the member. This mechanism prevents moral hazard and
impersonation.
Within ten days of discharge, the hospital has to send the claim Form A with
pre-authorisation number given by SSP office along with a photocopy of SSP card,
discharge summary, investigation reports, laboratory reports and the total bill along
with separate bills for the diagnostic and laboratory investigation. SSP office sends
the sanctioned amount by RTGS (Real Time Gross Settlement) to the hospital. In case
of treatment in non-network hospitals, insured can submit Form B for reimbursement.

65
vi. Claims adjudication and settlement
The network hospital has to send medical bills, discharge summary, reports
(investigation and diagnostic such as X-ray, CT scan, MRI, laboratory) within ten
days after the discharge of insured patient. Medical officers of SSP in the head office
scrutinise pre-authorisation forms, claim applications, investigation reports and
discharge summary. Office staff verifies the name, address and other details and
unclaimed total amount of benefit. The settlement of the sanctioned claim takes place
within 30 days of receipt of claim application using RTGS system.
The submission of Form C is required to claim special benefits. The
Supervisors, Project Officers in the region verify/ endorse it and send it to SSP head
office at Dharmasthala. The insurance company conducts audits and inspections.
Project officers of respective regions send medical team from SSP office to ensure
quality medical care to members of the scheme and take steps to prevent supplier and
member moral hazard.
vii. Client servicing and claim management process
SSP adopts a combination of linked and full service model of micro insurance.
It acts as an agent for a partner (insurance companies) in which agent takes the
responsibility for the delivery and marketing of products to the clients whereas partner
provides actuarial expertise and financial coverage and absorbs the risk of medical
component of the programme. The basis of medical benefit component of SSP Linked
model in which SSP collects premium, manages the claim processing and payment to
providers. Special benefit is a full service model in which SSP provides risk coverage
including claim management.
A Memorandum of Understanding between SKDRDP, insurance company and
network hospitals specifies the role and responsibilities of each party. The insurance
company issues a group health policy for a one time consolidated premium to SHGs
who enrol in the programme. SSP issues a membership card to each policy holding
family. The role of SSP in providing medical benefits are the registration of members,
collection of the premium, maintenance subscription records, handing over
subscription amount along with registration forms and consolidated statement to
insurance companies. It has to forward the approved claim forms to the insurance

66
companies,, coordinatee the pre-aauthorisatio
on with inssurance com
mpanies an
nd settle
cashless claaims with network
n hosspitals. SSP office sendds a debit noote to the in
nsurance
company. A debit notte should noot exceed 10 lakh. Iff it exceeds,, the submiission of
additional debit
d note for
fo 10 lakhh is required
d. In case off the shortagge of funds to make
payments to
t hospitals,, micro finaance divisio
on of SKDR
RDP extendds a credit advance.
a
The inter-trransfer of funds
f betweeen micro-ffinance andd micro inssurance is a special
feature of SSP
S that higghlights the advantage of nesting MHI
M in parent organisaation. In
providing special
s benefits, it hass to scrutin
nise and verrify treatmeent claims received
r
from hospittals and reim
mburse the members. Figure 3.2 shows the cclient serviccing and
claim manaagement proocess.
Nettwork hospitals are inclluded in SS
SP after the scrutinisatioon of the raange and
quality of services, cost
c of treeatment, loccation and proximity to membeers, and
preferencess of membeers as voicedd in monthlly or annuall meetings. Project Off
fficers of
SKDRDP would
w send a requisitioon form for inclusion inn network to various hospitals.
The form should
s be submitted
s b the hospiitals detailing the totaal number of beds,
by
general warrd beds, speecial consulltation facillities, diagnnostic equipm
ment, details of the
doctors/ sppecialists, charity woork of thee hospital, and rate list (for different
d
procedures, operationn charges, investigatio
on chargess, and surggery charges). An
undertakingg by the head of the hoospital to provide cashhless treatm
ment at conccessional
agreed ratees to beneficciaries of thhe scheme is
i necessaryy to reach aan agreemen
nt. Only
after the hoospitals sennd this conssent, a Mem
morandum of
o Understaanding betw
ween the
programmee and hospiitals is possible. Every
y fortnight,, SSP assistants reportt on the
duration off stay in thee hospital and
a unnecessary invesstigations. T
The report provides
p
informationn to monitoor the perfoormance of the hospitaal by the prroject officeer of the
concerned valaya. In case, the reeimbursemeent of amouunt spent foor medicinee bought
from the outside
o pharrmacy due to its unav
vailability inn the hospiital is requiired, the
concerned hospital haas to subm
mit the billl along witth claim foorm and diischarge
summary.

67
Sampoorna Suraksha
Programme

Executive Director
Project Officers
Medical officers
Accounts Department, SS
Assistant, Field Staff
Insurance Company Staff

Show ID of SSP
Provision of health care
Submission of authorisation form by hospital through
online
Scrutinisation and send approval by online to hospital
After discharge, submit filled Form 1A with documents
Pay by RTGS
Pay premium and submit registration form
Send a part of premium for medical coverage along with
registration form
Send debit note
Submit Form C for special benefits, Form B for
reimbursement claims
Payment by cheque for special benefits and reimbursement
claims
Payment by insurance company
Benefit design package by hospitals, SSP, Insurance
companies

Figure 3.2 ClientServicing and Claim Management Process

68
viii. The role of information technology
The programme owes its success partly to information technology that has
reduced a lot of paper work and helped the maintenance of member portfolio (household
information, premium, past claim data, loan details) and preparation of cumulative
reports at the project offices. All the records are computerized that includes members
name, address, their claim records, balance amount that can be claimed, payment made
to network hospitals, and forms submitted to insurance companies. The fax system
replaced the online submission of scanned forms that reduced claim-processing time.
ix. Source of revenue
The main source of funds is the premium revenue collected from the members.
SSP does not have any financial support from the government or other aid agencies.
x. Fraud prevention and detection mechanisms
SSP has implemented various fraud identification mechanisms namely
computerized identity card, verification of medical bill, payment of premium in advance,
visits to hospitals by Sampoorna Suraksha assistants to verify the admission of members
and pre-authorisation sanction by SSP office.
xi. Recruitment of staff
SSP does not have its own staff except a few office staff and medical officers.
Recruitment of the lower level field staff (sevanirathas) takes place once in a year. The
criteria are the age between 18-25 years with minimum qualification of pre-university
certificate course. The assessment of the candidates is based on the written test and
interviews. The grades will be determined based on their performance that determines the
selection of the suitable candidate, without any gender or caste discrimination. For
specialized positions, lateral recruitments are used. Project Officers and Supervisors,
based on 21 parameters, appraise the performance of field staff every year. An in-house
training institute, ‘Centre for Rural Excellence’ provides training to overcome the
weaknesses of the staffs as reflected in the appraisal. Project Officer, based on field
visits, suitability of the planned programmes, implementation of new programmes and
problem-solving skills, evaluate the supervisors.
69
 CHAPTER 4

IMPACT OF SAMPOORNA SURAKSHA

PROGRAMME ON FINANCIAL
PROTECTION
4.1 DESCRIPTION OF SURVEYED HOUSEHOLDS
The study analysed the survey data collected from 1146 households that included
information on 4961 individuals. The sample included 416 insured households (1850
individuals), 366 newly insured households (1594 individuals) and 364 uninsured
households (1517 individuals). Socio-economic characteristics of the sample households
did not differ except in terms of religion and distance to hospital. Almost 44 percent of
the insured households had access to health care services within one kilometer of their
residence compared to 29.6 percent of newly insured and 38.5 percent of uninsured
households. Newly insured had a longer distance to travel (average of 3.3 km) compared
to insured and uninsured households. It can be inferred from Table 4.1 that insured
members live near the hospital than uninsured and newly insured households (p<0.05).
Hindus formed the majority of target population, but newly insured had a higher percent
of Muslims. Two fifths of insured households were from semi-urban areas compared to
30.6 percent of newly insured and almost 32 percent of uninsured households.
Socio-economic characteristics of the head of the sample households show
homogeneity except the marital status. Nearly 85 percent of the head of the households
was married. A comparison of widow or divorce status among the groups’ shows that
18.4 percent of uninsured head of the households were widows/widowers/divorcees
compared to the head of the households of insured (11.8 %) and newly insured (13.1%)
families. One thirds of head of the households of insured sample, almost two fifths of
newly enrolled and 42 percent of uninsured sample were in the age group of 41 to 50
years (Table 4.2). Men were the head of the 84 percent of insured households, 84.7
percent of the newly enrolled households and 79.9 percent of uninsured households.
Forty-two percent of the head of the households completed the primary education (1st
standard to 7th standard) and almost 26 percent had completed the secondary education.
The main occupation of the head of the household was unskilled labour that included
daily labour and rolling of beedi (41.4%). The head of the households in the insured
group had a lower percent of unskilled labour and formal sector employment compared to

71
uninsurred and new
wly insuredd members. The averagge annual income
i of tthe head off the
househoold was 533,995 for thhe entire sam
mple.
Annual income of the entire houssehold for innsured housseholds wass 116,850, for
newly insured
i houuseholds 1002,630 and 107,926 for
f uninsureed households. There was
no diffeerence in thhe distributioon of incom
me in terms of the incom
me quintilee among sam
mple
groups. More thann three fouurths of insured housseholds, alm
most two thhirds in neewly
insured households and four fifths of un
ninsured hoouseholds haad a familyy size of 3 to 5
(p<0.055).
An analysis of assets owned by the samplee household
ds elucidatess the econoomic
conditioons of the sample hoouseholds. Table
T 4.3 ddepicts a higher
h perceent (89.2%) of
insured householdss owning thhe house compared to newly
n insurred (84.9%)) and uninsu
ured
househoolds (83.8%
%). A large number off uninsured households had cemeented red ox
xide
flooringg (73.9%) rrelative to newly insuured (66.9%
%) and insuured (63.5%
%) househo
olds.
Newly insured had a higher percent off the mud fflooring andd insured hhouseholds had
ceramicc tiles comppared to thee other twoo groups (p=0.00). Thee sample hoouseholds used
u
brick too construct the
t walls thhan mud; hoowever, most of the inssured houseeholds had mud
m
walls thhan other tw
wo groups and
a more of the uninsuured househholds had bbrick walls than
t
mud (p
p=0.00). Cllay tiles were
w the material
m used by almoost four fiffths of sam
mple
househoolds followeed by concrrete slabs in
n a higher percent
p in innsured households. Insu
ured
househoolds had moore rooms than
t uninsuured and neewly insuredd householdds. Almost half
of insurred househoolds had 3 too 4 rooms, whereas
w 41.3 percent of
o newly inssured had 1 to 2
rooms and
a 43.5 peercent of unninsured hadd 3 to 4 roooms. Nearlyy 85 percennt of the sam
mple
househoolds owned mobile phoones followed by fan annd televisioon.

72
 Table 4.1 Socio-Econo
S omic Characcteristics off Surveyed H
Householdss
Insureed Newlyy insured U
Uninsured Test valuee
Number off householdss 4166 3
366 364
Mean per ccapita annuaal income ( ) 270224 244730 26216 1.63911
SD 191660 111810 16550
Income quiintile (%) 4.57422
Q1 < 141000 18.55 2
20.5 21
Q2 141011- 19010 20.99 2
20.5 22.3
Q3 190111- 24000 19 22 18.4
Q4 240011- 34800 21.99 2
21.3 21.3
Q5 > 348000 19.77 1
15.7 17
Mean size of
o househollds (SD) 4.5 (1.774) 4.366 (1.77) 4.2 (1.48) 9.11211
Religion(%
%) 21.343*3
Hindu 95.77 8
82.2 94
Christian 0.7 2.5 1.9
Muslim 3.6 1
15.3 4.1
Area of residence (%) 36.222
Rural areass 52.22 5
55.2 56.2
Urban areas 7.2 1
14.2 12.1
Semi-urbann areas 40.66 3
30.6 31.7
Distance too hospital (kkm) Mean 2.3 (2.1) 3.33 (2.7) 2.4 (2.3) 42.647*1
(SD)
1 2 3
Kruskal Walliis test Pearson chi sqquare test Fishers exact test
SD- Standard D
Deviation 1 US $= almosst 61 Indian ruppees ( ) in Octoober 2013
*p<0.05

7
73
 T
Table 4.2 Basic Socio-eeconomic Characteristi
C ics of Heads of the Hou
useholds
Insu
ured New
wly insured Uninsureed Test vaalue
N=4
416 N
N=366 N=364
Age gro
oup: 20 to 330 6..2 6.2 3.6 221
12.02
31 to 40
0 22
2.4 24.3 23.4
41 to 50
0 33
3.4 38.5 42
More th
han 50 38 31 31
Marital Status: Maarried 86
6.5 86.1 81.6 16.92*
Unmarrried 1..7 0.8 0
Widow//widower/d
divorcee 11.8 13.1 18.4
Educatiion: Illiteratte 22
2.5 26.3 21.4 61
9.166
Primary
y education 42
2.8 40.9 40.4
Secondary educatio
on 23
3.6 23 25
Pre-univ
versity or degree
d 10
0.1 8.8 12.3
Vocatio
onal degree 1 1 0.9
Occupaation: Unskiilled laboureer 38
8.2 43.4 43.1 531
27.25

Skilled labour 18 16.9 15.7

Self-em
mployment 10 5.4 8.5
Formal sector emp
ployment 2..9 5.2 5.8
Unemployed 18
8.4 17.9 15.6
Unskilled salaried 5..8 5.2 5.2
Skilled salaried (innformal secttor) 3..1 2.2 3.6
Agricullture 3..6 3.8 2.5
Annual Income: M
Mean (SD) ( ) 602
200 4
49867 51054 63
0.766
(812
235) (4
40454) (43889))
1 2 3
Pearson chi
c square test Fishers exacct test Kruskkal Wallis test
SD- Standdard Deviation
*p<0.05
(In percen
ntages)

74
 Table 4.3 Assets Ownership of Surveyed Households
Insured Newly insured Uninsured Chi square value
House ownership (%) 89.2 84.9 83.8 4.356
Floor material (%) 24.565*
Mud 15.1 18 12.9
Ceramic tiles 13.2 5.7 5.8
Cement - red oxide 71.7 66.9 73.9
Marble 8.2 9.3 7.4
Wall construction (%) 20.632*
Mud 21.9 15 12.9
Brick 72.8 79.8 85.2
Cement blocks 5.3 5.2 1.9
Roof material (%) 4.21
Thatched 0.5 0.3 0
Clay tiles 79.3 77 79.1
Metal tin 6 8.7 8
Concrete 14.2 14 12.9
Types of toilet used (%) 5.556
Open 7.2 9.8 10.7
Private 92.3 88.8 88.8
Public 0.5 1.4 0.5
Source of water (%) 9.657
Piped into house 28.1 26.8 25.3
Public tap 21.6 27.3 30.5
Well 41.8 36.6 34.9
River/spring 2.4 2.5 2.7
Water tank 6 6.8 6.6
Electricity connection (%) 85.1 79.5 82.4 0.122
Assets(%): Land 16.1 13.7 13.2 1.579
Radio 8.9 9.3 6.3 2.554
Television 66.1 61.7 69.2 4.58
Bicycle 9.6 8.2 8 0.801
Fan 82.2 72.1 75.3 11.786
Bike 21.2 18 17.6 1.95
Mobile phone 84.9 86.9 81 4.859
Refridgerator 13.5 8.2 11.3 5.495
Car 0.7 0.8 0.3 1.024
*p<0.05

75
76
4.2 IMPACT OF SAMPOORNA SURAKSHA PROGRAMME ON FINANCIAL
PROTECTION- ACCESS TO CARE
4.2.1 Introduction
Financial protection denotes the reduction in medical expenses incurred by
insured individuals while increasing utilisation of health services. This chapter focuses on
the impact of SSP on financial protection provided to insured members. This study
compares insured, newly insured and uninsured individuals on various dimensions of
financial protection such as i) access to care, ii) health seeking behaviour, iii) utilisation
of inpatient facilities, iv) out of pocket expenditure, v) catastrophic health expenditure
and vi) risk coping strategies to assess the impact of SSP.
Access to care is one’s ability to obtain the health services when required. Due to
financial and non-financial barriers, people would not seek care during illness.
Households in the informal economy usually do not make financial provision for illness
due to low income of the individuals, unpredictability of the timing of illness and high
cost of care. When these individuals become sick, they have to either borrow at high
interest rate or postpone seeking care. The research question was whether SSP reduces
the financial barriers to access care. Since SSP provides the financial coverage, financial
risk associated with the cost of treatment would be lower. Health insurance coverage
compensates the low income of these households such that insured can access timely
care. Pre-determination of benefit package for each disease removes the psychological
barriers related to uncertainty about medical bills. Hence, the study hypothesised that SSP
increases access to care for insured individuals compared to uninsured and newly insured
individuals. The analysis uses the data on the incidence of illness, types of treatment
taken (outpatient and inpatient) and the socio-economic characteristics of sample
households. Firstly, a comparison of the proportion of insured, newly insured and
uninsured households who had reported sickness facilitates a better understanding of
access to care. After the discussion on the incidence of illness, the analysis focused on the
impact of SSP on access to care by comparing the proportion of individuals who sought
care upon illness. Various socio-economic characteristics of households highlight
77
important variables that shape access to care and the incidence of illness (Table 4.4 and
4.5). A comparison of the frequency of visits to the health facility between insured, newly
insured and uninsured individuals explicates the differences among these groups with
regard to access to care. Barriers to access care were explored using the data from the
focus group discussions (FGD).
4.2.2 Incidence of Illness in the Sample Households
In the total sample, 272 (65.4%) insured households, 256 (69.9 %) newly insured
households and 281 (77 %) uninsured households did not report illness. Thus, insured
households had a higher incidence of illness followed by newly insured households and
uninsured households (p<0.05) (Table 4.4).
Ill persons in insured and uninsured group had an average age of 43 years, higher
than that of newly insured (37 years) group. Types of illness was not a determinant of the
access to care. Insured reported a higher percent of hospitalisation compared to newly
insured and uninsured individuals. Households residing in rural areas reported higher
illness than those in semi-urban or urban areas.

78
 Table 4.4 Demographic and Health Related Characteristics of Ill Persons in the Sample
Insured Newly insured Uninsured Test
(N=161) (N=120) (N=90) value
Mean age of ill person 43 37 43 5.4611
(in years)
Gender of ill person (%) Male 50.9 49.6 50.6 0.9742**
Female 49.1 50.4 49.4
Types of illness (%) Acute 43.4 48.7 43.5 3.5712
Chronic 54.1 45.5 51.8
Maternity 2.5 6 4.7
Types of treatment (%) Outpatient 21.1 46.1 32.8 20.6562*
(N=76)
In patient 50.1 28.8 21.1
(N=285)
No treatment 20 30 50
(N=10)
Income quintile (%) Q1 22.2 23 26.4 4.9972
Q2 24.7 19.5 24.1
Q3 22.8 21.2 19.5
Q4 14.6 23 18.4
Q5 15.5 13.3 18.4
Area of residence (%) Urban 8.9 17.7 21.8 12.0262*
Semi-urban 38.6 38.9 25.3
Rural 52.5 43.4 52.9
1
ANOVA test
2
Pearson chi square test
*<0.05; **<0.1
(In percentages)

4.2.3 Access to Health Care

An analysis on access to care considered individual cases since there were
instances of more than one family member being ill. Hence, further analysis considers
161 individuals in 144 insured households, 120 individuals in 110 newly insured
households and 90 individuals in 83 uninsured households. Only two percent of the
individuals reporting illness did not seek care. Thus, out of 371 individuals reporting
sickness, 10 individuals did not seek treatment. Among 361 individuals who sought
79
treatment, 159 belonged to insured group, 117 to newly insured and 85 to uninsured
group. Intra-group analysis of those who did not access care reveals that almost half of
the individuals were from uninsured than insured or newly insured individuals (30%).
This finding was not significant suggesting no relationship between health insurance and
access to care (p>0.05). Hence, null hypothesis that SSP does not have any impact on
access to care was accepted.
Mean age of ill persons who accessed care was 41 and of those who did not seek
care was 44 years (Table 4.5). There was no gender difference in access to care in the
sample households. Chronic illness (50.7%) motivated individuals to access health care
than acute illness (45.1%). The duration of illness determines the access to care. Average
duration of illness of care seeking individuals was 15 days (p<0.05). Most of individuals
who sought care had men as the head of the households (82.5%) whereas 30 percent of
individuals who did not seek care had female head. A majority of the individuals from
low-income quintile (Q1, Q2 and Q3) did not seek care. The care-seeking individuals
lived away from the hospitals (median 2 km) and most of rural residents did not seek
care.
Higher number of visits (more than or equal to 2) was made by newly insured
(31.5%) than insured (26%) and uninsured (17.6%) individuals. Mann Whitney U test
suggests no difference between newly insured and insured group (p>0.05). However, the
difference was statistically significant between uninsured and newly insured individuals
(p<0.05) with majority of uninsured making one visit, rarely two or more visits.

80
 Table 4.5 Socio-economic Characteristics and Access to Care

Access to care Test value

No Yes
(N=10) (N=361)
Health insurance status (%)
Insured 20 44 4.1211
Fisher’s p =.156
Newly insured 30 32.5
Uninsured 50 23.5
Gender of ill person; Male (%) 50 50.4 0.1111
Age of ill person (in years) 44 41 0.7062
Types of illness (%)
Acute 70 45.1 2.55811
Fisher’s p =.335
Chronic 30 50.7
Maternity 0 4.2
Duration of illness (in days) 6 15 0.0132*
Gender of head of household; Male (%) 70 82.5 1.0481
Education of head of household 5 5 0.7082
(in years)
Income quintile (%)
Q1 30 23
Q2 30 22.7
Q3 0 21.6
Q4 30 17.7
Q5 10 15
Distance of hospital (in km) Mean (SD) 1.6 (3.8) 2.6 (2.2) 0.1642
Area of residence (%) 2.4931
Urban 0 15
Semi-urban 30 36.3
Rural 70 48.7
1
Pearson chi square test
2
Mann Whitney U test p value
* p <0.05

Further analysis to know the financing strategies adopted by the households

showed that 57.2 percent of insured, 79.5 percent of newly insured and 75.2 percent of
uninsured individuals borrowed to access care (p<0.05). Another strategy adopted by

81
insured, newly insured and uninsured individuals was to use the savings. Almost 32.7
percent of insured individuals, 24.7 percent of newly insured individuals and 35.3 percent
of uninsured individuals used the savings to access care. When faced with illness,
individuals sought care even if they did not have money to pay. They borrowed from
informal sources such as friends, relatives or neighbours or from formal sources such as
non-banking financial companies or MFIs.
4.2.4 Barriers to Access Care
Qualitative data gathered from FGDs with insured, newly insured and uninsured
respondents revealed that several factors resulted in not seeking care when sick. The
financial and non-financial factors are the broad categories of factors that act as barriers
to access care.
Among non-financial factors, poor quality of care at the hospitals and distance to
the hospitals were important,
“….bed for men and women are kept together…. there is no privacy..the ward is
not clean…”, “..good hospitals are in Kumta (a city in UK district) which is far away…”,
“..doctors do not see us well, we have to go to Hubli or Manipal for good hospitals (far
away city)..”.
Financial factors namely lack of money, high cost of health services and indirect
cost of care inhibit access to care,
“...we have too much loan to repay… We do not have money to pay….”, “…we do
not have much income….taxi is expensive...”, “..hospital bills have gone up… its
expensive....”. “..we have six people in the family but father only earns… we do not have
health card…....”. The other reasons were, “…going to hospital means you have to take
leave….one day’s earning will be lost…”, “…hospitals are too far…one day’s income
will be lost…we do not have money..we have too many loans already…..”.
Thus, lack of money to pay for health services, high indirect cost of care and
expensive health services were financial reasons highlighted by respondents. Poor quality
of care, long distance to hospital, lack of transport, and difficulty in absenting from work
were non-financial barriers.
82
4.2.5 Summary
One striking finding was the higher incidence of illness among insured
individuals compared to uninsured and newly insured individuals. This indicates the
hidden motive of insured individuals to claim from SSP. The hypothesis driven analysis
carried out in this section found no evidence to support the positive impact of SSP on
access to care. In the absence of health insurance, uninsured and newly insured
individuals borrowed or used savings to access to care. One more explanation for the
absence of the positive impact of SSP on access care lies in the ‘Jnana Vikasa’
programme. SKDRDP conducts educational programme to impart the basic knowledge
on various relevant issues including health education to its target population. Consequent
higher level of awareness on the timely access to care among SHG members removed the
non-financial barriers to access care. While there was no impact of SSP on access to care,
frequency of visits was quite different for insured, uninsured and newly insured
individuals. Insured and newly insured individuals had more number of visits than
uninsured individuals. Some individuals who did not seek treatment stressed lack of good
hospitals, high cost of treatment, lack of money, long distance to good hospitals and poor
transportation facilities, especially in rural areas as the barriers to access care. Of the
individuals who did not seek care, majority belonged to uninsured group. Thus, the study
accepts the null hypothesis that SSP does not improve the access to health services. The
question arises whether there is any difference in the treatment-seeking pattern of
individuals. Hence, the next section compares the health seeking experience of insured,
newly insured and uninsured individuals who accessed care during illness.

83
84
4.3 IMPACT OF SAMPOORNA SURAKSHA PROGRAMME ON HEALTH
SEEKING BEHAVIOUR
4.3.1 Introduction
Health seeking behaviour (HSB) denotes visiting health care facility such as
privately owned hospitals, public hospitals, private clinics, ayurvedic hospitals, nursing
homes or home medicine. Desirable HSB is the visit to formally recognised health care
facilities than self-care, traditional healers and unofficial medical channels. Formal health
care facilities include the hospitals (private and public) and nursing homes, clinics and
alternative systems of care such as ayurveda, homeopathy, and Unani. However, public
hospitals are known for low quality, lack of accountability and poor infrastructure
(Radwan L 2005; Mathiyazhgan 2006; Chuma et al. 2007; Klein 2011), prolonged
waiting period, long distance, inconvenient location and inadequate facilities (Patel et al.
2010) in India. Hence, the present study assumes that public hospitals provide low quality
of care and people seek care at private facilities than public hospitals.
Since SSP contracts with private hospitals, insured can get better services at an
agreed price. In addition, SSP brings down the financial barriers to access formal care.
Hence, SSP insured members would seek inpatient care from private facilities due to the
accessibility (large network of hospitals), acceptability (quality of care) and affordability
of care (claim benefits). Hence, insured individuals would be motivated to seek care from
the private facility than public hospitals or other types of treatment. Therefore, the
hypothesis was that insured members seek care from the private sector providers than
other providers (including clinic, public or informal care) compared to newly insured and
uninsured individuals.
This section analyses the findings of the study with an aim to understand the
impact of SSP on HSB. Firstly, to assess the impact on the pattern of HSB in the first and
second visits, treatment taken in different facilities was analysed. Home medicine, private
clinics, ayurvedic hospitals, government hospitals, private hospitals and nursing home
were the health care facilities visited by the sample individuals. Private hospital was
classified as district hospital (<100 beds) and regional hospital (>100 beds). Secondly,
85
study hypothesis was tested using discriminant analysis. Lastly, determinants of
hospitalisation in public or private hospitals were estimated using binary logistic
regression analysis using the following regression equation,
Prob (hospitalisation in private facility >0│ill) = β0+β1Mx +β2Xy+ ε, where Xy are
the variables that influence probability of hospitalisation in private hospitals; Mx
represents the mode of payment (SSP).The binary logistic regression model underwent a
number of specification and diagnostic tests, especially the Durbin-Wu-Hausman method
as explained in the following paragraph.
Probability of enrolment was estimated using a logistic regression model that
considered SSP status as a dependent variable and various instrumental variables as
independent variable to get the residuals of SSP health insurance variable.
Prob (Membership>0) = β0+β1Xβ +ε
Residual of SSP variable was included as an independent variable along with
other independent variables in HSB logistic regression.
Prob (Private >0│hospitalised) = β1Mx +β2Xy+ β3HI_res+ε, where Xy is a set of
variables that influence the probability of seeking care at the private facility; Mx
represents health insurance (HI). If β3 is significantly different from zero, then regression
is not consistent, making the coefficient of the health insurance biased (endogenous).
Accepting the null hypothesis (β3=0) suggests exogeneity of the health insurance in the
model (Ekman 2007a; Jutting 2003).
4.3.2 HSB by Insurance Status
In the survey, 371 individuals reported sickness of which 10 persons did not seek
treatment. Of 361 persons who sought treatment, 19 resorted to self-treatment (in the
first visit) and the remaining 342 individuals availed health care services with one or
more visits resulting in 429 visits (Figure 4.1). There were 384 visits made to the private
health service providers, 37 public services and 8 visits to traditional services (ayurvedic
medicine). Insured made 173 visits to private providers and 11 visits to public facilities.
Thus, the highest number of visits was made to private providers of health care compared
to public hospitals or ayurvedic treatment.
86
 Incid
dence of illness
n=371; n1=161,n2=120,n33=90

No treatment
Treaatment taken
n=10; n1=2,,,n2=3,n3=5
n=361,n1=159,n2=117,n3=85

Meedical services Self-tretameent

n=429* n=19;n1=3,n2=77,n3=9

Privatte providers of servicees Public services Traditiona

al services
n=384;; n1=173,n2=127,n3=884 n=37;n1=11,n
n2=13,n3=13 n=8; n1=2,n
n2=3,n3=3

Consultattion/clinic Nursing hoome District hospitaals Regional hospitals

n=75;n1=23,nn2=32,n3=20 n=25; n1=11,n2=
=11,n3=3 n=147;n1=72,n2=41,,n3=34 n=1137; n1=67,n2=43, n3==27

F
Figure 4.1 Overview
O of health carre seeking behaviour off individuals
*Number of visits to health facility
f by ill persons;
p n=totaal; n1=insured;; n2=newly inssured; n3=uninnsured

A ccomparison of insured, newly insu

ured and un
ninsured reggarding the HSB in thee
first and seecond visit (Table 4.6 and Table 4.7) show
ws strong evvidence of the
t positivee
SSP. In thee first visit, large proportion of in
impact of S nsured indivviduals visiited nursing
g
homes, disttrict and reg
gional hosp
pitals. Uninssured indiviiduals reliedd on self-treeatment and
d
governmen mes, districct hospitals and regionaal hospitals.
nt hospitals rather than nursing hom
When the total visits were
w analyseed, a large percent
p of uninsured
u in
ndividuals compared
c to
o
insured and
d newly in
nsured soug
ght ayurved
dic treatmen
nt (Figure 44.2). Least number off
uninsured iindividuals visited nurssing homes, which were usually eexpensive compared
c to
o
the district or regionaal hospitals.. Thus, therre was a significant diifference in
n HSB with
h
insured acccessing caree at the priv
vate hospitalls/ nursing homes thann uninsured individualss
(p<0.05).

8
87
 Table 4.6 Health Seeking Behaviour in the First Visit: Comparison by Insurance Status
Insured Newly insured Uninsured
Home medicine (N=19) 21.1 26.3 52.6
Clinic (N=64) 34.4 42.2 23.4
Nursing home (N=18) 55.6 38.9 5.5
Government hospital (N=31) 29 35.5 35.5
District hospital (N=118) 51.6 29.7 18.7
Regional hospital (N=108) 49.1 29.6 21.3
Ayurvedic hospital (N=5) 20 40 40
χ2 (12, N=361) =21.705, p=0.041
(Figures are percentages to total of each row)

Analysis of the data on the second visit to health facilities revealed a non-
significant difference (p>0.05) in HSB of the surveyed individuals. Uninsured and newly
insured more than insured individuals used clinic and government hospitals. A higher
percent of insured individuals went to district and regional hospital compared to
uninsured individuals (Table 4.7).
Table 4.7 Health Seeking Behaviour in the Second Visit: Comparison by Insurance Status
Insured Newly insured Uninsured
(N=27) (N=26) (N=12)
Clinic 3.7 15.4 25
Government hospital 7.5 7.8 16.6
District hospital 44.4 19.2 33.3
Regional hospital 44.4 42.4 16.7
Nursing home 0 11.5 0
Ayurveda hospital 0 3.8 8.4
(Figures are percentages to total of each column)

When the total visits were considered (Figure 4.2), a higher proportion of insured
individuals were observed to visit the district hospitals (38.1%) and the regional hospitals
88
(35.5%) thaan newly in
nsured (27.5
5% and 28.9 % respectively) and uninsured (31.2% and
d
24.8 % resspectively) individuals
i (p<0.05). Nearly
N 9 peercent of neewly insureed and 11.9
9
percent of uninsured compared to
t only 5.8 percent off insured inndividuals selected
s thee
nt hospitals (including
governmen ( p
primary heaalth centres)).
100%
90% 2.7
18.3 31.2 24.8
80% 11.9 2.8
70% 8.3
60% 6.7
27.5 28.9
50%
40% 21.5
8.7 2
30% Uninsured
4.7
20% 5.8 38.1 35.5 Newly insuured
10% 12.1 5.8 1.1
1.6 Insured
0%

Figure 4.2 haviour of Surveyed Inddividuals

4 Health Seeking Beh

4.3.3 Sequeence of heaalth seekingg behaviourr

Nott only the paattern of HS
SB was diffferent but allso the sequuence of visits to health
h
facilities inn the first and
a subsequ w uniquee (Figure 4.3). Insured
uent visits was d and newly
y
insured whho self-treaated on illn
ness visited
d private hospitals
h inn their seco
ond line off
treatment w
whereas unin
nsured conssulted doctoors at the cllinic, visitedd governmeent hospitalss
and districtt hospitals. A higher nu
umber of in
nsured sough
ht treatmennt in the disttrict and thee
regional ho
ospitals in their
t first visit
v comparred to unin
nsured indivviduals. Uniinsured and
d
insured wh
ho consulted
d doctors at the clinic and
a nursing
g home first time sough
ht care from
m
the district// regional hospitals
h an
nd the nursinng homes in
i the seconnd or third time. Sincee
only few inndividuals so yurvedic hospital is nott analysed (Figure 4.3).
ought the trreatment, ay
The govern he first choice for unin
nment hospiitals were th nsured durinng illness. Analysis off
the second visit reveaals that insu
ured individ
duals madee 27 visits, newly insu
ured had 26
6
visits and uuninsured haad 12 visits..

8
89
 Clinic D
District Hospital Governmment
Home Meedicine Nursing Home
H Regional Hosppital
Hospital

First visit
Firstt visit First visit First visit First visiit Firsst visit
n1=22
n1=44 n1=10 n1=61 n1=53 n1==9
n2=27
n2=55 n2=77 n2=35 n2=32 n2==11
n3=15
n3=110 n3=1 n3=22 n3=23 n3==11
Second visit
Secoond visit Seccond visit
=6
Insured; T3= Second visit Second vvisit
Insuured; Second visit Insuured;
T4=4,T5=1 Insured; T4=55
T4=1,T5=1 Insurred; Insured; T3=2 T3==1
Newly insureed; T1=11,T3=3,T4=2 Newly
Newwly insured; T2=1,T3=2,T T4=5, Uninsureed; Newwly
Newly insured; insured; insu
ured;
T1=2,T3=2,T4=1 T1=1
T5=2 T3=11,T4=1 T1=1,T4=4 T1==1,T4=1,
Uninnsured; Uninsured;
Uninsured; T2==2
T1=2,T3=2,T5=2 T4=1,T5=1
T3=2,T4=1

Figuree 4.3 Sequen

nce of Heallth Seeking Behaviour during Illneess
n1=Insureed individuals, n2=Newly
n insuured individuals, n3=Uninsuredd individuals

T1=Clinicc, T2=Nursing home,

h T3=Distrrict hospital, T44=Regional hosspital, T5=Goveernment hospitaal

4.3.4 Discriminan
nt Analysis of Health Seeking
S Behaviour off Insured an
nd Uninsurred
The study hypothesise
h es that insurred use priv
vate facilitiees (nursing homes, disstrict
and reg a uninsured individuuals. To test this
gional hospittals) compaared to newlly insured and
hypotheesis, discrim
minant anaalysis was carried out (Table4
4.8). Firstlyy, insured and
uninsurred (newly insured and
d uninsured
d) individuaals were tak
ken as groupp variables and
types of
o treatmen
nt such as home meedicine, clinics, goverrnment hosspitals, disstrict
hospitalls, regionaal hospitalss, nursing homes and
d ayurvediic hospitals as prediictor
variablees. The disscriminant function revealed
r a significant associationn between the
groups and all th
he predicto
ors with Wilk’s
W 08. Visits to the clin
lambbda p=0.00 nics,
ment hospiitals and home
governm h mediccine rather than accessing care at the disstrict
hospitalls and regioonal hospitaals differentiiated uninsu
ured from in
nsured indivviduals. Insu
ured
had a higher
h visitss to the priv
vate provideers than pubblic provideers or clinicc as they co
ould
get claim benefits from the neetwork hosp
pitals at thee district or regional leevel. The crross-
validateed classificaation shows correct classification of
o overall 55.8 percent of cases. Thus,
T
90
there was positive impact of SSP in health seeking behaviour of insured who sought care
at private hospitals than at government hospitals, clinics or self-treatment.
Secondly, structure matrix obtained by including insured and newly insured as a
grouping variable and types of health care as predictor variables shows interesting results
(Table 4.8). Visits to clinics, government hospitals, home medicine, and not accessing
care at district hospitals differentiated newly insured from insured individuals. Hence, the
study hypothesis that insured sought care at network hospitals than at government
hospitals, clinics or home medicine compared to newly insured individuals was accepted.
The cross-validated classification shows a high percent of cases being correctly classified
(59.8 %).
Thirdly, an analysis taking insured and uninsured as a grouping variable and the
types of health care as predictor variables corroborates earlier findings (Table 4.8). Visits
to regional hospitals rather than government hospitals, home medicine and ayurvedic
hospitals differentiated insured from the uninsured individuals. Insured had higher visits
to district and regional hospitals than public providers, ayurvedic or home medicine.
Hence, we reinforce the positive impact of SSP on health seeking behaviour of insured
individuals that resulted in more visits to private hospitals than informal care such as
home medicine or public hospitals. The cross-validated classification illustrates that 63.9
percent cases were correct classified.

91
 Table 4.8 Discriminant Analysis of Health Seeking Behaviour
Insured and uninsured Home medicine (.566), Government hospital (.399),
(both newly and uninsured) Private clinic (.565), District hospitals (-.397),
individuals1 Regional hospitals (-.370), Nursing homes (.239)
and Ayurvedic hospital -(.014)
Insured and newly insured Home medicine (.414), Government hospital (.321),
individuals2 Private clinic (.703), District hospitals (-.462),
Regional hospitals (-.237), Nursing homes (.197)
and Ayurvedic hospital (.211)
Insured and uninsured Home medicine (-.614), Government hospital
individuals3 (-.528), Private clinic (-.222), District hospitals
(.175), Regional hospitals (.446), Nursing homes
(.272) and Ayurvedic hospital (.034)
1
Box’s M 132.879, F 4.642, p=0.000; Wilk’s lambda p=0.008
2
Box’s M 92.260, F 3.203, p=0.000; Wilk’s lambda p=0.049
3
Box’s M 152.765, F 5.265, P=0.000; Wilk’s lambda p=0.034

4.3.5 Discriminant Analysis of Underlying Reasons for Health Seeking Behaviour

Quality of treatment was the main reason for majority of insured (40.4%) to
access care whereas trust in treatment was important reason for uninsured (33.3%) and
newly insured (30%) individuals (Table 4.9). Accessibility was the least motivator for the
selection of health facilities.

92
 Table 4.9 Health Seeking Behaviour– Reasons Given by the Surveyed Individuals
Insured Newly insured Uninsured
(N=159) (N=117) (N=85)
Accessibility 5.1 2.5 1.1
Lack of improvement 5.1 8.3 2.2
Lack of money 9.5 9.2 12.2
Quality of treatment 40.4 25 26.7
Low cost of treatment 13 13.3 7.8
Trust in treatment 24.2 30 33.3
Near to home 23 18.3 18.9
Severity of illness 8.7 7.5 16.7
Nature of illness 6 6.7 6.7
(Figures are percentages to total of each subgroup given in column)

Discriminant analysis helped to understand the important reasons for selecting

particular hospitals that could differentiate insured from uninsured and newly insured
individuals. Insured and uninsured (including newly insured) individuals were taken as
group variables. No improvement with treatment, lack of money, quality of treatment,
low cost, trust in treatment and nearness to home, severity and nature of illness were the
predictor variables. The discriminant function revealed a significant association between
the groups and all predictors with Wilk’s lambda p=0.02 (Box’s M 69.020; F 2.414,
p=0.00). Quality of treatment (.730) was the main factor that differentiated insured from
newly insured and uninsured individuals, followed by lack of money (-.591) and trust in
the treatment (-.368) provided by the health facility. This suggested a label of good
quality of care, affordability and low level of trust in the health facility used by insured
individuals. Near to home (.251), low cost of treatment (.148), severity of illness (-.097)
and no improvement from the previous treatment (-.012) were not loaded on the
discriminant function (59.8 percent of cases were correctly classified).

93
 The discriminant analysis focused on the factors that differentiate the selection of
hospitals by the surveyed individuals, regardless of the health insurance (Table 4.10).
Predictor variables were accessibility, no improvement with treatment, lack of money,
quality of treatment, low cost of treatment, trust in treatment, nearness to home, severity
of illness, and the nature of illness. The aim was to investigate the factors that
differentiate the selection of the private clinic, government hospitals, district hospitals,
and regional hospitals to identify which reason contributed more to group separation.

Table 4.10 Discriminant Analysis of the Factors Determining the Choice of Health
Facility
Private clinic1 Near to home (0.589), No improvement (0.556)
Government hospital2 Lack of money (0.812), Low cost of treatment (0.426)
District hospital3 Lack of money (-0.581), Severity of illness (0.478), Quality
of treatment (0.408), Always available (0.335)
Regional hospital4 Referred to a specialist (0.680), Lack of money
(-0.426), Nature of illness (0.426) Trust in treatment (-0.390)
1
Box’s M=184.961, F=2.648, p=0.000; Wilk’s lambda p=0.000
2
Box’s M=165.463, F=3.341, p=0.00; Wilk’s lambda p=0.000
3
Box’s M=222.102, F=3.254, p=0.000; Wilk’s lambda p=0.016
4
Box’s M=113.187, F=1.656, p=0.000; Wilk’s lambda p=0.02
i) Visit to private clinic was a dependent variable (yes or no). Structure matrix
correlations revealed near to home and no improvement to have the highest loadings,
which suggest a label of nearness and ineffective previous treatment as the function that
discriminate those visiting clinic and those who do not. The cross-validated classification
shows correct classification of 79.2 percent of the cases.
ii) Similarly, seeking care at the government hospitals was dependent variable
(yes or no), predictor variables remaining the same. Discriminate function revealed lack
of money (0.812) and low cost of treatment (0.426) to have the highest loadings which
suggest a label of low income and low price of the health services as the function that
discriminates those who visited government hospitals and those who did not (Table 4.10).

94
The cross-validated classification shows that 85.2 percent the cases were correctly
classified.
iii) District hospitals were used for severity of illness, quality of treatment,
accessibility and affordability (lack of money had negative loading). The results suggest
correct classification of overall 65.2 percent of the cases.
iv) Visit to regional hospitals was taken as dependent variable (yes or no).
Reference to a specialist, affordability (lack of money had negative loading), nature of
illness and trust in treatment (negative loading) had the highest loadings which suggest a
label of referrals, affordability, types of illness and low level of trust in the treatment of
hospitals as the functions that discriminate those who visited the regional hospitals and
those who did not. The cross-validated classification results suggest correct classification
of overall 66.6 percent of the cases.
To sum up, nearness to home and lack of improvement from the previous
treatment resulted in visits to private clinics and lack of money and low cost of treatment
were reasons for visits to government hospitals. Affordability of treatment, severity of
illness, good quality of the treatment and availability of services (for 24 hours in 7 days)
decided the treatment at district hospitals (network hospitals). Reference by the doctors,
high cost of care and nature of illness were the reasons for visits to the regional hospitals.
However, individuals who were treated in regional hospitals had low level of trust in the
treatment.
4.3.6 Econometric Estimation of HSB
The probability of hospitalisation in private facilities by insured, newly insured
and uninsured individuals was estimated using logistic regression analysis. To analyse the
impact of SSP on HSB towards the private hospitals, the study used binary logistic
regression analysis with SSP individuals coded as ‘1’ and newly insured individuals
coded ‘2’ and uninsured ‘3’. Individuals admitted in private hospitals were assigned a
code of ‘1’ and those in public hospitals had a code of ‘0’. Private health facilities
included the admission in nursing homes, district and regional hospitals.

95
 According to Birch’s model (2007), availability, acceptability and affordability
are the three A’s that determine access to health care and choice of the types of care.
Income class of the individuals measured affordability and the area of residence
determined availability of health facilities. Gender of ill person acted as a proxy to
measure the acceptability of care. Since the effect of a change in a variable depends on
the values of all variables in the model in logistic models, these variables were included
in the analysis in addition to SSP membership. The role of income class, gender of ill
person and the area of residence on the decision to get admitted in the private or public
facilities was assessed after classifying the individuals based on SSP membership status
for their income class, gender of ill person and the area of residence.
4.3.6.1 Income Related Equity in Health Seeking Behaviour
Income of the family influences the care sought by the sick individuals in
different health facilities. Usually, better-off people access private expensive care
whereas poor choose public facilities. There was a positive relationship between the
income quintile and HSB of insured individuals (p<0.05) (Table 4.11). A majority of
insured Q5 individuals sought care from the private hospitals. Newly insured individuals
from Q3, Q4 and Q5 and uninsured individuals (except Q5) accessed government
hospitals in higher proportion compared to insured individuals in the respective income
classes.
Table 4.11 Income Related Equity in Health Seeking Behaviour
Q1 Q2 Q3 Q4 Q5
Insured1 Government 19.4 2.8 9.7 4.8 0
Private 80.6 97.2 90.3 95.2 100
Newly insured2 Government 10 5 23.5 16 13.3
Private 90 95 76.5 84 86.7
Uninsured3 Government 12.5 15.8 30.8 21.4 0
Private 87.5 84.2 69.2 78.6 100
1 2
χ (4, N=159) =12.299, p=0.015 2χ2 (4, N=117) =3.064, p=0.547 3 2
χ (4, N=85) =4.794, p=0.309
(Figures represent percentages to the total of each income quintile in insured, newly insured and uninsured groups)

96
4.3.6.2 Gender Related Equity in Health Seeking Behaviour
Women usually seek care from the informal providers than the formal health care
system (Ahmed, 2003). However, participation in the micro-finance activities increases
the gender equity in treatment seeking. At a glance, it appears that higher percent of
insured women sought care from the private hospitals (94.4%) compared to newly
insured and uninsured women (Table 4.12). However, there was no statistical difference
between insured, newly insured and uninsured groups (p>0.05). Nevertheless, newly
insured women were admitted in the government hospitals more than private ones
compared to men of the same group (p<0.05).
Table 4.12 Gender of Ill Person and HSB in Private and Public Hospitals
Male Female
Insured1 Government facility 9.5 5.6
Private facility 90.5 94.4
Newly insured2 Government facility 6 21.3
Private facility 94 78.7
Uninsured3 Government facility 17.5 14.7
Private facility 82.5 85.3
1 2
χ (1, N=159) =0.756, p=0.384
2 2
χ (1, N=117) =4.872, p=0.027
3 2
χ (1, N=85) =0.106, p=0.745
(Figures represent percentages to the total of male and female sub-group in sample groups)

4.3.6.3 Area Related Equity in Health Seeking Behaviour

HSB depends on the types of the health facility near the residence. Table 4.13
illustrates the HSB of insured, newly insured and uninsured residing in urban, semi-urban
and rural areas. Insured and uninsured urban individuals visited government hospitals
(15.4%) than residents in semi-urban (7%) and rural areas (6.7%). Urban individuals
relied on public hospitals due to greater access compared to rural counterparts. However,
a significant relationship between HSB and the area of residence for the studied groups
was ruled out (p>0.05).

97
 Table 4.13 Health Seeking Behaviour in the First Episode of Illness
Urban Semi-urban Rural
Insured Private facility 84.6 93 93.3
Government facility 15.4 7 6.7
Newly insured Private facility 94.1 89.7 80.5
Government facility 5.9 10.3 19.5
Uninsured Private facility 75 89.5 84.6
Government facility 25 10.5 15.4
1 2 2 2
χ (2, N=159) =1.244, p=0.384 χ (2, N=117) =2.480, p=0.289
3 2
χ (2, N=85) =1.381, p=0.501
(Figures represent percentages to the total of each sub-group)

4.3.6.4 Results of Econometric Estimation

Binary logistic regression analysis was applied to estimate the probability of
hospitalisation in private facilities conditional on being ill (model 1). The hypothesis was
that SSP insured individuals seek care from the private hospitals than public hospitals.
Admission of insured individuals in public hospitals was lower than (6.9%) newly
insured (13.4%) and uninsured (16.2%) individuals. Sizeable percent of insured got care
from the private hospitals (93.1%) compared to newly insured (86.6%) and uninsured
(83.8%) individuals. Without considering the health insurance status, it was found that a
higher (12.5%) proportion of women sought care from public hospitals compared to men
(9.8%)(p>0.05). Similarly, analysis of the income quintile irrespective of health insurance
status revealed that almost 15 percent of individuals from Q1, 6.7 percent from Q2, 18
percent from Q3 visited public hospitals and almost 97 percent from Q4 and 89 percent
from Q5 sought care from private hospitals (p<0.1). Irrespective of the health insurance,
84.8 percent of urban individuals, 91.3 percent of semi-urban individuals and 88.4
percent of rural individuals were hospitalised in private facilities (p>0.05). Income
classes were coded into five dummy variables. SSP membership and area of residence
was coded into three dummy variables (Table 4.14).

98
 Table 4.14 Definition and Measurement of Variables
Variables Description
Health insurance 1= SSP insured (reference)
2= Newly insured
3= Uninsured
Gender of ill person 1 = Male, 0 =Female (reference)
Income quintile 1=Q1. 2=Q2, 3=Q3, 4=Q4, 5=Q5 (reference)
Area of residence 1= Urban if individual lives in urban area
2= Semi-urban if individual lives in semi-urban area
3=Rural if individual lives in rural areas (reference)

The estimated result on the relationship between hospitalisation in private

facilities and other independent variables is given in the Table 4.15. Evidence of insured
individuals being more likely to get hospitalised in private facilities than public facilities
compared to uninsured and newly insured individuals was found. The Odds Ratio (OR)
for newly insured and uninsured individuals was significantly less than 1, which implied
that newly insured and uninsured individuals were less likely to get admitted in private
facilities. The results indicate that newly insured was almost 0.4 times less likely to get
hospitalised in the private hospitals than public hospitals and uninsured was 0.373 times
less likely to get hospitalised in private facilities compared to insured individuals. The
odds of being admitted in private hospitals were higher (OR 4.676) if the individuals
were in Q5 than Q1. Thus, the results from the model indicate that HSB behave
according to theoretical expectations, especially income and health insurance. As the cost
of care at the private hospitals is high, better-off individuals have higher likelihood to
visit these facilities. Since SSP reduces the cost of care, insured are expected to visit the
private facilities. The model was checked for robustness by using the omnibus test of
model coefficients, Hosmer and Lemeshow test, -2 log likelihood ratios, Cox and Snell R
square and Nagelkerke R square. The results of these tests show that the model is
significant at the 0.05 leveland 88.9 percent of cases were correctly predicted by the
model. The model was checked for the possible endogeneity using Durbin-Wu-Hausman

99
 test. In this model, health insurance was found to be exogenous with prob (χ2) =0.994.
Residual analysis (specifically Cook’s Distance statistic) showed no outliers and Hosmer
and Lemeshow test value of 0.850 indicates good discrimination.
Table 4.15 Probability of Hospitalisation in Private Facilities (Model 1)

B S..E. Wald Df Sig. Exp(B) 95% C.I.

Variables Lower Upper
Health insurance 5.272 2 .072
(base= SSP insured)
Newly insured -.899 .469 3.672 1 .055 .407 .162 1.021
Uninsured -.985 .474 4.324 1 .038 .373 .148 .945
Area of residence .885 2 .643
(base=rural)
Urban .169 .511 .109 1 .741 1.184 .435 3.222
Semi-urban .408 .434 .884 1 .347 1.504 .642 3.525
Income quintile (base=Q1) 7.538 4 .110
Q2 .956 .588 2.643 1 .104 2.602 .822 8.239
Q3 -.202 .498 .165 1 .685 .817 .308 2.167
Q4 .383 .540 .501 1 .479 1.466 .508 4.229
Q5 1.542 .810 3.627 1 .057 4.676 .956 22.872
Gender of ill person .258 .377 .468 1 .494 1.294 .618 2.709
(base=female)
Constant 2.205 .547 16.235 1 .000 9.069
Number of observations: 285
Omnibus test model coefficient:Pearson chi square =18.56,df =10, p=0.046; -2 log likelihood = 201.982;
Cox and Snell R squared= 0.055; Nagelkerke R squared= 0.111
Hosmer and Lemeshow Pearson chi square =4.076, df =8, p=0.850
(Dependent variable: Hospitalisation in private facility; 1=yes)

4.3.7 Summary
Using the logistic regression analysis and discriminant analysis, the study
assessed the influence of SSP on HSB of insured, newly insured and uninsured
individuals. For insured individuals, health care was easily accessible due to the large
network of hospitals, affordable due to the claim from SSP and acceptable because of the

100
contract based purchasing that stipulated good quality of care. So, insured were expected
to choose private network hospitals rather than self- treatment or public hospitals.
Turning now to the evidence from the analysis of the data, we draw the following
conclusions. Insured individuals took treatment from the private hospitals or nursing
homes rather than home treatment, public hospitals or alternative treatment. Uninsured
and newly insured individuals, on the other hand, relied more on self-treatment, public
hospitals or private clinics. Logistic regression and discriminant analysis supported the
hypothesis that insured individuals sought care at private facilities than public hospitals
(H2). Income class was positively associated with the private care seeking behaviour.
There are several possible explanations to this result. Accessibility to the network
hospitals is one of the reasons to seek care at the district hospitals by insured individuals.
Moreover, financial barriers to access care reduced due to the insurance claims making
the private hospital care affordable. Unless, the private care is affordable, people seek
care from the government hospitals, self-treat or forego care. Acceptability of the quality
of care at the network hospitals (district and regional) was another factor that influenced
hospitalisation in private facilities. In contrast, uninsured and newly insured individuals
relied on the private clinics, home medicine and the government hospitals due to lack of
money and low cost of treatment. Thus, insured members for affordability (made possible
by SSP), acceptability (quality of treatment) and accessibility (always available) chose
SSP hospitals. Hence, SSP enabled insured individuals to overcome the financial barriers
to seek care at the private facilities.
When the choice had to be made among the different types of care, insured
individuals chose private hospitals and nursing homes compared to uninsured and newly
insured individuals in their first and later visits. However, newly insured and uninsured
individuals sought treatment in the private hospitals than treatment in public hospitals or
self-treatment in the second visits.
Equity in HSB was assessed based on the income class of the ill person. As SSP
provides financial coverage, insured individuals could afford expensive yet good quality
treatment in private hospitals. High-income people sought care at the district and regional
101
hospitals than self-treatment and treatment at public hospitals. Insured individuals from
low-income class, except Q1, sought care at private facilities compared to same class of
people in uninsured and newly insured group. Thus, SSP improved equity in access to
care as lower income (Q2 and Q3) individuals could seek timely and efficient treatment
in the private hospitals.
One disturbing finding was that large number of insured individuals in the lowest
income class (Q1) went to public hospitals rather than private hospitals compared to
those in uninsured and newly insured groups. Since SSP has not increased the benefit
amount over the years even when the cost of treatment escalated in India, poor had to
spend out of the pocket despite insurance. Further analysis revealed that poorest (in Q1)
among uninsured and newly insured individuals sought care from the private hospitals
and financed it through borrowing unlike insured individuals who borrowed less and
relied on public hospitals. Overall, the removal of financial barriers to access formal care
and good quality of care resulted in insured individuals visiting private providers rather
than self-treatment or treatment from informal providers and public hospitals. The next
issue related to the impact of SSP is utilisation of the health services by insured
individuals. Hence, the next section delves into utilisation impact of SSP.

102
4.4 IMPACT OF SAMPOORNA SURAKSHA PROGRAMME ON UTILISATION
OF HEALTH SERVICES
4.4.1 Introduction
Healthcare utilisation means the use of health care facilities. In this study,
utilisation is the admission in the hospitals including nursing homes since SSP provides
insurance coverage only for hospitalisation. As SSP covers hospitalisation expenses,
insured need not incur higher expenses compared to uninsured and newly insured
individuals. The research question was whether SSP improved hospitalisation of insured
individuals compared to uninsured and newly insured individuals.
Health insurance reduces the cost of treatment that drives up demand for health
services. There will be higher utilisation if SSP reduces the inpatient treatment cost.
Moreover, higher utilisation is possible due to better quality of care at the network
hospitals. SSP provides insurance coverage for inpatient treatment at 110 hospitals in 13
districts known for their quality of services, location and proximity to members. Hence,
the hypothesis was that SSP increases hospitalisation of insured members compared to
uninsured and newly insured individuals.
Framework given by Aday and Anderson (1974) helps to understand the
determinants of utilisation of the health services. These include predisposing factors
(gender of ill person, education and job status of the head and size of the household),
enabling factors (income, area of residence and health insurance) and need factors (types
of illness). This section evaluates the findings of the study on the impact of SSP on
utilisation using the data of the individuals who sought care during illness. The analysis
begins with the identification of the pattern of utilisation of insured, newly insured and
uninsured individuals. Later analysis focuses on the effect of the treatment as perceived
by the individuals. Lastly, probability of hospitalisation was estimated using binary
logistic regression equation as given below.

103
 Prob (hospitalisation>0│ill) = β0+β1Mx +β2Xy+ε
{1 if hospitalisation│Health Care Action>0, 0 otherwise}
Xy is a set of predisposing, enabling and need variables that influence probability
of hospitalisation; Mx represents the mode of payment (SSP). The logistic model was
subjected to a number of specification and diagnostic tests; especially the possible
endogeneity was tested using the Durbin-Wu-Hausman test. Model specifications were
changed to substantiate the findings.
4.4.2Utilisation of Health Services and Moral Hazard Behaviour
A higher percent of uninsured and newly insured individuals got treatment as
outpatient compared to insured individuals. Of the 285 individuals who availed inpatient
services, 143 belonged to insured group, 82 to newly insured group and 60 were
uninsured individuals. Intergroup comparison shows a picture of higher hospitalisation by
insured (88.8%) than newly insured (68.3%) or uninsured (66.7%). Outpatient treatment
was availed by a lower percent of (9.9%) insured compared to 29.2 percent of newly
insured and 27.8 percent of uninsured individuals. Thus, insured individuals were
hospitalised more than uninsured and newly insured individuals (p=0.00). Insured
individuals had higher probability of hospitalisation (0.34) than newly insured (0.22) and
uninsured individuals (0.17).
Another finding that draws our attention is the higher hospitalisation by insured
group despite homogeneity in the types of illness among the three groups. To check for
over-utilisation or moral hazard, the study compared the number of days spent in the
hospital by three groups. Insured spent more number of days on an average (19) than
newly insured (15) and uninsured individuals (12). Nevertheless, absence of such a
difference was proved by Kruskal Wallis test (p>0.05), hence there was no moral hazard
in SSP despite the higher utilisation by insured individuals. Lower utilisation by
uninsured and newly insured individuals suggests substantial barriers to hospitalisation,
even for chronic illness.

104
4.4.3 Patient Perceived Effect of the Treatment
Analysis of the effect of the treatment as perceived by patients would strengthen
the beneficial effect of SSP (Table 4.16). Almost half of the insured individuals felt
better after the treatment compared to 29 percent of newly insured and 34.7 percent of
uninsured individuals. However, almost 35 percent of insured did not feel better from the
treatment compared to a lower percent of uninsured individuals (24.5%).
Table 4.16 Effect of the Treatment Perceived by the Individuals
Better Slightly better Not better
(N=224) (N=69) (N=49)
Insured 50.4 29 34.7
Newly insured 27.7 37.7 40.8
Uninsured 21.9 33.3 24.5
χ2(4,N=342)=12.672, p=0.013
(Figures in percentages)

4.4.4 Econometric Estimation of the Probability of Hospitalisation

To estimate the probability of hospitalisation by insured, newly insured and
uninsured individuals, logistic regression analysis was used. SSP insured were assigned a
code of ‘1’. Newly insured and uninsured individuals were assigned a code of ‘2’ and ‘3’
respectively. Hospitalised individuals were given a code of ‘1’ and those who took
outpatient treatment had a code of ‘0’. The effect of a change in an independent variable
depends on the values of all variables in the model in binary logistic regression models.
Hence, certain variables were included in the analysis as enabling, pre-disposing and
need factors. Firstly, the role of types of illness, gender of ill persons, education and job
status of the head of the household, size of the household, income class and the area of
residence were analysed after classifying the individuals based on SSP membership.
Secondly, binary logistic model estimated the probability of hospitalisation and models
with different specifications substantiated the findings of the first (basic) model.

105
4.4.4.1 Need Factor Determining Utilisation
Types of illness, as evaluated by the health professional was taken as the need
factor that would influence utilisation. Illness can be acute, chronic and maternity care.
Inter-group analysis reveals that insured individuals sought inpatient care for these
illnesses more than the other two groups, except maternity. Newly insured and uninsured
individuals sought outpatient care for illnesses such as tuberculosis, dengue fever,
injuries, heart disease and fractures that otherwise required admission. Overall, 42.7
percent of acute illness, and 54.5 percent of chronic illness required inpatient care
(remaining was maternity). A higher percentage of chronically ill persons in insured
(53.6%) and uninsured group (50%) had admission (Table 4.18). Nearly half of the acute
patients in newly insured households chose hospitalisation. The logistic regression model
did not consider the maternity care as that invariably required admission. Hence, it would
contaminate the findings of the model. Relationship between the types of treatment and
types of illness irrespective of health insurance was significant (p=0.032).
4.4.4.2 Predisposing Factors Determining Utilisation
a. Gender of Ill Person
Women face barriers to access care due to less control over family resources and
restrictions due to the cultural factors and domestic responsibilities (Nanda 1999). SSP
would remove the financial barriers; hence insured women were expected to use inpatient
care at par with insured men. Uninsured and newly insured women may face financial
barriers that would lead to gender inequality in access to care. Analysis on the types of
treatment shows no statistical difference in access to care based on gender in insured and
uninsured group but not newly insured group (Table 4.17). Women in these three groups
had less hospitalisation and more of outpatient care than men.

106
 Table 4.17 Gender of Ill Persons and Types of Treatment

Male Female
Insured1 Outpatient(N=16) 37.5 62.5
Inpatient (N=143) 53.6 46.4
Newly insured2 Outpatient (N=35) 37.1 62.9
Inpatient (N=82) 58.7 41.3
3
Uninsured Outpatient (N=25) 40 60
Inpatient (N=60) 57.9 42.1

1 2 2 2 3 2
χ (1,N=159)=1.486, p=0.170 χ (1,N=117)=4.428, p=0.029 χ (1,N=85)=2.231, p=0.105
(Figures represent percentages to the total of each sub-group)

Irrespective of health insurance status, men were admitted in higher percent

(56%) than women (44%) (p=0.014). When the analysis was confined to the treatment
taken by women, it was observed that 62 percent of uninsured women, 58.5 percent of
newly insured women and 86.7 percent of insured women were hospitalised (p<0.05).
Mainly uninsured and newly insured women compared to insured women took outpatient
treatment. Thus, SSP resulted in higher utilisation of insured women compared to newly
insured and uninsured women.
b. Size of Household
Higher the number of members in a family, larger will be the probability of
admissions, ceteris paribus. Admitted insured individuals had an average size of 5
(median 4) higher than that of uninsured and newly insured individuals (median of 4
each) (p=0.00).
4.4.4.3 Enabling Factors Determining Utilisation
a. Income Class of the Household
Income can be a barrier to seek inpatient treatment for illness. Since SSP removes
this barrier, even the lowest income class can increase utilisation. However, the results
depict a different picture. In insured and uninsured group, a higher percent of low-income
class individuals (Q1 to Q3) sought admission. However, newly insured individuals from
107
all the income quintiles except Q5 got hospitalised (Table 4.18). Despite this finding,
there was no association between the income class and hospitalisation by the individuals
(p>0.05).
With regard to equity in utilisation based on the income quintile, vertical equity
was absent. Insured individuals from the poorest quintile did not utilise inpatient care
more than high-income individuals (p>0.05) (Table 4.18). Despite decrease in direct cost
of care, the poorest did not utilise SSP in larger proportion than other income classes.
This is due to indirect costs associated with seeking care, which can be sometimes high
and unaffordable by the poorer individuals. Horizontal equity was observed since insured
poor (Q1 and Q2 income quintile) was hospitalised in a higher proportion (93.1%)
compared to those in uninsured (64.3%) and newly insured groups (76%) (p=0.001).
b. Education of the Head of the Household
Higher education would influence the types of treatment, especially inpatient care.
To incorporate the effect of the education on utilisation, education of the head of the
household was included in the model. Median education of the head of the households
who accessed inpatient care was fifth standard for insured as well as uninsured
individuals and 7th standard for newly insured households (p>0.05). This finding suggests
no association between education and utilisation (p>0.05).
c. Job Status of Head of Household
Occupation of the head of the households determines the affordability of the
inpatient care. Those in formal sector employment (even salaried in informal sector), and
self-employed can afford expensive inpatient care compared to the wage earners,
unemployed or agriculturists due to seasonality of income. However, there was no
association between the job status and hospitalisation (p>0.05) (Table 4.18).
d. Area of Residence
Hospitalisation depends on the availability of the health care facilities such as
nursing homes, government hospitals and private hospitals that are easily accessible,
affordable and acceptable. Individuals living in rural areas have fewer hospitals to access
care. Moreover, indirect costs (transportation charges, boarding and food expenses)
108
would deter the households from seeking care at semi-urban or urban areas. SSP insured
households could claim the insurance benefit especially direct cost of seeking care that
removed financial barriers. Hence, insured rural individuals from rural areas were
expected to seek care in a higher proportion compared to those from uninsured and newly
insured groups. However, contrary to our expectations, individuals living in rural aeas
had higher admission compared to urban or semi urban areas (insured 51.8%, newly
insured 46.7% and uninsured individuals 51.8%) (Table 4.18) (p<0.05).
Table 4.18 Description of Predisposing, Enabling and Need Factors
Insured Newly insured Uninsured
(N=143) (N=82) (N=60)
Types of illnessa
Acute illness 43.5 49.2 45.6
Chronic illness 54 45 50
Maternity care 2.5 5.8 4.4
Gender of ill personb
Male 53.6 58.7 57.9
Job statusc
Unemployment 13.5 10.7 14.1
Labourer 57.4 66.9 57.6
Self employed 11.5 5.4 11.7
Agriculture 6.1 4.4 2.3
Formal sector 6.1 6.3 7.1
Salaried (informal sector) 5.4 6.3 7.2
Household sized
Median 4 4 4
Income classe
Q1 21.7 23.3 25.6
Q2 24.2 20 24.4
Q3 23 20 18.9
Q4 14.9 22.5 17.8
Q5 16.2 14.2 13.3
Area of residencef
Urban 7.9 14.7 21.4
Semi-urban 40.3 38.7 26.8
Rural 51.8 46.6 51.8
Pearson chi square; a= 3.562, p>0.05; b=0.633, p>0.05; c=6.494, p>0.05;e=2.006, p>0.05; f=8.467 p<0.1,
Kruskal Wallis chi square; d=4.131, p>0.05

109
4.4.4.4 Results of Econometric Estimation
Binary logistic regression analysis was carried out to estimate the probability of
hospitalisation conditional on being ill. It was hypothesised that SSP increases utilisation
of the health services due to the low cost of care, quality of services and wide network of
the hospitals. Irrespective of the health insurance status, it was found that 80.6 percent of
the individuals with chronic illness and 70.9 percent of the individuals suffering from
acute illness got inpatient treatment (p<0.05). Median education of the head of the
households who accessed outpatient and inpatient care was fifth standard (p>0.05). A
higher percentage of the individuals in which the head of the household worked as
agriculturist got hospitalised (84.4%), followed by formal sector employees (81.5%) and
labourers (76.4%). Salaried in informal sector (71.4%), business (71.2%) and
unemployed (71.2%) head of the households had lower proportion of admission (p>0.05).
Analysis of the entire sample on the basis of income class revealed that almost 22.4
percent of the households from Q1, 24.3 percent from Q2, 20.6 percent from Q3 and
almost 16.5 percent from Q4 and 16.2 percent from Q5 income quintile sought care as
inpatients (p>0.05). Median size of the households seeking outpatient and inpatient care
was four (p>0.05). The types of illness were coded into two dummy variables. Income
classes were coded into five dummy variables. SSP membership status and area of
residence was coded into three dummy variables. The gender was coded into two dummy
variables and the job status of head of the household was coded into six dummy
variables. Education of the head of the family was coded into four variables and size of
the household was coded into three dummy variables (Table 4.19).
The binary logistic regression analysis was performed to predict the determinants
of utilisation of health services. The basic model considered the cases of the ill individual
(model 2a). Model specifications were changed and best fit (model 2b), adult model
(model 2c) and women model (model 2d) were estimated to substantiate the findings of
the basic model. Model 2b considered the significant variables as predicted by the model
2a. Model 2c took cases of the adults and model 2d considered the cases of women and
used significant independent variables as estimated by model 2a.
110
 Table 4.19 Definition and Measurement of Variables
Variables Model 2a 2b, 2c, 2d
Health insurance
SSP insured=1 (reference)
Newly insured=2
Uninsured=3
Gender of ill person =1 if male, 0 if female
(reference)
Types of illness =1 if chronic, 0 if acute
(reference))
Occupation of the head of the household
1= Labourer if primary occupation is unskilled
worker, being paid daily wage
2= Business if engaged in self-employment
3= Agriculture if farmer including dairy farmer
4=Salaried in informal sector if unskilled
worker being paid monthly in unorganised
sector
5=Formal sector if skilled worker employed in
organised sector on a salary basis
0=Unemployed/not able to work(reference)
Income quintile
1=Q1, 2=Q2, 3=Q3, 4=Q4, 5=Q5 (reference)
Area of residence
1= Urban if individual lives in urban area
2= Semi-urban if individual lives in semi-
urban area
3=Rural if individual lives in rural areas
(reference)

The evidence of insured individuals being more likely to get admitted in case of
illness compared to newly insured and uninsured individuals was established by the
model 2a (Table 4.20). The Odds Ratio (OR) for newly insured and uninsured
individuals was significantly smaller than 1, which implied that insured individuals were
more likely to get admitted than newly insured and uninsured individuals. The odds of
admission compared to outpatient treatment were less for newly insured (OR 0.271) and
uninsured (OR 0.243) compared to insured individuals. Chronically ill individuals were
2.034 times more likely to get hospitalised compared to individuals with acute illness.
111
Men were 2.164 times more likely to get admitted than women. Income was a
determinant of utilisation with lower income quintile individuals had higher probability
of seeking outpatient treatment than higher income quintile. There was less likelihood of
Q3 (OR 0.329; at 10% significance) and Q4 (OR 0.291) income quintile individuals
being hospitalised compared to high income quintile (Q5) individuals. The education and
job status of the head of the household, size of the household and area of residence were
not associated with the probability of utilisation. Thus, SSP membership, types of illness,
gender of ill persons and income class determined the likelihood of hospitalisation.
Hence, the study accepts the hypothesis (H3) that SSP increases hospitalisation for
insured individuals compared to both newly insured and uninsured individuals.
The omnibus test of model coefficient, Hosmer and Lemeshow test, and -2 log
likelihood ratiosproved the robustness of the model. The results of these tests showed that
the model correctly predicts 77.4 percent of the cases. Residual analysis (specifically
Cook’s Distance statistic) showed no outliers and the model fits the whole set of
observation (Hosmer and Lemeshow test value of 0.717 indicates excellent
discrimination). The endogeneity test using Durbin-Wu-Hausman test found exogeneity
of health insurance (model 2a) with prob (chi2) =0.273. This confirms the positive impact
of SSP and the results are not due to any observable /unobservable characteristics that
would increase utilisation of health services.

112
 Table 4.20 Probability of Hospitalisation: Results of Model 2a

B S.E. Wald D Sig. Exp 95%CI

f (B) Lower Upper
Health insurance (base= SSP insured) 18.798 2 .000
Newly insured -1.306 .344 14.404 1 .000 .271 .138 .532
Uninsured -1.416 .366 14.937 1 .000 .243 .118 .498
Gender of ill person (base=Female) .772 .285 7.311 1 .007 2.164 1.237 3.785
Male
Types of illness (base: Acute) .710 .285 6.208 1 .013 2.034 1.164 3.555
Chronic
Education of head (base :Illiterate) 5.796 3 .122
Primary (1-7) -1.299 1.201 1.171 1 .279 .273 .026 2.869
Secondary (8-12) -.522 1.177 .197 1 .658 .593 .059 5.964
Graduate and above -1.050 1.159 .820 1 .365 .350 .036 3.396
Occupation 3.662 5 .599
(base= Unemployed)
Labourer .289 .381 .576 1 .448 1.336 .632 2.821
Business -.531 .747 .505 1 .477 .588 .136 2.541
Agriculture .799 .686 1.356 1 .244 2.223 .579 8.527
Salaried in informal sector .751 .665 1.275 1 .259 2.120 .575 7.812
Formal sector .786 .829 .897 1 .343 2.194 .432 11.147
Household size (base 1-3) .709 2 .702
4-6 .195 .556 .124 1 .725 1.216 .409 3.616
7 and above .367 .511 .515 1 .473 1.443 .530 3.927
Income quintile (base=Q5) 5.309 4 .257
Q1 -.992 .591 2.817 1 .093 .371 .116 1.181
Q2 -.762 .593 1.651 1 .199 .467 .146 1.492
Q3 -1.113 .575 3.754 1 .053 .329 .107 1.013
Q4 -1.235 .577 4.578 1 .032 .291 .094 .902
Area of residence (base=Rural area) 1.733 2 .420
Urban -.348 .400 .758 1 .384 .706 .323 1.545
Semi-urban .211 .330 .407 1 .523 1.234 .646 2.358
Constant 2.508 1.299 3.726 1 .054 12.281
Number of observations: 348; Omnibus test model coefficient:Pearson chi square =58.914, df =20 p=0.000;
-2 log likelihood = 335.843; Cox and Snell R squared= 0.152; Nagelkerke R squared= 0.227
Hosmer and Lemeshow Pearson chi square =6.196, df =8, p=0.625
(Dependent variable: Hospitalisation; 1=yes)
113
 To confirm the above findings, the best-fit model that considers the significant
variables predicted by model 2a such as health insurance status, income quintile, gender
of ill persons and types of illness was estimated (Table 4.21). Again, the odds of being
hospitalized was less if the individual was newly insured (OR 0.260) and uninsured (OR
0.246) compared to insured individuals. Men suffering from illness were 1.995 times
likely to be hospitalised than women. Individuals suffering from chronic illness were
1.873 times likely to be admitted compared to individuals with acute illness. The odds of
admission compared to outpatient treatment were high for high income (Q5) individuals
(OR 2.520) compared to the low income (Q1) individuals, at 10 percent significance
level. The model correctly predicts 77.3 percent of the cases.
Table 4.21 Probability of Hospitalisation: Results of Model 2b

B S.E. Wald Df Sig. Exp 95% C.I.for

(B) Exp (B)
Lower Upper
Health insurance
(base= SSP insured) 21.073 2 .000
Newly insured -1.347 .329 16.764 1 .000 .260 .136 .496
Uninsured -1.404 .346 16.423 1 .000 .246 .125 .484
Gender of ill person .690 .272 6.452 1 .011 1.995 1.171 3.398
(base=Female) Male
Types of illness .627 .269 5.423 1 .020 1.873 1.104 3.175
(base= Acute) Chronic
Income quintile 4.763 4 .313
(base=Q1)
Q2 .343 .392 .767 1 .381 1.409 .654 3.035
Q3 .019 .378 .002 1 .961 1.019 .486 2.136
Q4 -.106 .387 .075 1 .784 .899 .421 1.921
Q5 .924 .518 3.183 1 .074 2.520 .913 6.954
Constant 1.243 .384 10.486 1 .001 3.466
Number of observations: 348
Omnibus test model coefficient:Pearson chi square =43.632, df =8, p=0.000; -2 log likelihood = 351.126
Cox and Snell R squared= 0.115; Nagelkerke R squared= 0.172
Hosmer and Lemeshow Pearson chi square =4.885, df =8, p=0.770
(Dependent variable: Hospitalisation; 1=yes)

114
 Model 2c confirms the findings of models 2a and 2b by using the cases of adults,
defined as ill persons aged more than 17 years. Of the total individuals who accessed
care, 12 percent were children and 88 percent were adults. Table 4.22 illustrates the
results of the model that shows significant relationship between SSP and utilisation.
There was less likelihood of hospitalisation for newly insured individual (OR 0.249) and
uninsured individuals compared to insured individuals (OR 0.256). The odds of
hospitalisation were high if the individual was chronically ill rather than acute illness,
controlling for other socio-economic variables (OR 1.91). The odds an individual from
high income group (Q5) being hospitalised were 2.834 times the odds a low income
group (Q1) individual admitted. The odds of the admission compared to the outpatient
treatment was high for male (OR 1.686) than female, at 10 percent significance level. The
model correctly predicts 73.1 percent of the cases.
Table 4.22 Probability of Hospitalisation: Results of Model 2c

B S.E. Wald df Sig. Exp(B) 95% C.I.for

Exp (B)
Lower Upper
Health insurance
(base= SSP insured) 19.631 2 .000
Newly insured -1.388 .347 15.972 1 .000 .249 .126 .493
Uninsured -1.361 .356 14.631 1 .000 .256 .128 .515
Gender of ill person
(base=Female) Male .523 .286 3.328 1 .068 1.686 .962 2.956
Types of illness (base:
Acute) Chronic .647 .281 5.284 1 .022 1.910 1.100 3.316
Income quintile (base=Q1) 5.445 4 .245
Q2 .531 .417 1.622 1 .203 1.701 .751 3.853
Q3 .195 .409 .228 1 .633 1.216 .545 2.711
Q4 .001 .410 .000 1 .997 1.001 .449 2.235
Q5 1.042 .537 3.860 1 .049 2.834 .989 8.120
Constant 1.131 .401 7.937 1 .005 3.097
Number of observations: 315
Omnibus test model coefficient:Pearson chi square =39.573, df =8 p=0.000; -2 log likelihood = 316.967
Cox and Snell R squared= 0.117; Nagelkerke R squared= 0.174
Hosmer and Lemeshow Pearson chi square =3.300, df =8, p=0.914
(Dependent variable: Hospitalisation; 1=yes)

115
 Model 2d considered the cases of women to substantiate earlier findings (Table
4.23). Newly insured (OR 0.258) and uninsured (0.240) women had lower likelihood of
hospitalisation than insured. Income class and types of illness were not significantly
associated with hospitalisation of women. The model correctly predicts 76.9 percent of
the cases.
Table 4.23: Probability of Hospitalisation: Results of Model 2d

B S.E. Wald Df Sig. Exp(B) 95%C.I.for Exp

(B)
Lower Upper
Health
insurance(base= SSP 12.137 2 .002
insured)
Newly insured -1.353 .442 9.360 1 .002 .258 .109 .615
Uninsured -1.426 .469 9.250 1 .002 .240 .096 .602
Types of illness (base: .535 .362 2.189 1 .139 1.707 .840 3.469
Acute) Chronic
Income quintile 3.282 4 .512
(base=Q1)
Q2 .615 .535 1.322 1 .250 1.850 .648 5.281
Q3 .342 .499 .469 1 .493 1.408 .529 3.746
Q4 .259 .536 .234 1 .629 1.296 .453 3.708
Q5 1.390 .849 2.683 1 .101 4.014 .761 21.176
Constant 1.052 .483 4.738 1 .029 2.863
Number of observations: 171
Omnibus test model coefficient:Pearson chi square =22.389, df 7 p=0.002; -2 log likelihood = 186.015
Cox and Snell R squared= 0.123; Nagelkerke R squared= 0.174
Hosmer and Lemeshow Pearson chi square =2.765, df =8, p=0.948
(Dependent variable: Hospitalisation; 1=yes)

4.4.5 Summary
The positive impact of SSP on utilisation of health services was obvious from the
results of the models 2a, 2b, 2c and 2d. The evidence of insured individuals being more
likely to get admitted in case of illness compared to newly insured and uninsured
individuals was established. By removing financial barriers to access care, SSP could
make hospitalisation affordable. Chronic illness increased the likelihood of
hospitalisation than acute illness. High-income class individuals had higher admissions

116
compared to the low-income individuals. The area of residence, household size, job status
and education of head of household were not associated with utilisation. SSP
membership, gender of ill person, types of illness and income class did influence the
likelihood of hospitalisation. According to the likelihood ratio test (-2LL), model 2b is
superior to all the models (2a,2c and 2d). Overall, these models predict higher utilisation
of health services by insured individuals. Hence, the current study accepts the hypothesis
(H3) that SSP increases utilisation of health services.
The results from these models indicate that insured individuals behave according
to the theory. Demand for health services is price elastic; hence, any price decrease would
enhance utilisation. SSP benefits reduce the price of care; there by increase utilisation of
inpatient services. Moreover, certain design features such as availability of drugs in
network hospitals, better quality of services at network hospitals, streamlined claim
procedure and coverage of inpatient treatment in benefit package increases the utilisation
of health services. Given the similarities in the pattern of illness, higher proportion of
hospitalisation in insured group implies over-utilisation. The results of the study did not
show over-utilisation defined as the number of days spent in the hospital. This was
possible due to certain design features of SSP (strict monitoring by SSP assistants and
pre-authorisation process) that curtailed moral hazard behaviour.
Of the various variables studied, income of the household was directly associated
with hospitalisation. Elasticity of demand for health care is income elastic and high-
income individuals have a higher propensity to use the health services. In this study,
vertical equity in utilisation was absent; hence, higher proportion of insured poorest did
not hospitalise compared to the higher income individuals. However, the horizontal
equity was seen in which insured poorest were hospitalised more than their counterparts
in uninsured and newly insured groups. Indirect cost of treatment might have prevented
the poorest to utilise health services more than high-income individuals. Nevertheless,
compared to individuals in similar income class in uninsured and newly insured group,
insured poorest were hospitalised due to the claim benefits from SSP that reduced the
financial burden of inpatient treatment.
117
 Equity in utilisation based on the gender revealed a different picture. Although
there was no gender related vertical equity in utilisation, the study found socially
desirable result (horizontal equity) as evident from the model 2d. Insured women had
more episodes of admissions compared to uninsured and newly insured women. As the
financial barrier to access inpatient care was low, insured women had higher admissions.
The study finding (model 2d) brings to light the positive impact of health insurance on
the women empowerment in health. Regardless of insurance status, men were found to
get hospitalised (model 2a and 2c) more than women, a general finding in India since
women tend to ignore their health and postpone treatment or resort to home medicine
(Asfaw, 2010).
The model findings reveal the role of evaluated and perceived need with
chronically ill persons seeking hospitalisation. Chronic illness usually has longer duration
resulting in bad health status. However, insured did seek inpatient treatment even for
acute illness. On the contrary, uninsured individuals did not seek inpatient care even for
chronic illness. At the end, positive effect of SSP on hospitalisation was evident owing
to the risk coverage of inpatient treatment. Horizontal equity in utilisation based on the
gender and income was present. Vertical equity in utilisation based on the income and
gender was absent. Higher hospitalisation among insured individuals is due to the low
cost of treatment made possible by SSP. To know the cost impact, further analysis on the
impact of SSP needs to be undertaken. Hence, the next section concerns with the impact
of SSP on the out of pocket expenses incurred for medical treatment.

118
4.5 IMPACT OF SAMPOORNA SURAKSHA PROGRAMME ON OUT OF
POCKET EXPENDITURE
4.5.1 Introduction
One of the measures used to study the impact of MHI on financial protection is
OOPE incurred by insured members. Medical expenses push the poor into poverty and
impoverish the households. The research question was whether SSP reduced OOPE
incurred for medical treatment for insured individuals compared to newly insured and
uninsured individuals. By providing the financial assistance during hospitalisation, SSP
reduces medical expenses. Hence, the hypothesis was that SSP reduces OOPE associated
with illness due to the claim benefits. Testing of the hypothesis used the multiple linear
regression analysis.
Direct cost of medical services on illness includes hospital expenses (registration
fees, consultation fees, admission charges, and diet charges), medicine costs, diagnostic
charges and laboratory or investigation charges. Other costs were food expenses, lodge
charges and transportation expenses. Indirect cost of accessing care includes the lost
wages due to illness (multiplying daily wages with number of working days lost) and
interest amount paid on the loan taken to pay for medical expenses. Total expenses
include direct and other expenses and indirect costs. Firstly, OOPE incurred by insured,
newly insured and uninsured individuals was assessed considering total and direct cost
before and after the claim from SSP to know the differences in OOPE owing to claim.
Secondly, OOPE as a percentage of annual expenditure, before and after claim, was
analysed for the studied groups. Lastly, determinants of OOPE were estimated using the
multiple linear regression analysis.
Log (OOPEi│HCAi|ill)= β0+β1Mx+β2Xy+ε
OOPEi│HCAi is OOPE conditional on health care action upon illness. Mx
represents the mode of payment (SSP). Xy is a set of variables that influence probability
of OOPE. The model was subjected to a number of tests namely variance inflation factor,
correlation matrix, Cook’s D statistic and Dfits statistic.

119
 The data off 361 individuals who sought the treatment on
o illness, which
w inclu
uded
159 inssured and 2202 uninsurred (117 neewly insured
d and 85 uninsured
u inndividuals) was
used fo PE mainly the types of illness and
or the anallysis. Deterrminants of the OOP
treatmeent, health insurance (SSP),
( dayss spent in the hospitaal, age andd gender of ill
personss, income cllass, size off family and area of resiidence weree analysed.
4.5.2 Out
O of Pockeet Expendiiture Incurrred for Heaalth Care Services
S
Direct, ind
direct and total expen
nses incurrred by insu
ured, newlly insured and
uninsurred individu
uals before and
a after thee claim from
m SSP are given
g in Tabble 4.24. Beefore
the claiim, direct median
m cost incurred by d and insureed was 50000, which was
y uninsured
higher than newlyy insured ( 4500). Direct cost ass percent of
o the total treatment cost
(exclud
ding indirectt cost) was 92.6 percen
nt for insureed, 92.5 perccent for new
wly insured and
91.4 peercent for uninsured
u in
ndividuals. Other expeenses, on an
n average were
w 952
2 for
insured, 1234 fo
for newly insured an
nd 1366 for
f uninsurred individduals (p>0.05);
howeveer, the diffeerence in th
he amount was
w not sig
gnificant. Bo
oth uninsurred and insu
ured
spent allmost 35 peercent of th
he total costt (direct, oth
her cost and
d indirect ccost) as indiirect
cost com
mpared to nnewly insurred individu
uals (40 %)). Indirect cost
c had tw
wo componeents.
Average wages losst due to illlness includ
ding days oof admission
n or outpatiient visits were
w
high fo
or newly in
nsured ( 97
760) compaared to insuured ( 6913
3) and uninnsured ( 68
899)
individu
uals (p>0.05). Mean in
nterest paym
ments on thhe loan takeen to pay fo
or medical bills
b
was thee highest forr newly insu
ured ( 1837
7) and unin
nsured ( 1602) and low
west for insu
ured
individu
uals ( 905) (p<0.05).
After the cllaim from SSP,
S median
n total OOP
PE for insurred was mucch less ( 49
950)
than neewly insured ( 8875) and uninsu
ured ( 83755) (p<0.00)). Direct co
ost of treatm
ment
was low
wer for in
nsured (med
dian 2800
0) compareed to uninssured and newly insu
ured
individu
uals (p=0.00
03). Before the claim, there
t was siimilarity in the direct aand total cosst of
illness between innsured, new
wly insured and uninsu
ured individ
duals. How
wever, afterr the
claim, there
t was a significant difference in
i OOPE in
ncurred by th
hese individ
duals.

120
 Tablee 4.24 Cost of Medicall Care
In
nsured ( ) Newly
y insured ( ) Uninsurred ( )
1
Direct ccost before claim
c 1
12429(3116
64) 1567
78 (34453) 14600 (28567)
(
2
Total co
ost before cllaim 2
22469(5755
50) 28533(50919) 24473((39271)
Direct ccost after claaim3 1
10202(3141
17) 15678(34453) 14600((28567)
ost after claiim4
Total co 1
19099(5676
60) 28533(50919) 24473((39271)
1
Kruskal Walliis χ2 (2) =1.715, p =0.424
2
Kruskal Walliis χ2 (2) =1.232, p =0.540
3
Kruskal Walliis χ2 (2) =11.815, p =0.003
4
Kruskal Walliis χ2 (2) =19.222, p =0.000
Mean expensess (standard deviiation in brackeet)

4.5.3 Out of
o Pocket Expenses
E ass a Percentaage of Annual Consum
mption Exp
penditure
Undderstanding the impactt of health expenses
e on
n householdd consumptiion requiress
the deliberaation on thee direct OO
OPE as a perrcentage off annual connsumption expenditure
e .
Figure 4.4 depicts thee positive im
mpact of SSP on OOP
PE incurredd by insured
d members.
The percen
ntage of in
nsured indiividuals whho spent more
m than 15 percentt of annuall
consumptio
on expenditu
ure on OOP
PE reduced drastically after the claaim.

40

35 34..8

30
Percentage of individuals

24.4
25 25
21.9 Insured beforee
20 the claim
21.1 18.1
16.8
15
13 Insured after
11.8
10 claim
8.1
5
2.5 2.5
0
N
No <5% 5 to <10% 10 too <15% 15 to <25%
% >25%

OOP as percentage
p of annnual consumptioon expenditure

Figure 4.4 Direct Outt of Pocket Expenses as

a a Percen
ntage of thee Annual Co
onsumption
n
Expenditurre Incurred by
b Insured Members
M

12
21
 A comparison of insured with newly insured and uninsured individuals shows
lower percent of insured (13%) individuals allocating more than 25 percent of annual
consumption expenditure for medical illness (Figure 4.5).
40

35
Percentage of individuals

30

25

20

15

10

0
5 to 10 to 15 to
0 <5% >25%
<10% <15% <25%
Insured after claim 16.8 34.8 21.1 11.8 2.5 13
Newly insured 3.3 33.3 19.2 12.5 5.8 25.8
Uninsured 4.4 31.1 18.9 14.4 4.4 26.7

Figure 4.5 Direct Out of Pocket Expenses as Percentage of the Annual Consumption
Expenditure
From Figure 4.5, we infer that newly insured and uninsured groups had higher
percentage of individuals who paid more than 15 percent compared to insured members.
This corroborates the hypothesis of positive impact of SSP on OOPE (H4).
The next question was whether the total OOPE as a percentage of annual
consumption expenditure was low for insured individuals after SSP claim. Confirmative
positive reply to this question by Kruskal Wallis statistical tests (Table 4.25) proves a
significant decrease in the total OOPE as a percentage of annual consumption
expenditure for insured members (p<0.00) compared to uninsured and newly insured
individuals (15%). Despite SSP, average total OOPE was 8 percent for insured
individuals owing to high indirect costs in the form of lost wages and interest payments
on the loan taken to meet the excessive medical expenses and outpatient treatment costs.

122
Table 4.25 Out of Pocket Expenses as a Percentage of the Annual Consumption
Expenditure
Insured Newly insured Uninsured
1
Direct cost before the claim 7 7 6
Direct cost after claim2 4 7 6
Total cost before Claim3 12 15 15
Total cost after claim4 8 15 15
1
Kruskal Wallis χ2 (2) =2.504, p =0.286
2
Kruskal Wallis χ2 (2) =17.216, p =0.000
3
Kruskal Wallis χ2 (2) =2.165, p =0.339
4
Kruskal Wallis χ2 (2) =21.430, p =0.000

By providing financial coverage for hospitalisation, SSP could reduce the burden
of health expenses for insured households who otherwise had to incur additional four
percent of the annual consumption expenditure for medical treatment (reduced from 12 %
to 8%).
4.5.4 Econometric Estimation of the Probability of Out of Pocket Expenses
To estimate the probability of OOPE by insured, newly insured and uninsured
individuals, multiple regression analysis was used. The following sub-sections deal with
the analysis of various characteristics of individual, household (size of the household and
income class) and community (area of residence) determining OOPE, after classifying
individuals based on SSP membership status.
4.5.4.1 Characteristics of Individuals
Some of characteristics of the sick individuals that influence OOPE were types of
illness and treatment, health insurance, days spent in the hospital, age and gender of ill
persons.
a. Age of the Ill Persons
The study expects younger individuals to have lower expenses than older persons.
As age increases, people are more prone to illness; hence, the treatment related
expenditure would be high. The results reveal mean age of uninsured ill persons to be the

123
highest (44 years)) compared
d to newly insured (377 years) an
nd insured iindividuals (43
years) (p>0.05).
(
b. Typees of Illnesss
ute illness and materrnity
Chronic illness relateed OOPE are higherr than acu
g maternity as types of illness is a misnomer. In this stud
expensees. However, including dy, it
was inccluded as maternity
m reequires the admission and medicaal care just like any other
o
illness and
a a persoon has to in
ncur expensees to avail health serviices. The annalysis on total
t
OOPE related to the different types off illness po
oints at a significant
s association
n for
insured and newly
y insured ind Table 4.26). After the claim, insuured individ
dividuals (T duals
spent a median OO
OPE of 463
30 for acutee illness, 66100 for chrronic illnesss and 1525
5 for
materniity admissio
on. Newly insured haad 5420, 12875 and
d 6550 reespectively and
uninsurred spent 66800, 10050 and 97
775 respectiively. Treatm
ment of thee chronic illn
ness
was exp
pensive thann acute illn
ness or mateernity care. After the claim,
c for innsured famiilies,
b 87 percent, chronic illness exp
materniity expensess reduced by penses by 166.4 percent and
acute illlness expen
nses reduceed by 18 percent.
p Reggardless of health insuurance, chro
onic
illness ( 33036) w
was associated with a higher averaage OOPE th
han acute illlness ( 143
363)
or mateernity expennses ( 6805) (Kruskal Wallis
W χ2 p<0.05).
Taable 4.26 Total Expensses Incurred
d for Differeent Illness

Insuured (beforee Insured

d (after Newly inssured Uniinsured
claiim) ( )1 claim) ( )2 ( )3 ( )4
Acu
ute 145526 (45867)) 12158 (45868)
( 14641 (32
2871) 177727 (38401)
Chro
onic 262
207 (66035)) 25487 (64894)
( 46143 (63
3677) 319902 (40027)
Maternity 103
31( (2367) 1625 (1
1731) 9772 (791
17) 67994 (7309)
1 2
Kruskal Walliis χ (2) =5.712, p =0.057
2
Kruskal Walliis χ2 (2) =6.536, p =0.038
3
Kruskal Walliis χ2 (2) =9.181, p =0.01
4
Kruskal Walliis χ2 (2) =4.596, p =0.1
Mean (Standarrd deviation in bracket)
b

124
c. Gender of Ill Perso
ons
A positive
p asssociation beetween the gender and
d median tootal expensses incurred
d
with womeen ( 6300) spending lower
l amou
unt than men
m ( 8478)) (p<0.05) for insured
d
group was found befo
ore the claim
m from SSP
P (Table 4.2
27). Newly insured and
d uninsured
d
women speent a mediaan OOPE off 7050 an
nd 6900 reespectively (newly insu
ured men
10275 and uninsured men
m 8550
0). After thee claim from
m SSP, wom
men spent 25.3
2 percentt
lower (med
dian 4180)) than actuaal expenses compared to
t men (red
duced by 10
0 %, median
n
of 5678). Hence, the OOPE redu
uced more for
f insured women
w com
mpared to men.
m
T
Table 4.27 Gender
G of Ill Persons annd Total Ou
ut of Pockett Expenditurre

Insured
d (before Insured
I (aftter New d Uninsurred ( )4
wly insured
claim) ( )1 c
claim) ( )2 ( )3
Male 30537 (76445) 27354 (76249) 318
854 (54951) 28157 (47360)
(
Femalee 14095 (24200) 10530 (20622) 252
212 (46770) 19419 (28088)
(
1
Mannn Whitney U test = 2643.5, Z=2.014, p =0.0444
2
Mannn Whitney U test = 2759, Z=11.623, p =0.1055
3
Mannn Whitney U test = 1622.5, Z=0.932, p =0.3552
4
Mannn Whitney U test = 919.5, Z==1.241, p =0.2155
Meann expenses (stanndard deviationn in bracket)

Anoother matterr that drew the attentio

on was high
her total exp
penses for in
nsured men
n
than womeen. Women had higher percent of chronic illn
ness (53.2%
%) in contraast to higherr
percent of acute illnesss suffered by men (677.6%) (p<0.05). The fact
fa that chrronic illnesss
gives way to high OOPE
O is inaapplicable hhere. Furth
her probe in
nto this incciting issuee
he cause to
revealed th o be the in
ndirect costs. Prolongeed illness ((median 15
5 days) and
d
subsequentt loss of wag
ges due to no
n work (meedian 30 daays) raised O
OOPE for men.
m Almostt
30 percent of the men
n were daily
y labourers. To meet the
t cost of illness, men
n borrowed
d
(52.2%) inn higher percent
p thaan women giving risse to largeer interest payments.
Subsequenttly, averagee indirect costs
c were more for men ( 115521) compaared to thee
women ( 66854). Averrage direct medical coosts were allso higher ffor men ( 17704)
1 than
n
women ( 11435) as men spent more dayss (average15 days) in hospital th
han women
n
(average off 10 days) (M
Mann Whitn
ney U test p <0.05).
12
25
 Inter-groupp analysis considering
c g sick wom
men to know
w the impact of SSP
P on
OOPE for
f women shows no significant
s a
association b
before the claim
c (Krusskal Wallis χ2
χ p
>0.05). After the claim,
c insurred women incurred low
wer OOPE than the otther two gro
oups
(Kruskaal Wallis χ22 p<0.05) (T
Table 4.28). Average direct and in
ndirect expennse incurred
d by
insured women was mpared to uninsured and newly insured woomen. Oveerall,
w low com
ntal equity was presen
horizon nt since in
nsured wom
men spent less OOPE
E compared
d to
uninsurred and new
wly insured counterparts
c s.
Table 4.28 Averaage Out of Pocket
P Expeenses Incurrred by Wom
men
Insured
d (after claim
m) ( ) Neewly insured ( ) Uniinsured ( )
Dirrect cost 9958 11399 10200
Lost wages 1665 10331 7113
Intterest on loaan 625 1527 1084
Total cost1 10530 25212 19419
1
Kruskal Walliis χ2 (2) =10.1884, p =0.006

Median dayys of illnesss was the hiighest for unninsured wo

omen (30 ddays) and neewly
insured (15 days) than insureed women (10
( days). T
Though 34 percent
p of insured
i wom
men
borroweed compareed to a loweer percent of
o uninsured
d (26.4%), source of bborrowing were
w
informaal such as reelatives and
d friends an
nd the borro
owed amoun
nt was less. Newly insu
ured
women borrowed
d the mostt (38.8%) that gavee rise to large interrest paymeents.
quently, indiirect cost of
Conseq o care was more for uuninsured and
a newly iinsured gro
oups.
SSP beenefit and lower
l indirrect expenses resulted
d in less OOPE for innsured wom
men.
Regardlless of the health insu
urance, meen were fou
und to havee higher OOPE ( 247
747)
comparred to womeen ( 17829)) (p<0.05).
d. Typees of Treatm
ment and the
t Duratio
on of Illnesss
OOPE dep
pends on th
he types of treatment taken. Inpaatient treatm
ment is usu
ually
expensiive than ou
utpatient treeatment. Beefore the claim,
c mediian total exxpenses due to
hospitallisation weere more fo
or newly in
nsured ( 10275) and uninsured individualss (
There was a significannt differencce in
10300) compared to insured individualss ( 8450). T
the typees of treatm
ment and OOPE
O with outpatient treatment being
b less expensive than
t
126
inpatient trreatment (p<
<0.05). Thee differencee between outpatient
o aand inpatien
nt treatmentt
costs for sttudied grou
ups before the
t claim was
w also sig
gnificant (p < 0.00) (T
Table 4.29).
Strikingly, the discrepancy ceased
d to exist affter the claim P (p>0.05) and insured
m from SSP d
spent mediian total OO
OPE of 5
5200 for hoospitalisation. This finnding is notteworthy ass
profound buurden of ho
ospitalisation
n was low ffor insured members.
m
Table 4.299 Average Out of Po
ocket Expen
nses Incurrred for Ou
utpatient an
nd Inpatientt
Treatment
Insureed (before Insured (aafter wly insureed Uninsurred
New
1 2
claim)) ( ) claim) ( ) ( )3 ( )4
Outpatieent 2606
6 (3655) 2606 (36
628) 40
040 (33858)) 3588 (21848)
(
Inpatien
nt 16106
6 (60691) 12394 (59
9957) 230
019 (56937) 22280 (44825)
1
Mannn Whitney U test = 465.5, Z==3.885, p =0.0000
2
Mannn Whitney U test = 844.5, Z==1.715, p =0.0866
3
Mannn Whitney U test = 537.5, Z==5.344, p =0.0000
4
Mannn Whitney U test = 386.5, Z==3.507, p =0.0000
Meann expenses (stanndard deviationn in bracket)

Irrespective off the health

h insurance,, inpatient treatment ((average 2
26426) wass
associated with
w higherr OOPE than
n outpatientt treatment (average 114581) (Man
nn Whitney
y
U test p =0
0.017). Day t hospital was anotheer determinnant of OOP
ys spent in the PE (Kruskall
Wallis χ2 p =0.00). In
nsured spen
nt more num
mber of daays in the hospital
h (aveerage of 19
9
days) than newly insu
ured (15 daays) and unninsured (12
2 days) inddividuals. Despite
D that,
insured incuurred lowerr OOPE owing to the claim benefits from SSP
P.
4.5.4.2 Chaaracteristiccs of Houseeholds
Incoome class an
nd size of th
he householld would deetermine OO
OPE.
a. Income Class
Indiividuals fro
om the hig
gh-income class
c d have higgh OOPE, since thesee
would
individuals can afforrd treatmen
nt at expensive privaate facilitiees. Howeveer, indirectt
expenses would
w be low
wer as they
y need not bborrow. Thiis study obsserved an in
nsignificantt
relationshipp between income and OOPE (T
Table 4.30)). Intra-income class analysis off
insured grooup shows lack of equiity before an
nd after thee claim (p>00.05). Beforre the claim
m
from SSP, individualss from Q4 income quuintile spent higher am
mount, med
dian 9570,,
12
27
followeed by indiv
viduals from
m Q2 ( 747
75), Q1 ( 7200),
7 Q3 ( 7000) annd Q5 ( 69
900)
(p>0.05
5). After thee claim, therre was a disscernible, ho
owever non
n-significantt, change in
n the
OOPE incurred by
y various in
ncome quin
ntiles (p>0.0
05). Individ
duals from Q
Q4 had hig
ghest
median OOPE ( 66270) follow
wed by ind
dividuals fro
om Q1 ( 5500),
5 Q3 ( 4860), Q5
5(
4825) and
a Q2 ( 4325) class. On the equ
uity in the distribution
d n of claims, analysis off the
percentage changee in OOPE due to the claim in vaarious incom
me quintile helps in beetter
understanding. Inddividuals frrom Q4 quiintile claimeed the mostt with 57 peercent decreease
in mediian OOPE followed by
b Q2 class. Individuaals from Q5
5 class got the least cllaim
followeed by Q3. H
Hence, the sttudy could not establissh equity in
n claims from
m SSP, defi
fined
as the laargest beneffit to the po
oorest.
Tablle 4.30 Totaal Out of Po
ocket Expennses by Inco
ome Quintilee
Insured (beforre Insured
d (after N
Newly insureed Uninsuured ( )4
claiim) ( )1 claim) ( )2 ( )3
Q
Q1 32172 (70759
9) 29112 (71247)
( 10304 (1015
59) 205300 (25595)
Q
Q2 12518 (19124
4) 9429 (17286) 4
41446 (7063
32) 25751 (31335)
Q
Q3 21272 (44150
0) 16937 (38307)
( 2
27773 (4038
80) 223855 (43299)
Q
Q4 13873 (14554
4) 9785 (12934) 34047 (5478
81) 260933 (57798)
Q
Q5 33881 (35408
8) 31798 (28356)
( 32641 (6177
72) 196833 (41903)
1 2
Kruskal Walliis χ (4, N=361)) =1.324, p =0.8851
2
Kruskal Walliis χ2 (4, N=361)) =2.215, p =0.6696
3
Kruskal Walliis χ2 (4, N=361)) =4.7, p =0.3199
4
Kruskal Walliis χ2 (4, N=361)) =1.132, p =0.8889
Mean expensess (standard deviiation in brackeet)

To exploree equity in OOPE, th wly insured and

he study coompared inssured, new
uninsurred individu
uals in Q1 and
a Q2. Th
here was no
o difference in OOPE incurred
i beefore
SSP claaims amongg the lowesst income classes
c in th
hese groupss. After the claim, insu
ured
memberrs had to pay
p a lowerr average OOPE
O ( 188738) comp
pared to new
wly insured
d (
24677) and uninsuured individ
duals ( 246
630) (p<0.0
05). Therefo
ore, the im
mpact of SSP
P in
reducin
ng the OOPE for the lo
owest incom
me classes compared to those off newly insu
ured
and uniinsured grou
ups demonsstrates horizzontal equitty in financial protectioon. Irrespecctive
of the health insurance, a comparison
n of the in
ncome classs and OO
OPE showss no
128
relationshipp between them (Kru
uskal Walliss χ2 p>0.05). Individuuals from Q5 incomee
quintile haad the high
hest averagee OOPE ( 29891) followed by Q4 ( 2443
33), Q2 (
24154), Q33 ( 21738) and
a lastly Q1
Q ( 21510) individualls.
b. Househoold Size
Sizee of the hou
usehold determines OO
OPE since large
l families have to spend moree
than smalleer ones. Thee median to
otal cost off care for th
he householdds with lesss than threee
members was
w 6800. Five mem
mbers housseholds speent a mediian of 63
370 and 10
0
members OOPE
O spen
nt 7000. It
I was 86
600 for more than 10 members householdss
(Kruskal Wallis
W χ2 p =0.860).
=
4.5.4.3 Chaaracteristiccs of the Co
ommunity
a. Area of Residence
Ressidential areea would determine th
he cost of care
c especiaally indirecct and otherr
expenses. Rural
R areass have feweer hospitalss forcing th
he people to
t seek carre from thee
hospitals siituated in urrban or sem
mi-urban areeas. This wo
ould increasse indirect cost of caree
along with other expeenses. Afterr the claim, lower med
dian OOPE was incurrred by rurall
residents from
f insureed ( 7825) group com
mpared to uninsured ( 10280) and newly
y
insured inddividuals ( 13125) (p>
>0.05). Even
n insured seemi-urban rresidents (m
median of
4560) and uurban resid
dents ( 2775
5) had loweer OOPE co
ompared to newly insu
ured ( 7263
3
and 5900
0 respectively) and uninsured
u g
group ( 64
400 and 9150 respeectively) in
n
respective areas.
a

12
29
 Table 4.31 Area of
o Residencce and Totall Out of Poccket Expensses
Insured (b
before Inssured (after Newly insured
i U
Uninsured
1 2 3
claim) ( ) claaim) ( ) ( ) ( )4
Urban area 22936 (40470) 20219 (39316
6) 25931 (50806)
( 113862 (2143
38)
Semi-u
urban area 17862 (35408) 12977 (30450
0) 18632 (32490)
( 330287 (5479
97)
Rural area
a 25661 (71226) 23259 (71617
7) 39331 (63159)
( 225371 (3463
39)
1 2
Kruskal Walliis χ (2) =0.769, p =0.681
2
Kruskal Walliis χ2 (2) =2.039, p =0.361
3
Kruskal Walliis χ2 (2) =2.263, p =0.323
4
Kruskal Walliis χ2 (2) =2.552, p =0.279
Mean expensess (standard deviiation in brackeet)

After the cclaim, burd

den of OOP
PE for sem
mi-urban ressidents decrreased by 27.6
2
percent in contrasst to a low
wer reductiion for urb
ban (9%) and rural areas (11.8
8%).
Regardlless of the health insu
urance, residents in ruural areas ( 28301) hhad to spen
nd a
higher average
a totaal OOPE thaan urban ( 20204) or ssemi-urban areas ( 180047).
4.5.4.4 Results of Econometrric Estimattion
The study used
u multip
ple linear reegression an
nalysis to kn
now the im
mpact of SSP
P on
OOPE. It was hypoothesised th
hat SSP decrreases OOP
PE since insured individduals can cllaim
he programm
from th me for hospitalisation. The log tran
nsformed am
mount of O
OOPE, age of
o ill
person and days sspent in thee hospital were
w used for the reg
gression anaalysis. Dum
mmy
variablees for the types of illlness, typees of treatm
ment, gendeer of ill peersons, sizee of
househo
olds, area of
o residencee, SSP statuss and incom
me class werre defined.
Model 3a used the caases of illn
ness reporteed by all households,
h irrespectivee of
health insurance
i sttatus and model
m 3b too
ok cases of hospitalisaation. Results of modeel 3a
are giveen in Tablee 4.32 with
h the signifiicant results at the end
d of the Taable. Backw
ward
eliminaation stepw del at the 9th step after
wise regresssion estimaated the rrobust mod a
eliminaating insign m the modeel. The anaalysis begann with the full
nificant variiables from
model considering
c g the indepeendent variaables that inncluded age and genderr of ill perssons,
types of treatment and illnesss, days spen
nt in the ho
ospital, size of the houusehold, income
class, area
a of resid
dence and health insu
urance statu
us. In modeel 1a, OOPE
E would bee 35
percent higher for newly insu
ured individ
duals [exp (.302)=1.35
( 5] and 29.6 percent hig
gher
130
for uninsured individuals [exp (.268)=1.296] than insured individuals. As given earlier,
insured individuals spent less compared to newly insured and uninsured individuals.
Moreover, OOPE would be 9.6 percent higher for men than women [exp (.095)=1.096].
Chronically ill individuals would have 41 percent more OOPE than individuals suffering
from acute illness [exp (.134)=1.41]. A one percent increase in the length of stay in the
hospitals would yield a .41 percent increase in OOPE. Semi urban individuals would
have 17.8 percent less OOPE and rural individuals would have 16.2 percent more OOPE
compared to urban individuals. Age of ill person, income class, and size of the household
and area of residence did not contribute to the OOPE. The regression estimates confirmed
the positive impact of SSP; hence, the finding proves the hypothesis that SSP reduces
OOPE for insured individuals compared to newly insured and uninsured individuals
(H4). Variance Inflation Factor did not suggest any multicollinearity since the value was
one for all the significant independent variables, less than cut off 10. Correlation matrix
did not show any significant correlation between independent variables. Cook’s D
statistic detected no outliers (all cases had values <0.014) and Dfits statistic (< 1.0) did
not suggest any observation that strongly influenced the model. F value was 18.159 (p <
0.05).
Model 3b considered the cases of hospitalisation to explore the impact of SSP on
OOPE (Table 4.33) taking all the variables included in model 3a. Backward elimination
stepwise regression estimated the robust model by eliminating insignificant variables
from the model at the 11th step. The finding of the model is in conformity with model 3a.
Since SSP covers IP services, it is not surprising to observe similar independent variables
as predictors of OOPE in this model. OOPE would be 33 percent higher for newly
insured individuals [exp (.288) =1.33] and 24.6 percent more for uninsured than insured
individuals [exp (.220)=1.246]. Moreover, OOPE would be 10.6 percent higher for men
than women [exp (.101=1.106]. Chronically ill individuals would have 13 percent more
OOPE than individuals suffering from acute illness [exp (.123=1.13]. A one percent
increase in the length of stay in the hospitals would yield a 0.455 percent increase in
OOPE.
131
 Table 4.32 Estimation of Model 3a: Health insurance and OOPE
Standardized
Coefficients
Model 3a Beta t Sig.
1 (Constant) -1.845 .066
Newly insured (=1, 0 otherwise) (base: insured) .301 5.760 .000
Uninsured (=1, 0 otherwise) .261 5.053 .000
Gender of head: Male (=1, 0 otherwise) (base: female) .082 1.713 .088
Semi urban (=1, 0 otherwise) (base: urban) -.170 -1.896 .059
Rural (=1, 0 otherwise) .154 1.718 .087
Household size 1-3 (=1, 0 otherwise) (base: 7 & above) -.052 -.613 .540
Household size 4-6 (=1, 0 otherwise) -.100 -1.236 .217
Q1 (=1, 0 otherwise) (base: Q5) .037 .554 .580
Q2 (=1, 0 otherwise) .004 .064 .949
Q3 (=1, 0 otherwise) .027 .438 .662
Q4 (=1, 0 otherwise) -.014 -.228 .820
Chronic (=1, 0 otherwise) (base: Acute) .117 2.370 .018
Maternity (=1, 0 otherwise) -.062 -1.265 .207
Log (Age of the head) -.046 -.944 .346
Log (Days spent in hospital) .424 8.617 .000
9 (Constant) -4.599 .000
Newly insured .302 5.961 .000
Uninsured .260 5.108 .000
Male .092 1.990 .047
Semi urban area -.164 -1.860 .064
Rural area .151 1.713 .088
Chronic .123 2.610 .009
Log (Days spent in hospital) .411 8.502 .000
Adjusted R2 .250
Standard error of the estimate .865
(Dependent variable in log transformed form)
Number of observations: 361

132
 Table 4.33 Estimation of Model 3b: Health insurance and OOPE

Standardized
Coefficients
Model 3b Beta t Sig.
1 (Constant) -1.547 .123
Newly insured (=1, 0 otherwise) (base: .289 5.322 .000
insured)
Uninsured (=1, 0 otherwise) (base: insured) .210 3.894 .000
Gender of head: Male (=1, 0 otherwise) .092 1.765 .079
(base: female)
Semi urban (=1, 0 otherwise) (base: urban) -.123 -1.224 .222
Rural (=1, 0 otherwise) .050 .496 .620
Household size 1-3 (=1, 0 otherwise) (base: -.103 -1.098 .273
7 & above)
Household size 4-6 (=1, 0 otherwise) -.135 -1.503 .134
Q1 (=1, 0 otherwise) (base: Q5) -.003 -.041 .968
Q2 (=1, 0 otherwise) -.011 -.153 .879
Q3 (=1, 0 otherwise) -.015 -.216 .829
Q4 (=1, 0 otherwise) -.047 -.713 .476
Chronic (=1, 0 otherwise) (base: Acute) .114 2.140 .033
Maternity (=1, 0 otherwise) -.052 -.954 .341
Log (Age of the head) -.047 -.894 .372
Log (Days spent in hospital) .471 8.931 .000
11 (Constant) -6.484 .000
Newly insured .288 5.490 .000
Uninsured .220 4.215 .000
Male .101 2.038 .043
Chronic .123 2.448 .015
Log (Days spent in hospital) .455 9.027 .000
Adjusted R2 .314
Standard error of the estimate .813
(Dependent variable in log form; Hospitalisation only)
Number of observations: 282

133
4.5.5 Analysis
A of Sampoorna
S a Suraksha
a Claims
The analysis on SSP claims
c usin
ng the household data shows averrage amoun
nt of
claim to be 3447. The max
ximum amo
ount reimbuursed for the
t studied sample waas
45,000. SSP has reimbursed
r 549,750 to
t the samplle insured individuals. Out of the 159
ons, 16 persons got ou
ill perso utpatient treeatment (10%) and 57 inpatient trreatment claaims
were reejected (36%
%). There were 143 inpatient
i addmissions, of which 886 got cash
hless
claim from
f the prrogramme. The numb
ber of claim
ms made by
y the entire insured (416
(
househo
olds) group in the life of
o the mem
mbership rev
veals that 21
1.4 percent ddid not claim
m at
all. Alm
most 49 perccent claimeed once and
d 17.6 perceent claimed twice. Tweelve househ
holds
surveyeed claimed tthrice or mo
ore times.
4.5.6 Su
ummary
SSP decreaased OOPE associated with treatm
ment of illneess compareed to uninsu
ured
and neewly insureed groups. Without SSP
S benefits, insured individualls would have
h
incurred
d high leveel of OOPE
E. Before th
he claim frrom SSP, th
here was no differencce in
direct cost,
c other expenses and
a indirect expenses among inssured and bboth uninsu
ured
groups. Because off SSP, theree was a redu
uction in thee direct costt and lowerr burden of total
t
OOPE on
o insured individuals.
i . Regression
n analysis cconfirmed th
he result of Kruskal Wallis
W
test thaat insured incurred lo
ower OOPE
E compareed to uninssured and newly insu
ured
individu
uals. Hencee, study pro ypothesis oof positive impact of SSP on OO
oves the hy OPE
(H4).
OOPE calcculated as a percentagee of annuall household
d expenditurre confirmss the
positivee impact of SSP. Of thee 100 ann
nual househo
old expendiiture, uninsu
ured and neewly
insured had to sp
pend 15 for medical expensees compared to just 8 of insu
ured
uals. Due too the limitattions of ben
individu nefit packagge, that exccluded outpaatient treatm
ment
and cerrtain diseasees, reductio
on of eight rupees
r for iinsured ind
dividuals waas not possiible.
Moreov
ver, SSP excludes indirrect costs (llost wages) or transporrtation expeenses and has
h a
ceiling on the risk covered ( 5000)
5 which
h is too low
w, given the high cost oof treatment..
oteworthy finding
Another no f wass the effect of the inpaatient treatm
ment on OO
OPE
due to SSP memb osts were inncredibly large
bership. Before the claaim, hospittalisation co
134
causing similarity in OOPE for insured, uninsured, and newly insured individuals. By
lowering OOPE related to hospitalisations, direct cost of treatment reduced along with
removal of the differences in the outpatient and inpatient treatment costs. While
uninsured still faced higher inpatient treatment costs, it was no longer a significant
variable influencing OOPE for insured individuals. Regression model (3b) on
hospitalised individuals confirms the positive impact of SSP on OOPE.
Days spent in the hospital, chronic illness and gender of ill persons influenced the
amount of OOPE. Chronic illness was associated with more utilisation and high OOPE
for insured, uninsured and newly insured individuals. SSP had minimal impact on the
OOPE related to chronic illness. Due to fixed benefit package, individuals with chronic
illness incurred higher OOPE than those with acute illness.
Another finding that needs deliberation was the absence of income related equity
in OOPE for insured individuals. Lowest income individuals spent the most in nominal
rupees compared to other income classes within insured group and the claim benefits
went to those in the moderate poor (Q3) and non poor (Q4). Despite SSP, the poorest
had high level of OOPE as a percentage of consumption expenditure compared to other
income classes. Nevertheless, horizontal equity in the benefit of SSP was present. Insured
poor had less burden of OOPE compared to their counterpart in newly insured and
uninsured groups. The poor insured individuals might not have benefited from SSP
compared to the high-income insured individuals (absence of vertical equity); however,
they were better compared to the poor in uninsured and newly insured groups (presence
of horizontal equity).
Gender related impact of SSP has two distinct parts. Firstly, men incurred higher
expenses compared to women in the studied households. For insured individuals, the
difference in OOPE between men and women disappeared after the claim from SSP.
Moreover, the reduction in OOPE was high for women compared to men. Secondly, the
study observed a lower OOPE for insured women compared to uninsured and newly
insured women. In addition, insured women had lower indirect expenses (lost wages and
interest payments) and borrowed from low cost informal sources of funds. Women and
135
poorest insured individuals had to pay lower OOPE compared to their counterparts in
newly insured and uninsured groups.
Regarding the distribution of SSP benefits, urban insured individuals got the least
benefit from SSP and semi-urban residents got the most. The explanation for this
phenomenon relies on the discussion in section 4.3.6 that establishes difference in health
seeking behaviour of individuals in different areas. Urban residents relied more on home
medicine, clinics and government hospitals compared to residents of other areas. Larger
number of semi-urban residents visited network hospitals; as a result, they got the
maximum benefits.
The combination of findings (section 4.3.7 and 4.4.5) provides the support for the
positive impact of SSP on the financial protection. One of the issues that emerge from
these findings is the need to evaluate the relative impact of SSP that nullifies the
differences in household size. Hence, the next section focuses on the catastrophic health
expenditure, a relative measure of financial protection to assess the impact of SSP.

136
4.6 IMPACT OF SAMPOORNA SURAKSHA PROGRAMME ON
CATASTROPHIC HEALTH EXPENDITURE
4.6.1 Introduction
Out of pocket expenses consider the absolute cost of treatment without regard to
income of the households. The same amount of OOPE would be catastrophic for poorer
than richer households. Hence, a relative measure to compare the impact of health
expenses on insured, newly insured and uninsured households is required that uses a
common denominator such as income. Given the positive impact of OOPE, we set out to
determine the impact of SSP on the relative measure of financial protection, termed as
catastrophic health expenditure (CHE). This study used annual per capita income to
adjust for household size while calculating CHE. The research question was whether SSP
reduces the incidence of CHE for insured individuals compared to newly insured and
uninsured individuals.
As SSP reduces the direct cost of treatment, there would be lower incidence of
CHE for insured individuals. Therefore, health expenses calculated as a percent of annual
per capita income would be less for insured individuals. Hence, the hypothesis was that
SSP reduces CHE for insured individuals compared to newly insured and uninsured
individuals. Pearson chi square test and logistic regression analysis were used to test the
hypothesis. Firstly, an analysis of the CHE incurred by insured, newly insured and
uninsured individuals considered the total and direct cost before and after the claim from
SSP. Secondly, the study explored an association between the number of visits to health
facility and the CHE. Lastly, determinants of the CHE were estimated using binary
logistic regression analysis.
Prob (CHEi│HCAi)= β0+β1Mx + β2Xy+ ε
{1 if CHE│HCA >0, 0 otherwise} CHEi│HCAi is catastrophic health expenditure
conditional on Health Care Action. Mx represents the mode of payment (SSP). Xy is a set
of variables that influence probability of catastrophic payments.Substantiation of the
basic model required changes in the specifications of the model resulting in four models
of the study.
137
 Catastrophic health expenditure has been defined in the literature as any health
care payment that is more than 10 percent of annual income of the household (Ranson
2002; Pradhan and Prescott 2002). This study used this definition in classifying the
households as experiencing CHE or not. Based on OOPE and annual income of the
households, analysis at the individual level focused on the impact of SSP on CHE. There
was considerable debate whether to take household or individual cases for the
classification of CHE. There were children, homemakers and old parents who accessed
health care facility for treatment. As they were not the earning members, it would be
difficult to calculate catastrophic health expenses. Whenever a family member becomes
sick, usually the other members collectively spend for treatment. Hence, household level
analysis is justifiable. However, in case of the household with more than one member ill,
classification of households into incurring the CHE or not becomes complicated. If the
household was taken as a unit of analysis, explicit description of the factors associated
with the CHE could not be analysed such as types of illness, number of visits, treatment
taken and gender of ill persons, which were individual characteristics of sick person.
Hence, the study used the data on sick individuals and household income. There was a
possibility that adding the percentage of annual income spent on illness for each
individual would result in the CHE for the entire family. However, non- CHE households
did not face CHE even after adding the percentage of annual income spent for illness by
each sick individual.
4.6.2 Catastrophic Expenditure among Insured and Uninsured Households
Analysis on the impact of SSP on CHE considered direct cost (hospital cost and
transportation costs) and total cost of treatment at the hospitals. Before SSP claim, there
was no association between CHE and health insurance status. Close to 45 percent of
insured individuals would have felt CHE compared to a lower percent of newly insured
(39.8%) and uninsured individuals (43%). Inclusion of claim data in the estimation of the
CHE provides a contrasting result. The number of individuals who faced the CHE
drastically reduced from 72 to 42 (from 44.8% to 26.5 %). Because of insurance, only
one quarter of sick insured individuals incurred CHE compared to two fifths of uninsured

138
individuals (Table 4.34
4). Thus, th
here was a significant
s d
difference in
n the CHE incurred by
y
insured andd uninsured
d (p<0.05). CHE relateed to the diirect expensses of illness confirmss
the above findings.
f Beefore the claaim, there was
w no sign
nificant diffeerence in th
he incidencee
of CHE am
mong insurred, uninsu
ured, and newly insureed individu
uals. SSP reduced
r thee
incidence oof CHE for insured
i indiividuals by 42 percent and only 233.3 percent of them feltt
CHE after the
t claim (p
p<0.05).
Thee effect of insurance
i w partial as there were
was w one fouurth of the individualss
facing CHE
E even with
h health insu
urance. Furtther analysiis on individ
duals who experienced
e d
CHE despiite insurance reveals th
hat 10 percent of indiv
viduals tookk outpatien
nt treatment,,
which was not claimaable under the prograamme. Ano
other eight percent of individualss
om diseases like heartt attack, canncer, strokee and disorrders that reequire hugee
suffered fro
money for treatment and
a the prog
gramme couuld give a maximum
m oof 35000 on a family
y
floater basiis. Six perceent of them
m had excludded diseases like feverr, cholera, and
a diabetess
related diso
orders.
phic Health Expenditure by Insuran
Table 4.34 Catastrop nce Status
CHE Insured
I Newly
N insured Uninsu
ured
(
(n=159) (n
n=117) (n=85))
1
Dirrect cost (Beefore claim)) Yes 4
40.3 40.2
4 41.2
No 5
59.7 59.8 58.8
Dirrect cost (Affter claim)2 Yes 2
23.3 40.2
4 41.2
No 7
76.7 59.8 58.8
Tottal cost (Beffore claim)3 Yes 4
44.8 39.8 43
No 5
55.2 60.2
6 57
4
Tottal cost (Aftter claim) Yes 2
26.5 39.8 43
No 7
73.5 60.2
6 57
1 2 2 2
χ (22) =0.025, p=0.9988 χ (2) =122.105, p=0.002
3 2 4 2
χ (22) =0.19, p=0.91 χ (2) =100.759, p=0.005

13
39
4.6.3 Number of Visits to Health Care Facilities and CHE
When ill, people avail health care services one or more times depending on the
effectiveness of the previous treatment and disease (Table 4.35). For the sample
individuals, higher number of visits to health facilities increased the occurrence of CHE.
An analysis of the number of visits and CHE before the claim for insured shows no
significant relationship. However, after the claim a positive association between CHE and
the number of visits exists. SSP did not have a visible impact on insured individuals who
had second or third visits to health facility. Insured paid a average visit of 1.3, newly
insured had 1.4 visits and uninsured had 1.2 visits but median was one for three groups
(p>0.05). Despite the absence of any difference in the number of visits among these
groups, number of insured with one visit had lower occurrence of CHE than uninsured
and newly insured individuals (at 10 % significance level).

Table 4.35 Catastrophic Health Expenses and Number of Visits to Health Facility
Insured Newly Uninsured
(after claim) insured
First visit1 (N=83) 32.5 33.7 33.7
Second visit2 (N=41) 31.7 43.9 24.4
Third visit3 (N=5) 40 60 -
1 2
χ (2, N=361) =5.481, p=0.065
2 2
χ (2, N=65) =4.195, p=0.123
3 2
χ (2, N=30) =0.625, p=0.429
(Figures represent only the cases of CHE in percentages of each subgroup)

4.6.4 Econometric Estimation of the Probability of CHE

Binary logistic regression analysis was used to estimate the probability of CHE
incurred by insured, newly insured and uninsured individuals. Individuals with SSP were
differentiated from those without it by assigning a code of ‘1’ for SSP insured, ‘2’ for
newly insured and ‘3’ for uninsured individuals. Individuals incurring CHE were
assigned a code of ‘1’ and those who did not have CHE had a code of ‘0’. Certain
variables expected to determine CHE were included in the analysis. Firstly, analysis

140
focused on various characteristics of individual (health insurance, types of illness,
duration and types of treatment and gender and age of ill person), household (job status of
head of the household, size of the household and income class) and community (area of
residence), after classifying individuals based on SSP membership status. Secondly,
binary logistic model estimated the probability of incurring CHE.
4.6.4.1 Characteristics of Individuals
Some of characteristics of the sick individuals that influence CHE were health
insurance, gender and age of ill persons, types of illness, duration of treatment and types
of treatment.
a. Age of Ill Persons
Higher the age of the ill persons, larger would be the felt need to seek care.
Hence, the likelihood to face CHE is high for the aged compared to younger persons.
Irrespective of the health insurance status, median age of ill persons who incurred CHE
was found to be 43 years, and for those without CHE was 41 years (p>0.05). Among the
persons who incurred CHE, insured individuals had a median age of 46 years, higher than
that of newly insured (42 years) and uninsured (40 years) individuals (p>0.05).
b. Gender of Ill Persons
Irrespective of health insurance status, a higher percent of men incurred CHE
compared to women (Table 4.36). An analysis of data before and after the claim for
insured individuals highlights that insurance reduced the incidence of CHE more for
female members than for male members. Almost 57 percent of men had CHE compared
to nearly 36 percent of women before the claim; this result was significant confirming a
difference in CHE for men and women. After insurance claim, there was a reduction in
CHE for men by 41 percent and for women by 43 percent. The difference in the episode
of CHE for men and women was still observed at 10 percent significance level after the
claim (p<0.1). The explanation for this lies in the indirect cost of illness. Men had to
forego work due to illness; hence, they had to borrow in higher proportion due to the low
income and absence from work. Even direct cost was high for them owing to prolonged
days of admission in the hospitals compared to women. Regardless of these inherent
141
differences among men and women, SSP did reduce the occurrence of CHE for both the
men and women.
Table 4.36 Catastrophic Health Expenditure and Gender of Ill Persons
CHE Male Female
1
Insured (before claim) No 43.2 64.1
Yes 56.8 35.9
Insured (after claim)2 No 66.7 79.5
Yes 33.4 20.5
Newly insured3 No 52.6 61
Yes 47.4 39
Uninsured4 No 55.8 57.1
Yes 44.2 42.9
1 2
χ (1, N=159)=6.971, p=0.008
2 2
χ (1, N=159)=3.310, p=0.069
3 2
χ (1, N=117)=1.586, p=0.453
4 2
χ (1, N=85)=0.015, p=0.992

Analysis on the cases of men would help us to know the impact of illness on CHE
among insured, newly insured and uninsured groups. Before the claim, 56.8 percent of
insured men had CHE compared to a lower proportion of newly insured (47.4%) and
uninsured men (44.2%) (p>0.05). After the claim, there was substantial decline in CHE
for insured men (33.4%) (p>0.05). Insured women did benefit from SSP compared to
their counterparts in newly insured and uninsured groups. They had lower occurrence of
CHE before the claim (p>0.05) than other two groups (Table 4.36). After the claim, there
was considerable drop in the event of CHE for insured women (reduced from 35.9% to
20.5%) that widened the disparity between them and newly insured/ uninsured women
(p<0.05). Insured women had relatively lower expenses and they benefited from SSP
compared to both insured men and newly insured/ uninsured women. Without the
consideration of SSP membership status, men (62.8%) had to face CHE compared to
women (37.2%) (p<0.1) (Table 4.44).

142
c. Types of Illness
Types of illness (acute, chronic and maternity related care) determine CHE. Intra-
group analysis before the claim shows that nearly 48 percent of insured faced CHE due to
acute illness, and 46.5 percent of individuals for chronic illness. After the claim, CHE
due to acute illness reduced to 26.1 percent and chronic illness decreased to 29.1 percent
(Table 4.37). Individuals experiencing CHE due to maternity got complete financial
protection since there was cent percent reduction in CHE after the claim. A clear pattern
emerged for the uninsured and newly insured individuals (p<0.05). Almost one thirds of
acutely ill newly insured and one fifths of uninsured individuals ended in CHE and three
fifths of chroniclly ill faced CHE. Even maternity caused 57.1 percent of newly insured
individuals to incur CHE whereas it was zero for insured. Unlike uninsured and newly
insured, chronic illnesses did not expose insured to undesirable consequences of CHE.
Table 4.37 Catastrophic Health Expenditure and Types of Illness
CHE Acute Chronic Maternity

Insured (Before claim)1 No 52.2 53.5 75

Yes 47.8 46.5 25
Insured (After claim)2 No 73.9 70.9 100
Yes 26.1 29.1 0
3
Newly insured No 78.9 37.7 42.9
Yes 21.1 62.3 57.1
Uninsured4 No 67.6 40.9 100
Yes 32.4 59.1 0
1 2
χ (2, N=159)=0.792, p=0.673
2 2
χ (2, N=159)=1.694, p=0.429
3 2
χ (2, N=117)=9.875, p=0.000
4 2
χ (2, N=85)=9.172, p=0.01

Inter-group analysis substantiated the positive impact of SSP on chronic illness

and resultant CHE. Before the claim from SSP, chronic illness caused CHE for the
studied individuals. After the claim, it was less for insured individuals (p<0.05) compared

143
to uninsured and newly insured individuals. Regardless of SSP, chronic illness caused
CHE for majority of the studied individuals (64.9%) (Table 4.44).
d. Types of Treatment
Health expenditure depends on the types of treatment availed by ill persons.
Inpatient treatment generally costs more that may cause CHE; however, there are
exceptional cases. Sometimes, outpatient treatment also causes CHE. At a glance, it
appears that inpatient treatment was associated with CHE for insured, uninsured, and
newly insured individuals (Table 4.38). Intra-group analysis before the claim shows that
almost half of individuals availing inpatient services incurred CHE. However, due to
insurance claim, only 28.7 percent of hospitalised insured had to suffer from CHE
whereas 53.7 percent of newly insured and 55 percent of uninsured inpatients had to deal
with CHE. There was a reduction by 57 percent in the proportion of individuals
experiencing CHE due to inpatient treatment (p<0.05). Not only insured used more of
inpatient services as analysed earlier (section 4.4.5) but also they paid less and had lower
incidence of CHE. Regardless of SSP, admitted individuals had higher percent of CHE
(90.8%) than outpatient (Table 4.44) (p<0.05). Hence, hospitalisation determines CHE.

Table 4.38 Catastrophic Health Expenditure and Treatment

CHE Treatment
OP IP
Insured (after claim)2 No 87.5 71.3
Yes 12.5 28.7
3
Newly insured No 85.7 46.3
Yes 14.3 53.7
Uninsured4 No 80 45
Yes 20 55
1 2
χ (1, N=159) =8.286, p=0.004
2 2
χ (1, N=159) =1.907, p=0.167
3 2
χ (1, N=117) =15.623, p=0.000
4 2
χ (1, N=85) =8.745, p=0.003

144
e. Duration of the Treatment
The duration of treatment is the number of days spent in hospital (even for
outpatient visits) by the ill persons. An increase in the duration of treatment would
increase the risk of CHE. The result shows a positive association between the duration of
treatment and CHE (Mann Whitney U test, p=0.00). Irrespective of SSP, the average
duration of treatment for CHE incurring individuals was 19 days which was higher than
that of non-CHE individuals (10days) (p<0.05).
4.6.4.2 Characteristics of the Household
Household characteristics such as job status of the head of the household, size of
the household and income class would determine CHE. Each of these characteristics were
analysed in detail to explicate their association with CHE.
a. Job Status of the Head of the Household
Irregularities in income and low earning capacity expose the households to higher
incidence of CHE. Thus, working as a labourer, or in informal sector increases the
possibility of CHE in contrast to employment in formal sector. Using the CHE based on
total cost, Table 4.39 exhibits the job status of head of households incurring CHE (before
the claim). Agriculturist head of the household in uninsured and newly insured
households had the lowest incidence of CHE.
Table 4.39 Association between Job Status and CHE
UN L SE FE IS A
Insured (N=159) 13.2 63.2 2.6 7.9 7.9 5.2
Newly insured (N=117) 9.3 65.1 9.3 4.7 9.3 2.3
Uninsured (N=85) 13.9 47.2 19.4 5.6 11.1 2.8
UN-Unemployed
L- Wage labour (beedi roller, daily labourer)
SE- Self employment
FE- Formal sector employment
IS-Salaried (informal sector)
A-Agriculture
(Only cases of CHE given in percentages)

145
 The study analysed the cases of unskilled labourer to explore the intensity of CHE
among the households with head working as unskilled labourer. SSP brought down the
incidence of CHE for labourers by 30 percent. Based on direct cost, almost 81 percent of
insured had CHE before the claim, which reduced to 57.1 percent after the claim (Table
4.40). Consequently, insured individuals with head of household working as labourer had
lower episodes of CHE compared to their counterparts in uninsured and newly insured
group.
Table 4.40 Head of the Household as Unskilled Labourer and CHE
Insured Newly insured Uninsured
(N=83) (N=74) (N=46)
Direct cost (before claim) 1 81.2 84 63
Direct cost (after claim) 2 57.1 84 63
Total cost (after claim) 3 29.8 42.7 47.8
1 2
χ (2, N=203)=8.108, p=0.017
2 2
χ (2, N=203)= 13.953, p=0.001
3 2
χ (2, N=203)=4.947, p=0.084
(Only cases of CHE given in percentages)

Irrespective of SSP, families with labourer head of the household had higher
percent of CHE compared to other job status (Table 4.44) (p>0.05).
b. Household Size
Larger families would have more income than smaller families. Hence, they
would incur less CHE than families with fewer members due to the pooling of resources.
Regardless of SSP, families that incurred CHE had lower mean size (4) than families
without CHE (average size 5) (p<0.05). Moreover, there was a significant difference in
the family size of insured (mean 5) and newly insured (mean 4) and uninsured
households (mean 4) (p<0.05) who experienced CHE.
c. Income Class of the Household
CHE may not be uniform across income classes. Analysis of intra-income class
(Table 4.41) reveals a significant difference in the CHE experienced by different income
classes with lower income classes (in Q1 and Q2) incurring CHE more than upper
146
income class individuals (Q3, Q4 and Q5). Three fourths of lowest income (Q1) insured
individuals faced CHE compared to three fifths of uninsured individuals from the same
class before the claim. A higher incidence of CHE for lowest income class and lower
incidence of CHE for high-income class highlights the importance of affordability of
care. Resource poor households usually have problems meeting the cost of treatment
associated with illness compared to resource rich counterparts. To the contrary, high-
income (Q5) individuals in newly insured households faced CHE in sizeable proportion
compared to Q5 individuals in other two groups. The excessive medical expenses related
to illnesses such as paralysis, heart attack, kidney failure and caner exposed six
individuals in high-income class in newly insured group to CHE.
With the claim, incidence of CHE on the Q1 insured individuals reduced by 28
percent, for Q2 by 52.2 percent, for Q3 by 50 percent, for Q4 by 57 percent and no effect
for Q5 individuals. Thus, reduction in CHE was larger for Q4 and Q2 individuals
compared to the poorest (Q1) individuals.
Table 4.41 Catastrophic Health Expenditure: Intra-Income Class Comparison
CHE Q1 Q2 Q3 Q4 Q5
Insured (Without claim)1 No 26.5 39.5 62.2 70.8 88.5
Yes 73.5 60.5 37.8 29.2 11.5
Insured (With claim)2 No 47.1 71.1 81.1 87.5 88.5
Yes 52.9 28.9 18.9 12.5 11.5
3
Newly insured No 40.7 47.8 54.2 81.5 62.5
Yes 59.3 52.2 45.8 18.5 37.5
Uninsured4 No 36.4 45 70.6 57.1 91.7
Yes 63.6 55 29.4 42.9 8.3
1 2
χ (4, N=159)=29.804, p=0.000
2 2
χ (4, N=159)=18.854, p=0.001
3 2
χ (4, N=117)=10.598, p=0.032
4 2
χ (4, N=85)=12.117, p=0.017

147
 An assessment of inter-income class impact of insurance involves a comparison
of CHE experienced by different income classes within each sub-group (with and without
CHE). The findings (Table 4.42) were in line with the expectation of a direct relationship
between CHE and income class when SSP claim was not included in the analysis. When
the claim data was included in the analysis, proportion of individuals experiencing CHE
increased for the lowest (Q1) and highest income quintiles (Q5) and decreased for
middle-income class individuals. So, Q2, Q3 and Q4 individuals benefited more than the
poorest or rich income class.
Table 4.42 Catastrophic Health Expenditure: Inter-Income Class Comparison
CHE Q1 Q2 Q3 Q4 Q5
Insured (Without claim)1 No 10.3 17.2 26.6 19.5 26.4
Yes 34.7 31.9 19.5 9.7 4.2
Insured (With claim)2 No 14.7 23.3 25 17.2 19.8
Yes 39.5 25.6 18.6 9.3 7
Newly insured3 No 16.4 16.5 19.4 32.8 14.9
Yes 32 24 22 10 12
Uninsured4 No 16.7 18.7 25 16.7 22.9
Yes 37.8 29.8 13.5 16.2 2.7
(Figures represent percentages of each subgroup across income quintiles)

These results show that SSP does not provide financial protection to the lowest
income individuals who need the greater benefits. However, SSP does include lower
income individuals in Q2 and Q3 class who benefited the most with more than 50 percent
reduction in CHE. In case of newly insured, those in Q5 had CHE more than Q4 whereas
uninsured Q4 individuals had higher incidence of CHE than those in Q3. These were the
exceptions to the result of direct association between CHE and income class. Irrespective
of SSP, individuals from Q1 had high percent of (36.6%) CHE compared to 12.2 percent
of Q4 and 7.7 percent of Q5 individuals (p<0.05) (Table 4.44).

148
4.6.4.3 Characteristics of the Community
Rural areas have less access to hospitals than urban and semi-urban areas in India
(Gumber 2001). They usually travel to nearby towns, which would increase the total
costs of care and CHE. Surprisingly, the chi square test did not show any significant
difference between the area of residence and CHE experienced by insured and newly
insured individuals (Table 4.43). However, higher percent of the uninsured individuals
living in rural areas (59.5%) experienced CHE than those in urban areas (15.8%).
Irrespective of SSP, semi-urban and rural area individuals tend to have higher incidence
of CHE compared to individuals in urban areas, but this finding was not significant
(Table 4.44).
Table 4.43 Catastrophic Health Expenditure and Area of Residence
CHE Urban Semi-urban Rural

Insured (Before claim)1 No 50 45 60

Yes 50 55 40
Insured (After claim)2 No 78.6 68.3 75.3
Yes 21.4 31.7 24.7
Newly insured3 No 61.9 57.4 57.1
Yes 38.1 42.6 42.9
Uninsured4 No 84.2 58.3 40.5
Yes 15.8 41.7 59.5
1 2 2 2
χ (1, N=159)=3.225, p=0.196 χ (1, N=159)=1.109, p=0.574
3 2 4 2
χ (1, N=117)=0.152, p=0.927 χ (1, N=85)=10.247, p=0.006

Another question related to community variable that drew our attention was the
distance to hospital. Since the transportation and other costs determine the total cost,
distance would contribute to CHE. The analysis revealed that mean distance to hospitals
was 3.3 km for newly insured and 2.3 km for insured and 2.4 km for the uninsured
individuals. For insured and uninsured individuals, distance to hospitals did not result in
CHE. Newly insured individuals staying far away from hospitals faced CHE (p<0.05).

149
The finding was not surprising as those living away from hospitals had to incur higher
transportation, lodging and food costs compared to those staying near the hospitals.
Regardless of insurance status, an association between distance to hospital and CHE has
been found (p<0.05). The average distance to hospitals for the individuals incurring CHE
was 2.8 km and for those without CHE was 2.5 km.
Table 4.44 Independent Variables Included in CHE Binary Logistic Regression Model
CHE No (N=231) Yes (N=130)
Types of illnessa
Acute illness 53.4 32.1
Chronic illness 42 64.9
Maternity 4.6 3
Gender of ill personb
Male 46.6 62.8
Female 53.4 37.2
Types of treatmentc
Outpatient 27.7 9.2
Inpatient 72.3 90.8
Job statusd
Unemployment 12.5 10.7
Labourer 54.2 60.2
Self employed 5 3.1
Agriculture 10 6.9
Formal sector 4.6 3.1
Salaried (informal 7.1 8.4
sector)
Income classe
Q1 15.8 36.6
Q2 21.7 25.2
Q3 22.5 18.3
Q4 21.3 12.2
Q5 18.7 7.7
Area of residencef
Urban 74.1 25.9
Semi-urban 63.4 36.6
Rural 62.8 37.2
Pearson chi square; a= 15.761, p>0.05; b=3.310, p>0.05; c=17.083, p<0.05; d=3.256, p>0.05; e=27.858, p>0.05; f=3.7, p>0.05

150
4.6.4.4 Results of Econometric Estimation
Binary logistic regression analysis was carried out to estimate the probability of
CHE. It was hypothesised that SSP decreased the cost of treatment; hence CHE would be
less for insured individuals. Table 4.45 displays the coding of variables in total cost
(model 4a), direct cost (model 4b), hospitalisation (model 4c), low income (model 4d)
and women model (model 4e). Model 4a considered CHE based on the total cost (direct
cost, other expenses and indirect cost) of treatment. Model 4b took CHE based on the
direct cost of treatment. Model 4c used the cases of hospitalisation and calculated CHE
based on the related direct cost while considering significant variables as estimated in
model 4b. Model 4d took cases of low income (Q1 and Q2) to know the significance of
the independent variables on CHE for poor and model 4e considered the cases of women
and significant independent variables estimated by model 4a. Types of treatment and
gender of ill person were coded into two dummy variables. Each SSP membership, types
of illness, size of household and area of residence were coded into three dummy
variables. The job status of the head of the household was coded into six dummy
variables. Age of ill persons and duration of the treatment were continuous variables.
Income quintiles were coded into five dummy variables.

151
 Table 4.45 Measurement of Independent Variables: CHE Models
Variables Model Model Model Model
4a, 4b 4c 4d 4e
Individual characteristics
Health insurance
SSP insured=1 (reference)
Newly insured=2
Uninsured=3
Gender of ill person =1 if male, 0 if female
(reference)
Types of illness =1 if chronic, 2 if maternity, 0
if acute (reference)
Types of treatment =1if inpatient, 0 if outpatient
Household characteristics
Occupation of household head
1= Labourer if primary occupation is unskilled
worker being paid daily wage
2= Business if engaged in self-employment
3= Agriculture if farmer including dairy farmer
4=Salaried in informal sector if unskilled
worker being paid monthly in unorganised
sector
5=Formal sector if skilled worker employed in
organised sector on a salary basis
0=Unemployed/not able (reference)
Income quintile
1=Q1, 2=Q2, 3=Q3, 4=Q4, 5=Q5 (reference)
Community characteristics
Area of residence
1= Urban if individual lives in urban area
2= Semi-urban if individual lives in semi-urban
area
3=Rural if individual lives in rural areas
(reference)

A strong evidence for insured individuals being less likely to incur CHE due to
illness compared to uninsured and newly insured individuals was found (Table 4.46).
152
The Odds Ratio (OR) for newly insured and uninsured individuals was significantly
larger than 1, which implied that newly insured and uninsured individuals were more
likely to experience CHE than insured individuals. The odds of CHE compared to not
incurring CHE was high for newly insured (OR 3.725) and uninsured (OR 4.738)
individuals compared to insured individuals. Hence, the study hypothesis (H5) that SSP
decreases the likelihood of CHE for insured individuals compared to both newly insured
and uninsured individuals is proven. Chronically ill individuals had higher likelihood of
facing CHE compared to those with acute illness (OR 2.975). For each day of
hospitalisation, the likelihood of CHE increased by 1.019 times. Outpatient treatment
decreased the likelihood of CHE compared to hospitalisation (OR 0.193). The individuals
living in urban areas had lower likelihood of CHE (OR 0.467) compared to those in rural
areas, at 10 percent significance level. Income was a determinant of CHE with lower
income quintile individuals had higher probability of incurring CHE than high income
quintile. The odds of experiencing CHE compared to not facing it were high for Q1 (OR
9.195) and Q2 individuals (OR 3.102) compared to highest income quintile (Q5)
individuals. SSP membership, chronic illness, longer days of treatment, inpatient
treatment, lower income class and rural area of residence determines the likelihood of
CHE. Age and gender of ill persons, job status of the head of the household and size of
the household were not associated with the probability of CHE.
A number of specification and diagnostic tests checked the robustness of the
model; especially the possible endogeneity has been tested using Durbin-Wu-Hausman
test. In this model, health insurance was found to be exogenous with prob(χ2) =0.867.
The model fit was assessed using the omnibus test of model coefficients, Hosmer and
Lemeshow test, -2 log likelihood ratio, Cox and Snell R square and Nagelkerke R square.
The results of these tests showed that the model fits well and 75.1 percent of cases were
correctly predicted by the model. Residual analysis (specifically Cook’s Distance
statistic) showed no outliers.
Model 4b considered the direct cost of treatment for calculating CHE. The logic
was that SSP coverage was limited to the direct cost; hence, certain independent variables
153
such as area of residence were not significant (Table 4.47). Again, newly insured (OR
5.208) and uninsured individuals (OR 5.290) had higher likelihood of incurring CHE
compared to insured individuals, controlling for other socio-economic variables.
Individuals with inpatient treatment were almost 5 times more likely to face CHE
compared to those with outpatient treatment. For each day of admission, sick individuals
were 1.020 times more likely to incur CHE. Individuals suffering from chronic illness
were 3.011 times likely to experience CHE compared to individuals with acute illness.
The odds of having CHE compared to not having it were high if the individuals belonged
to Q1 (OR 12.3), Q2 (OR 3.914) and Q3 (OR 3.238) compared to Q5 individuals. Age
and gender of ill person, job status of the head of household, household size and area of
residence were not significantly associated with probability of CHE. Thus, health
insurance, types of treatment, days spent in the hospital, types of illness and income class
were found to significantly determine CHE. The results of these tests showed that the
model fits well and 76.5 percent of cases were correctly predicted by the model.

154
 Table 4.46 Probability of Catastrophic Health Expenditure: Estimation of Model 4a

B S.E. Wald df Sig. Exp 95% C.I.Exp (B)

(B) Lower Upper
Health insurance (base= SSP insured) 24.214 2 .000
Newly insured 1.315 .327 16.221 1 .000 3.725 1.964 7.065
Uninsured 1.556 .355 19.206 1 .000 4.738 2.363 9.500
Age of ill person -.005 .008 .354 1 .552 .995 .980 1.011
Gender of ill (base=female) Male .447 .272 2.697 1 .101 1.564 .917 2.668
Types of illness (base: Acute) 14.854 2 .001
Chronic 1.090 .285 14.685 1 .000 2.975 1.703 5.196
Maternity .105 .709 .022 1 .882 1.111 .277 4.460
Types of treatment (base: Inpatient) -1.643 .388 17.896 1 .000 .193 .090 .414
Outpatient
Duration of treatment .018 .008 5.818 1 .016 1.019 1.003 1.034
Job of head (base= Unemployed) .786 5 .978
Labourer -.058 .355 .027 1 .870 .943 .471 1.891
Business -.262 .798 .108 1 .743 .770 .161 3.678
Agriculture -.105 .603 .031 1 .861 .900 .276 2.934
Salaried (informal sector) -.284 .584 .237 1 .627 .753 .240 2.365
Formal sector .385 .749 .264 1 .607 1.470 .338 6.382
Household size (base:1-3) .870 2 .647
4-6 -.526 .578 .829 1 .362 .591 .190 1.834
7and above -.360 .533 .456 1 .500 .698 .246 1.983
Income quintile (base=Q5) 23.700 4 .000
Q1 2.219 .533 17.345 1 .000 9.195 3.237 26.124
Q2 1.132 .512 4.884 1 .027 3.102 1.137 8.467
Q3 .896 .518 2.998 1 .083 2.450 .888 6.757
Q4 .393 .523 .566 1 .452 1.482 .532 4.128
Area of residence (base=Rural area) 3.259 2 .196
Urban -.762 .427 3.184 1 .074 .467 .202 1.078
Semi-urban -.071 .299 .056 1 .812 .931 .518 1.674
Constant -2.535 .786 10.397 1 .001 .079
Number of observations 361
Omnibus test model coefficient:Pearson chi square =104.506, df =20, p=0.000; -2 log likelihood = 361.135;
Cox and Snell R squared= 0.253; Negelkerke R squared= 0.348;
Hosmer and Lemeshow Pearson chi square =5.749, df =8, p=0.675

155
Table 4.47 Estimation of Model 4b: Probability of Catastrophic Health Expenditure

B S.E. Wald df Sig. Exp(B) 95% C.I.Exp (B)

Lower Upper
Health insurance 29.426 2 .000
(base= SSP insured)
Newly insured 1.650 .340 23.507 1 .000 5.208 2.673 10.148
Uninsured 1.666 .369 20.421 1 .000 5.290 2.569 10.897
Age of ill person -.007 .008 .735 1 .391 .993 .977 1.009
Gender of ill person
(base=Female) Male .233 .284 .670 1 .413 1.262 .723 2.202
Types of illness (base: 14.654 2 .001
Acute)
Chronic 1.102 .302 13.336 1 .000 3.011 1.666 5.441
Maternity -.293 .778 .142 1 .707 .746 .162 3.429
Types of treatment .
(base: outpatient) Inpatient 1.645 420 15.368 1 .000 5.183 2.277 11.800
Duration of treatment .020 .008 6.561 1 .010 1.020 1.005 1.035
Occupation of household 1.943 6 .925
head (base= Unemployed)
Labourer -.312 .429 .527 1 .468 .732 .316 1.699
Business -.419 .846 .246 1 .620 .658 .125 3.449
Agriculture .119 .653 .033 1 .855 1.127 .313 4.050
Salaried in informal sector -.147 .623 .056 1 .813 .863 .254 2.929
Formal sector -.623 .950 .430 1 .512 .537 .083 3.453
Household size .022 .087 .064 1 .801 1.022 .862 1.213
Income quintile (base=Q5) 27.125 4 .000
Q1 2.510 .576 18.994 1 .000 12.300 3.979 38.022
Q2 1.365 .553 6.085 1 .014 3.914 1.324 11.573
Q3 1.175 .551 4.539 1 .033 3.238 1.099 9.542
Q4 .357 .572 .389 1 .533 1.429 .466 4.386
Area of residence 2.413 2 .299
(base=Rural area)
Urban -.700 .452 2.398 1 .121 .497 .205 1.204
Semi-urban -.171 .310 .305 1 .581 .843 .459 1.547
Constant -4.79 1.036 21.376 1 .000 .008
Number of observations 361
Omnibus test model coefficient:Pearson chi square =114.589, df =21, p=0.000; -2 log likelihood = 343.103
Cox & Snell R squared= 0.272; Negelkerke R squared=0.379;
Hosmer & Lemeshow Pearson chi square =6.491, df =8 p=0.592
(Dependent variable: Catastrophic health expenditure; 1=yes)

156
 Models 4c considered the cases of hospitalised individuals (Table 4.48). As SSP
provides the inpatient coverage, knowledge on the impact of SSP in reducing CHE for
hospitalised individuals would substantiate the earlier findings (model 4a). There were
285 cases of hospitalisation in the study. CHE due to admission in the hospital was
highest for newly insured adults (91.6%) than uninsured (80%) and insured (54.5%)
individuals. Chronic illness had significant impact on CHE for 81.3 percent of uninsured
and 95 percent of newly insured and was lowest for insured individuals (56.4%). The
days of treatment were highest for insured (average of 19 days) and lowest for uninsured
(average of 12 days). Newly insured had average days of admission of 15 days. There
was a discernible pattern in the incidence of CHE among the income classes. High-
income classes (Q5) had lower incidence of CHE (16.7 % for insured, 17.1 % for newly
insured and 20.8% for uninsured). It was highest for Q1 (23.1 % for insured, 21.1 % for
newly insured and 22.9 % for uninsured) and Q2 (24.6 % for insured, 25 %for newly
insured and 22.9 % for uninsured) income class.
Newly insured (OR 10.899) and uninsured (OR 3.810) had higher likelihood of
CHE compared to insured individual. For each additional day of admission, the
likelihood of CHE increased by a factor of 1.028. The results of these tests showed that
the model fits well and 72.4 percent of cases were correctly predicted by the model. Thus,
lack of health insurance and longer duration of treatment increased the likelihood of CHE
due to hospitalisation.

157
 Table 4.48 Estimation of Model 4c: Probability of Catastrophic Health Expenditure
B S.E. Wald Df Sig. Exp(B) 95%C.I.for
Exp(B)
Lower Upper
Health insurance(base= 35.057 2 .000
SSP insured)
Newly insured 2.389 .445 28.754 1 .000 10.899 4.552 26.095
Uninsured 1.338 .373 12.838 1 .000 3.810 1.833 7.920
Types of illness (base: 1.181 2 .554
Acute)
Chronic .097 .299 .104 1 .747 1.101 .612 1.981
Maternity -.644 .681 .896 1 .344 .525 .138 1.994
Duration of treatment .028 .012 5.404 1 .020 1.028 1.004 1.052
Income quintile (base=Q1) 2.668 4 .615
Q2 -.569 .465 1.500 1 .221 .566 .227 1.408
Q3 -.160 .461 .121 1 .728 .852 .345 2.104
Q4 -.602 .480 1.572 1 .210 .548 .214 1.404
Q5 -.204 .513 .158 1 .691 .816 .298 2.229
Constant .036 .412 .008 1 .930 1.037
Number of observations 285
Omnibus test model coefficient:Pearson chi square =54.630, df =9, p=0.000: -2 log likelihood = 291.680
Cox and Snell R squared= 0.174; Negelkerke R squared= 0.248
Hosmer and Lemeshow Pearson chi square =3.583, df =8, p=0.893
(Dependent variable: Catastrophic health expenditure; 1=yes)

Model 4d (Table 4.49) took the cases of low income insured, newly insured and
uninsured individuals belonging to Q1 and Q2 income class. Only significant variables
estimated by the model 4a were included as independent variables in the regression
analysis. Among these income classes, a higher percent of uninsured poor (59.5%) had
episodes of CHE, followed by newly insured (54%) and insured (38.9%) poor. Chronic
illness propelled CHE in 60.7 percent of insured, 59.3 percent of newly insured and 60
percent of uninsured poor individuals. Hospitalisation was the cause of CHE for 92.9
percent of insured poor, 85.2 percent of newly insured and 84 percent of uninsured poor.
Moreover, insured individual had highest average days spent in the hospital (18 days)
compared to newly insured (8 days) and uninsured (7 days).

158
 The odds of incidence of CHE was high for newly insured (OR 3.103) and
uninsured (OR 4.813) compared to insured individuals. Chronic illness increased the
likelihood of CHE by 2.090 times than acute illness. Hospitalisation was highly
associated with the probability of CHE compared to the outpatient treatment. For every
additional day spent in the hospital, odds of incurring CHE rather than not having it
increased by a factor 1.032. Thus, the episode of CHE was higher if the person had
chronic illness, inpatient treatment and longer duration of treatment and SSP reduced the
incidence of CHE. Hosmer and Lemeshow test value of 0.710 indicated that the model
correctly predicts 69.5 percent of the cases.
Table 4.49 Estimation of Model 4d: Probability of Catastrophic Health Expenditure

B S.E. Wald Df Sig. Exp(B) 95%C.I.for

Exp(B)
Lower Upper

Health insurance (base= 12.573 2 .002

SSP insured)
Newly insured 1.132 .431 6.897 1 .009 3.103 1.333 7.222
Uninsured 1.571 .471 11.134 1 .001 4.813 1.912 12.113
Types of illness (base: 4.963 2 .084
Acute)
Chronic .737 .356 4.294 1 .038 2.090 1.041 4.199
Maternity 1.305 1.172 1.239 1 .266 3.687 .371 36.685
Types of treatment (base: 1.060 .484 4.797 1 .029 2.888 1.118 7.458
outpatient)
Inpatient
Duration of treatment .032 .013 5.611 1 .018 1.032 1.005 1.060
Constant -2.459 .597 16.985 1 .000 .086
Number of observations 164
Omnibus test model coefficient: Pearson chi square =28.071, df =6, p=0.000; -2 log likelihood = 199.184
Cox and Snell R squared= 0.157; Negelkerke R squared= 0.210
Hosmer and Lemeshow Pearson chi square =5.435, df =8, p=0.710
(Dependent variable: Catastrophic health expenditure; 1=yes)

159
 The impact of SSP on the CHE faced by insured women would help us to draw
conclusion on the gender equity of MHIs (model 4e in Table 4.50). There were 179
women in studied groups, of which 56 had experienced CHE due to medical treatment.
Higher percent of uninsured women (42.9%) had episodes of CHE, followed by newly
insured (37.3%) and insured (20.5%) women. Chronic illness had resulted in CHE than
acute illness in all the women. Nearly 63 percent of insured, 68.2 percent of newly
insured and 83.8 percent of uninsured chronically ill women had CHE. Inpatient
treatment in 93.8 percent of insured women, 86.4 percent of newly insured and 83.3
percent of uninsured women caused CHE. Inpatient insured women spent an average of
13 days, higher than that of newly insured (10 days) and uninsured (6 days) in the
hospitals. Women in Q1 income class had higher percent of CHE in insured (31.3%) and
uninsured (38.9%) groups. Q3 income class in newly insured individuals had highest
incidence of CHE (36.4%) followed by Q1 class (27.3%).
Regression results support the hypothesis of positive impact of SSP on CHE. The
probability of incidence of CHE was higher for newly insured (OR 5.115) and uninsured
(OR 6.851) women compared to insured women. The duration of treatment increased the
probability of CHE by a factor of 1.03 (at 10 % significance). Hospitalisation increased
the likelihood of incurring CHE by 5.042 times compared to outpatient treatment. The
model is robust with 76.5 percent of the cases correctly predicted.

160
Table 4.50 Estimated results of Model 4e: Probability of Catastrophic Health Expenditure

B S.E. Wald Df Sig. Exp(B) 95%C.I.for

Exp(B)
Lower Upper
Health insurance (base= 16.095 2 .000
SSP insured)
Newly insured 1.632 .488 11.164 1 .001 5.115 1.963 13.323
Uninsured 1.924 .528 13.279 1 .000 6.851 2.434 19.288
Types of illness (base: 7.373 2 .025
Chronic)
Acute -.187 .799 .055 1 .815 .829 .173 3.969
Maternity .955 .768 1.545 1 .214 2.598 .577 11.702
Types of treatment
(base: outpatient) 1.618 .562 8.291 1 .004 5.042 1.676 15.165
Inpatient
Duration of treatment .029 .015 3.612 1 .057 1.030 .999 1.061
Income quintile 12.526 4 .014
(base=Q5)
Q1 1.363 .692 3.878 1 .049 3.907 1.006 15.167
Q2 .597 .674 .782 1 .376 1.816 .484 6.810
Q3 .527 .709 .551 1 .458 1.693 .422 6.800
Q4 -1.08 .832 1.714 1 .190 .337 .066 1.718
Constant -4.43 1.106 16.055 1 .000 .012
Number of observations 179
Omnibus test model coefficient:Pearson chi square =52.722, df =10, p=0.000; -2 log likelihood = 169.725
Cox and Snell R squared= 0.255; Negelkerke R squared= 0.359
Hosmer and Lemeshow Pearson chi square =8.45, df =8, p=0.390
(Dependent variable: Catastrophic health expenditure; 1=yes)

4.6.5 Summary
Our results do confirm that SSP did successfully reduce the incidence of CHE for
insured members. Before the claim from SSP, there was no difference in CHE of insured,
newly insured and uninsured individuals. After the claim, analysis revealed a drastic
reduction in CHE for insured individuals. Logistic regression analysis on the household
survey data confirmed the hypothesis of the study (H5) that uninsured and newly insured
individuals had higher incidence of CHE compared to insured individuals. The models
with different specifications substantiate the findings of the basic model 4a and confirm

161
the positive impacct of SSP (H
H5) on insu
ured individduals. A nu
umber of sppecification and
diagnosstic tests prooved them robust
r and in
nsurance vaariable to bee exogenouss.
The most striking
s resu
ult to emerg
ge from thee models is that MHI ddoes reducee the
CHE fo
or insured individualss. SSP undoubtedly prrovided fin
nancial prottection sincce it
reduced
d CHE for in
nsured indiv owever, the effect of inssurance wass partial, as one
viduals. Ho
fourth of
o individuaals still had to face CHE
E even with
h insurance. Certain design featurees of
heme resulted in parrtial financial protectiion. Exclusion of ceertain diseaases,
the sch
outpatieent treatmen
nt and a ceeiling of 5
5000 expossed certain individuals to CHE allbeit
there were payments orr deductibless imposed in SSP. Sincce SSP did not
w no co-p n have access
to any financial aiid from extternal dono
ors and the programmee was incurrring huge loss
since th n the beneffit package does not arrise.
he last few years, the question off increase in
Given these
t limitattions, SSP did
d reduce the
t incidencce of CHE for
f hospitaliised individ
duals
to a con
nsiderable extent.
e
Model 4a, 4b and 4d predicted the
t chronic illness to be
b a determ
minant of CHE.
C
Chronicc illness nott only increeases utilisaation and OO
OPE but also exposes the individ
duals
E. A possiblle explanatiion for thiss result is thhe recurrence of illnesss that warrrants
to CHE
frequen
nt access to health serv
vices and th
hereby OOPE and CH
HE. Chronicc illness, iff not
treated would threaaten the lon
ngevity of sick person or
o cause dissability thatt would sho
orten
producttive years off life. Howeever, SSP reeduced CHE
E for a sizeaable numbeer of individ
duals
with ch
hronic illnesss. Moreoveer, CHE rellated to acu
ute and mateernity was lless for insu
ured
comparred to newlly insured and uninsu
ured individ
duals. The duration off treatment did
influencce CHE, with
w longer duration po
ositively reesulted in CHE.
C Even hospitalisaation
increaseed the probability of CHE.
C All th
he models (44a to 4e) with
w differen
nt specificattions
supportt the days off treatment and hospitaalisation as important
i d
determinants
s of CHE.
Gender off ill person
ns and areaa of the reesidence were
w not prredicted as the
determiinant of CH
HE, except in
i Model 4aa. The area of residencce influencees the total cost
of care,, especially indirect an
nd other exp
penses. Thee models co
onsidered thhe direct cosst of
care (m
model 4b, 4c); hence, area of the
t residennce proved to be insiignificant. The
possibillity of CHE
E was higherr if the persson was malle rather thaan female annd insured men
m
162
had higher incidence of
o CHE eveen after the claim. Men
n had to inccur higher indirect costt
due to abseence from work
w (sectio
on 4.5.4.1c) and they had
h to borrow in higherr proportion
n
compared to
t women to
t meet thee daily needs. Even th
he direct coost was hig
gh for them
m
owing to prrolonged daays of admisssion in hosspitals comp
pared to woomen. Hencee, gender off
ill person w
was a significant deterrminant in the
t model 4a
4 based onn total costt, but not in
n
other modeels. Horizon
ntal equity in CHE w
was present since insurred women
n had lowerr
incidence of CHE co
ompared to
o uninsured
d and new
wly insured
d women (model
( 4e).
Moreover, SSP reduceed the incid
dence of CH
HE more fo
or female m
members thaan for malee
members.
dels 4a and 4b predicteed a direct relationship
Mod r between thhe income of the family
y
and CHE with lowerr incidence for high-income classs. Howeveer, when th
he cases off
hospitalisedd individuaals and wom
men were aanalysed, it failed to bbe a predictor of CHE.
One of the objectives of SSP is to
o promote equity
e in fin
nancial proteection. Nev
vertheless, itt
failed to acchieve this objective since the pooorest did not
n get highher financial protection
n
than other income classes due to certain design feattures of thee scheme. Subsequentt
analysis rev
vealed that poorest (Q
Q1) individu
uals utilised
d outpatientt services (n
not covered
d
by SSP), haad illness th
hat required
d costly treaatment such
h as paralysis, heart atttack, kidney
y
failure and cancer and < 14000). Hence, effeective proteection given
d had lowesst income (< n
to low income class was lower than otherr income class
c p. However,
in insuured group
poorest insu
ured individ
duals had lo
ower inciden
nce of CHE
E compared to their cou
unterparts in
n
uninsured aand newly insured indiividuals (moodel 4d) thaat reveals hoorizontal eq
quity impactt
of SSP. To
o sum up, the
t evidencce from this study sug
ggests lack of verticall equity butt
presence of horizontaal equity in the incidence of CHE. On the issue of eq
quity in thee
distributionn of claim, SSP had better
b impacct on the hiigh (Q4) annd lower in
ncome (Q2))
individuals than the po
oorest (Q1)..
Theere was a po
ositive link
k between thhe number of visits maade to a health facility
y
and incidennce of CHE
E. Higher th o visits, laarger is the chances off CHE sincee
he number of
the individ
dual has to
o incur ad
dditional ex
xpenses. Bu
urden of C
CHE reducced for thee
individuals who visiteed health faccility once compared to
t the individuals who had two orr
16
63
three visits. Again, certain features of SSP as highlighted above resulted in the absence of
positive impact on CHE owing to two or three visits.
The present results are significant in at least two major respects. SSP reduces the
cost of care measured in absolute (OOPE) and relative terms (CHE) and it increases
utilisation of health services in private hospitals. However, the ignorance of the risk
coping strategies that compensate lack of health insurance is a major problem of this kind
of analysis. Hence, there is an increasing concern that any evaluation of financial
protection should consider the impact of MHI on the risk coping strategies of the
households. The next section assesses the impact of SSP on the risk coping strategies of
the households.

164
4.7 IMPACT OF SAMPOORNA SURAKSHA PROGRAMME ON THE RISK
COPING STRATEGIES
4.7.1 Introduction
Illness is a major risk factor that jeopardizes the normal life of people with long-
term negative effect. When faced with illness, the households usually seek treatment
rather than postpone the treatment, especially when illness is severe or impairs normal
life. Iatrogenic poverty resulting from illness is transient if the affected household has
certain ex-ante and ex-post measures to tackle the health risks. Ex-ante strategies include
health insurance, ex-post strategies involves self-insurance and survival strategies. Self-
insurance can occur in two ways; i) use of savings ii) informal risk sharing arrangements
within family, friends or neighbours for consumption smoothing during the episode of
illness. Survival strategies involve the sacrifice of human capital (sending additional
household member for work), sale of the productive assets, and borrowing from the
banks and charity in the times of health crisis (Dercon 2002). However, some of these
strategies have adverse impact on the future consumption as the household would have
less income due to the sale of productive assets and the repayment of loan.
Financial protection provided by the MHI reduces the reliance on the risk coping
strategies such as borrowing, sale of assets and the use of savings. Hence, this study
considered the impact of SSP on the risk coping strategies of insured members. SSP
membership has resulted in less OOPE and lower incidence of CHE. This positive
impact would lead to less reliance on the risk coping strategies such as borrowing, sale of
assets and savings. The research question was whether SSP reduced the reliance on other
risk coping strategies for insured individuals compared to uninsured and newly insured
individuals. Health insurance reduces the negative consequences of such strategies by
meeting a major part of the total medical cost and stabilises the expenditure that would
fluctuate due to illness. Hence, the study hypothesised that SSP reduces reliance on other
risk coping strategies for insured individuals compared to newly insured and uninsured
individuals. The hypothesis driven analysis used binary logistic regression model to know
the impact of SSP on the incidence of borrowing and the use of savings. We know from
165
the previous sections that certain design features of SSP gives partial protection to
insured individuals. Hence, insured individuals rely on risk coping strategies to some
extent. Nevertheless, amount of funds mobilised from these strategies would be less as
SSP claims would bring down the cost of treatment. Thus, the study hypothesised insured
individuals to mobilise fewer funds compared to uninsured and newly insured
individuals. To test this hypothesis, the amount of funds mobilised from borrowing and
savings was analysed using multiple linear regression analysis.
Firstly, an analysis on the availability of money to meet medical expenses gives
information on the need for risk coping strategies. If the funds were available to meet cost
of medical treatment, the necessity to mobilise money from various sources does not
arise. Since SSP meets the direct cost of hospitalisation, insured individuals would have
more funds compared to newly insured and uninsured individuals. Second, risk coping
strategies used by individuals were elucidated. As the cost of treatment would be less for
insured individuals, they rely less on the other risk coping strategies compared to newly
insured and uninsured individuals. Thirdly, determinants of borrowing and savings were
estimated.
a. Binary logistic regression equation to determine the incidence of borrowing is as
follows;
Prob (Borrowi| HCAi>0)=β0+ β1Mx+ β2Xy + ε
{1 if Borrow│HCA >0, 0 otherwise}
Borrowi│HCAi isthe probability of borrowing conditional on health care action.
Mx is the dummy variable for health insurance status (SSP) and Xy is a set of covariates
that determine borrowing. Model specification was changed to corroborate the findings.
b. Binary logistic regression equation to determine the use of savings is as follows;
Prob (Savingsi| HCAi>0)=β0+β1Mx+ β2Xy + ε
{1 if Savings used│HCA >0, 0 otherwise}
Savingsi│HCAi is probability of use of savings conditional on health care action.
Mx is the dummy variable for health insurance status (SSP) and Xy is a set of covariates
that determine savings. The model fit was assessed using omnibus test of model
166
coefficients, Hosmer and Lemeshow test, -2 log likelihood ratio, Cox and Snell R square
and Nagelkerke R square.
c. The determinant of the amount of borrowing were estimated by using multiple linear
regression model.
Log (Amount borrowi│HCAi)= β0+β1Mx + β2Xy+ ε
Amount borrowi │HCAi is amount of borrowing conditional on health care action.
Mx represents the mode of payment (SSP). Xy is a set of variables that determines the
amount of borrowing.
d. The determinant of the amount of savings were estimated by using multiple linear
regression model.
Log (Amount of savingsi│HCAi)= β0+β1Mx + β2Xy+ ε
Amount of savingsi │HCAi is amount of savings conditional on health care
action. Mx represents the mode of payment (SSP). Xy is a set of variables that determine
the amount of savings. These models were subjected to a number of tests namely
variance inflation factor, correlation matrix, Cook’s D statistic and Dfits statistic.
4.7.2 Access to Self-Finance during Health Crisis
Available funds in the family determine the ability to pay medical bills without
resorting to risk coping methods. If the ill person is a minor, old or not working, family
income acts as the source of funds to pay for the bills. Since insured can get the benefit
from SSP, they could afford medical treatment compared to newly insured and uninsured
individuals. The current study shows that a higher percent of insured had the financial
resources to meet medical expenses compared to newly insured and uninsured individuals
(Table 4.51). Nearly thirty six percent of insured met medical expenses without resorting
to negative risk coping strategies. Just about one fifths of newly insured and 27 percent of
uninsured could afford the treatment without borrowing, using the savings or sale of
assets (p=0.019).

167
 Table 4.51 Availability of Money to Pay Medical Expenses
Insured Newly Uninsured
insured
Yes 57 (35.8) 24 (20.5) 23 (27.1)
No 102(64.2) 93 (79.5) 62 (72.9)
χ2 (2, N=371)=7.896, p=0.019
(Percentage given in bracket)

4.7.3 Risk Coping Strategies during Health Crisis

Risk coping strategies adopted by the families to meet medical expenses were
borrowing, use of the savings and sale of assets or valuables and other household assets.
Many households used two or three strategies to meet annual medical expenses. Ex-post
strategies such as low return and low risk economic activities and lower consumption
spending were absent. Predominantly, households used the asset-based strategies such as
the sale of assets, utilisation of the savings, borrowing or health insurance (by insured).
Sale of assets mainly consisted of crop or valuables like jewellery or two wheeler
vehicles. There was no change observed in the portfolio of income sources like engaging
school going children and women in income generating activities and sending additional
members of the family to the labour market.
When the analysis was carried out considering both ex-post and ex-ante strategies,
it was found that health insurance (65.4%) and borrowing (57.2 %) was the major option
utilised by a large percentage of insured individuals followed by savings (32.7%) and
lastly the sale of assets (5.6%). Borrowing was opted by a higher percentage of newly
insured individuals (79.5%), savings was the second most used alternative (24.7 %) and
sale of assets was the least opted option (2.6%). Uninsured individuals too had similar
pattern (Table 4.52). A lower proportion of insured borrowed (57.2%) compared to
uninsured and newly insured groups. Assets sale is time consuming and less liquid
especially in rural areas; it was the last option exercised by the individuals faced with
health shock. Seven percent of studied individuals used both the strategies of borrowing
and savings to meet the treatment costs. Moreover, 64 percent of insured members
borrowed even to pay the premium amount to SSP.

168
 Tablle 4.52 Sourrce of Finan
ncial Resourrces during Crisis - Exx Ante Strateegies
Insured nsured
Newly in U
Uninsured
(n=159) (n=11
17) (n=85)
Borrowing
g 91 (57.2) 93 (79
9.5) 664 (75.2)
Sale of assets 9 (5.6) 3 (2.6) 6 (7)
Savings 52 (32.7) 29 (24
4.7) 330 (35.3)
Health inssurance 104 0 0
χ2 (4, N=361) =17.7773, p=.000

From
m the data in the Figurre 4.6, it is apparent th
hat there waas no differeence among
g
uninsured, newly insu
ured and in
nsured indivviduals regaarding the reliance on
n other risk
k
coping straategies such as borrowiing, sale of assets and use 0.05). Thus,
u of the savings (p>0
the study aaccepts the null
n hypoth SP does not reduce relliance on neegative risk
hesis that SS k
coping strattegies (H6).

79.49
75.29
80
70 57.23
60
Percentage

50 Borrowing
32.70 35.29
40
24.79 Sale of
30
assets
20 5.66
7.06
2
2.56 Savings
10
0
Insured Newly insured
i Unninsured

F
Figure 4.6 Risk g Strategies of Sample Individuals
R Coping I - Ex Post Strategies
S

Furtther analysiis of the rissk coping sttrategies in terms of thhe amount of

o resourcess
mobilised through
t borrrowing, saavings and sale of asssets shows an interestiing picture.
Insured inddividuals’ borrowed
b lo
ower amouunt (median
n of 50000) compared to newly
y
insured andd uninsured
d ( 6000 eaach) groupss (Table 4.5
53). The am
mount of saavings used
d
was less ffor insured
d ( 1500) compared to uninsurred ( 20000) and new
wly insured
d
individuals ( 3000). Therefore,
T the future consumptio
on or incom
me generatin
ng capacity
y
was impairred for unin
nsured and newly insurred familiess since theyy have to reepay higherr
16
69
amountt of loan. Thhe amount of
o assets so
old by insureed individuaals (mediann of 8500) was
lower th
han that of newly
n insurred ( 10250
0). Howeveer, it was hig
gher than thhat of uninsu
ured
( 4000) individualls. These reesults were not statisticcally signifi
ficant suggeesting that th
here
was no significant difference in the amo
ount mobilissed from th
he sale of asssets or sav
vings
by insurred, newly insured and
d uninsured individualss. Insured mobilised
m lesss funds in total
t
(median
n of 3000) compared
d to newly insured
i and
d uninsured
d individuals (median of
o
5000 eaach) (p<0.05
5).
Tablee 4.53 Amou
unt of Moneey Mobiliseed to Pay for Health Caare
Insu
ured ( ) Newly insured
i ( ) Uninsurred ( )
Borrowinga
B 131
130 (31974)) 20314
4 (37105) 16830 (29639)
Savingsb 28
840 (3193) 4713
3 (7810) 2900 (3002)
Sale of assetsc 284
438 (34928)) 17625
5 (23507) 4417 (3720)
T
Total amoun ntd 964
45 (30129) 17784
4 (36168) 13478 (26429)
a
Krusskal Wallis χ2 (2,
( N=361) =20.983, p =0.00
b
Kruskal Wallis χ2 (2,
( N=361) =2.8881, p =0.237
c
Krusskal Wallis χ2 (2,
( N=361) =2.0059, p =0.357
d
Kruskal Wallis χ2 (2,
( N=361) =15.843, p =0.00
Meann amount (standdard deviation in
i bracket)

Above find
ding drew th
he attention to study eaach risk copiing strategyy separately and
exploree the differeences in borrowing, usse of saving
gs and salee of assets aamong insu
ured,
newly insured and uninsured individuals.
i .
4.7.4 Relationship
R p between Borrowing
B h Insurance Status
and Health
Borrowing was a majo
ority strateg
gy to cope w
with health expenses,
e aas it was eassy to
access due
d to amplle sources (fformal and informal so
ources), sim
mple proceduure and flex
xible
repaym
ment terms. As SSP pro
ovided finaancial beneffits, the neeed to borroow was lesss for
insured compared tto uninsured and newly
y insured in
ndividuals. Hence,
H studdy hypothessised
that inssured borrow
w less comp
pared to un
ninsured and
d newly inssured indiviiduals. A lo
ower
percentage of insured indiviiduals (57.2
2%) relied on borrow
wing comppared to neewly
insured (79.5 %) oor uninsured
d individuaals (75.2 %)) (p<0.05). Binary loggistic regression
model facilitated testing
t of hypothesis.
h In addition
n, multiple linear regrression mo
odels
helped to
t understan
nd the impaact of SSP on
o the amou
unt of borrow
wing.

170
4.7.4.1 Determinants of borrowing for treatment
The study used binary logistic regression analysis to estimate the likelihood of
borrowing for individuals in insured, uninsured, and newly insured groups. Individuals
with SSP were differentiated from those without it by assigning a code of ‘1’ for SSP
insured, ‘2’ for newly insured and ‘3’ for uninsured individuals. Borrowed individuals
were assigned a code of ‘1’ and those who did not borrow were assigned a code of ‘0’.
Certain variables expected to determine borrowing were included in the analysis. Firstly,
analysis considered the various characteristics of the individual (types of treatment and
health insurance), household (age, gender and job status of head of the household, size of
the household and income class) and community (area and district of residence), after
classifying individuals based on SSP membership status. Secondly, binary logistic
regression analysis was used to predict the probability of the borrowing.
4.7.4.1a Characteristics of Individuals
Health insurance and types of treatment would determine the probability of
borrowing for studied individuals.
i. Types of Treatment
Health expenditure depends on the types of treatment availed by ill persons.
Inpatient (IP) treatment is generally expensive than outpatient (OP) treatment, which
would result in borrowing. Almost nine of ten admitted individuals from insured group
borrowed compared to seven of ten newly insured and eight of ten hospitalised
individuals in uninsured group. OP treatment resulted in borrowing in 8.8 percent of
insured, 26.9 percent of newly insured and 20.3 percent of uninsured individuals. Thus,
there was a significant difference in the incidence of borrowing for the types of the
treatment (p<0.05). Irrespective of SSP, inpatient (70.8%) treatment resulted in more
borrowing than OP (63.8%) treatment (Table 4.58). OOPE and borrowed amount was
found to be positively related (p<0.05); higher OOPE resulted in higher amount
borrowed.

171
4.7.4.1b Characteristics of Households
Age, gender and job status of head of household, size of household and income
class would influence borrowing. Hence, the following section dissects these variables in-
depth.
i. Age of the Heads of the Household
Higher the age of the person, more will be the assets at disposal or savings that
decreases the need to borrow. Hence, the age of a person and borrowing would have
inverse relationship, as the amount of borrowing would be less for the elderly person
compared to younger person. Median age of the head of the households which borrowed
was 47 years and of those without borrowing was 50 years irrespective of insurance
status (Mann Whitney U test p<0.1). The median age of the head of the households in the
newly insured who borrowed was 47 years, which was lower than that of insured and
uninsured (49 years) households.
ii. Gender of the Heads of the Household
Gender of the heads of the household would influence various strategies adopted
to face health shocks. This study analysed the gender differences in borrowing strategies
to explore this relationship. The results indicate no visible gender difference in the
borrowing strategy adopted by the heads of the household (p>0.05) (Table 4.58).
Regardless of SSP, borrowing was less for women compared to men. Insured families
with women as the head of the households had lower episodes of borrowing (57.7%)
compared to newly insured (75%) and uninsured (63.6%) families (p>0.05). Moreover,
insured families with men as the heads of the household had lower incidence of
borrowing (56.3%) compared to newly insured (78.8%) and uninsured households
(78.1%) (p<0.05). One of the reasons for this finding is the CHE. Insured experienced
lower percent (28.1%) of CHE than newly insured (40.4%) and uninsured (43.8%)
(p<0.05). Hence, the necessity to borrow was less for insured families.

172
iii. Job Status of the Heads of the Household
The occupation in informal sector either as a labourer or as a monthly salaried
worker would increase the possibility of borrowing in contrast to employment in formal
sector due to seasonality of income. To test this assumption, the study analysed the job
status and borrowing by the sample households. Table 4.54 shows the job status of head
of households who borrowed to pay for medical expenses, classified based on SSP
member status. Among the various job statuses, majority of the borrowing was from
labourer households followed by unemployed heads. Families with self-employed and
formal sector employed head of the households had lowest incidence of borrowing (Table
4.58).
Table 4.54 Borrowing and Job Status of Heads of the Household
UN L SE A IS FE
Insured (N=91) 23.1 57.1 2.2 7.7 6.6 3.3
Newly insured (N=93) 17.2 62.4 4.3 4.3 4.3 7.5
Uninsured (N=64) 17.2 56.3 3.1 12.5 10.9 0
χ2 (5, N=361)=13.596, p =0.327
L- Wage labour (beedi roller, daily labourer) UN-Unemployed
SE- Self employment FE- Formal sector employment
IS-Salaried (informal sector) A-Agriculture
(Figures in percentages)

To delve into the magnitude of borrowing among the households with heads
working as unskilled labourer, further analysis focused on the cases of unskilled labourer.
Insured borrowed less (58.7%) than newly insured (79.3%) and uninsured families
(79.2%). Consequently, insured individuals with heads of the household working as
labourer had lower episodes of borrowing compared to their counterpart in uninsured and
newly insured groups (p<0.05).
iv. Household Size
The pooling of resources in large families would reduce the need to borrow.
However, median size of both borrowed and not borrowed families was four in all three
studied groups (Mann Whitney test, p=0.202). Hence, household size may not influence
borrowing.

173
v. Income Class of the household
The individuals from high-income classes usually use savings than borrowing
compared to low-income classes. The study revealed that high-income individuals (Q5)
borrowed less compared to low-income individuals (Q1 to Q3). Since these differences
were not significant, income may not be a determinant of borrowing.
Table 4.55 Borrowing in Income Class: Comparison by Health Insurance Status
Borrowed Q1 Q2 Q3 Q4 Q5
Insured1 No 20.6 25 26.1 14.2 14.1
Yes 22 23.1 19.1 16.5 19.3
Newly insured2 No 16.7 20.8 12.5 29.2 20.8
Yes 24.7 19.4 22.6 21.5 11.8
Uninsured3 No 23.9 19 23.8 9.5 23.8
Yes 26.5 25 18.8 18.8 10.9
1 2
χ (4, N=159)=1.624, p=0.805
2 2
χ (4, N=117)=3.12, p=0.538
3 2
χ (4, N=85)=3.16, p=0.531

The study considered poorest individuals in the sample to know the difference in
borrowing among them. The poorest (Q1 and Q2) in insured group borrowed less
(56.9%) than their counterparts in newly insured (82%) and uninsured (78.6%) groups.
Moreover, of those individuals who incurred CHE, insured borrowed less (75%
borrowed) than newly insured (88.9% borrowed) and uninsured individuals (92%
borrowed). Therefore, SSP reduced the impoverishing impact of illness by reducing the
need to borrow for the poor individuals.
4.7.4.1.c Characteristics of the Community
a. Area of Residence
It is apparent from previous discussions that people in rural areas incur higher
OOPE and CHE that may result in borrowing compared to those in urban and semi-urban
areas. On the contrary, a higher proportion of urban individuals borrowed. More of
insured individuals in urban (64.3%) and semi-urban (60%) areas borrowed in contrast to
newly insured and uninsured individuals. Rural residents from newly insured (67.3%)
and uninsured groups (69%) largely borrowed than insured individuals (54.1%)(Table
174
4.56). However, there was no difference in the incidence of borrowing for insured and
uninsured individuals (p>0.05), but not in case of newly insured individuals (p<0.05).
Irrespective of insurance status, 79.6 percent of urban individuals borrowed which was
quite high compared to 74 percent of the semi-urban individuals and 61.4 percent of rural
individuals (p<0.05) (Table 4.58).
Table 4.56 Borrowing and Area of Residence
Insured1 Newly insured2 Uninsured3
(N=159) (N=117) (N=85)
Urban 64.3 90.5 78.9
Semi-urban 60 87.2 83.3
Rural 54.1 67.3 69
1 2
χ (4)=0.809, p=0.667
2 2
χ (4)=7.714, p=0.021
3 2
χ (4)=1.851, p=0.396
(Figures represent borrowed individuals in each group as percentages)
b. District of Residence
In all three groups studied, people from Dakshina Kannada (DK) had lower
borrowing than those of Uttara Kannada (UK) and Gadag (Table 4.57) (p>0.05).
Irrespective of SSP, individuals from UK (81%) had higher borrowings than those of DK
(60.9%) and Gadag (60%) (p<0.05).
Table 4.57 Borrowing and District of Residence
Insured1 Newly insured2 Uninsured3
(N=159) (N=117) (N=85)
DK 47.6 71.7 73.9
UK 74 88.9 75.9
Gadag 55.6 50 80
1 2
χ (4, N=159)=8.908, p=0.012
2 2
χ (4, N=117)=9.375, p=0.009
3 2
χ (4, N=85)=0.171, p=0.918
(Figures represent only borrowed individuals in each group as percentages)

175
 Table 4.58 Description of Independent Variables of Borrowing Model
Borrowed No (N=113) Yes (N=248)
Types of treatmenta
Outpatient 39.5 60.5
Inpatient 29.1 70.9
Gender of head of householdb
Male 83.2 82.3
Female 16.8 17.7
Job statusc
Unemployment 17.6 19
Labourer 51.1 59.3
Self employed 6.8 3.2
Agriculture 11 7.9
Formal sector 4.2 4
Salaried (informal sector) 9.3 6.6
Income quintiled
Q1 20.4 24.2
Q2 23 22.2
Q3 18.6 23
Q4 17.7 18.1
Q5 20.3 12.5
Area of residencee
Urban 20.4 79.6
Semi-urban 26 74
Rural 38.6 61.4
DKf 39.1 60.9
UK 19 81
Gadag 40 60

Pearson chi square; a= 2.989, p<0.1; b=0.046, p>0.05; c=5.868, p>0.05; d=4.427, p>0.05; e=9.146,p>0.05, f=16.45, p<0.05

4.7.4.2 Econometric Estimation of the Incidence of Borrowing

The probability of borrowing due to healthcare was studied using binary logistic
regression analysis. It was hypothesised that SSP decreased the cost of treatment; hence
chances of borrowing would be less for insured individuals. Table 4.59 displays the
coding of variables included in the borrowing and savings models. Model specifications
were changed in the borrowing models to substantiate the findings of the basic model.
176
Hence, the basic model (model 5a), labourer model (model 5b), hospitalisation (model
5c) and low income model (model 5d) were estimated. Model 5e considered actual
amount of borrowing (log transformed) and used multiple regression analysis to assess
the impact of SSP. Model 5a considered cases of borrowing due to illness episode and
model 5b took the cases of only labourer head of the households. Model 5c used the cases
of inpatient treatment and considered the significant variables estimated by model 5a.
Model 5d took the cases of low income (Q1 and Q2) to know whether SSP makes any
impact on borrowing by the poor people and considered the independent variables
estimated by model 5a as significant. Model 5f considered the use of savings as
dependent variable in which individuals using savings were coded as ‘1’ and ‘0’
otherwise. Binary logistic model was used to find the determinants of the use of savings.
Model 5g was based on the amount of saving used by individuals; hence multiple
regression model (log transformed) was used. The types of treatment and gender of the
head of the household were coded into two dummy variables. The job status of head of
the household was coded into six dummy variables. Age of the head of the household was
a continuous variable and size of the household was coded into three dummy variables.
Income quintiles were coded into five dummy variables. SSP membership status and area
of residence were coded into three dummy variables each.

177
 Table 4.59 Measurement of Independent Variables Included in the Regression Analysis
Variables Model Model Model Model Model
5a, 5f 5b 5c 5d 5e, 5g
Health insurance
SSP insured=1 (reference)
Newly insured=2
Uninsured=3
Types of treatment =1if inpatient, 0 if
outpatient
Gender of head of household=1 if male,
0 if female (reference)
Job status of the household head
1= Labourer if primary occupation is
unskilled worker being paid daily wage
2= Business if engaged in self-
employment
3= Agriculture if farmer including dairy
farmer
4=Salaried in informal sector if
unskilled worker being paid monthly in
unorganised sector
5=Formal sector if skilled worker
employed in organised sector on a
salary basis
0=Unemployed/not able (reference)
Income quintile
1=Q1, 2=Q2, 3=Q3, 4=Q4, 5=Q5
(reference)
Area of residence
1= Urban if individual lives in urban
area
2= Semi-urban if individual lives in
semi-urban area
3=Rural if individual lives in rural areas
(reference)
District of residence
1=Dakshina Kannada
2=Uttara Kannada
3= Gadag (reference)

178
 Binary logistic regression analysis was performed to predict the determinants of
borrowing (Table 4.60). The evidence of insured individuals being less likely to borrow
due to illness compared to newly insured and uninsured individuals was found. The odds
of borrowing compared to not borrowing was high for newly insured (OR 3.122) and
uninsured (OR 2.972) individuals compared to the insured. Age of the head of the
household was another significant determinant of borrowing. The likelihood of
borrowing decreased (OR 0.969) for every increase in age. Inpatient care had higher
likelihood of borrowing compared to outpatient treatment (OR 3.013). Income was a
determinant of borrowing with lower income quintile individuals had a higher probability
of borrowing than high income quintile. Individuals from the middle-income quintile
(Q3) were 2.279 times more likely to borrow compared to high-income quintile (Q5), at
10 percent significance level. Individuals in the households with unemployed heads (OR
4.821) had higher likelihood of borrowing compared to heads employed in formal
sectors. Gender of head of the household, size of the household, area and district of
residence were not associated with probability of borrowing. Lack of health insurance,
younger head of the household, inpatient treatment, labourer class household heads and
middle income class increased the likelihood of borrowing. Hence, the study accepts the
hypothesis that SSP decreases the likelihood of borrowing for insured individuals
compared to both newly insured and uninsured individuals (H7a). The tests on model
fitness showed that 73.7 percent of the cases were correctly predicted by the model. The
model was subjected to endogeneity test (Durbin-Wu-Hausman test) and health insurance
was found to be exogenous with prob (χ2) =0.984. Hence, the effect of the unobservable
variables was absent. Residual analysis (specifically Cook’s Distance statistic) showed no
outliers.
The model 5b considered the employment of the heads of the household and took
only the cases of labourers and the significant independent variables of model 5a. Any
MHI should measure its performance in terms of the inclusion of the less privileged
persons in the financial protection. Hence, the study considered the cases of labourer
class, one of the less privileged classes in India. Of the 206 cases of labourers, 67.8
179
percent borrowed to meet medical expenses. Insured individuals borrowed less (62.7%)
compared to newly insured (78.4%) and uninsured individuals (78.3%). The median age
of labourer class head of the households who borrowed was 45 years and those who did
not borrow were 48 years. Hospitalisation in these households resulted in borrowing for
74.2 percent of the families. Hospitalised insured individuals borrowed less (63.3%) than
newly insured (80.8%) and uninsured persons (90.6%). Individuals living urban (76.9%)
and semi-urban (81%) areas had a higher percent of borrowing than those in rural areas
(60.4%).
The estimation results (Table 4.61) confirm the positive impact of SSP as
analysed in the previous section. Newly insured individuals had 2.935 times and
uninsured individuals had 3.334 times higher likelihood of borrowing compared to
insured individuals. Inpatient treatment increased the likelihood of debt by a factor of
3.978 than outpatient treatment. For every year of the age of the heads, the likelihood of
borrowing decreased by a factor of 0.996. Individuals living in urban areas were 2.789
times and those living in semi-urban areas were 3.783 times more likely to borrow
compared to individuals living in rural areas. Thus, health insurance, age of heads of the
household, inpatient treatment and area of residence were significantly associated with
probability of borrowing. The results of these tests showed that the model fits well and
74.1 percent of cases were correctly predicted by the model.

180
 Table 4.60 Probability of Borrowing: Estimated Results of Model 5a

B S.E. Wald Df Sig. Exp(B) 95% C.I.

Lower Upper
Health insurance (base= SSP insured) 16.757 2 .000
Newly insured 1.139 .323 12.450 1 .000 3.122 1.659 5.877
Uninsured 1.089 .339 10.349 1 .001 2.972 1.531 5.772
Types of treatment 1.103 .327 11.358 1 .001 3.013 1.587 5.723
(base: outpatient) Inpatient
Age of head of household -.032 .013 5.782 1 .016 .969 .944 .994
Gender (base=Female) Male .000 .353 .000 1 .999 1.000 .501 1.997
Job of head (base= Formal sector) 9.277 5 .099
Unemployed 1.573 .795 3.912 1 .048 4.821 1.014 22.917
Labourer .946 .700 1.824 1 .177 2.575 .653 10.161
Business .555 .880 .397 1 .528 1.741 .310 9.771
Agriculture .025 .773 .001 1 .974 1.026 .225 4.669
Salaried in informal sector .527 .815 .418 1 .518 1.694 .343 8.373
Household size (base: 7 and above) 2.769 2 .250
1-3 -.310 .533 .339 1 .560 .733 .258 2.085
4-6 -.647 .475 1.858 1 .173 .524 .207 1.328
Income quintile (base=Q5) 4.343 4 .362
Q1 .301 .470 .410 1 .522 1.351 .538 3.392
Q2 .192 .442 .188 1 .664 1.211 .509 2.882
Q3 .824 .439 3.525 1 .060 2.279 .964 5.386
Q4 .392 .439 .794 1 .373 1.479 .625 3.501
Area of residence (base=Rural area) 3.402 2 .182
Urban .724 .452 2.567 1 .109 2.062 .851 4.999
Semi-urban .446 .311 2.058 1 .151 1.563 .849 2.876
District of residence (base= Gadag) 5.753 2 .056
DK -.313 .397 .623 1 .430 .731 .336 1.592
UK .462 .448 1.063 1 .303 1.587 .660 3.816
Constant -.097 1.039 .009 1 .925 .907
Number of observations 361
Omnibus test model coefficient:Pearson chi square =59.321, df =20, p=0.000;-2 log likelihood = 389.398
Cox and Snell R squared= 0.093; Negelkerke R squared= 0.131
Hosmer and Lemeshow Pearson chi square =6.302, df =8, p=0.609
(Dependent variable: Borrowed; 1=yes)

181
 Table 4.61 Probability of Borrowing: Estimated Results of Model 5b

B S.E. Wald Df Sig. Exp 95% C.I.

(B) Lower Upper
Health insurance 9.162 2 .010
(base= SSP insured)
Newly insured 1.077 .419 6.593 1 .010 2.935 1.290 6.678
Uninsured 1.204 .473 6.486 1 .011 3.334 1.320 8.424
Age of heads of household -.035 .017 4.206 1 .040 .966 .935 .998
Types of treatment (base: 1.381 .426 10.515 1 .001 3.978 1.727 9.165
outpatient) Inpatient
Area of residence 13.022 2 .001
(base=Rural area)
Urban 1.026 .543 3.568 1 .059 2.789 .962 8.087
Semi-urban 1.325 .384 11.923 1 .001 3.763 1.774 7.984
Constant .239 .909 .069 1 .793 1.269
Number of observations 206
Omnibus test model coefficient:Pearson chi square =33.023, df =6, p=0.000; -2 log likelihood = 213.049
Cox and Snell R squared= 0.149; Negelkerke R squared= 0.213
Hosmer and Lemeshow Pearson chi square =7.225, df =8, p=0.513
(Dependent variable: Borrowed; 1=yes)

Models 5c considered the cases of hospitalised individuals only. Since SSP covers
hospitalisation, the cost of treatment would be less for insured individuals. There were
285 cases of hospitalisation in the study. Borrowing due to the admission in the hospital
was the highest for uninsured individuals (85%) than newly insured (82.9%) and insured
(58%) individuals. Irrespective of insurance, 70.9 percent of the admissions ended in
borrowing. The median age of the hospitalised individuals who borrowed was 45 years.
Individuals living in urban (77.8%) and semi-urban (73.1%) areas had a higher percent of
borrowing than those in rural areas (66.2%). Regardless of the insurance, individuals
from the families where the heads of the household worked as a wage labourer had the
highest borrowing (59.6%) than any other occupation followed by the unemployed
(11.3%) head of the households.
Table 4.62 illustrates the results of the model that shows a significant relationship
between SSP and borrowing. The likelihood of borrowing was high (OR 3.373) if the
individual was newly insured and if the individual was uninsured (OR 4.423) rather than
182
insured. Higher the age of the head, lower was the probability of borrowing (OR 0.971).
Individuals in households with head working in formal sector (OR0.225) and salaried in
informal sector (OR0.272 at 10% significance level) had a lower likelihood of borrowing
compared to unemployed heads. The area of residence was not a determinant of
borrowing. The results of these tests showed that the model fits well and 72.3 percent of
cases were correctly predicted by the model. Thus, SSP, age and job status of the head of
the household determined the likelihood of borrowing due to hospitalisation.
Table 4.62 Probability of Borrowing: Estimation of Model 5c

B S.E. Wald Df Sig. Exp(B) 95% C.I.

Lower Upper
Health insurance 20.536 2 .000
(base= SSP insured)
Newly insured 1.216 .351 11.992 1 .001 3.373 1.695 6.712
Uninsured 1.487 .412 12.993 1 .000 4.423 1.971 9.927
Age of head of household -.030 .014 4.377 1 .036 .971 .944 .998
Occupation of household 6.921 6 .328
head (base=
Unemployed)
Labourer -.426 .684 .388 1 .533 .653 .171 2.495
Business -.624 .635 .963 1 .326 .536 .154 1.862
Agriculture -.778 .878 .786 1 .375 .459 .082 2.566
Salaried in informal -1.303 .741 3.091 1 .079 .272 .064 1.161
sector
Formal sector -1.490 .748 3.972 1 .046 .225 .052 .976
Area of residence 2.193 2 .334
(base=Rural area)
Urban .457 .499 .839 1 .360 1.580 .594 4.205
Semi-urban .422 .310 1.853 1 .173 1.525 .831 2.798
Constant 2.301 1.048 4.823 1 .028 9.980
Number of observations 285
Omnibus test model coefficient:Pearson chi square =27.957, df =5, p=0.000: -2 log likelihood = 315.894
Cox and Snell R squared= 0.093; Negelkerke R squared= 0.133
Hosmer and Lemeshow Pearson chi square =5.527, df =8, p=0.700
(Dependent variable: Borrowed; 1=yes)

Model 5d (Table 4.63) considered the lowest income (in Q1 and Q2) insured,
newly insured and uninsured individuals. The significant variables estimated by the
183
model 5a were included as independent variables in the regression analysis. Among these
income classes, a higher percent of newly insured poor (82%) had higher borrowing,
followed by uninsured (79%) and insured (62%) poor. Hospitalisation related borrowing
was high in uninsured group (92.6%) than newly insured (86.8%) and insured (58.2%)
groups. Irrespective of insurance, 73.5 percent of the individuals borrowed for inpatient
treatment. The median age of the head of the household with borrowing was 45 years and
those who did not borrow were 48 years. The poor urban individuals (89.5%) borrowed
more compared to their counterparts in semi-urban (75%) and rural areas (61%).
The results of the model confirm the positive impact of SSP on the borrowing
strategies of insured individuals. Newly insured had 5.075 times higher likelihood of
borrowing and uninsured had 5.980 higher likelihood of borrowing compared to insured.
Hospitalisation increased the likelihood of borrowing by 5.737 times than outpatient
treatment. The odds of borrowing were high for individuals living in urban areas (OR
8.291) and semi-urban areas (OR 2.025) rather than rural areas. Thus, borrowing was
higher if the person did not have SSP, lived in urban or semi-urban areas and had
inpatient treatment. SSP reduced the incidence of borrowing for poor people. Hosmer and
Lemeshow test value of 0.960 indicate that model correctly predicted excellent
discrimination in 74.4 percent of cases.

184
 Table 4.63 Probability of Borrowing: Estimation of Model 5d

B S.E. Wald Df Sig. Exp(B) 95% C.I.

Lower Upper
Health insurance 15.199 2 .001
(base= SSP insured)
Newly insured 1.624 .500 10.539 1 .001 5.075 1.903 13.530
Uninsured 1.788 .550 10.592 1 .001 5.980 2.037 17.558
Types of treatment 1.747 .526 11.010 1 .001 5.737 2.044 16.101
(base: outpatient) Inpatient
Age of head of household -.013 .016 .620 1 .431 .987 .957 1.019
Area of residence 7.972 2 .019
(base=Rural area)
Urban 2.115 .838 6.374 1 .012 8.291 1.605 42.831
Semi-urban .705 .398 3.137 1 .077 2.025 .928 4.419
Constant -1.824 .604 9.103 1 .003 .161
Number of observations 164
Omnibus test model coefficient:Pearson chi square =30.282, df =5, p=0.000; -2 log likelihood = 169.741
Cox and Snell R squared= 0.169; Negelkerke R squared= 0.239
Hosmer and Lemeshow Pearson chi square =1.992, df =8, p=0.960
(Dependent variable: Borrowed; 1=yes)

4.7.4.3 Econometric Estimation of Amount of Borrowing

Multiple regression analysis was performed to know the impact of SSP on the
amount of borrowing due to illness. It was hypothesised that SSP decreases the amount of
borrowing since insured individuals can claim from the programme for hospitalisation.
The regression analysis considered the log transformed borrowed amount and the age of
the heads of the household. Backward elimination stepwise regression estimated the
robust model by eliminating insignificant variables from the model at the 13th step. The
analysis began with the full model considering certain independent variables namely
types of treatment, age, gender and job status of the heads of the household, income class,
district of residence and health insurance status. Insured individuals had lower amount of
borrowing (Table 4.64). The amount of borrowing would be 65 percent less for insured
individuals [exp (-.088)=1.65] than uninsured individuals. As given earlier, insured
individuals spent less compared to uninsured individuals. Moreover, borrowing would be

185
19.6 percent higher for inpatient treatment than outpatient treatment [exp (.179=1.196].
Individuals from smaller households (household size 1-3) would borrow 9.7 percent less
amount compared to individuals from larger households (7 and above) [exp (-
.093=1.097]. A one percent increase in OOPE would yield a .70 percent increase in the
amount borrowed. The model gives strong evidence to confirm the hypothesis of the
study that SSP reduces the amount of borrowing (H7b) for insured individuals.
Variance Inflation Factor test did not suggest any multicollinearity since the value
was one for all the significant independent variables, less than cut off 10. Correlation
matrix did not show any significant correlation between independent variables. Cook’s D
statistic detected no outliers (all cases had values <0.16) and Dfits statistic (< 1.0) did not
suggest any observation that strongly influenced the model. F value was 88.928 (p=0.00).

186
Table 4.64 Estimation of Model 5e: Health insurance and Amount of Borrowing

Standardized
Coefficients
Beta t Sig.
(Constant) -1.767 .079
Insured (=1, 0 otherwise) (base: uninsured) -.079 -1.448 .149
Newly insured (=1, 0 otherwise) .015 .288 .774
Gender of head: Male (=1, 0 otherwise) .021 .463 .643
(base: female)
Treatment: Inpatient (=1, 0 otherwise) .170 3.696 .000
(base: outpatient)
Log (Age of the head) -.050 -1.106 .270
Chronic (=1, 0 otherwise) (base: Acute) .036 .785 .433
Maternity (=1, 0 otherwise) .040 .869 .386
Log (OOPE) .696 15.014 .000
Urban (=1, 0 otherwise) (base: Rural) -.042 -.882 .378
Semi urban (=1, 0 otherwise) -.051 -1.096 .274
Q1 (=1, 0 otherwise) (base: Q5) .022 .314 .754
Q2 (=1, 0 otherwise) .000 .007 .994
Q3 (=1, 0 otherwise) .025 .368 .713
Q4 (=1, 0 otherwise) .051 .825 .410
Household size 1-3 (=1, 0 otherwise) -.032 -.655 .513
(base: Household size 7 & above)
Household size 4-6 (=1, 0 otherwise) -.094 -1.971 .050
Estimation at 13th step
(Constant) -4.697 .000
Insured -.088 -2.067 .040
Treatment .179 4.196 .000
Log (OOPE) .708 16.752 .000
Household size 1-3 -.093 -2.243 .026
Adjusted R2 0.590
Standard error of the estimate 0.642
Dependent Variable: Log transformed amount of loan
Number of observations: 253

187
4.7.5 Relationship between the Health Insurance Status and the Use of Savings
This section explores the association between savings used for different types of
treatment by insured, newly insured and uninsured individuals. The hypothesis was that
insured use fewer savings compared to uninsured and newly insured individuals since
SSP meets most of the direct expenses of hospitalisation. Contrary to our expectation,
higher percent of insured used savings (32.7 %) in comparison with newly insured
(24.7%), however it was less than that of uninsured (35.3%) individuals.
4.7.5.1. Determinants of the Use of Savings
Binary logistic regression analysis estimated the probability of the use of savings
for individuals in insured, uninsured and newly insured groups. Individuals with SSP had
a code of ‘1’. Newly insured and uninsured individuals were coded ‘2’ and ‘3’
respectively. Individuals who used savings were assigned a code of ‘1’ and those who did
not use savings had a code of ‘0’. Certain variables expected to determine savings were
included in the analysis. Firstly, the analysis considered various characteristics of
individual (types of treatment and health insurance status), households (age, gender and
job status of head of the household, size of the household and income class) and
community (area and district of residence), after classifying individuals based on SSP
membership status. Secondly, estimation of binary logistic model with varied
specifications was used to test the hypothesis. Model 5f was a binary logistic model that
estimated the determinants of use of savings. Model 5g was a multiple linear regression
model to know the determinants of the amount of savings. Table 4.59 provides the
coding of variables.
4.7.5.1.a Characteristics of Individuals
i. Types of Treatment
Insured individuals used more savings for (88.5%) inpatient treatment than
uninsured (64.3%) and newly insured individuals (Table 4.65). Thus, there was
significant difference in the incidence of savings for the types of the treatment (p<0.05).
Irrespective of SSP, inpatient (72.5%) treatment resulted in the use of higher savings than
OP (27.5%) treatment (p<0.05).
188
4.7.5.1.b Characteristics of Households
i. Age of the Heads of the Household
The age of the heads of the household and savings is directly related. The study
explored this assumption by including it as an independent variable. It was found that the
median age of the head of household who used savings was 51 years and of those without
savings was 47 years irrespective of insurance status (Mann Whitney U test p<0.1). The
median age of newly insured head was 46 years, which was lower than that of insured (53
years) and uninsured (50 years) households.
ii. Gender of the Heads of the Household
There is no established relationship between gender of the heads of the household
and use of savings. The current study shows an interesting finding. Households with men
as the head used more savings compared to households with women as the heads (Table
4.65) (p<0.05). Regardless of SSP, there was no difference in the use of savings among
men or women head households (p>0.05) although the general trend was that households
with men as the head used more savings.
iii. Job status of the Heads of the Household
Majority of labourer households in three groups (Table 4.65) used the savings,
followed by families with agriculture, self-employed, salaried and formal sector
employed as heads of households (p>0.05).
iv. Household Size
Median size of families that used savings and did not use savings was four in all
three studied groups (Mann Whitney test, p >0.05). Hence, the household size and
savings were not related.
v. Income Class of the Household
Individuals from high-income class would use savings compared to those from
low-income class. To the contrary, a higher percent of the Q1 (24.8%) used more savings
than Q5 (22%). Small proportion of Q2 (19.3%), Q3 (17.4%) and Q4 (16.5%) individuals
used savings (p>0.05). Insured individuals from the Q2 income quintile (52.2%) used
higher savings compared to Q5 individuals (Table 4.65). Q4 (50%) individuals from
189
newly insured group used savings more than Q5 (25%). In uninsured group, higher
percent of Q1 (31%) class used more savings than Q5 (p>0.05).
4.7.5.1.c Characteristics of the Community
a. Area of Residence
Higher proportion of rural individuals used savings in three groups (Table 4.65).
Rural individuals from newly insured (63.6%) and uninsured groups (60.7%) largely used
savings than insured (57.7%) (p>0.05). Regardless of SSP, individuals in rural (64.3%)
areas used more savings than those in semi-urban (27.5%) and urban areas (13.8%)
(p<0.05).
b. District of Residence
In addition to area of residence, district of residence was considered since FGD
revealed probable influence of the district of residence on savings. In all three groups
studied, people from Dakshina Kannada (DK) used more savings than those of Uttara
Kannada (UK) and Gadag (Table 4.65) (p>0.05). Irrespective of SSP, individuals from
DK (60.0%) used higher savings than those of UK (29.4%) and Gadag (10.1%) (p<0.05).

190
 Table 4.65 Description of Independent Variables Included in the Savings Model
Insured Newly insured Uninsured
(N=50) (N-31) (N=28)
Types of treatmenta
Outpatient 11.5 42.4 35.7
Inpatient 88.5 57.6 64.3
Gender of head of householdb
Male 90.4 87.9 64.3
Female 9.6 12.1 35.7
Job statusc
Unemployment 9.6 6.1 17.9
Labourer 51.9 69.7 46.4
Self employed 9.6 3 3.6
Agriculture 11.5 6.1 3.6
Formal sector 9.8 8 17.8
Salaried (informal sector) 7.7 9.1 10.7
Income quintiled
Q1 25 21.2 32.1
Q2 23.1 15.2 21.4
Q3 15.4 18.2 17.9
Q4 15.4 27.3 3.6
Q5 21.2 18.2 25
Area of residencee
Urban 9.6 18.2 14.3
Semi-urban 32.7 18.2 25
Rural 57.7 63.6 60.2
Dakshina Kannada (DK)f 61.5 57.6 64.3
Uttara Kannada (UK) 26.9 30.3 28.6
Gadag 11.5 12.1 7.1
Pearson chi square; a= 11.479, p<0.05; b=9.957, p<0.05; c=12.456, p>0.05; d=7.176, p>0.05; e=2.900,p>0.05, f=.628, p>0.05

4.7.5.2 Econometric Estimation of the Probability of Use of Savings

The estimation results on the relationship between SSP and savings is depicted in
Table 4.66. The evidence of insured individuals being less likely to use savings compared
to newly insured and uninsured individuals was not evident. Hence, study accepts the null
hypothesis that SSP does not reduce the use of the savings for insured individuals (H8a).
The age of the heads of the household was another significant determinant. The odds of

191
using savings increases by a factor of 1.035 for each year of the age of the heads of the
household. Individuals with hospitalisation were 0.5 times less likely to use savings
compared to individuals with outpatient treatment. Income was a determinant of savings
with lower income quintile individuals had the lower probability of use of savings than
high income quintile. Individuals belonging to low income quintile (Q3) were 0.389
times less likely to use savings compared to the highest income quintile (Q5). Individuals
residing in semi-urban areas had lower likelihood of the use of savings compared to those
in rural areas (OR 0.534). People from Dakshina Kannada had higher likelihood of the
use of savings compared to Gadag residents (OR 2.642). Thus, older heads of the
household, outpatient treatment, high income class, living in rural areas and Dakshina
Kannada district increased the likelihood of theuse of savings.
The model was subjected to endogeneity test (Durbin-Wu-Hausman test) and
health insurance was found to be exogenous with prob (χ2) =0.874. Hence, this result
rules out the effect of the unobservable variables on the study findings. Residual analysis
(specifically Cook’s Distance statistic) showed no outliers. The model correctly predicted
70 percent of the cases.

192
 Table 4.66 Probability of the Use of Savings: Estimated Results of Model 5f

B S.E. Wald df Sig. Exp(B) 95% C.I.

Lower Upper
Health insurance (base= SSP insured) .874 2 .646
Newly insured -.284 .303 .873 1 .350 .753 .415 1.365
Uninsured -.111 .317 .123 1 .726 .895 .481 1.664
Types of treatment (base: outpatient) -.869 .316 7.543 1 .006 .420 .226 .780
Inpatient
Age of head .034 .013 7.047 1 .008 1.035 1.009 1.061
Gender (base=Female) Male -.195 .344 .324 1 .570 .822 .419 1.613
Job status of head (base= Unemployed) 3.570 5 .613
Labourer -.667 .815 .671 1 .413 .513 .104 2.533
Business -.142 .744 .036 1 .849 .868 .202 3.726
Agriculture .266 .904 .087 1 .768 1.305 .222 7.671
Salaried in informal sector .162 .829 .038 1 .846 1.175 .231 5.972
Formal sector -.001 .848 .000 1 .999 .999 .190 5.260
Household size (base 1-3) .473 2 .790
4-6 .305 .506 .364 1 .546 1.357 .504 3.655
7 and above .309 .453 .467 1 .494 1.363 .561 3.308
Income quintile (base=Q5) 7.335 4 .119
Q1 -.215 .448 .231 1 .631 .806 .335 1.942
Q2 -.709 .433 2.681 1 .102 .492 .211 1.150
Q3 -.943 .424 4.950 1 .026 .389 .170 .894
Q4 -.677 .427 2.511 1 .113 .508 .220 1.174
Area of residence (base=Rural area) 4.170 2 .124
Urban -.412 .406 1.031 1 .310 .662 .299 1.468
Semi-urban -.627 .309 4.106 1 .043 .534 .291 .980
District of residence (base= Gadag) 6.483 2 .039
DK .971 .419 5.377 1 .020 2.642 1.162 6.004
UK .501 .466 1.154 1 .283 1.650 .662 4.112
Constant -1.812 1.122 2.606 1 .106 .163
Omnibus test model coefficient: Pearson chi square =35.120, df =18, p=0.00; -2 log likelihood = 384.192
Cox and Snell R squared= 0.11; Negelkerke R squared= 0.134;
Hosmer and Lemeshow Pearson chi square =5.148, p=0.742
(Dependent variable: Savings; 1=yes); Number of observations: 361

193
4.7.5.3 Econometric Estimation of the Amount of Savings Used for Treatment
Multiple regression analysis was performed to know the impact of SSP on the
amount of savings. As SSP provides financial protection, we hypothesised that SSP
decreases the amount of savings for insured individuals. Regression analysis used the log
transformed amount of savings and age of the head of the households. Backward
elimination stepwise regression estimated the robust model by eliminating insignificant
variables from the model at the 15th step. The analysis began with the full model
considering certain independent variables namely types of the treatment, age, gender and
job status of the head of the household, income class, district of residence and health
insurance status. In model 5g, amount of savings used would be 17.4 percent less for Q4
individuals compared to Q5 individuals [exp (-.161)=1.174]. A one percent increase in
OOPE would yield a .48 percent increase in savings amount used. Amount of savings
was not significantly different for insured individuals compared to uninsured and newly
insured individuals (Table 4.67). SSP status, gender and job status of the head of the
household were not associated with the amount of savings used. Hence, we reject the
study hypothesis that SSP reduces the amount of savings used for the treatment (H8b).
Multicollinearity was measured using Variance Inflation Factor that did not
suggest any collinearity since the value was 1 for all the significant independent
variables, less than cut off of 10. Correlation matrix did not show any significant
correlation between independent variables. Cook’s D statistic detected no outliers (all
cases had values <2) and Dfits statistic (< 1.0) did not suggest any observation that
strongly influenced the model. The model fitted well with F value of 17.562 (p=0.00).

194
Table 4.67 Estimated Results of Model 5g: Health insurance and Amount of Savings

Standardized
Coefficients
Beta t Sig.
(Constant) .567 .572
Insured (=1, 0 otherwise) (base: uninsured) .012 .095 .924
Newly insured (=1, 0 otherwise) .141 1.230 .223
Gender of head: Male (=1, 0 otherwise) -.038 -.360 .720
(base: female)
Treatment: Inpatient (=1, 0 otherwise) -.116 -1.098 .276
(base: outpatient)
Log (Age of the head) .132 1.271 .207
Acute (=1, 0 otherwise) (base: Chronic) -.098 -.916 .363
Maternity (=1, 0 otherwise) -.013 -.118 .906
Log (OOPE) .493 4.681 .000
Urban (=1, 0 otherwise) (base: Rural) .065 .600 .551
Semi urban (=1, 0 otherwise) .129 1.242 .218
Q1 (=1, 0 otherwise) (base: Q5) -.227 -1.568 .121
Q2 (=1, 0 otherwise) -.088 -.723 .472
Q3 (=1, 0 otherwise) -.133 -.988 .326
Q4 (=1, 0 otherwise) -.306 -2.431 .017
Household size 1-3 (=1, 0 otherwise) .011 .103 .918
(base: Household size 7 & above)
Household size 4-6 (=1, 0 otherwise) -.095 -.824 .413
Estimation at 15th step
(Constant) 1.662 .100
Q4 -.161 -1.787 .077
Log (OOPE) .487 5.416 .000
Adjusted R2 0.265
Standard error of the estimate 0.857
Dependent Variable: Log transformed amount of savings
Number of observations: 113

195
4.7.6 Sale of Assets to Pay for Medical Expenses
Usually, sale of assets take place when the households find it extremely difficult
to pay from pocket, borrow or use their savings to meet high cost of medical care. It also
depends on the saleable assets held by the households. The most common assets sold by
agriculture families were crop (whether harvested or not) whereas families of informal
workers sold consumer durables and sometimes jewellery or land/ house if illness was
catastrophic. Sale of assets was a last resort used by most of the individuals. Since SSP
provides financial coverage for hospitalisation, the need to sell assets for insured was not
as much as that for uninsured or newly insured individuals. To test the hypothesis that
insured sold fewer assets compared to uninsured and newly insured individuals, Pearson
chi square test was used.
Sale of assets was higher among insured (5.6%) than newly insured (2.6%) but
lower than uninsured (7%) individuals. There was no statistical difference among
insured, newly insured and uninsured individuals in the sale of assets (p>0.05). Hence,
null hypothesis that sale of assets does not differ among insured, newly insured and
uninsured individuals was accepted (H9). Due to small size of the sample that sold assets,
regression analysis was irrelevant.
4.7.7 Summary
Individuals in the survey used multiple risk coping strategies to meet medical
expenses. Predominantly, they used asset-based strategies such as sale of assets, use of
savings, borrowing money or health insurance (by insured). The use of savings and sale
of assets were less frequent than borrowing. Strategies with potential negative impact on
the portfolio of income sources like engaging school going children, women in income
generating activities and sending additional members of the family to labour market was
not seen. There was no difference in the ex-post risk coping strategies adopted by
insured, newly insured and uninsured individuals. Nevertheless, overall mobilised
amount from other risk coping strategies was low for insured compared to newly insured
and uninsured individuals.

196
 The availability of money had positive association with the savings used by the
individuals and negatively related to the borrowing. A small percent of the individuals
used savings in addition to borrowing. Individuals used savings mainly to meet outpatient
treatment costs. Saving was used in the households in which heads of the household was
older and belonged to high-income class. Rural individuals used more savings compared
to individuals in semi-urban areas and those living in DK used more savings than those of
Gadag. SSP was not a significant predictor of the savings (use and amount) in the logistic
and multiple linear regression models. Hence, the study rejects the hypothesis that SSP
reduces reliance on savings for insured individuals (H8a and H8b).
The study found sale of assets, such as crop or valuables like jewellery or two
wheeler vehicles, to be the least used health financing strategy. The target population of
SKDRDP is poor in informal sector. These households did not have any assets to sell
other than television, motor bikes and dwelling house. Thus, the study found statistical
significance for the incidence and amount of borrowing but not for savings and sale of
assets. However, average amount realised from sale of assets for insured was the highest
due to expensive cost of treatment for illness such as dialysis, kidney operation,
angiogram etc. faced by three to four insured individuals who had to pay almost one lakh
each.
Logistic regression analysis for the household survey data confirmed the
hypothesis of the study on the impact of SSP on borrowing to meet medical expenses.
The results indicate that insured individuals had lower borrowing compared to newly
insured and uninsured individuals. Models on hospitalisation, low-income class, labourer
head of the households and amount of loan support the hypothesis of positive impact of
SSP in reducing the incidence and amount of borrowing for insured individuals (H7a and
H7b). The results from these models indicate that insured rely less on the borrowing as
postulated in the theory. Since SSP brings down OOPE and CHE, the need to borrow is
less for insured individuals. Insured individuals in lower income class (model 5d) and
head of the households working as labourer (model 5b) had lower incidence of borrowing
compared to their counterparts in uninsured and newly insured groups. Consequently,
197
SSP could provide financial protection to vulnerable sections of the society against health
risks leading to impoverishment in such households.
One unanticipated finding was that SSP did not have any impact on the use of
savings and sale of the assets. The use of savings depends on the income of the family
and accumulated savings. Since the target population belongs to the poor section of the
society mostly in the informal sector, they can be expected to have fewer savings that can
be used to pay for medical expenses. Despite SSP claims, insured individuals had to bear
the indirect cost of care and outpatient treatment. This nullified the effect of SSP with
regard to the use of savings. In-depth analysis on the sale of the assets could not be
carried out due to the small sample size. Hence, we cannot substantiate the lack of impact
of SSP on the sale of assets.
The results of the study indicate that age of the head of households was a
significant determinant of the incidence of borrowing and savings. Since the younger
person in general has lower income compared to the elder person with similar socio-
economic background, the elderly head of the households would have more savings and
rely less on the borrowing. In this study, older head of the households used savings and
borrowed less than younger heads. The regression models with different specifications
(5a to 5d and 5g) confirmed this finding. Hospitalisation did positively influence
borrowing since the cost of treatment (direct and total) would be high compared to
outpatient treatment. In addition, individuals used fewer savings (amount and incidence)
for hospitalisation compared to the outpatient treatment.
Income class proved to be a determinant of the incidence and amount of
borrowing with an exception to individuals with labourer head of the households.
Moderately poor (Q3) individuals borrowed in higher proportion than high-income class
(Q5). Similarly, the job status explained the observed differences in the incidence of
borrowing in the basic model (5a) and hospitalisation model (5c). The unemployed heads
had a higher probability of borrowing compared to formal sector employees. The area of
residence was a determinant in two models (poor and labourer head of the household).
Urban and semi-urban residents had higher likelihood of borrowing and lower use of the
198
savings com
mpared to rural
r areas. Smaller households haad lower likkelihood off borrowing.
Higher OOPE resulted
d in higher amount
a of borrowing
b as well as usse of savings.
Anoother strikin
ng finding that
t needs fu
further explaanation is thhe highest amount
a and
d
incidence of
o borrowin
ng by newly
y insured ind
dividuals. Newly
N insurred individu
uals had thee
high cost of
o treatment (both OOP
PE and CHE
E); an averaage of sevenn percent off the annuall
consumptio
on expenditu
ure comparred to uninsu
ured and insured individuals. The head of thee
householdss in this grou
up were you
unger, prim
marily lived in
i urban areeas and belo
onged to Q3
3
income claass. These factors con
ntributed to
o the higheest incidencce and the amount off
borrowing for newly in
nsured indiv
viduals. On
n the other hand,
h insured had lowerr borrowing
g
since majorrity of them
m had inpatient treatment covered by SSP andd had lowerr OOPE and
d
CHE.
Hav
ving establiished the lo
ower incideence and amount
a of borrowing
b for insured
d
individuals, the next isssue concerrns the abseence of com
mplete finanncial protecttion by SSP
P
since insureed used varrious risk co
oping strateegies to meet the cost of illness. The
T reasonss
point at ceertain design features of SSP. Fiirstly, SSP covered hoospitalisatio
on expensess
only. Indireect costs (in
ncluding losst wages du
uring treatment and travvelling, food
d expenses))
had to be borne
b by th
he individu
uals. Persons with outp
patient treattment had to
t use theirr
money for the treatm
ment. Certain
n excluded
d diseases from
f the beenefit packaage namely
y
common aiilments such
h as fever, and cough forced sick heir pocket.
k people to pay from th
Secondly, tthe benefit package
p 00) comparred to the esscalated treaatment cost.
waas low ( 500
In case of some
s illness requiring expensive drugs or prrocedures (ddiagnostic or
o surgical),
individual had
h to use own
o funds. Thirdly, a large
l percen
nt of insuredd household
ds borrowed
d
to pay the ppremium. They
T borrow
wed money from the neeighbours orr friends orr used creditt
facility of SSP to paay the prem
mium. This suggests that
t insuredd household
ds are cash
h
constrainedd even to paay the annu
ual premium
m. Hence, th
he amount oof savings available
a to
o
pay for the care was lo
ow. Thus, th
he present sstudy could not find anny differencce in the usee
of savings among
a insu
ured, newly insured and
d uninsured individualss.
SSP
P as a risk coping strrategy would complem
ment other risk coping
g strategiess
rather than
n making th
hem redund
dant. Yet, itt would red
duce the neegative effeect of thesee
19
99
strategies by reducing excessive reliance on them. The finding revealed in this section
confirms the positive impact of SSP on financial protection. Thus, direct measure of
financial protection namely OOPE, CHE and comprehensive measures of risk coping
strategies do support the study hypothesis that SSP indeed provided financial protection
to its members. The findings points out the potential positive contribution that SSP can
make to reduce the reliance on risk coping strategies. This would decrease the possibility
of impoverishment in the poor households and improve their quality of life. However, it
is interesting to know whether the poorest are included in SSP membership. If SSP
excludes the poorest, the positive impact of the programme would not be welfare
promoting. Hence, the next chapter finds out the determinants of the enrolment and social
inclusion in SSP to promulgate such MHI schemes as a poverty reduction strategy.

200
 CHAPTER 5

ENROLMENT IN SAMPOORNA
SURAKSHA PROGRAMME
5.1 Introdu
uction
Enrrolment in SSP
S influen
nces the riskk coping caapability of the househ
holds during
g
the health crisis.
c It play n mitigating poverty indduced by the expensivee
ys an imporrtant role in
medical carre, especiallly for the poor
p househ
holds. How
wever, we hhave seen th
hat SSP hass
excluded some
s low--income ho
ouseholds and includ
ded high-inncome hou
useholds ass
members (ssection 4.1)). The research question was wh
hether pooreest was exccluded from
m
bership or not. This chaapter analysses the facttors determiining SSP membership
SSP memb m p
and reasonss for enrolm
ment in SSP
P. Firstly, th
he study con
nsidered thee incidence of illness in
n
the previouus year of the study in insured,, newly inssured and uuninsured households
h .
Second, th
he factors determinin
ng enrolmeent were estimated using binaary logisticc
regression model. Speecifications of the bassic model were
w changged to subsstantiate thee
findings an
nd to test th
he robustneess of the model.
m Thirrdly, the facctors for en
nrolment ass
perceived by
b the resp
pondents were
w studied
d using facctor analysis. Fourthly
y, the study
y
explored thhe performaance of SS
SP in sociall inclusion.. Lastly, thhe analysis focused on
n
adverse selection in SS
SP.
SSP
P, initiated by
b SKDRD
DP is a socioo-economicc development program
mme targetss
the poor hoouseholds in
n the inform
mal sector. Hence, it has
h to incluude the majority of thee
poorest in the risk po
ool. Althoug
gh most off the target populationn falls below
w or at thee
border of BPL
B (below
w poverty liine), this sttudy defined
d the destittute (annuall income off
less than 14000) as extremely
y poor or ppoorest. Thu
us, the hyp
pothesis waas that SSP
P
includes thhe poorest in large proportion than high
h-income ffamilies in the targett
population.. Various determinant
d ts of enrolment were estimated using binaary logisticc
regression analysis,
a inccome being
g one of them
m to test stu
udy hypotheesis.
Advverse selecction in MHI
M threateens the lo
ong-term viiability and financiall
sustainabiliity. SSP en
nrols entire household as the unit of enrolm
ment to curttail adversee
selection. Due to th
he constantt flow of informatio
on among the peoplle in rurall
communitiees, informattion asymm
metry will bee less prevaalent and muuch less thee possibility
y
of adverse selection. However,
H S does not
SSP n insist on
n the waitinng period an
nd includess
pre-existingg diseases in
i the coverrage. Moreoover, the up
pper age lim
mit for enro
olment is 80
0
20
01
years that encourages older high-risk individuals to enrol in SSP. Hence, the study
hypothesises that SSP has adverse selection.
The hypothesis of the study was tested using the data collected from 1146
households (4961 individuals); 782 insured households (3444 individuals that includes
newly insured who enrolled in 2011-12), and 364 uninsured households (1517
individuals). Individual (head of household), household and community characteristics
of insured households were compared with uninsured households. The probability of
obtaining a membership in SSP was found out by using the binary logistic regression
model as given below;
Prob (membership>0) = β0+β1Xy +ε, where Xy stands for a set of independent
variables like income, characteristics of the household head, household characteristics,
community characteristics that affect membership. Factor analysis was carried out to
understand the reasons for enrolment in SSP. Social inclusion and adverse selection were
tested using binary logistic regression model and Pearson chi square test.
5.2 Incidence of Illness among Insured and Uninsured Households
Enrolment in SSP was associated with the incidence of illness in the previous year
of enrolment. A higher percent of insured (38.2% of insured and 32.5% of newly insured)
reported the incidence of illness compared to one fourths of uninsured (p<0.05).
Table 5.1 Incidence of Illness and Enrolment in SSP
Incidence of illness
Yes No
Insured (N=416) 38.2 61.8
Newly insured (N=366) 32.5 67.5
Uninsured (N=364) 23.5 76.6
χ2 (2, N=1146) =20.008, p =0.00

5.3 Determinants of Enrolment in the Surveyed Households

Binary logistic regression analysis was done to estimate the likelihood of
membership in SSP. Individuals with SSP were differentiated from those without it by
assigning a code of ‘1’ for SSP insured (and newly insured) and ‘0’ for uninsured
individuals. The factors determining enrolment in SSP can be classified into enabling

202
factors (income and education), predisposing factors (gender, marital status, age and the
job status of head of the household, area of residence, distance to hospitals and size of the
household) and need factors (chronic illness in the household). Firstly, the present study
analysed various factors that would determine enrolment after classifying individuals
based on SSP membership status. Secondly, binary logistic model estimated the
determinants of enrolment with a different model specification to substantiate the
findings of the basic model.
5.3.1 Enabling Factors
a. Education of Head of the Household
The education level would determine the enrolment with educated seeking
enrolment than the uneducated. A higher percentage of the head of insured households
were illiterate or primary educated and uninsured group had higher proportion of
secondary school educated, pre-university or degree holders (Table 5.3). Uninsured head
of the households had an average education of six years, higher than that of insured
(average of 5 years) (p>0.05). This suggests that better educated head of the households
tend to stay away from SSP.
b. Income Class
Social inclusion objective of SSP would be achieved if the poorest income class
were represented more than high-income class. SSP’s target population is those below or
near the poverty line. Hence, the study expects poorest (below the poverty line) to be
included in the programme (H10). It should be kept in mind that high-income families in
this study are still poor when we consider the definition of the income quintiles given by
Planning Commission on all-India basis. However, our analysis considered the income
quintiles to find out the inclusion of the poorest in SSP.
Intra-group analysis (Table 5.3) illustrates that SSP included households from
different income classes in almost equal proportion. A higher proportion of insured
households was from Q4 (21.6%) followed by Q2 (20.7%) and Q3 (20.5%). Uninsured
households had relatively higher percentage of Q4 and Q1 quintiles and lowest from Q5.
SSP excluded lowest income quintile (Q1) (only 19.4 percent) whereas a higher
203
represen
ntation wass observed from Q2 and Q3 qu
uintile, whiich denotess partial so
ocial
inclusio
on.
5.3.2 Prredisposingg Factors
a. Age of
o Head of the Househ
holds
Age of the head of thee household
d would dettermine enrrolment with
h older heaad of
househo
olds joiningg the schem
me compared
d to youngerr head of th
he households. The med
dian
age of the
t head off insured ho
ouseholds was
w 46 yearss and that of
o uninsuredd was 48 yeears.
Hence, householdss with the younger
y heeads joined SSP than households
h with the older
o
heads. However,
H g was not significant (M
thhese finding Mann Whitn
ney test, p>00.05).
b. Gend
der of Head
d of the Ho
ouseholds
Women headed houseeholds usuallly have low
w level of education
e aand income that
would result
r in no
on-enrolmen
nt. Women were the heads
h of hou
usehold in 20.1
2 percen
nt of
uninsurred househo
olds whereass only 15.8 percent of insured
i hou
useholds hadd women ass the
head (T
Table 5.3). The gend
der of the head of th
he househo
olds indeedd contributee to
enrolmeent status in
n SSP (p<0.1). Uninsurred had a hiigher percen
ntage of wo
omen as heaad of
the hou
useholds witth lower inccome and ed
ducation levvel compareed to househholds with men
m
as the head. The median an
nnual incom
me of malee-head housseholds was 93,600 and
female--head houseeholds was 76,800 an
nd this differrence was significant
s ((Mann Whittney
U test, p=0.00). Ev
ven a differrence was found
fo in thee education of head of the househ
holds
(Mann Whitney U test, p<0
0.05). Nearly half off the womeen (47.7%) were illiteerate
comparred to only 18.8
1 percen
nt of men wh
ho head thee household
ds. These diffferences in
n the
basic ch
haracteristiccs of male and
a female head
h of the householdss might havee contributeed to
the diffe
ferences in enrolment
e in
n SSP.
c. Mariital Status of
o the Head
d of the Ho
ouseholds
The head of the hou
useholds un
ndertake thee responsib
bility of thee family in
n the
social fabric
f of Kaarnataka. Iff the head of
o the houssehold is maarried, he hhas to meett the
financiaal needs and
d take decissions relateed to risks facing
f the household
h inncluding heealth
risk. Hence,
H marrried heads of the hou
usehold wo
ould enrol in
i SSP moore than sin
ngle
(unmarrried, widow
wed, divorceed) head off the househholds. The size
s of famiily ranged from
f
204
three to six members for married head of the households than single heads (2-3
members). There was a significant difference in the marital status of the heads and
enrolment in SSP (p<0.05). Nearly 87 percent of insured had married head of the
households compared to 81 percent of uninsured.
d. Job Status of Head of the Households
The job status of the head of the households largely determines the seasonality of
the income of the family, thereby the purchase of health insurance. Employment in
formal sector, self-employment (business) or skilled salaried job in informal sector is
associated with better income and certainty of income than labour (skilled or unskilled)
and agriculture. On the contrary, a higher proportion of insured were found to be self-
employed (8.2%), salaried in informal sector (8.8 %) or employed in formal sector
(private and government) (5.8%) in contrast to uninsured households (7.8%, 8.1 % and
4% respectively) (p>0.05) (Table 5.3). Hence, an unequivocal conclusion could not be
arrived.
e. Size of Households
Number of members in the household determines the enrolment decision. Large
families denote higher informal insurance as they can rely on each other during the health
crisis, which reduces the demand for MHI. Yet they have higher risks and health
expenditure; hence, they would be inclined to secure the family against risks of ill health.
Nuclear family demands MHI due to the lower buffer in the form of informal insurance
and financial insecurity (Abel-Smith 1992). It is difficult to predict the impact of
household size on enrolment in SSP. There is no theoretical support to know the impact
of household size on SSP enrolment. Let us know what the data speaks.
Although median size of the households in insured and uninsured group was four,
there was a significant association between the household size and membership in SSP
(Mann Whitney U test p<0.05). Nearly 32 percent of uninsured and 27.6 percent of
insured households had family size of 1 to 3. Almost 56 percent of uninsured and 53.4
percent of insured had family size of 4 to 5. Twelve percent of uninsured group had 6-10
members whereas insured group had 18 percent and almost one percent had 11-19
205
members. This stresses that larger families tend to enrol in SSP coverage, due to financial
insecurity, higher risk of ill health and adverse effect of huge health expenses. Even the
average cost of premium was low for large families. Hence, safeguarding the large family
from unforeseen consequences of ill health motivated enrolment in SSP.
f. Area of Residence
Geographical location of the households is an important determinant of enrolment
in MHI. A comparison of both groups reveals that insured were more from semi-urban
areas (35.9%) than uninsured households (31.9%) although target population resided
mainly in rural areas (55%) (Table 5.3).However, no significant association between area
of residence and SSP membership was possible (p>0.05). Of the available hospitals in the
semi-urban areas, one fourth was network hospitals whereas just one tenth of total
hospitals were SSP hospitals in rural and urban areas.
g. Distance to Hospitals and Enrolment in SSP
This study has established a positive relationship between the distance to hospitals
and CHE in the section 4.6. If the households stay far away, they would incur higher
expenditure for treatment. Since SSP brings down the direct cost of treatment, the
households would incur lower total cost of treatment. Thus, families staying far away
from the hospitals would be motivated to enrol in SSP. The study results indicate that
insured had to travel an average of 2.8 km to network hospitals compared to uninsured
(mean distance 2.4 km) households. Hence, distance to hospitals encourages households
to purchase SSP (Mann Whitney U test p<0.05).
5.3.3 Adverse Selection
Risk of ill health motivates individuals to enrol in MHI resulting in adverse
selection (need factor). Many observable and non-observable factors determine health
risk. However, measurement of health risk is difficult. Self-reported health status (from
very good to very bad) and the health expenditure as measures of health risk are usually
used. However, these indicators are highly subjective varying according to the perception
and understanding of the respondents. Higher health expenses may be due to over
utilisation. Another measure is to consider age, gender or job status of members of the
206
family and classifying households with members in jobs involving high level of health
risk, presence of women or elderly in the family as high-risk households. However, the
study did not consider it a measure of health risk as household was the unit of analysis,
hence these characteristics could not be used to measure health risk. Thus, in this study,
health risk, defined as the bad medical situation, acts as an indicator of adverse selection.
Adverse selection (AS) or health risk was defined as the prevalence of chronic
illness in the family such as hypertension, diabetes, asthma, cancer from which the person
suffers for longtime. Since SSP insists on the family enrolment, health risk of the
household rather than individuals measured the adverse selection. Due to certain design
features of SSP (inclusion of pre-existing illness, no waiting period, and lack of screening
of members before enrolment), we can expect adverse selection to be present in SSP. To
test the hypothesis (H11), adverse selection was included as independent variable in the
enrolment logistic regression model.
Nearly 66 percent of uninsured households did not report any chronic illness
compared to 60.7 percent of insured household. These findings were not significant
(p=0.112). Further assessment considered the prevalence of adverse selection in different
income quintiles to know whether it has come from low-income households or not. SSP
members in low-income quintile (Q1, Q2, Q3) had higher health risk compared to those
in high income class (Q5) (Table 5.2). However, the conclusion on the welfare promoting
impact of SSP could not be reached (p>0.05).
Table 5.2 Intra-Income Comparison of Health Risk
Q1 Q2 Q3 Q4 Q5
1
Non members
Yes 28.2 39.7 42.3 26.7 35.9
No 71.8 60.3 57.7 73.3 64.1
Members2
Yes 42.3 42.5 38.1 37.5 35.6
No 57.7 57.5 61.9 62.5 64.4
1
χ2 (4, N=1141) =6.072, p =0.194
2
χ2 (4, N=1141) =2.519, p =0.641

207
 Table 5.3 Basic Characteristics of Insured and Uninsured Households
Insured Uninsured
(N=782) (N=364)
Gender of head of family: Male a 84.2 79.9
Education of head of family b
Illiterate 24.7 22
Primary 42.7 40.7
Secondary 23 24.5
Pre-university/graduate 8.9 11.8
Others 0.7 1
Job statusc
Unemployment 15.8 18.3
Labourer 56.2 56
Self employed 8.2 7.8
Agriculture 5.2 5.8
Formal sector 5.8 4
Salaried (informal sector) 8.8 8.1
Income classd
Q1 19.4 22
Q2 20.7 19.8
Q3 20.5 18.4
Q4 21.6 22.3
Q5 17.8 17.5
Area of residencee
Urban 10.5 12.4
Semi-urban 35.9 31.6
Rural 53.6 56
Pearson chi square; a= 3.217, p<0.1; b=3.417, p>0.05; c=3.774, p>0.05; d=1.480, p>0.05; e=2.421, p>0.05

5.3.4 Econometric Estimation on the Determinants of Enrolment in SSP

Binary logistic regression analysis was applied to estimate the determinants of

enrolment in SSP. Table 5.4 displays the coding of thevariables that were included in the
model (model 6a), and vulnerable group model (model 6b). Model 6b took cases of
unemployed and labourer households to explore the determinants of enrolment in these
households. The gender and marital status of the head of the households and chronic
illness in the family was coded into two dummy variables. The job status of head of the
households was coded into six dummy variables. Age of the head of the households and

208
distance to hospital were continuous variables. SSP membership status, size of the
household and area of residence were coded into three dummy variables each.
Table 5.4 Measurement and Coding of Independent Variables
Variables Model 6a Model 6b
Gender of head of household=1 if male, 0 if female
(reference)
Marital status of head of household=1 if married, 0 if
single
Job status of the household head
1= Labourer
2= Business
3= Agriculture
4=Salaried in informal sector
5=Formal sector
0=Unemployed/not able (reference)
Chronic illness in the family=1 if yes, 0 if no
Income quintile
1=Q1, 2=Q2, 3=Q3, 4=Q4, 5=Q5 (reference)
Area of residence
1= Urban if household lives in urban area
2= Semi-urban if household lives in semi-urban area
3=Rural if household lives in rural areas (reference)

Age, gender, marital status, education and job status of head of households and
income class were not significantly associated with enrolment (Table 5.5). The
households living in the semi-urban areas were 1.35 times more likely to enrol than rural
areas. Families living far away from hospitals were 1.083 times more likely to enrol in
SSP. Thus, households living in semi-urban areas and away from hospitals were more
likely to enrol in SSP. The results of these tests showed that the model fits well and 68.5
percent of cases were correctly predicted by the model.
Model 6b estimated that households in semi-urban areas influenced the likelihood
of enrolment by a factor of 1.621 than rural areas (Table 5.6). There was a positive
association between distance to hospital and enrolment in SSP (OR 1.110). Household
size was not a determinant of enrolment for unemployed/labourer families. 69 percent of
cases were correctly predicted by the model.
209
 Table 5.5 Probability of Enrolment: Results of Model 6a
B S.E. Wald Df Sig. Exp(B) 95% C.I.
Lower Upper
Age of household head -.008 .008 1.081 1 .299 .992 .978 1.007
Gender of head (base=Female) Male .112 .313 .127 1 .721 1.118 .606 2.064
Education of head (base: Illiterate) 3.606 3 .307
Primary (1 to 7 std) .505 .374 1.824 1 .177 1.658 .796 3.451
Secondary (8 to 12 std) .367 .352 1.090 1 .297 1.444 .725 2.876
Graduate and above .162 .345 .220 1 .639 1.176 .598 2.312
Marital status (base: Married) Single -.387 .314 1.520 1 .218 .679 .367 1.257
Occupation of household head (base= 3.740 5 .587
Unemployed)
Labourer -.282 .216 1.698 1 .193 .754 .494 1.153
Business -.264 .319 .687 1 .407 .768 .411 1.434
Agriculture .116 .442 .069 1 .792 1.123 .473 2.669
Salaried in informal sector -.263 .301 .767 1 .381 .769 .426 1.385
Formal sector -.604 .387 2.443 1 .118 .546 .256 1.166
Chronic illness in the family (base: -.230 .137 2.804 1 .094 .795 .607 1.040
Yes) No
Household size (base: 1-3) 1.188 2 .552
4-6 -.292 .295 .975 1 .323 .747 .419 1.333
7 and above -.165 .271 .369 1 .543 .848 .498 1.444
Income quintile (base=Q5) .808 4 .937
Q1 -.034 .233 .021 1 .885 .967 .612 1.526
Q2 .086 .223 .150 1 .699 1.090 .704 1.688
Q3 .126 .220 .329 1 .566 1.134 .737 1.745
Q4 .012 .210 .003 1 .953 1.012 .671 1.526
Area of residence (base=Rural area) 4.224 2 .121
Urban .090 .218 .172 1 .678 1.095 .714 1.678
Semi-urban .300 .147 4.187 1 .041 1.350 1.013 1.799
Distance to SSP hospital .080 .030 6.949 1 .008 1.083 1.021 1.149
Constant -.518 .554 .873 1 .350 .596
Number of observations 1146
Omnibus test model coefficient: Pearson chi square =54.227, df =21, p=0.00; -2 log likelihood = 1399.754
Cox and Snell R squared= 0.026; Negelkerke R squared= 0.036
Hosmer and Lemeshow Pearson chi square =3.643, df =8, p=0.888

210
 Table 5.6 Probability of Enrolment: Estimated Results of Model 6b
B S.E. Wald Df Sig. Exp(B) 95% C.I.
Lower Upper
Education of head (base: 5.226 3 .156
Illiterate)
Primary (1 to 7 std) 1.184 .866 1.869 1 .172 3.268 .598 17.847
Secondary (8 to 12 std) .970 .860 1.275 1 .259 2.639 .490 14.227
Graduate and above .738 .865 .727 1 .394 2.092 .384 11.402
Gender of head .075 .354 .045 1 .833 1.078 .538 2.158
(base=Female) Male
Age of head -.004 .008 .276 1 .599 .996 .981 1.011
Marital status (base: -.386 .361 1.143 1 .285 .680 .335 1.379
Married) Single
Chronic illness in the -.222 .159 1.953 1 .162 .801 .587 1.093
family (base:Yes) No
Household size(base= 1-3) 1.476 2 .478
4-6 -.265 .342 .600 1 .438 .767 .392 1.500
7 & above -.055 .314 .031 1 .861 .947 .512 1.752
Area of residence 8.176 2 .017
(base=Rural area)
Urban .206 .269 .588 1 .443 1.229 .726 2.082
Semi-urban .483 .169 8.170 1 .004 1.621 1.164 2.258
Distance to SSP hospital .104 .036 8.479 1 .004 1.110 1.035 1.190
Income quintile (base=Q5) 4.823 4 .306
Q1 .265 .228 1.355 1 .244 1.304 .834 2.039
Q2 .230 .240 .915 1 .339 1.259 .786 2.016
Q3 .097 .239 .165 1 .684 1.102 .690 1.762
Q4 -.223 .262 .727 1 .394 .800 .479 1.336
Constant -.239 1.076 .049 1 .824 .787
Number of observations 800
Omnibus test model coefficient: Pearson chi square =31.895, df =16, p=0.01; -2 log likelihood = 1046.664
Cox and Snell R squared= 0.036; Negelkerke R squared= 0.051
Hosmer and Lemeshow Pearson chi square =7.035, df =8, p=0.533
(Dependent variable: Enrolled; 1=yes)

211
5.4 Rea
asons for Enrolment in SSP
The factor analysis focused
f on the variouus reasons that motivvate enrolm
ment.
Mutual help, reducction in finaancial barrieers, access to
t good hosspitals and bbenefit pack
kage
he main reasons to join
were th n SSP (Figu
ure 5.1). Fig
gure 5.2 sh
hows the reaasons given
n for
enrolmeent by insurred and new
wly insured households
h .
May need in Benefit Peacee
future Paackage of
26% 14% mind
7%
Let otherrs Access to
benefit good
17% hospitals
15%
Need not All members
worry abouut in the group
money have enrolled
16% 5%

Fiigure 5.1 Reeasons for Enrolment

E in
n SSP
Newly insu
ured joined d hospitals and to seccure the fam
d SSP to access good mily
financiaally againstt any illneess in futurre. They reegarded mu
utual help to be ano
other
importaant reason. O
Other than securing
s thee future agaainst risk off illness, redducing finan
ncial
barrierss and benefit packagee of SSP attracted
a in
nsured mem
mbers of SS
SP. There was
significcant differen
nces in the reasons giiven by inssured and newly
n insurred respond
dents
(p<0.05
5).

277.7
30 25.3
25
18.7
20 17.3
Percentage

16.66 16.1 15 15.9

13.8
15 Renewed insured
9.2
Neewly insured
10 6.7 7.3
4.8 4.2
5
0
SSP Benefit Peace of Access to All Need nott Let others May need in
Package mind good members in worry abouut benefit futuree
hospitals the group money
have
enrolled

Figure 5.2 Reason

ns for Enrolm
ment: Com
mparison of R
Renewed In
nsured and Newly Insu
ured
holds
Househ

212
 To understand the most relevant reasons for enrolment, factor analysis was carried
out. KMO sampling adequacy was 0.580 and Barlett’s test of sphericity was significant
with p <0.05. Rotated component matrix highlights ‘SSP benefits’ (.741), ‘May need in
future’ (.751) as the first component, ‘Let others benefit’ (.730) and ‘Peace of mind’
(.635) as second component, ‘All members in the group have enrolled’ (.667), and ‘Need
not worry about money’ (.771) as third component and ‘Access to good hospitals’ (.951)
as fourth component (Table 5.7).
Table 5.7 Rotated Component Matrix: Factors Underlying Enrolment in SSP
Component
1 2 3 4
SSP benefits .741 -.064 .104 -.152
Peace of mind .300 .635 .116 -.047
Need not worry about money -.149 .338 .771 .191
Go to good hospitals .024 -.118 -.002 .951
All members in the group have enrolled .081 -.301 .667 -.198
Let others benefit -.213 .730 -.102 -.089
May need it in future .751 .083 -.205 .234

Rotated component matrix revealed hedging risk through SSP benefits as the first
factor for enrolment, financial security for oneself and others as the second factor, group
influence to avert the cost of illness as the third factor and access to good care as the
fourth factor for enrolment in SSP. Thus, some components of social capital namely
group influence and mutual financial security influenced enrolment.
5.5 Non-Enrolment in SSP
Several factors that resulted in non-enrolment in SSP are demand side factors
(individual and household characteristics) and supply side or scheme related factors. The
demand side factors were low income of family, unrealised benefits, multiple enrolments
in health insurance schemes, lack of family support, withdrawal from self-help groups,

213
lack of time to fill forms, family members being healthy, not aware of SSP and cultural
beliefs. Supply side factors were complex claim procedure, inadequate benefit package,
inflexible timing of collection of the premium and lack of good health care facilities.
5.5.1Demand Side Factors
1. Low Income of the Family
The target population of SKDRDP is poor households with low income.
“...we have too much loan to repay… We do not have money to pay premium...”,
“…we do not have much income….. I am planning to leave SHG as we have too many
loans..”, “..prices have gone up… we have outstanding loans and payment of weekly
installment of the loan itself is difficult..”
Low income of the family and increase in premium amount in 2011 have resulted
in unaffordability of premium;
“..they (insurers) have increased the premium this year… It is difficult to pay..”.
A credit facility provided by SSP to pay the premium for those who cannot
otherwise afford it to enlarge the risk pool. However, lack of awareness among the
participants about this facility prevented many from joining SSP;
“…it’s difficult to borrow to pay the premium..”, when asked were you aware of
borrowing facility to pay the premium; “..no one has told us...If we knew, we would have
enrolled in Suraksha..”.
2. Unrealised Benefits
About the benefits of insurance, participants expected to claim from SSP at least
once;
“…we have been doing Suraksha for many years..we did not get any benefits so
far… we do not want to continue”, “No one fell sick in the last two years…why waste
money by paying the premium?..”, “…we did not get any benefit in these five years..
wedo not want to continue..”, when asked, “..did not that mean having illness? …Do you
want to fall sick to claim?..”, the reply was, “..wedo not want illness…but we do not want
to spend money for something which does not benefit us..”, when asked again, “…you
may benefit in future if someone fall sick in your family..”, reply was, “..we believe in
214
Lord Manjunatha (of Dharmasthala, a well-known temple in DK district)…He will keep
us healthy..”
3. Multiple Memberships in Other Health Insurance Schemes
Multiple memberships in other health insurance schemes were expressed as
another factor responsible for non-enrolment.
“…we have ESI..wedo not want Suraksha…”,“….we bought Suraksha card
although we had Yashaswini…We did not get benefit from both... No one was ill in the
family..”.
4. Lack of Family Support
The most important social factor highlighted by the participants as an important
factor for not joining or not renewing their membership was lack of support from family,
mainly objection by husband;
“…my husband was ill last year..he did not get benefit… He told me not to do this
time..”, “..I do not have approval from home..” The other reason was; “..my husband told
me not to do.. we need money...we can use premium amount for some other need..”
5. Cultural Beliefs
The culture shapes people’s perception about health insurance and ayurvedic
system of treatment, home medicine (using herbs, shrubs and spices available at home to
treat illness). Many participants stated;
“…we trust ayurveda medicines..it does not have side effects…”, “..if we buy
insurance, we will get illness..”
6. SHG Membership Status
The formation of new SHG after the enrolment time and inability to continue
SHG membership was another reason;
“..my family does not allow me to continue in group (SHG), they have told me to
pay back loan as quickly and leave the group..” “..we joined the group (SHG) in May, we
could not join Suraksha..”, “…we have to join in February, we were not in group then..”

215
7. Other Reasons
Inability to make time to attend meetings or pay the premium was highlighted; “..I
was away from home when the enrolment took place in my village…”.
The domestic responsibilities kept many participants busy;
“..we have to cook, wash clothes and take care of elderly, so we take medicines
from pharmacist or drink ‘kashaya’ (home medicine) if we fall sick”, “even if we have
Suraksha, we have to forgo the days’ work but we cannot do that as we are poor”.
Since the household was the unit of enrolment specified by SSP, large families
complained of their inability to pay the premium;
“..we have 6 people in the family but father only earns… we cannot pay premium
for the entire family..”; “…I want to enrol my parents..others are healthy, then why enrol
all?..”
5.5.2 Scheme Related Factors
1. Inadequate Benefit Package
The participants stressed the need to include outpatient treatment in the benefit
package and the need to increase the claim benefits;
“..we get illness like fever, cough, we cannot get benefit as outpatient is not
covered..”. “…we always go to a private clinic…it does not come under network…”,
“…amount of benefit is too less, what will you get with Rs.5000?...they (insurers) should
increase benefit amount and include common ailments…”
Exclusion of many diseases from the coverage was another concern;
“…Suraksha does not cover many diseases which are common here…why insure
when we cannot get the benefit?..”
2. Complex Claim Procedure
Lack of knowledge about the rules for submitting the pre-authorisation forms and
procedure to claim benefits was a problem as highlighted by field staff;
“…many eligible claims were rejected as insured members did not submit
Suraksha card within 24 hours after admission…”, “..people do not bother about the

216
name they give in the card and in the hospital… we have suggested them to give alias
names and correct age..”
3. Lack of Good Health Care Facilities
Perceived quality of care at health facilities such as cleanliness, absence of
medicines, and delay in payment to hospitals influenced non-enrolment;
“….bed for men and women are kept together…. There is no privacy..the ward is
not clean…”, “..doctors do not discharge even if we are better as money is not sent to
them by Suraksha office”, “..drugs are not available …”.
The distance factor was expressed to be another concern;
“..good Suraksha hospitals are in Kumta (a city in UK district) which is far
away…”, “..doctors do not see us well, we have to go Hubli or Manipal for good
hospitals (far away city)..”.
4. Timing of Collection of Premium
Availability of money or time during the enrolment period (February of every
year) stands out as a key factor affecting enrolment in SSP. There were opposing views
among participants on the time of enrolment, it was suggested that;
“…Suraksha should be kept open throughout the year”, “…no, it should be done
in February, we will keep postponing if it can be done any time in the year..”, “..we do
not have money in March season…if it was monsoon, we get more money working in
fields..”.
5.6 Summary
The incidence of illness was the highest in insured households than uninsured
households in the previous year of the study. This may have influenced enrolment and
may jeopardise the financial health of the programme. To prevent this, usually health
insurance scheme incorporates mechanisms to detect and control pre-existing illnesses,
owing to its negative effect on the financial sustainability of the scheme. Even SSP had
the household as a unit of enrolment to curtail the inclusion of high-risk individuals.
Certain features to curtail adverse selection such as waiting period, exclusion of pre-
existing illnesses and reference system were absent in SSP. However, adverse selection,
217
measured by the presence of chronic illness, was found to be absent in SSP (regression
analysis). Moreover, the incidence of illness alone would not push SSP towards financial
difficulties. Previous research findings confirm the absence of adverse selection in the
presence of higher illness episodes among insured households in MHI schemes (Gumber,
2001). Hence, the finding of this study adds to the literature that advocates important role
of MHI in health financing contrary to those who highlight information asymmetry as
one of the major barriers to any MHI service to the poor.
In addition to risk (need) factors, certain demand factors (predisposing and
enabling) and supply (features of SSP) determines enrolment in MHI. However, other
than distance to hospital and the area of residence, various household characteristics such
as education and marital status of the head of the household, size of the household did not
determine enrolment. In addition, age, gender and job status of the head of the household,
income class and chronic illness was not associated with enrolment. To confirm these
findings, the present study made changes to model specification by analysing the cases of
the households of unemployed and labourer head of the families. In these cases, the
probability of enrolment was associated with longer distance to hospital and semi-urban
area of residence.
The study found higher representation of semi-urban residents in SSP due to the
presence of large number of network hospitals. Moreover, higher likelihood of enrolment
was associated with longer distance to hospitals, which contradicted theoretical
expectations as documented by Schneider and Diop (2001) and Msuya (2004). Our
finding supports the fact that living away from hospitals increases the cost of treatment,
which encourages families to seek alternative mechanism to reduce the cost such as
health insurance.
Social inclusion in SSP is absent since there was no higher representation of the
poorest households. Despite the credit facility to pay the premium, poorest households
(Q1) stayed out of SSP. Probable cause for this finding is the Rashtriya Swasthya Bhima
Yojana (RSBY) started in 2010. Central government of India introduced a national level
MHI scheme (RSBY) targeted at the families below the poverty line (BPL). These
218
families could buy a health
h card for
f 30. Heence, larger section of the
t poorest householdss
did not enrrol in SSP. Various
V dessign featurees of SSP namely
n highh premium, withdrawall
of subsidy and inflex
xibility in the
t paymen
nt of the prremium alsso deterred poor from
m
j
joining SSP
P. This fiinding is baacked by th
he FGDs th
hat revealeed inability to pay thee
premium, sseasonality of income and lack of
o flexibility
y in the colllection of premium
p ass
factors conttributing to non-enrolm
ment in SSP
P.
SSP
P membersh
hip to somee extent cann be attributted to somee componen
nts of sociall
capital nam
mely solidaarity (group
p influencee; ‘all SHG
G members are mem
mbers’) and
d
reciprocity (concern fo
or others; ‘let others beenefit’). Hou
useholds ennrolled in SS
SP to hedgee
risk using bbenefits of SSP, to pro
ovide financcial security
y for onesellf and otherrs, influencee
of the grouup to curtaill medical ex d to access good hospiitals in casee of a need.
xpenses and
Since SSP is nested in own socio ecconomic deevelopment programmee
n SKDRDP, a well-kno
in Karnatakka, trust bu
uilding was not a challenging task
t for SS
SP. The sollidarity and
d
concern forr others and
d mutual hellp underlyin
ng enrolmen
nt can be atttributed to pre-existing
p g
trust amongg the SHG members.
m T memberrs were willling to crosss-subsidise the risk and
The d
income, an important ingredient
i o any insurance mechaanism.
of
Thuus, families living in seemi-urban areas
a and aw
way from th
he hospitalss were moree
likely to ennrol in SSP
P. Some co
omponents of social capital
c playyed significcant role in
n
influencingg group mem
mbers to enrrol in SSP. R
Risk aversion in the fo
orm of safeg
guarding thee
family agaiinst medicaal expenses in future was
w observed
d. SSP beneefit packagee motivated
d
many housseholds to join
j SSP; at the sam
me time, cerrtain featurres of beneefit packagee
namely infl
flexibility in
n the collecttion of the premium an
nd high preemium prev
vented somee
from joininng the progrramme. Poo
orest were excluded
e du
ue to high premium
p co
oupled with
h
low incomee and the availability
a of other lesss expensiv
ve alternativves (RSBY)). Memberss
did not parrticipate in the decisio
on making oof benefit package
p or setting prem
mium. Thatt
might hav
ve contribu
uted to low
wer enrolm
ment since members could not link theirr
involvemennt in SSP with
w broadeer goal of ‘h
health for all
a memberrs’ which was
w possiblee
through th
heir premiu
um contribu
ution. This section prrovides strong eviden
nce for thee
exclusion of
o the pooreest due to certain
c desiggn features.. Hence, it bbecomes im
mperative to
o
21
19
explore the association between design characteristics and the performance of SSP.
Chapter 6 explicates this relationship to provide a deeper understanding of the working
and outcome of SSP.

220
 CHAPTER 6

RESOURCE MOBILISATION AND

EFFECT OF FEATURES OF SSP ON THE
OUTCOME
6.1 Introduction
Resource mobilisation determines the financial sustainability and viability of the
MHI scheme. It is directly measured by the cost recovery ratio (financial sustainability),
amount of resources mobilised (premium collection) and indirectly by quality impact on
healthcare. Moreover, various characteristics of SSP (technical, management,
organisational and institutional) affect the performance of SSP in terms of resource
mobilisation (RM), social inclusion (SI) and enrolment, financial protection (FP),
financial sustainability (FS) and viability of the programme. So far, we know that SSP
provides partial financial protection in terms of reduction in OOPE, CHE and less
reliance on borrowing. It is also proven that poorest of the target population was excluded
from insurance coverage. This prompted us to explore the factors that prevented complete
financial protection and social inclusion. i) What is the amount of resources mobilised by
SSP? ii) Is SSP financially sustainable? iii) What is the perception of quality of care by
members and non-members of SSP? iv) What aspects of technical characteristics shaped
the performance of SSP in terms of FP, enrolment and RM? v) What are the management
related factors that determined enrolment and RM? vi) What is the role of organisational
characteristics in RM and FS? vii) What are the institutional characteristics that influence
the viability of SSP?
Incurred claims ratio, expense ratio, combined ratio and net income ratio
measures financial sustainability. Incurred claims ratio was calculated by dividing the
claims by the premium collected and it denotes the extent of financial protection given to
members. A higher ratio means better financial protection but lower financial
sustainability. Expense ratio was calculated by adding the expenses (administrative/
operating expenses and taxes for the insurer) and dividing it by the premium collected.
Net income ratio was calculated by adding the claims and expenses and deducting it from
the premium collected and the resulting number was divided by premium amount. Net
operating profit was calculated by adding claim amount, operating expenses, tax payment
by the insurer and deducting it from the premium. Combined ratio was calculated by
adding incurred claims ratio and expenses ratio.
221
 Quality of care of the network hospitals can enhance enrolment in SSP and retain
the existing members due to satisfaction from the treatment. SSP can pressurise the
providers to improve the quality of services through the stipulations in the contract with
them. To understand the perception of the quality of care offered at network hospitals,
this study compared insured, uninsured and newly insured individuals. It considered
certain criteria of quality of care namely cleanliness of hospitals, expertise of doctors,
expertise of nurses, friendliness of staff, availability of facilities, availability of medicines
and time taken by the doctors to examine the patients. SSP selects the hospitals based on
certain criteria that include the availability of basic infrastructure and good treatment.
Hence, the perception of a good quality of care at the network hospitals by insured
individuals is expected.
The data collected by interviewing SSP administrators, project officers and field
staff comprises the primary data (qualitative and quantitative). The perception of insured,
newly insured and uninsured individuals on the quality of care of the hospital was
assessed through a five point Likert rating scale (1=highly unsatisfactory; 5=highly
satisfactory). Secondary data was collected from the brochures, annual reports and
promotion materials of SSP. Firstly, the analysis focused on the resources mobilised
(premium collected) by SSP over the years and claim benefits disbursed. Secondly, the
study ascertained financial sustainability by calculating claims ratio and related ratios.
Thirdly, quality of care provided at health facilities as perceived by insured, uninsured
and newly insured respondents were analysed using discriminant analysis. Fourthly, the
study explored an association between various characteristics (technical, management,
organisational and institutional) and the outcome of SSP (FP, SI and RM).

222
6.2 RESOU
URCE MO
OBILISATIION

Ressource mobiilisation refflects attracttiveness of MHI and ddetermines its

i viability.
The amounnt of revenu
ue mobilised
d depends oon the premium per inddividual and
d number off
the househholds enrollled in MH
HI. In addittion, the nu
umber of insured mem
mbers as a
percentage of target population gauges
g the ppopularity of
o MHI. Thhis section presents
p thee
ngs on the revenue
study findin r mo
obilisation, financial su
ustainabilityy and quality
y of care ass
perceived by
b sample in
ndividuals.
6.2.1 Reven
nue Mobiliisation
Revvenue colleection deterrmines the resource mobilised
m bby SSP an
nd financiall
sustainabiliity. This seection analy
yses the prremium stru
ucture, prem
mium colleection sincee
inception, premium paid
p to in
nsurance coompanies and
a benefitts sanction
ned by thee
programmee.
6.2.1.1 Preemium Stru
ucture of SS
SP
Thee premium collected from
f the m
members forrm the main source of
o financiall
resource fo
or SSP. Gov
vernment su
ubsidies andd external aiid were not availed by SSP. Tablee
6.1 depictss the premiu
um structurre for a fam
mily of two
o, three, fouur, five, six
x and seven
n
members. The premiu
um payable for the firsst member of
o a family was 350 in 2011-12.
There has been
b a sign
nificant chan
nge in the ppremium co
ontribution and margin
nal cost perr
member sin
nce 2007. Until
U 2007, marginal cost for a member w
was higher for a smalll
62) than a large family
family ( 16 09), but the changes made in 2007
y ( 132) (Shhetty N. 200 7
removed suuch a differrence. The premium
p am
mount increeased from 220 in 20
010-11 to
350 in the yyear 2011-1
12. A familly with five members had
h to pay aan additionaal 250 perr
year and a family
f of seeven membeers had to ppay an addittional 310 per year. Small family
y
had to pay more than a large fam
mily in term
ms of premiium burdenn compared to the yearr
2010-11. A
Average cost per memb
ber decreaseed with sizee of family. It was 262
2 for two--
member fam
mily and 200 for sev
ven-memberr family in 2011-12.
2

22
23
 Table 6.1 Description of the Premium and Eligible Limit for Cashless Treatment
Number Annual Annual Marginal Average Yearly Eligible
of family contribution contribution cost cost Percentage limit
members (2010-11) (2011-12) ( 2011-12) (2011-12) change in
premium
One 220 350 - 350 59.1 5000
Two 365 525 175 262 43.9 10000
Three 510 700 175 233 37.3 15000
Four 625 875 175 219 33.6 20000
Five 800 1050 175 210 31.3 25000
Six 945 1225 175 204 26.6 30000
Seven 1090 1400 175 200 28.4 35000
Source: SKDRDP’s ‘Sampoorna Suraksha’ Micro-Insurance Campaign Material (for 2010 -11 and 2011-12)
(In Indian rupees)

6.2.1.2 Enrolment and Premium Collection

SSP is one of the successful MHI programmes in terms of enrolment and renewal
of membership. There was a phenomenal growth in the number of families and
individuals enrolling in the scheme since inception (Table 6.2). During the first year in
2004-05, 1.86 lakh members from 54,000 families joined SSP. In 2010-11, 16 lakh
members from 418,956 families joined the scheme. However, recent (2011-12) increase
in enrolment was negative (-0.11 %). The enrolment as a percentage of target population
ranged from 32 percent to 54 percent with an average rate of 41.6 percent.

224
 Table 6.2 Premium
P Co
ollection and
d Coverage of Familiess under SSP
P
Year Numbeer of Number of Premiuum
familiees covered memberss collectted ( )
2004-05 54000 186000 168129933
2005-06 77078 195600 286754467
2006-07 146722
2 403828 574423349
2007-08 223389
9 721203 1069000589
2008-09 252542
2 932682 1541700730
2009-10 294374
4 1177325 1680833995
2010-11 419979
9 1662089 2783388765
2011-12 420302
2 1660185 3640855225
Total 1174510053
S
Source: SKDRD
DP’s ‘Sampoorrna Suraksha’ M
Micro-Insurancee Campaign Maaterial (2011-12
2)

Figu
ure 6.1 dep
picts yearly
y growth inn membersh
hip in term
ms of both number off
families and number of
o members since incepption. Numb
ber of membbers enrolleed increased
d
by 42.7 perrcent in the second yeaar of operatiion (2005-0
06), by 90 percent in th
he third yearr
(2006-07), but the grow
wth was slo
ower in the llater years until
u 2010-111. Numberr of familiess
enrolled haad a similar pattern witth highest yearly
y increase in 20066-07 and a steep
s fall in
n
2008-09 an
nd 2011-12. Premium mobilised increased
i b 100 perccent in 2006
by 6-07, by 86
6
percent in 22007-08, 44
4 percent in 2008-09 an
nd by 65 perrcent in 20110-11. The scheme
s wass
not successsful in mob
bilising enough premiu
ums in the year 2009-10, as the incrementall
increase waas just 9 perrcent.

22
25
 1200
106.4
1000

800 78.5 Nuumber of

9
90.3 fam
milies
600 covvered

Nuumber of
400 52.2 26.2
41.1
meembers
42.7 29.3 42.6
200
166.5
13
5.1 -00.11
0
0.07
2005-06 20006-07 2007-08 2
2008-09 2009-10 2010-11 2011-112
-200

Figuure 6.1 Grow

wth Rate off Enrolmentt over a Periiod of Timee
6.2.1.3 Allocation of Premium to Insurance Comp
pany and SSP
S
In 2004-055, SSP colleected a totaal premium of 16,812
2,933 and iit transferreed
10,762,208 to Unitted Insuran
nce Compan
ny for medical benefitss componennt. It retaineed
6,050,725 to proviide the speccial benefitss and to meeet the operaating expensses (Table 6.3).
6
In the next
n year, it paid 59.8 percent
p to IC
CICI Lombbard Generaal Insurancee Company Ltd.
and retaained the reemaining am
mount. Nearrly sixty-two
o percent was
w paid in 22006-07 and
d 64
percent in 2007-08
8 to ICICI Lombard General
G Insurance Com
mpany Ltd. In 2008-09, it
paid 54 percent of premium to
o Reliance Insurance
I C
Company. Siince 2009-110, it contraacted
ur general iinsurance co
with fou ompanies namely
n Unitted Insurancce Companyy Ltd., Orieental
Insuran
nce Compan
ny Ltd., New
N India Assurance Company Ltd. and National In
ndia
Insuran
nce Compan
ny Ltd. In 2009-10, itt paid nearlly half of the
t premium
m to insuraance
nies and the retained am
compan mount was used
u to meeet special beenefits claim
ms.

226
 Tablle 6.3 Premiium Shared
d with Insuraance Compaanies
Yearss Paid to insurer ( ) Rettained at SS
SP ( )
2004-05 10762208 (64.1%
%) 605
50725 (35.99%)
2005-06 17125483 (59.8%
%) 115
549984 (40.2%)
2006-07 35207731 (61.3%
%) 222
234618 (38.7%)
2007-08 68514286 (64.1%
%) 383
386303 (35.9%)
2008-09 83322145 (54%)) 708
848585 (45%
%)
2009-10 87000000 (51.8%
%) 810
083995 (48.2%)
2010-11 153249342 (55.1
1%) 125
5089423 (444.9%)

6.2.1.4 Ben
nefits Proviided by SSP
P since Inception
Thee analysis of
o SSP sincce inceptionn in provid
ding medicaal and speccial benefitss
(Table 6.4) reflects its perform
mance in ffinancial prrotection. T
The amoun
nt of claim
m
sanctioned and disburrsed increaased from 26,421,143
3 in 2004-05 to 51,,122,218 in
n
2006-07 annd 289,317
7,396 in 2010-11. Overr the years, two thirds of the claim
med amountt
was for hosspitalisation ne fifths for delivery alllowances aand a lower percent forr
n, nearly on
death, conssolation for natural callamities andd rest allow
wances. In teerms of thee amount off
claims settlled, there was
w an increease of 1033.6 percent in 2006-077 compared to 2004-05
5
and 78.5 percent in 2007-08,
2 77
7 percent in
n 2008-09 and
a 52 perccent in 2009
9-10, 114.6
6
percent in 22010-11.
Heaalth treatmeent benefit increased
i b 104 perccent during the period 2004-05 to
by o
2006-07, delivery
d y 122 percent, death consolationn by 57 percent, and
alllowance by d
domiciliaryy treatment allowance by
b 330 perccent. Death
h consolatioon benefit in
ncreased by
y
223 percentt in 2008-09
9 compared
d to the prev
vious year. Increase
I in the amountt sanctioned
d
for health trreatment an
nd delivery allowance
a w
was moderaate in 2008-009.

22
27
 Tabble 6.4 Beneefits Given Under
U the S
Scheme sincce Inceptionn
Year H
Health Delivery Death Domiiciliary Consolation Total
C
Treatment allowancee Consolaation treatm
ment- f natural
for
Rest c
calamities
allow
wances
2004- Number 7737 2593 227 360 8893 11810
05 (65.5) (22) (1.9) (3) (
(7.6)
Amount 226421143 7017268 11350000 2020990 6
648061 354235622
( ) (74.6) (19.8) (3.1) (0.6) (
(1.8)
2005- Number 8587(65.4) 2761(21) 208 (1.55) 518(44) 1
1061(8) 13135
06
Amount 224989658 6012452 11750000 2231665 4416392 328166677
( ) (76.1) (18.3) (3.6) (0.7) (
(1.3)
2006- Number 16274 6045 308 882 2
232 23746
07 (68.5) (25.4) (1.3) (3.7) (
(1)
Amount 51122218 133723677 18395000 9586000 2
260520 675532055
( ) (75.67) (19.8) (2.7) (1.4) (
(0.4)
2007- Number 229326 10202 525 1644 4
475 42171
08 (69.5) (24.2) (1.2) (3.8) (
(1.1)
Amount 94236027 217615600 31690000 11690009 3
397800 1207333996
( ) (78) (18) (2.6) (1) (
(0.4)
2008- Number 447006 13166 4680 2549 3
380 67781
09 (69.34) (19.42) (6.92) (3.76) (
(0.56)
Amount 172696021 272183577 115843550 16982200 3
342550 2135394778
( ) (80.87) (12.75) (5.43) (0.8) (
(0.2)
2009- Number 58734 15993 6290 1871 4
463 83351
10 (70.47) (19.18) (7.55) (2.24) (
(0.54)
Amount 2225103867 344574388 152078000 11553370 3
375200 2762996775
( ) (81.47) (12.47) (5.5) (0.41) (
(0.15)
2010- Number 70952 40553 19420 2168 8
869 133962
11 (53) (30.3) (14.5) (1.6) (
(0.6)
Amount 2289317396 110110432 537398118.6 1497157 8
828821.2 4554936225
( ) (63.5) (24.2) (11.8) (0.32) (
(0.18)

2011- Number Speciial 2387

12 Amount beneffits 5433086
( ) arecoonsidered
Source: S
SKDRDP’s ‘Saampoorna Surakksha’ Micro-Inssurance Campaiign Material (2011)

Amount off benefits per

p claim (Figure
( 6.22) shows an
n increasinng trend in the
benefitss sanctionedd from inception to 2010-11, exceept in the yeear 2005-066. Total amo
ount
per claiim was 3,0
000 in 2004
4-05, that reduced to 2,500 in 2005-06
2 butt increased to
t

228
3,400 in 20
010-11. Heaalth benefit per claim was
w 3,415
5 in 2004-055 and 4,07
78 in 2010--
11; delivery
y allowancee was the highest in 200
04-05 ( 2,7
706) and 2010-11 ( 2,7
715).
4500

4000 4078
38833
3674
3500 34155 3400
3214 33315
Average claim benefits

3141 3150
3000 30000 2910 2845 2863
27066 2715
2500 2500 Health
2177 2212 2133 2154
2000 2067 Delivery

1500 Total

1000

500

0
2004-05 2005-06 2006-007 2007-08 20008-09 2009-10 2010-11

Fiigure 6.2 Am
mount Sancctioned by SSP
S per Claiim
6.2.2 Finan
ncial Sustaiinability off SSP
Thee sustainabillity of MHII depends on nable balancce of funds maintained
n the reason d
(Zhang et al. 2009). At the sam
me time, huge
h surplu
us of funds underminees financiall
protection provided
p obilises the funds from
to the membeers. SSP mo m premium contribution
c n
to provide claim ben
nefits. The amount of benefits directly vaaries with health caree
utilisation. Albeit, thiss study did not observve a high co
orrelation between
b adm
mission ratee
and claims ratio (r=0.32). Admisssion rate w
was 5.67 peercent in 20004-05, reacched lowestt
07-08, thereeafter increeased to 6.95 percent in
level in 200 n 2008-09. It was 7.88
8 percent in
n
2010-11, hiighest sincee inception.
Incuurred claim
ms ratio, net
n income ratio, com
mbined ratiio and exp
pense ratio
o
measures fiinancial susstainability. Incurred cllaims ratio was
w very high (208%) in 2004-05,
but it reducced significcantly (113%
%) in 20077-08. It detteriorated too an alarmiing level off
197.5 perceent in 2009
9-10. On an
n average, th
he claims raatio was 1007.5 percentt in the lastt
six years, which
w denottes that the scheme paiid 107.5 in
n claims foor a premium
m of 100.
Medical claaims ratio declined
d fro
om 245.5 percent in 2004-05 to 1160.7 perceent in 2008--
09(Table 66.5). Speciaal claims raatio declineed from 14
48.8 percen
nt in 2004--05 to 49.5
5
22
29
percent in 2008-09. Special claims werre high in 2010-11, however,
h m
medical beneefits
claim drastically
d reeduced to 189
1 percent in the sam
me year to brring down tthe total claaims
o 163.6 perrcent. The net operatiing loss for the progrramme deccreased from
ratio to m
22,286,165 in 2004
4-05 to 10
0,071,113.9 in 2005-066, but increaased 20,4994,666 in 20
006-
07, 33,,819,808 in
n 2007-08, 35,603,,786 in 20
008-09. It sharply inncreased to
o
186,869
9,494 in 20009-10 and 219,224,48
81 in 2010-11.
Combined ratio declin
ned from 239.5 percentt in 2004-05
5 to 140.3 ppercent in 20
007-
08, butt later increeased to 15
59.8 percentt in 2008-0
09 and 187
7 percent inn 2010-11. The
averagee combined
d ratio since inception
n was 176.8
85 percent. It was higgh for med
dical
claims (213.2%)
( th
han special claims (91.8%). Net inncome ratio on an averaage for med
dical
w -194.3 percent, fo
claims was or special claims it wass 32.9 perceent and totaal (both med
dical
and speecial claims)) was-192.8
8 percent.
The claimss ratio, calcu
ulated on th
he special cclaims, decllined from a high leveel of
148.8 percent
p in 22004-05 to nt in 2007--08 and 49.5 percent in 2008-09
o 69 percen 9. It
increaseed to 132.8
8 percent in
n 2010-11. Special
S claiims coverag
ge risk wass borne by SSP
that inccurred an un
nderwriting loss of 4
40,001,252 in 2004-05 but it earneed a profit of
o
1,971,6
679 in 2005-06, 3,996
6,723 in 20
006-07, 8,6619,418 in 2007-08 annd 28,799,,354
in 2008
8-09. Howevver, it had to
o suffer a lo
oss of 54,7
743,586 in the
t year 20110-11.

230
 Table 6.5 Incurred Claims Ratio, Incurred Expense Ratio and Combined Ratio of SSP
Particulars 2004- 2005- 2006- 2007- 2008- 2009- 2010-
05 06 07 08 09 10 11
Admission 5.67 5.9 5.6 5.55 6.95 6.88 7.88
rate (%)
Incurred Medical 245.5 146 145.2 137.5 160.7 318 189
claims ratio claims
(%) Special 148.8 67.8 73.9 69 49.5 68.1 132.8
claims
Total 208.8 115.8 117.7 112.9 138.7 197.6 163.6
Incurred Medical 24.4 24.4 24.4 24.4 16.6 17.5 18.5
expense claims
ratio (%) Special 6.3 6.1 3.2 3 4.5 4.7 4.9
claims
Total 30.7 30.5 27.6 27.4 21.1 22.2 23.4
Net income Medical -170 -172.6 -101.5 -89.9 -106.3 -378.4 -341.3
ratio (%) claims
Special -66 17 18 22.5 40.6 22.2 -43.8
claims
Total -236 -155.6 -83.5 -67.4 -65.7 -356.2 -385.1
Combined Medical 269.9 170.4 169.6 161.9 177.3 335.5 207.5
ratio (%) claims
Special 155.1 73.9 77.1 72 54 72.8 137.7
claims
Total 239.5 146.3 145.3 140.3 159.8 219.8 187
(Assumed 16 % service tax rate and 10 % of premium as operating expenses for insurer till 2007-08)

6.2.3 Perceived Quality of Care of Hospitals

The perception of the cleanliness of the hospitals was high among insured
respondents (mean 3.98) than newly insured (3.69) and uninsured (3.83) respondents
(Table 6.6). Regarding the perception of cleanliness of hospitals that respondents
frequently visit, the percentage mean for insured members was 79.57 which was higher
than that of newly insured (73.88) and uninsured respondents (76.65). Hence a significant
difference in the perception of cleanliness of hospitals by insured and uninsured and
newly insured individuals was evident (p<0.05). Perception of expertise of doctors by
newly insured individuals (mean 3.96) was better than that of insured (mean 3.95) and
uninsured respondents (mean 3.91). However, this finding was not significant suggesting
231
lack of difference among the studied individuals. Similarly, no difference in the
perception of care by nurses was noted (p>0.05). Insured members perceived hospital
staffs to be friendly (percentage mean 77.26) compared to other two groups. Insured
members perceived good quality of care measured by the availability of medicines and
facilities (percentage mean 70.67 and 72.55 respectively). Insured members felt that
doctors at network hospitals spent more time compared to other two groups.
Table 6.6 Quality of Care at Hospitals: Comparison of Insured and Uninsured Groups
Mean Standard Percentage Test
deviation mean value
Cleanliness of hospitals Insured 3.98 0.927 79.57 25.322*
Newly insured 3.69 0.921 73.88
Uninsured 3.83 0.892 76.65
Expertise of doctors Insured 3.95 0.96 79.09 1.316
Newly insured 3.96 0.856 79.13
Uninsured 3.91 0.909 78.19
Care given by nurses Insured 3.73 0.954 74.62 5.282**
Newly insured 3.59 0.996 71.86
Uninsured 3.61 0.969 72.14
Friendliness of staff Insured 3.98 0.882 77.26 26.145*
Newly insured 3.56 0.939 71.26
Uninsured 3.62 0.962 72.42
Availability of facilities Insured 3.63 1.04 72.55 49.347*
Newly insured 3.22 1.102 64.43
Uninsured 3.11 1.118 62.2
Availability of medicines Insured 3.53 1.132 70.67 14.768*
Newly insured 3.28 1.088 65.68
Uninsured 3.28 1.141 65.66
Time taken to examine by Insured 3.55 0.995 71.01 50.453*
doctors
Newly insured 3.4 1.047 67.98
Uninsured 3.04 1.049 60.77
Kruskal Wallis *p<0.05
Further probe using discriminant analysis was performed to know various quality
factors that differentiate insured and uninsured (cases of newly insured were included)
individuals. Predictor variables were cleanliness of hospitals, expertise of doctors, care
232
given by nurses, friendliness of staff, availability of facilities and medicines and time
taken for physical examination by the doctors. The aim was to investigate the quality
factors that differentiate insured from uninsured individuals. It is known that insured
individual’s visit district and regional hospitals than home medicine, government
hospitals and private clinic.
The structure matrix considered insured and uninsured (including newly insured)
individuals as a grouping variable and the factors that determine quality as predictor
variables. The discriminate function revealed a significant association between the groups
and all predictors with Wilk’s lambda (p=0.00) (Box’s M= 76.053; F=2.752, p=0.000).
Availability of facilities such as laboratories and X-ray (.781), examination time (.587),
friendliness of staff (.533), cleanliness of hospital (.501) and availability of medicines
(.401) differentiated insured and uninsured individuals. Expertise of doctors (.052) and
care by nurses (.265) were not loaded on the discriminant function.
Discriminant analysis showed that operational quality of care (laboratory, X ray
and diagnostic equipment and cleanliness of hospitals, availability of medicines),
friendliness of support staff, and doctor quality of care (examination time) differentiated
uninsured from insured households. Thus, insured individuals perceived better quality of
care at network hospitals. SSP selected network hospitals by applying stringent criteria
that stipulates basic facilities, cost of treatment and good treatment. This has resulted in
better perception of quality of care at network hospitals.
6.2.4 Summary
High level of incurred claims ratio implies two things: i) financial protection to
those who needed it the most, and ii) insufficient premium collection. In 2004-05, claim
sanctioned by SSP was high (reflected in high claim ratio, combined ratio and net loss).
However, it reduced significantly and average claims ratio was healthy in 2009-10. The
performance deteriorated in 2010-11 with very high claims ratio, higher average rupee
sanctioned per claim, combined ratio, net income ratio and net loss. Moreover, in the
year 2010-11, incremental increase in membership was negative. Medical claims was
the highest in 2009-10, special benefits reached its peak in 2010-11. Hence, lower
233
premium collection coupled with high claims ratio, combined ratio and net loss threaten
the financial sustainability of SSP. Insurance companies absorbed the loss incurred due
to medical benefits. Due to low coverage of target population, insufficient premium
collection, high claims and low income of members, resource mobilised by SSP was
low.
The membership base of SSP declined owing to changes in the policy in 2011-12.
From 2008-09 to 2011-12, BPL families were given a special concession in the premium
amount. Since then, SSP removed the distinction between BPL (below poverty line) and
APL (above poverty line) while determining the premium amount. The removal of BPL
concession in the premium coupled with introduction of universal health insurance
(UHS) adversely affected enrolment. Since 2009-10, Yeshasvini scheme of Karnataka
state was offered to the members of co-operative institutions at very low costs. Many of
the SKDRDP members were also the members of these co-operative societies; hence,
they could enrol in these schemes resulting in non-renewal of their membership in SSP.
In addition, central government conceptualised Rashtriya Swasthya Bhima Yojana
(RSBY) in 2008, which came into effect in Karnataka in 2010. This scheme provided
inexpensive insurance benefits to BPL families. In the same year, premium amount was
increased to meet higher claims and other cost of operation including inflation. These
factors might have contributed to negative or very low increase in membership in SSP in
2011-12.
Net profit was earned by special benefits component from 2005-06 to 2009-10.
However, SSP special benefits had to incur huge loss in the year 2010-11 due to two
times increase in delivery allowances and death consolation. The performance of medical
benefits covered by the insurance companies suggests unhealthy financial status since
2009-10. As the incurred claims ratio for medical benefits was relatively high throughout
the years since inception, many insurance companies did not contract with SSP. Since
2009-10, overall performance deteriorated as incurred claims ratio, incurred expenses
ratio, net income ratio, and combined ratio increased to an alarming level. This
necessitated external financial assistance in the form of a loan from SKDRDP. Any
234
deficit in special com
mponent of SSP wass financed through a credit fro
om MFI off
SKDRDP.
Theere are several plausiblle explanatioons to the high
h claim rratio in SSP
P. The main
n
factor is SSP technical design features
f succh as exclu
usion of outtpatient treatment thatt
might havee motivated
d insured individuals
i to seek in
npatient carre. SSP did
d not havee
deductibless, co-paymeents or waaiting period
d, which reemoved thee financial barriers to
o
access caree and increased utilissation. Eveen the cash
hless system
m of paym
ment to thee
providers eased
e the diifficulties in
n claiming from SSP for
f insured members. Field staffss
were suppoortive to the members in
n solving prroblems facced during cclaibursal frrom SSP.
Thee study obseerved higher utilisationn of health services
s resuulting in thee high levell
of claim ratio. Howev
ver, the poorrest individ
duals do nott seek care jjust to claim
m from SSP
P
as hospitaliisation invo
olves other expenses su
uch as transsportation aand food in addition to
o
the loss off wages du
ue to absen
nce from th
he work. Hence,
H SSP promoted welfare by
y
increasing uutilisation and
a lowerin
ng the burdeen of OOPE
E for the pooor househo
olds but nott
over-utilisaation. A hig
gher claims ratio denotees better fin
nancial prottection to th
he memberss
who got more than wh
hat they con
ntributed to
o SSP. The welfare im
mpact of SSP
P cannot bee
nce it proviided financcial protectiion to almo
ignored sin ost one andd half lakh individualss
(claims givven 45.5 crore
c in 2010-11) who
o otherwisee had to sufffer from th
he financiall
consequencces of mediccal illness. Moreover,
M S did mo
SSP obilise huge resources (almost
( 36
6
crore in 2011-12) whicch would no
ot be possib
ble in its abssence.
Ressource mob
bilisation in
ndirectly deepends on the qualityy of care of network
k
hospitals. IInsured ind
dividuals so
ought care at network
k hospitals and perceiived certain
n
quality variiables such as cleanlin
ness, availabbility of faccilities (labo
oratory and diagnostic))
and medicines, friendlliness of staaff and timee spent by th
he doctor inn examining
g the patientt
to be higheer in those hospitals. Hence,
H the results
r of th
he study suubstantiate and
a providee
support for active role of SSP in quality
q imprrovement off the hospitaals.
SSP
P does not have
h reservee funds to reduce
r the risk
r of deficcit and ensu
ure financiall
sustainabiliity. Lack off reserves reeflects high level of claaims ratio aand persistent deficit in
n
revenue co
ollection. One
O option is to impprove the revenue
r thrrough increease in thee
23
35
premium, expand the risk pool through higher enrolment or plug the deficit through
external donor assistance. Before taking a decision on increase in premium, scheme
administrators should consider its probable negative effect on enrolment. This requires a
suitable trade-off between premium and enrolment since increase in the cost of insurance
would adversely affect the expansion of risk pool. Another welfare reducing option is to
either bring about changes in benefit package, reduction in claim amount or remove
certain benefit covered by SSP to make it financially sustainable. The third option is strict
monitoring of supplier and insured members’ moral hazard through gatekeeper system or
referral mechanism.
If the financial sustainability of SSP were in danger in the years to come, potential
benefits of SSP would be marred. Nevertheless, financial sustainability would not hinder
the viability of SSP due to i) nesting of SSP in SKDRDP which provides administrative
and managerial assistance thereby brings down the cost of operation and ii) health risk
coverage by public insurance companies and iii) trust of members that would strengthen
loyalty of members. The long-term feasibility of SSP calls for strict control over costs,
better revenue collection and restriction on the claims sanctioned under special benefits.

236
6.3 EFFECT OF FEATURES OF SSP ON FINANCIAL PROTECTION,
ENROLMENT AND RESOURCE MOBILISATION
Technical characteristics such as revenue collection, risk pooling and strategic
purchasing affects the performance of MHI in terms of FP, enrolment and RM (Preker et
al, 2004). Certain technical design features affect the enrolment thereby revenue
collection. Management characteristics namely staff (leadership and capacity in terms of
management skills), culture (style of management and structure) and access to
information (on financial resources, health information and behaviour) determine RM
and enrolment. Forms of organisation (economies of scale and scope, contractual
relationships), incentive regime (extent of decision rights, market exposure, financial
responsibility, accountability, and coverage of social functions) and linkages (extent of
horizontal and vertical integration or fragmentation) are organisational characteristics that
influence RM and financial sustainability. Certain institutional characteristics such as
stewardship (strategic and operational decisions, regulations), governance (ownership
arrangements), insurance markets (rules on revenue collection, pooling, and transfer of
funds) and factor/ product markets determine viability and performance of SSP (Preker et
al, 2004). This section focuses on the role of these characteristics on the outcome of SSP
in terms of RM, FP, enrolment and financial sustainability.
6.3.1 Technical Design Characteristics
6.3.1.1 Revenue Collection
The effectiveness of SSP depends on the resources mobilised which in turn
depends on a) coverage of target population b) level of prepayment compared with
OOPE, c) whether contributions are compulsory or voluntary, d) degree of progressivity
of contributions, and e) subsidies for the poor (Preker et al. 2004). Coverage of target
population as measured by enrolment depends on certain technical design factors namely
i) affordability of contributions, ii) unit of membership, iii) distance to hospitals, iv)
timing of collection of premium, v) quality of care and vi) trust in the scheme
administration (Carrin 2005).

237
a) Coverage of Target Population
The percentage of population covered by SSP denotes general attractiveness of
the programme. The coverage of target population was 53.4 percent in 2004 that reduced
to 47.7 percent in 2005 and again by 5.5 percent to 42.2 percent in 2006. There has been
successive decline from 2007 to 2011. It was 48.2 percent in 2007, 38.7 percent in 2008,
36.5 percent in 2009, 34.1 percent in 2010 and 32 percent in 2011. This calls for further
investigation into the declining risk pool. In terms of absolute number, both the
membership in micro-credit and SSP increased since inception. However, the percentage
increase in SSP membership as compared to the previous year increased at a decreasing
rate. Declining membership has adversely affected the enrolment in SSP and RM.
b) Ratio of Prepaid Contributions to Healthcare Costs or Claims
Higher prepaid contributions would generate sufficient revenue that enables the
programme to provide better and sustainable financial protection to insured members.
The ratio of premium to health care costs covered by the programme varied from 0.47 in
2004-05 to almost 0.88 in 2005-06. It declined to 0.72 in 2008-09 and 0.61 in 2010-11.
This denotes higher financial protection as the prepayment was less than claims.
Nevertheless, the financial consequence on SSP was drastic. Insurance companies had to
suffer heavy losses and SSP had to obtain funds from MFI wing of SKDRDP to meet the
deficit. If the programme continues to incur losses, it would dissuade insurance
companies from issuing group policies to the members of SSP.
Financial sustainability improves by increasing the revenue collection or by
curtailing the expenditure. Since the claim benefits and administrative expenses consume
the revenue earned, there is a need to curb them. Hence, in an attempt to understand the
viability of SSP, the study assessed the expenses involved in providing the special
benefits coverage. The administrative costs as a percentage of premium borne by SSP to
provide insurance coverage was high in the initial two years (6.2 % in 2004-05, 6.1% in
2005-06), but it declined by half in 2006-07 (3.1%) and 2007-08 (3.1%). It was 4.5
percent in 2008-10. It reached a high level of 4.9 percent in 2010-11 owing to higher cost

238
of various resources. A reduction in administrative cost is highly needed given the low
level of revenue collection and high claims ratios.
c) Nature of Contribution
SSP membership is voluntary for SHG members and their families. SSP did not
coerce or put pressure on the SHG members to enrol in the programme. However, it
insisted premium payments in cash. Voluntary membership has positive and negative
effects. It can encourage the adverse selection among the members as those with pre-
existing illness would join whereas healthy people would stay out. In contrast,
compulsory membership would increase RM and FS due to enlarged risk pool. However,
an attempt to curtail adverse selection through compulsory membership for all SHG
members hampers market mechanism by limiting the opportunities available for them.
d) Degree of Progressivity of Contributions
The premium charged was a flat amount without a concession to the low-income
households or the poorest, which appear to be regressive. Until 2007, the marginal cost
for additional member was high for a small family than a large one. The marginal cost
was uniform for all the members regardless of the household size since 2007.
Nevertheless, average cost per member was lower for the large families compared to the
small families. Poorest had to incur higher premium as a percent of annual income
compared to the middle or high-income household. Thus, social inclusion in the form of
larger representation of the poorest section of target population was not achieved (section
5.6).
e) Subsidies for the Poor
There was no concession in the premium, irrespective of caste, religion and
income since the year 2011-12. This is regressive, as the poor will have to pay higher
percentage of annual income compared to non-poor. SSP contracted with public sector
insurance companies and removed the distinction between families below poverty line
and above poverty line while determining the premium amount. Such a policy change
might have adversely affected enrolment and RM. As the target population is the poor in

239
the informal sector, regressive premium would discourage many to join SSP or renew
their membership.
f) Technical Design Features Determining Enrolment
i. Affordability of premium
The premium charged by SSP was on an average 1.17 percent of annual income
of the surveyed households. It was the highest for low-income families (Q1 2.2%; Q2
1.32% and Q3 1.01%) than high-income families (Q4 0.78% and Q5 0.48%). Despite the
credit facility to pay the premium, coverage of the target population was low. In this
regard, FGD identified lack of awareness on the borrowing facility in some karyakshetras
as the primary reason for non-enrolment, in addition to a high level of premium.
ii.Unit of membership
Another determinant of enrolment is the unit of membership. SSP insisted on
family enrolment rather than individual memberships to encourage the participation of
the entire household in addition to cross subsidise the benefits of risk pooling. This has in
fact reduced adverse selection as discussed in section 4.2.4. Larger pooling and cross
subsidisation of the risk took place since the high risk as well as the low risk individuals
in a family enrolled.
iii. Timing of premium collection
Membership in SSP depends on the timing of collection of premium (monthly,
quarterly or yearly). SSP enrolment takes place in February of every year; hence, timing
of the collection of the premium is inflexible. To encourage larger participation and to
overcome the inflexibility, SSP offered credit facility. This curtailed the negative
influence of seasonality of income on enrolment, to some extent. The repayment of such
a loan took place along with other financial transactions namely savings and credit
repayments in weekly meetings. This not only brings down transaction cost but also
improvesthe affordability of premium. Almost 64 percent of insured members borrowed
to pay the premium. However, FGD identified inflexibility to be one of the reasons for
non-enrolment. Whatever may be the effect of inflexibility on enrolment, there was a
positive effect on the adverse selection. Usually, the demand for health insurance will be
240
high when an individual falls sick. If the enrolment can take place at any time, the
possibility of the adverse selection would be high. By restricting the enrolment on the
incidence of illness, SSP created barriers to adverse selection.
iv. Trust in SSP
SKDRDP enjoys clientele due to trust in the competence of its management;
hence, SSP could leverage the trust of its parent organisation. Supportive field staffs that
were always available to provide information on the pre-authorisation, network hospitals
and sanction of claim benefits strengthened the pre-existing trust. Hence, the staffs’
responsiveness to non-medical expectations of members was high. The viability of SSP
largely depends on the people’s confidence and trust in the management. Since SSP
enjoys the patronage of the religious temple and trust of its members, it was in a better
position to harness information and monitor the behaviour of members that enhances
viability of the programme.
v. Quality of care
Quality of care has been identified as another factor determining the enrolment.
Availability of laboratory, X- Ray and diagnostic equipments, cleanliness of hospitals,
availability of medicines, friendliness of support staff and time taken by doctors to
examine the patient were considered the most important features of quality of care of the
network hospitals. The study reiterates the perception of good quality of care at the
network hospitals by insured households compared to that of newly insured and
uninsured households (explained in section 6.1.3). Since SSP selects the network hospitals
by applying stringent criteria, the quality perception is high among insured group. Thus,
higher enrolment and retaining of existing members are possible. This study did not
explore the quality impact on the enrolment as it is beyond the scope of the identified
objectives.
vi. Distance to network hospitals
Distance to hospitals was another crucial determinant of the enrolment in SSP
(section 5.6). The households residing far away from the network hospitals had higher
likelihood of enrolment in SSP compared to those living near the hospitals. By
241
encouraging the enrolment from the people residing far away, SSP could reduce the
financial barriers to access care.
6.3.1.2 Risk Pooling
Risk pooling is determined by trust in SSP management and the mechanisms of
cross subsidisation that facilitate transfer of income from rich to poor and risk from
healthy to the sick. Risk pool of SSP in terms of the membership consists mainly of poor
families (70 % of the target population was BPL). This socially desirable objective has
restricted the mobilisation of resources and designing of a comprehensive benefit
package since the poor cannot afford high amount of premium. Despite the shortcomings,
the number of members was over 16 lakh in 2011-12, one of the highest in MHI industry
in India and other countries where membership ranged from 1000 to 2 million
(Devadasan 2006).
a) Trust in the Management of SSP
Trust in the integrity and competence of the management of the programme has
greatly contributed to the viability of SSP. Trust was built by providing adequate
information on the programme, acting upon the feedback from members by the
management, member-friendly approach of field staff and good rapport developed due to
many years of association with SKDRDP micro finance programme.
b) Mechanisms to Enlarge the Risk Pool
Financial sustainability improves when the membership base expands. SSP
aimed at the larger risk pool from the very start by targeting the population of entire
district rather than specific taluks that has not only enhanced risk pool but also gave rise
to economies of scale in membership base, cost of administration and transaction. SSP
has penetrated into new markets in 2011-12 in two districts where it launched micro
finance programme. However, enrolment in these new districts seems to be low as SSP
was novel to these members (as observed in FGD).
Adverse selection, moral hazard and fraudulent activities due to information
asymmetry prevent the cross subsidisation and larger risk pool in any MHI scheme. SSP
has implemented various fraud identification mechanisms namely inflexibility in the
242
timing of enrolment, computerised identity card, verification of medical bills, limits on
the benefit package, visits to the hospitals by Sampoorna Suraksha assistants to verify the
admission of members and scrutiny of pre-authorisation procedure by SSP office.
However, lack of screening for pre-existing illness and absence of waiting period to claim
benefits increases the scope for information asymmetry. Nevertheless, waiting period is
not justifiable in SSP as the enrolment takes place only once in a year. Such a qualifying
period is required in schemes that are open throughout the year.
SSP curtailed moral hazard by a unique feature uncommon to other MHI
schemes. It has a standard protocol to approve the claims (known as pre-authorisation
procedure) that clearly defines the steps involved in sanctioning the claim. Admissions
with the pre-authorisation from the certified medical officers of SSP were approved and
cashless payment was made directly to the hospitals. In-house medical officers appraise
the discharge summary, days of hospitalisation, diagnostic procedure and drugs
prescribed before making payment to hospitals. In case of discrepancies, the erring
hospitals would be accountable. Moreover, the designated staff of SSP made daily visits
to hospitals to check for fraud or prolonged stay in the hospital in addition to the
verification of the admission and scrutinisation of the identity card. By this way,
impersonation to claim the benefit as a third party was difficult.
6.3.1.3 Strategic Purchasing
The purchasing of health care services is a vital function that includes contracting
with the hospitals, deciding payment mechanism, system of referrals and waiting period
requirement. SSP practised the strategic purchasing to some extent. In addition to routine
payment of the hospital bills for specified services, SSP had contractual relationship with
hospitals.
a) Selection of Network Hospitals
SSP adopted active purchasing based on the quality, accessibility and cost
criterion in selecting the network hospitals. It sends the information on the benefit
package and price of care to the hospitals. If the hospitals agree to the conditions
specified in the agreement, a memorandum of understanding would be signed between
243
the project officers and the director of the hospital. The hospitals did not exert monopoly
power during price and payment negotiations. If the terms of contract were not agreeable,
they could refuse to be part of network hospitals. The project officers removed the
hospitals from the network that inflated medical bills or involved in fraudulent activities
and did not take any action despite many reminders. Thus, moral hazard and fraudulent
practices were curtailed. However, the lengthy and complex claim procedure would
reduce enrolment as the majority of the target population comprises less educated
individuals.
b) Claim Disbursement Procedure
Claim disbursement follows a predetermined procedure as explained in the
section 3.6.2. The insurance company and medical team from SSP office conduct audits
and inspections to ensure quality medical care to the members of the scheme and to
prevent the supplier and member moral hazard. The absence of referral system may not
result in over-utilisation of health care facilities due to the opportunity cost (indirect cost)
associated with accessing health care. Supplier moral hazard on the part of hospitals was
indirectly curtailed by persuading them to restrict the bill amount to pre-determined
package as per the contract.
c) Benefit Package
Based on the target population’s willingness and ability to pay, SSP determined
the premium. Benefit package was fixed considering the cost of health care services in
the state of Karnataka. SSP covered inpatient health services in the benefit package and
excluded outpatient (OP) treatment and common ailments. One attractive feature of
benefit package was the inclusion of life and health risk, maternity treatment, and death
compensation. However, the real effectiveness of the benefit package was low since the
cost of health care services has gone up drastically whereas the amount of benefit did not
change since inception. This was evidenced in the survey as some insured individuals had
to rely on other risk coping strategies such as borrowing, sale of assets and use of
savings. Nevertheless, increasing the amount insured invariably necessitates higher
premium, which is not affordable by a majority of the target population. Albeit, SSP
244
sanctions additional amount, higher than sum insured, if the treatment was enormously
expensive such as cancer, heart and other vital organ surgeries. In addition, insured can
get a loan from the ‘Pragatibandhu’, MFI of SKDRDP to meet any expense that exceeds
insured amount. These additional provisions could remove some of the limitations of
benefit package.
6.3.2 Management Characteristics
6.3.2.1 Staff
The religious leader of the Dharmasthala temple (President) leads SKDRDP and
Board of Trustees manages its operations. Although directors and project officers do not
have the management qualification, they have experience in implementing various socio-
economic development programmes. Field staffs (Sevanirathas) motivate SHG members
to enrol in SSP by educating them the importance of health insurance. They monitor
moral hazard behavior due to the close proximity to the members. An experienced
management implements SSP using the administrative set-up of SKDRDP. The
programme had staff with skills required to formulate benefit package, contract with
providers and process claims in addition to collecting premium and creating awareness.
SSP could make use of pre-existing network of grass-root member households and a
large team of field staff with the knowledge of local community and tradition. As SSP
expands, complexity in administration and management would arise that necessitates
investment in management information system (MIS) and professional training of the
staff.
6.3.2.2 Culture
A hierarchical organisation structure of SKDRDP has the President and Board of
Trustees as the top-level management who delegate the authority to four directors that
supervise and guide project officers in each valaya. Project officers oversee the work of
supervisors and field staff. The office staffs carry out claim processing, maintaining
accounts and record keeping. Hierarchical structure stresses top down management with
least participation of insured members in the management. The President interacts

245
continuously with staff that helps them to identify with the ideology and values of the
SKDRDP such as charity, philanthropy and mutual aid.
6.3.2.3 Access to Information
The monthly SHG joint meetings communicate the information on SSP namely
benefits package, excluded diseases, claim procedure, nearby network hospitals and
rejection of claims to the members. Frequent information flows among the members and
staff of SSP built trust and curtailed moral hazard largely. The ‘Jnana Vikasa’
Programme imparts knowledge to the SHG members on various issues including health
that removed non-financial barriers to access care (section 4.2.5).
SSP has computerized data recording system at various regions that stores
members’ basic information and data on utilisation of benefits (name of member and
hospitals, duration of stay, amount of hospital bill, claims sanctioned). However, the
valaya maintains the records and does not analyse them to assess the performance of the
programme. Lack of MIS (management information system) would threaten the viability
of SSP when the programme expands to a large number of districts in Karnataka.
Management and administration of large risk pool requires quick access to information.
Hence, implementation of MIS becomes a necessity. To conclude, SSP has the parent
organisation that provides stable leadership, management skill, information systems,
infrastructure, access to rural network and financial resources. This would enhance
viability of SSP.
6.3.3 Organisational Characteristics
6.3.3.1 Forms of Organisation
Since SSP is embedded in SKDRDP, it could utilise the work force, office
infrastructure and established network to provide MHI services resulting in a lower cost
of operation. However, SSP had to incur the additional expense of medical staff who
handle the pre-authorisation procedure and operating/maintenance cost of computers and
other office equipments in SSP head office. The economies of scale and scope were
possible since the parent organization had a broad range of services namely micro- credit,
bundled insurance and savings. However, economies of scale didnot increase resource
246
mobilisation and higher enrolment. The potential to increase enrolment is high since
SKDRDP has a good clientele that can be motivated to join SSP for better viability and
financial sustainability. Contractual agreements with the insurance companies and
hospitals are the backbone of SSP in which the programme acts as the Third Party
Administrator (TPA) and manages the administration and implementation. These
agreements make SSP viable as insurance companies absorbed the loss since inception
and the hospitals strive to provide better care to insured members.
6.3.3.2 Incentive Regime
An exploration on the extent of the decision rights reveals that operations were
decontrolled from the board. The Executive Director, the Executive Committee
consisting of the directors and project officers managed the operations. Field supervisors
and field staffs were not involved in any major decision making. Office staffs handle
pre-authorisation and claim settlement procedure and kept accurate member records
including accounts. There is an audit wing to scrutinise the records of SSP, detect
fraudulent activities and prepare financial statements. Systematic and organised
administrative framework made every staff accountable and responsible that has
enhanced trust among members.
The main source of funds was the revenue collected from the members. External
funding in the form of grants or donations or financial support from the government or
other aid agencies was absent. Moreover, SSP did not maintain reserves that exposed the
programme to higher financial risk. SSP has incurred loss since inception, but parent
organization supported it, out of conviction. However, the threat of financial
sustainability is impending due to declining enrolment resulting in inadequate resource
mobilisation and high level of claims.
6.3.3.3 Linkages
Vertical integration through contractual agreement with the providers could
provide treatment to members at concessional rates. The Executive director in
consultation with the insurance companies, Board of Trustees and the President sets the
premium and determines the benefit package. The director of SSP guides the
247
implementation through project officers who select the hospitals, contract with providers,
and follow-up quality of care and supervise the implementation. Since insurance
companies borne medical claims by underwriting the risks, they enhanced financial
sustainability of SSP. Nevertheless, long-term viability is doubtful if these companies shy
away from covering risks because of loss as evident in high claim ratio.
6.3.4 Institutional Characteristics
6.3.4.1 Stewardship
The management of SSP and insurance companies without the intervention of
local, state or national government took up stewardship function. The government, both
central and state, do not play any role in SSP design, risk coverage or implementation.
SSP collected the premium amount, transferred a part of it to the insurance companies,
recorded members’ data, implemented pre-authorisation procedure and made cashless
payment to hospitals whereas insurance companies provided risk coverage, verified the
pre-authorisation forms and disbursed sanctioned claim amount to SSP. Thus, risk and
servicing the clients were shared between SSP and insurance companies.
The top management consisting of the President, Board of Trustees, the executive
director and SSP director took strategic decisions after consultation with the insurance
companies. Project officers and supervisors took operational decisions and field staff
implemented them.
Regarding the regulation of SSP, IRDA (Micro-Insurance Regulations, 2005)
establishes the rules and regulations that are abided by the insurance companies. This Act
recognises SHGs as the distribution agents who can carry out the functions of the
premium collection, claims administration and distribution of policies. Hence, SSP chose
the partner-agent model in which SKDRDP acts as an agent for a partner (the insurance
companies). Since private insurance companies have to tie up with MFIs or other
channels to meet the statutory requirements, SSP seems viable in the long run.

248
6.3.4.2 Governance
SSP and insurance companies share the ownership and governance of the
programme. The objective of SSP is to provide financial protection against unforeseen
contingencies and improve the access to health care services. The ownership and
governance arrangements support the achievement of these objectives.
6.3.4.3 Insurance Markets
As per IRDA Act, private insurance companies should mobilise seven percent of
total premium from rural and social sector of the country in the 8th year of operation.
They have tied up with MFIs or NGOs to meet the statutory requirements to reduce
transaction cost. These companies issue a negotiated custom designed group insurance
policies to SSP that include the coverage for pre-existing illness. The custom-designed
package meets the local needs of the target population and plays an important role in
enrolment.
In MHI market, currently there is a limited competition since the evidence base on
the positive impact is yet to build up. Besides, the insurance companies are sceptical of
covering risks, at the same time MFIs hesitate to diversify into non-core activities like
insurance. Barriers to entry into MHI market are high which reduces the threat of new
competitors. At the same time, its commitment to the welfare of the underprivileged
people makes the exit from the market difficult.
There is threat from RSBY for SSP and a visible impact is the deterioration in
enrolment in 2011-12. From the frying pan to the fire, finding insurance underwriting
partners every year is challenging, as the programme has incurred huge loss since its
inception. Soliciting insurance partnership in the midst of uncertain environment is quite
difficult.
6.3.4.4 Factor and Product Markets
There is a limited competitive pressure in the product market as the schemes
aimed at the poor in the informal sector are rare. A few government programmes that
target the poor are Yeshasvini, UHS and RSBY. In Yeshasvini programme, only surgical
hospitalisation is covered and it acts as a standalone insurance programme, not embedded
249
in any development programme. However, SSP provides bundled product covering the
risks of health, life and natural calamities and is entrenched in SKDRDP, thereby utilises
the outreach and experience in providing financial services to people. It is one of the few
MHI schemes in India having a membership base of over 10 lakh individuals. At present,
SSP, as a MHI product, has little competition in the insurance market.
In the factor markets, SSP has to compete with other companies that employ
people with basic education. SSP hires local people who have completed 10 years of
basic education, and trains them. Since labour market is abundant with such people in
Karnataka and cut-throat competition does not exist. It does not hire professional
managers to perform various functions; instead uses internal promotions to fill these
positions. SSP does not own many hospitals to provide health services to members.
Despite that, SSP has significant market power through contract with providers that
specify the quality of care and payment mechanism. Non-compliance with specifications
of the contract can lead to the deletion of hospitals from the list of network hospitals.
Thus, SSP has indirect influence on the providers of health care. Table 6.7 depicts the
link between features of SSP and its impact on the performance with future implications.

250
 Table 6.7 Effect of Characteristics of SSP on the Outcome of the Programme
Characteristics Features of SSP Impact on Future Implications
Performance
1. Technical
Revenue Collection
Coverage of target 53.4 % in 2004-5; Decline in Shrinking risk pool
population 32 % in 2011-12 enrolment and RM

Ratio of premium to 0.47 in 2004-05, Higher FP; reduced FS affected

healthcare costs and 0.61 in 2010-11 CHE

Nature of contribution Voluntary; low Poor used credit To some extent, poor
interest credit facility to pay the were given a option
facility premium to enrol

Degree of progressivity Regressive, average Enrolment of large Social exclusion

of contributions cost was less for families; poor paid
large families higher premium

Subsidies for the poor No subsidies since Negative growth in Adverse effect on
2011-12 enrolment enrolment

Risk pooling Cross subsidisation One of the largest

across income and risk pools in India
risk
Trust in the Part of SKDRDP, Faith in the May encourage
management of SSP well known NGO in integrity and enrolment
Karnataka competence of the
management in
Karnataka
Mechanisms to enlarge Target population Economies of scale Higher scope to
risk pool was the entire in administration enhance enrolment
district; coverage of and transaction as percentage of
health risks by costs; decline in target population
general insurance administrative costs
companies
Strategic purchasing
Selection of network Active purchasing Perception of better Good quality of
hospitals based on quality, quality of care treatment would
accessibility and cost result in renewal and
criterion higher enrolment

251
Claim dissbursement No waiting
w perio
od; Morral hazard annd No tthreat to FS due
proceduree pre-auuthorisationn frauddulent practtices to m
moral hazardd
requirred; strict curtaailed
monittoring of
claim
mants
Benefit paackage Inpatiient coveragge High her Partial impact on
o FP
of 5000; cashleess hosppitalisation,
treatm
ment memmbers relied d on
risk coping
strattegies
2. Manag
gement

Staff Experrienced and d Impoortant role in

i Delaay in claim
hard working
w staaff; enro
olment and apprroval, investment
frequeent field vissits motiivated poor to in m
managementt
to mo
onitor memb bers join SSP; no mo oral inforrmation sysstem
hazaard and professionaal
trainning requireed
Culture Ideoloogy and vallues Conv veyed Mem mbers trust in
i
such as
a charity, transsparency annd SSPP, may enhan nce
philannthropy andd trustt to every enroolment and RM R
mutuaal aid stakeeholder
Access to
o information Disseemination off Trusst was built due Infoormation
inform
mation week kly to frequent asymmmetry probblems
meetiings; flow of
o inforrmation; Mo oral weree minimum; lack
inform
mation amo ong hazaard curtailed
d; of M
MIS act as th
hreat
memb bers; Minimmal data managemeent to effficient
MIS was not effectivve operrations.

3. Organisational

Forms off organisatiion Partneer-agent andd Econnomies of scale Goood model off MHI
mutuaal model thaat and scope; in In
ndia; econommies
leveraaged trust of enhaanced of opperations co
ould
SKDR RDP; bundlled enroolment; not iincrease RM
M
creditt, savings an
nd reduuction in
insuraance adm
ministrative cost
c
Incentivee regime Auditting of recorrds, No fraudulent
f Threeat of FS stiill
transp
parency in activvity; huge lo
oss existts due to hig
gh
accouunts; premiu um as thhere was no leveel of claims and
m source of
was main exterrnal fundingg inaddequate RM M and
fundss lackk of externall

252
 funding
Linkages Formal contracts Revenue of the Insurance companies
with insurance network hospital ensure viability
companies and increased; huge
hospitals; risk losses absorbed by
underwritten by insurance
insurance companies companies
4. Institutional

Stewardship Played by both SSP Transparency and Enhanced viability

and insurance ethical practices of SSP
companies;
regulated by IRDA

Governance Ownership and Resource sharing Better viability of

governance was and economies of SSP; management
entrusted with SSP scope and scale know-how limited
and insurance were achieved;
companies administrative costs
were brought down

Insurance markets Limited competition As competition was RSBY is a threat;

in the MHI market; less, enrolment till barriers to entry and
acted as agent of 2010 was good exit high; monopoly
general insurance in rural areas
companies

Factor and product Limited competitive Significant market There is a potential

markets pressure in the power through for higher enrolment
product market; contract with
factor market had providers; product
surplus human of SSP unique
resource

6.3.5 Summary
Effective design and management are critical to the success of MHI schemes.
This study identified certain technical, management, organisational and institutional
characteristics that influenced enrolment, resource mobilisation, financial protection and
social inclusion. Technical features of SSP such as credit facility to pay premium,
additional loan to insured members to meet medical expenses, bundling of medical and
253
life insurance benefits, cashless treatment, higher benefit compared to other MHIs and
wide network of network hospitals encouraged higher participation of the target
population. Although enrolment in absolute number has increased, the growth in
enrolment declined over the years since inception. In 2010, RSBY was introduced in
Karnataka, which attracted many of these members. In the same year, subsidies for the
poor were withdrawn and the premium was hiked which adversely affected enrolment.
Despite the positive role of social capital (mutual help, solidarity and concern for others)
in enhancing enrolment, there was negative growth in membership base. Certain
undesirable design features like increase in the premium, availability of cheaper options
(RSBY), removal of subsidies for the poorest families, inflexibility in the collection of
premium, regressively charged premium and low benefit amount can be attributed to
decline in participation rates. However, credit facility to pay the premium removed many
of the design constraints.
Regressive premium, lack of subsidy coupled with low income resulted in the
exclusion of poorest target population from enrolment, especially poor (mainly in
seasonal occupation) could not afford the premium. SSP members from the poorer
households had to spend 2.2 percent of annual household income to pay the premium.
Thus, the design of SSP aimed at rural middle-income class than poor since certain
features like the absence of a sliding scale, exemption policy, payment in- kind and flat
rate of the premium limited the participation of the poorest in SSP.
SSP acted as a strategic purchaser of the health services largely by negotiating the
price of care with providers and selecting the hospitals with basic facilities. It monitored
the provider behaviour through pre-authorisation requirements that checked the line of
treatment and probable cost of care before effecting payment. This curtailed moral hazard
(from members) and fraudulent practices to large extent. However, it did not attempt to
improve the quality of care, except selection of the hospitals with basic facilities. Certain
strategic mechanisms namely gate keeping and drug formularies, referral practice,
financial incentives to providers of care and insured to encourage the use of specific
providers was absent. It did not negotiate favourable prices for essential drugs. Since the
254
primary caare and reeferral systtem was ignored,
i m
members weere motivaated to gett
hospitalisedd even for acute
a illnesss that resulteed in high claims.
c
Lacck of gatekeeeper mech
hanism was another faactor respon
nsible for high
h claims.
SSP did not encouragee the implem
mentation of standard treatment
t prrotocols inccluding drug
g
formulariess and physsician proffiling (trackking of the physiciann treatmen
nt patterns).
However, monitoring
m utilisation of servicess by the meembers faciilitated the removal off
fraudulent hospitals frrom the list of network
k hospitals. SSP used fee-for-serv
f vice system,,
known for the cost esccalation and
d administraative compllexity along
g with higheer incentivee
to over-serrvice and over-prescrib
o be. It shouuld be recognised thatt a scheme would nott
sustain finaancially if sttrict referraal system orr gate keepin
ng were nott practised. SSP can see
primary heealth centre and tier sy
ystem of Inndian health
h care system to impllement gatee
keeping.
Thee enrolment of memberrs as a perceent of target populationn has declin
ned over thee
years; thereeby the risk
k pool has shrunk
s jeopardizing fin
nancial sustainability and
a viability
y
of SSP. Inssufficient reevenue colleection has rresulted in huge
h lossess since inception. Even
n
with the hiigh claims, financial protection
p w partial owing
was o to loow benefit package (
5000), whicch was unch
hanged desp
pite an increease in the cost
c of meddical treatmeent. Hence,
insured members had to incur CH
HE and ado
opt other risk coping sttrategies. Ho
owever, thee
burden of severe casees on broad
der health care
c system
m reduced. Increasing the benefitt
package is not a solutiion to partiaal financial protection. This necesssitates high
her premium
m
that adverssely influen
nce enrolmeent. If prem
mium increaases, people with low health risk
k
would refraain from enrrolment giv
ving rise to adverse selection. In aaddition, poo
orest would
d
stay away from the programme.
p . Hence, a tradeoff iss required tthat balancees financiall
sustainabiliity and risk
k pool. On the other hand,
h SSP has
h an advaantage over other MHII
scheme since it is nestted within broader
b deveelopment prrogramme w
with adequaate financiall
hat can bail out SSP in times of tro
resources th ouble.
Houusehold as a unit of membership
m p, inflexiblee period off enrolmentt and betterr
informationn flow amon
ng memberrs due to cloose proximiity reduced selective en
nrolment off

25
55
ill persons in the family, especially enrolment after illness. Pre-authorisation rules,
scrutinisation and monitoring of hospitalisation reduced moral hazard behaviour.
Being embedded in SKDRDP, SSP enjoys clientele due to faith in the integrity
and competence of management. Senior management of SSP was committed to the
programme and determined to continue it out of conviction, despite financial difficulties.
Moreover, SKDRDP increased the income of poor families in its area of operation
through micro-finance and other developmental activities. This enhanced the ability and
willingness of SHG members to enrol in SSP. Since SSP met their priority needs (health),
readiness to participate and support the the programme was high.
The member orientation and strong community networks facilitates the viability
of SSP. Some of management factors that shaped the success of SSP are contracting with
providers, determining the appropriateness of care provided and its pricing, accounting
and bookkeeping, monitoring, peoples’ confidence and trust in the management. Relevant
information disseminated to members in the monthly SHG meetings conveyed
transparency and trust that premium amount belonged to members’ betterment. This
positively shaped the renewal and enrolment decisions of members and indirectly
increased resource mobilisation. However, certain hindrances namely lack of
professional management with requisite skills in marketing, and communication,
actuarial science, lack of member participation in the management and absence of any
negotiation with providers for better quality of care would affect the programme
adversely. In addition, the management of data and creation of electronic database was
insufficient. This would limit the revenue collection, containment of administrative cost,
and quality of health services.
Organisational characteristics of the scheme such as contractual linkages between
SSP and providers stipulated the nature and scope of the services the providers should
offer to the members. Thus, yearly contracts ensured flexibility to change the providers
(include or delete from the list of network hospitals) based on their performance. Even
the contractual relationship with insurance companies defined the role and
responsibilities of the parties concerned. The insurance companies absorbed the loss of
256
medical component of the benefit package; which ensures the viability of the programme
in the long time.
Parent organisation SKDRDP facilitated resource sharing and economies of scope
and scale. Offering of MHI services through partner-agent and mutual model leveraged
the trust that SSP enjoys among SHG members and enhanced the enrolment of target
population. Moreover, bundling the insurance services with credit or savings and using
the existing infrastructure to provide service and collect premium reduced administrative
cost. Regular auditing of the financial records and preparation of annual reports that are
publicly available helped to build up the credibility.
The premium was the main source of funds. SSP did not seek external financial
assistance or aid to cover the losses of special component of benefit package. Hence,
threat of financial sustainability looms around the programme due to the high level of
claims, inadequate RM and lack of external funding.
The government did not play a stewardship role by providing subsidies or
administrative assistance to SSP. Instead, insurance companies and SSP played the role
of stewardship by sharing the risk of coverage and servicing the clients. Moreover, the
government did not monitor, regulate and accredit the providers; hence, SSP developed
the technical skills to conduct these activities. SSP and the insurance companies jointly
had the ownership and governance responsibilities that facilitated resource sharing.
The competition in the job market was not intense as there was surplus labour
with required qualification. The competition in health care market becomes irrelevant
since SSP does not own all the network hospitals to provide health care facilities to the
members. The rivalry in MHI market was minimal as the high level of entry and exit
barriers to MHI market would prevent a large number of players from entering the
industry. However, SSP has to face the threat from the recently introduced RSBY and
schemes of other MFIs.
Taken together, these results suggest that SSP is viable owing to i) Nesting within
SKDRDP ii) Tie up with insurance companies iii) Dedicated staff and management iv)
High potential for greater penetration. However, self-financing of SSP is limited due to
257
several features; limited population coverage, low cost recovery rates and membership
limited to poorest groups. Unless these issues are addressed, SSP cannot be considered
an exclusive health-financing alternative, rather it can be considered as supporting
mechanism that complements the government efforts to provide health care to all the
population. SKDRDP has to capitalise its monopoly in many parts of rural Karnataka to
make SSP a self-financing MHI scheme.

258
 CHAPTER 7

SUMMARY OF FINDINGS, CONCLUSION

AND SUGGESTIONS
7.1 Introduction

Illness is the second most frequent risk in rural areas after the crop failures. It is
the single largest cause of perpetual poverty in many of the poor households. There is a
strong link between health and income that makes the poor most susceptible to health
shock. Given the inadequate public funding as well as inefficient delivery of public health
services and lower penetration of private health insurance, MHI is identified as a
potential insurance mechanism to mitigate iatrogenic poverty. Moreover, IRDA
regulations impose rural and social sector obligations on private insurance companies to
provide insurance benefits to the poor that unlocks a huge market for MHI industry.
MHI aims to provide adequate financial resources to ensure timely access to
health care services. The most convincing argument in favour of MHI would be the
tangible proof that it can do what it claims and provide protection against the financial
consequences of health risks to the population. Scaling up of MHI to provide insurance
coverage to larger population is not advisable without the evidence on the impact of
schemes on the target population. Hence, this study was carried out to understand the
impact of Sampoorna Suraksha Programme, a well-known MHI scheme in Karnataka.
This descriptive cross sectional survey collected data using questionnaire and qualitative
instruments from 1146 households selected randomly applying multi-stage cluster
sampling design. Logistic/multiple linear regression analyses and chi square test were
used to test the hypothesis of the present study.
7.2 Summary of the Findings
This section summarises the main findings that draws together results presented in
different sections and discusses these findings in the context of previous research on the
impact of MHI. It highlights managerial implications and provides policy suggestions
which are of interest to the scheme management and policy makers.

259
7.2.1 Findings on the Impact of SSP on Financial Protection (Chapter 4)
Access to Care: SSP did not have any impact on access to health care (H1). This
result is in agreement with World Health Organisation (2000) findings, which
documented lack of impact of MHI in improving the access to health care system.
Uninsured and newly insured individuals could overcome the financial barriers by
borrowing, sale of assets or use of the savings to pay for health services. ‘Jnana Vikasa’
programme of SKDRDP, which educated SHG members on the importance of health,
removed the non-financial barriers.
Health Seeking Behaviour: SSP insured members sought treatment in private
network hospitals rather than self-treatment or public hospitals compared to uninsured
and newly insured individuals (H2). Accessibility to network hospitals, affordability (due
to SSP claims) and acceptability (quality of care at network hospitals) influenced insured
members to seek care in private network hospitals. The present study confirms the
previous findings (Jowett 2004; Jutting and Tine 2000; Chankova et al. 2008) and
contributes additional evidence that suggests the role of income in HSB in addition to the
positive impact of MHI in India.
Utilisation of Health Services: This study has gone some way towards
enhancing our understanding of the MHI impact on utilisation of health services. Insured
individuals utilised health services in higher proportion compared to uninsured and newly
insured individuals (H3). Income class, gender of ill persons and types of illness
determines hospitalisation. The most obvious finding to emerge from this study is the
absence of vertical equity based on income and gender, but horizontal equity was evident.
Out of Pocket Expenditure: The present study provides additional evidence with
respect to the positive impact of MHI on financial protection. SSP decreased out of
pocket expenses associated with treatment for illness for insured individuals compared to
uninsured and newly insured individuals (H4). This result is consistent with those of
Jutting (2003) and Schneider and Diop (2001), but contradicts the findings from the
Indian studies (Ranson 2001; Gumber 2001). The days spent in the hospital, SSP, chronic
illness, area of residence and gender of ill persons emerged as reliable predictors of out of
260
pocket expenses. Certain design features prevented complete financial protection of SSP
in the form of zero out of pocket expenses. Insured poor individuals might not have
benefited from SSP compared to better-off insured individuals (absence of vertical
equity); however, they had lower expenses compared to uninsured and newly insured
individuals (presence of horizontal equity). The study did not find gender based vertical
equity as insured men incurred higher expenses compared to insured women. However,
the findings support horizontal equity in which insured women had lower out of pocket
expenses compared to uninsured and newly insured women.
Catastrophic Health Expenditure: One of the more significant findings to
emerge from this study is that SSP successfully reduced the incidence of catastrophic
health expenditure (CHE) for insured members (H5). However, the impact was partial, as
one fourth of households still had to face CHE even with health insurance. The partial
effect of SSP occurred due to certain limitations of benefit package (smaller benefit
package, exclusion of outpatient treatment and certain diseases from coverage). Partial
protection of MHI as found in this study is in accordance with the earlier studies by
Ranson (2002) and Devadasan (2007) in India and Zhang (2010) in China.
There was gender based equity among women since insured women had a lower
probability of CHE compared to uninsured and newly insured women. Gender based
equity in claim distribution was detected as SSP reduced the incidence of CHE more for
female than for male members. Even vertical equity based on income was absent but
horizontal equity was present. The binary logistic regression analysis estimated SSP
status, income quintile of the household, chronic illness, hospitalisation and duration of
treatment to be significant predictors of CHE.
Risk Coping Strategies: This study could not establish a significant difference in
the broad category of risk coping strategies among insured, uninsured, and newly insured
individuals (H6). However, it clearly demonstrated the relevance of MHI in reducing
illness-related borrowing (H7a, H7b). Younger and unemployed heads of the household,
low income, SSP status and hospitalisation predicted the likelihood of borrowing. SSP
insured individuals borrowed less amount.
261
 The presen
nt study cou
uld not iden
ntify a clearr benefit off MHI in thee prevention of
sale of assets
a (H9) or use of th
he savings (H8a,
( H8b) to meet thee cost of illnness. Hence, the
study findings
fi do not supporrt the previious researcch of Dekk
ker and Willms (2009) and
Aggarw
wal (2010) w
who provided evidencee that health
h insurancee reduces reeliance on other
o
risk coping strategies. Certaain design features off SSP such
h as the exxclusion off the
outpatieent treatmennt, certain diseases,
d traansportationn costs and limited
l bennefit package (
5000) were
w responnsible for not eliminaating the reeliance on risk copingg strategiess by
insured individualss.
7.2.2 Fiindings on the Impactt of SSP on
n Social Incclusion (Chapter 5)
The findinggs on the social inclu
usion are rrather disap Contrary to the
ppointing. C
expectaations, this study did not
n find larg
ge proportio
on of the poorest
p in thhe memberrship
base (H
H10). Althouugh these reesults differred from som
me other em
mpirical studdies (Chank
kova
2008; Ranson
R 200
01; Gumberr 2001), they are consistent with
h those of Msuya (20
004),
Jutting (2003) andd Schneider and Diop
p (2001). S usion was evident due to
Social exclu
certain supply sidee factors (d
design featu
ures) such as
a fixed perriod of enroolment, lack
k of
pation in ddecision maaking (ben
particip nefit packag
ge or settiing premiuum), regresssive
premium
m and lack of subsidy in addition to low incoome and con
nsequent noon-affordab
bility
of the premium.
p
There was no evidencce for the presence
p SSP (H11). The
of adverse sellection in S
distancee to the hoospital and area of ressidence predicted the membership in SSP. One
unanticipated findiing was thatt various co
omponents oof the sociall capital nam
mely solidaarity,
reciproccity and muutual aid infl
fluenced enrrolment.
7.2.3 Fiindings on Resource Mobilisatio
M on of SSP (C
Chapter 6)
Resource mobilised
m b SSP oveer the yearss was insuffficient to ccover the cllaim
by
amountt disbursed. Lower leveel of resourrce mobilisaation was allso due to loow coverag
ge of
target population,
p insufficientt premium collection
c a decreasee in membeership (negaative
and
in 2010
0-11). Techhnical desig
gn featuress of inpatieent coverag
ge in addittion to lack
k of
deductib
bles/ co-paayments, no
o waiting period andd absence of referral or gatekeeeper
o pre-existting diseasees, cashlesss treatment and easier reimbursem
system,, coverage of ment
262
procedure resulted
r in a high leveel of claimss. At the saame time, hhigh level of
o premium,,
lack of subbsidy to low
w income faamilies and inflexibilitty in enrolm
ment period along with
h
competitionn from gov
vernment sponsored
s RSBY and
d Yeshasvinni schemess adversely
y
affected en
nrolment. Reeasonable balance
b of funds
f was difficult
d sincce claims were
w alwayss
higher thann revenue co
ollection. Fiinancial susstainability is doubtful if the deficits continuee
at the samee or higher rate.
r
7.2.4 Findings on Ch
haracteristtics of SSP
P and Its Effect
E on E
Enrolment, Financiall
Protection and Resou
urce Mobiliisation (Ch
hapter 6)
P partner ag
SSP gent model facilitated
f b
better beneffit package aat a lower cost
c through
h
integrated ddistribution network off SKDRDP.. Since the target
t popuulation of SS
SP was ninee
districts of Karnataka unlike smaall commun
nities in most of the Affrican schem
mes, higherr
enrolment was possib
ble. Enrolm
ment was drriven by ceertain insureed-friendly features off
SSP such as credit facility
fa to pay
p premiu
um, bundlin
ng of mediccal and life
fe insurancee
benefits, caashless treaatment, bettter benefit package
p co
ompared to other MHIIs and widee
network off SSP hospittals. The ho
ousehold as a unit of en
nrolment annd inflexiblee enrolmentt
period reduuced adverrse selectio
on and preevented sellective mem
mbership of
o high-risk
k
individuals that enhan
nced cross subsidisatio
on of risk. Social incluusion was not evidentt
owing to llack of affo
fordability of
o the prem hdrawal of subsidies, flat rate off
mium, with
premium w ding scale and lack of premium exemptionn policy. SSP
with no slid S benefitt
package faailed to prov
vide compllete financial protectio
on since it is unchang
ged ( 5000))
even when the cost off care has gone
g up. SS
SP was incu
urring lossess due to shrrinking risk
k
pool and hiigh level of claims threeatening finaancial sustaainability an
nd viability.
Som
me of the managemen
nt characteeristics thatt ensured viability
v off SSP weree
members’ trust
t in the integrity an
nd competeence of the managemennt, commitm
ment of thee
managemen
nt, nesting in
i MFI (SK
KDRDP), strrong commu
unity netwoorks and con
ntracts with
h
providers aand insurancce companies, dedicateed field stafff, transpareency in acco
ounting and
d
bookkeepin
ng. This po
ositively shaaped the renewal and enrolment decisions of
o memberss
and increassed resourcce mobilisattion. Howeever, lack of
o professio
onal manag
gement with
h
requisite skkills in marrketing and
d communiccation, actu
uarial sciencce, lack of negotiation
n
26
63
with providers for better quality of care, absence of local management, accountability and
monitoring by members would limit revenue collection and containment of cost.
Organisational characteristics of the scheme such as contractual linkages between
SSP, providers and insurance companies enhanced the viability of the programme. SSP
could leverage the trust that members had in SKDRDP programs and use the
infrastructure of the parent organisation, which reduced administrative cost and improved
enrolment. Since financial records were audited and available publicly, transparency was
ensued that limited fraudulent activities. Nevertheless, financial unsustainability would
jeopardize the programme due to inadequate resource mobilisation and lack of external
funding.
Since SSP did not seek any assistance, government did not play stewardship role
by providing subsidies or training and administrative assistance. Insurance companies
and SSP itself played the stewardship role by sharing ownership and governance, risk
coverage and service to the clients. Regarding product and factor market, there was
limited competition. SSP has monopoly in rural areas, as few MHI products are available.
However, it faces major threat from government sponsored RSBY since 2010-11. Human
resource, especially lower level personnel required to implement the programme was
adequate given the surplus of labour in India, especially in rural areas.
7.3 Main Findings and Conclusion
1. The reliance on ex-post risk coping strategies compensates lack of MHI for uninsured
and newly insured individuals.
2. There was diversion of demand for care from traditional and public facilities to private
hospitals.
3. Insured individuals had higher utilisation measured by admission rate; however, moral
hazard behaviour was absent.
4. MHI provides effective financial protection against out of pocket health expenses.
5. A sizeable share of households experiencing catastrophic health expenditure had
payments less than 10 percent of annual per capita income due to SSP.

264
6. SSP reduced the excessive reliance on borrowing but it had no effect on the use of
savings or sale of assets.
7. Horizontal equity based on the income and gender in utilisation of health services, out
of pocket expenses and catastrophic health expenditure exists.
8. There was no adverse selection but members had higher incidence of illness.
9. Social capital components of solidarity, reciprocity and mutual aid determined
enrolment in SSP
10. Intensive monitoring of admitted insured members for any fraudulent activities,
fixing of price for each disease in consultation with hospitals and regular audits to
detect any financial irregularities were some of the factors that shaped the success of
SSP.
11. Experienced and well-established parent organisation (SKDRDP), contractual yet
amicable relationship with insurance companies and providers of care, trust of the
target population in SSP and dedicated management and the staff (office and field)
increases the viability of SSP. However, financial sustainability needs to be addressed
because the enrolment has declined and claims ratio has been very high since its
inception.
12. SSP faces financial constraints to provide absolute financial protection due to the
limited coverage of the target population, low cost recovery and membership base
(low income) that restricts premium collection. There was no external financial
support from the government or aid agencies.
To sum up, the beneficial effect of MHI on financial protection was evident from
this study. However, on the contrary, there was inadequate resource mobilisation and
social exclusion. This discrepancy is due to certain characteristics of SSP such as high
premium, introduction of RSBY, lack of subsidies for the poor, stagnant benefit package,
exclusion of outpatient treatment and regressive nature of premium structure.
SSP contributed to the achievement of ultimate objective of the health system
especially reduction in impoverishment and equitable utilisation. There was lower
incidence of borrowing, OOPE and CHE by insured individuals; hence, SSP reduced
265
impoverishing effect of illness. Equitable utilisation, especially horizontal equity based
on income and gender was observed with insured poorest and women using the health
services more than those of uninsured and newly insured households. There was no
evidence for the sustainability of resource mobilisation. Hence, the facilitating role of the
government in the form of clear policies and provision of subsidies, especially for the
poorest is required.
7.4 Managerial Implications
Management of SSP and similar MHI schemes either new or already in operation
should consider the following to improve the performance and outcome of the scheme.
1. To widen the membership base and enhance financial sustainability, SSP management
should disseminate information on the various aspects of SSP to create awareness among
the target SHG members. Certain mechanisms can be used to expand risk pool and ensure
sustainability of SSP. These are,
i) Deeper penetration in the existing and new districts through an intensive
awareness programme to enrol higher percent of target population.
ii) Financial assistance to the poorest through financial assistance or grants from
corporate donors.
iii) Use established rural network to penetrate into existing untapped areas.
iv) Effective use of infrastructure and staff of parent organisation (SKDRDP) to
enrol/retain members.
v) Motivate members to join SSP through a sense of community belongingness
and credit facilitates of SKDRDP.
2. To overcome distance barrier, transportation charges can be included in the benefit
package. Many services such as outpatient treatment and wage loss can be covered to
make SSP more enticing. Since outpatient (OP) treatment is excluded, there is a tendency
to be hospitalised to claim from SSP. Inclusion of OP would prevent over-utilisation and
encourage proper channelising of limited health care resources to the pressing needs.
This can be rolled out on a pilot basis to assess its impact on financial sustainability.

266
3. Complete financial protection necessitates changes in certain design features, subjected
to the availability of financial resources and affordability of premium. Modifying benefit
package by increasing benefit amount requires huge funds. The programme is making
losses over the years, yet it is not advisable to increase premium due to its negative effect
on enrolment and revenue collection. Hence, a rational trade-off is warranted that
balances the interests of members and the viability of the programme. To meet the cost of
additional benefits and include the poorest, financial assistance from the government,
corporate sector and other international or national aid agencies can be sought. This may
seem impossible; consistent efforts should be made so that poorest are included in the
risk pool.
4. Supply side interventions in the form of standard treatment protocols, drug formularies
and primary health care facilities are essential to increase financial protection. Provider
control mechanisms should be implemented to detect unnecessary investigations,
fraudulent practices or inflated billing. This is necessary in India where the providers are
mostly unregulated and there are many incentives for supplier (hospital) moral hazard in
the health system. Referral system or strict gate keeping can be implemented to increase
efficiency. The management should be actively involved in the strategic purchasing by
educating members about their rights to seek good quality care at hospitals in monthly
meetings and stipulations in the contract.
5. Recent changes in health insurance aimed at the poor (RSBY) and schemes initiated by
other MFIs would limit the growth of budding MHIs including SSP. In the long run,
intensive propaganda and marketing of SSP is required to maintain or enhance
membership base. Otherwise, SHG members may enrol in other MHI schemes with
greater benefits at lower premium which would decrease enrolment in SSP. The active
participation of members to incorporate community preferences in the benefit package of
SSP should be practised. This would motivate people to renew the membership and
improve satisfaction. Elements of social capital embedded in MFI programme of
SKDRDP namely solidarity, reciprocity and feeling of mutual aid and faith in the
integrity and competence of the management of SSP can be used to achieve wider
267
coverage, penetration into untapped areas and reduction in adverse selection and moral
hazard behaviour.
6. Financial sustainability of SSP requires immediate attention. If it continues to make
huge losses, it will be difficult to tie up with private for-profit insurance companies in the
future. This calls for either increase in membership, premium, or pruning benefit
package. These options require certain trade-offs to be made. To enhance enrolment, the
premium should be low and benefit package is to be generous. This throttles resource
mobilisation and endangers financial sustainability. Increase in premium decreases
membership base, hence reduces revenue collection. Curtailing certain services or
exclusion of some expensive diseases defeats the very purpose of SSP.
7. MHI schemes should provide training in actuarial science and management to improve
their technical expertise and establish MIS to increase effectiveness and efficiency.
Moreover, the impact of changing job market on the staff of SSP needs to be assessed.
Field staff may leave the organisation due to inadequate salary, long working hours and
frequent transfers. Dilution of the religious reverence may take place when SSP expands
to other districts far away from the influence of Dharmasthala temple. Hence, the
programme has to build trust through action and customer responsive policies and
products. Replicability of SSP in other states or districts depends on the financial and
managerial support from the external parties. The development of micro-credit has taken
three decades to grow to present status. Micro insurance too has to undergo a long
journey by synchronizing the expertise from insurers, distributors (MFIs or NGOs),
service providers and build capacity for scaling up of MHI.
7.5 Policy Implications
Taken together, the findings of the present study suggest a greater role for the
government and the corporate sector. The following points highlight the active role of the
policymakers to make a MHI scheme viable and sustainable.
1. MHI provides financial protection to vulnerable sections of the society, hence scaling
up of MHI to penetrate into remote rural areas is required. Since MHI removes financial

268
barriers to access care and thereby facilitates treatment at good hospitals, policymakers
should promote it.
2. Poorest were not largely represented in membership pool due to lack of affordability.
Since MHI can be the most important mechanism, policymakers should support it
financially and regulate its operations. The poorest can be motivated to be a member of
MHI by subsidizing the premium. The debate is whether policymakers should stress
nation-wide health insurance scheme or strengthen health care system by establishing
quality hospitals in remote and rural areas. If insured members cannot access care due to
its unavailability of facilities, health insurance is purposeless. Given the moderate
performance of the government since independence in setting up health facilities,
provision of health insurance might be a better option.
3. Intensive monitoring of admitted insured members for any fraudulent activities, fixing
of price for each disease in consultation with hospitals, regular audits to detect any
financial irregularities, creation of awareness on MHI and its benefits to target population
were some of the factors identified as responsible for the success of SSP. These factors
are essential for the effective implementation of any MHI scheme including RSBY.
4. Self help groups and other community organisations are to be promoted to scale up
MHI schemes for faster information dissemination, local knowledge and awareness.
These SHGs are the target population of MHIs, hence larger risk pool and deeper
insurance penetration is achieved through their promotion. Hence, ‘financial inclusion’
can lead to ‘insurance inclusion’.
5. There is no single solution to any problem, even in the case of health care financing.
Different elements, actors and mechanisms are to be judiciously combined to achieve the
Alma Ata declaration of ‘health for all’. Hence, MHI can provide financial protection if it
is implemented with accountability, dedication and strict monitoring of various
participants that includes regulators, insurance companies, hospitals, members and
administrators of the scheme.
At the end, it should be remembered that MHI is just a health financing
mechanism and not a magic potion for all evils in the health system. As a supporting
269
mechanism, it complements the government efforts to provide health care to all
population. This innovative mechanism should be utilised by the members to the fullest
through better understanding of its benefits to oneself and others in the community.
7.6 Limitations
A number of limitations need to be considered. Firstly, the study findings are not
generalisable to the entire population of India. Nevertheless, the study findings are
applicable to similar MHI schemes initiated by MFIs in Karnataka. The plausible
explanation centers on the fact that there is less divergence in socio-economic
characteristics of SHGs such as income, occupation, education and area of residence.
Secondly, the present study was not able to analyse non-financial barriers such as lack of
health facilities, transportation or intra-household dynamics. The current research was not
designed to evaluate factors that cause non-financial barriers. Third limitation was the
recall bias related to income, expenditure and treatment costs. This bias was minimised
by asking the households to recall each episodes of illness and produce medical bills.
Moreover, the questionnaire collected data on medical expenses, drug costs,
transportation costs, lodging charges, interest on the amount of borrowing, wages per day
and number of work days lost separately. Thus, treatment cost was the aggregate of
multiple items. This controlled the recall bias largely. Fourthly, family income was
calculated by adding the income of every member of the family (weekly income in case
of seasonal worker/agriculturist). There might be a tendency for the families to provide
incorrect income data. However, the field staffs cross verified the income data since they
knew the income pattern of the households clearly. Fifthly, geographical access to care
determines utilisation and affects the pattern of seeking care. Hence, a comparison of
insured, newly insured and uninsured individuals on the health seeking behaviour and
utilisation may be biased. Nevertheless, the study design minimised such bias by
selecting the individuals from the same location/ karyakshetra. Lastly, the findings on the
sale of assets as one of the risk coping strategies should be interpreted carefully. Since the
sample size was small, the findings cannot be applied to a larger target population.

270
7.7 Suggestions for Future Research
This study has thrown up many questions in need of further investigation. It is
recommended that more research be undertaken in the following areas.
1. This study could identify some aspect of social capital such as mutual aid, concern for
others well-being, and solidarity during illness. Further research might explore the impact
of MHI on various dimensions of social capital.
2. It would be interesting to compare the risk management techniques of different
schemes to build up the repertoire on the best practices.
3. A better understanding of the impact of MHI on intra-household dynamics, in
particular women is needed.
4. Further research on the equity impact of MHI using experimental study design can be
undertaken.

271
272
APPENDIX I: ENGLISH HOUSEHOLD QUESTIONNAIRE USED FOR THE

SURVEY

Dear Participant,

Greetings!

I am doing a doctoral study on the ‘Impact of Micro Health Insurance on Sampoorna

Suraksha Programme in Karnataka’. The purpose of the study is to understand the impact
of Sampoorna Suraksha on insured. I have selected you as one of the participant for this
study. I request you to kindly fill this questionnaire which will take 20 minutes. The
instructions for completing the questionnaire are given on the form itself. I ensure you
that the information given by you will be kept strictly confidential and your identity will
not be disclosed anywhere. Your participation represents a valuable contribution to my
research. I appreciate your co-operation for this study.

Q1. Has anyone not been completely well within the last one year? (Include any kind of
illness suffered, as well as problems of pregnancy and childbirth, even if treatment was
not sought. Include outpatient treatment in the last 3 months also. If any member suffered
more than one illness or more than one hospitalisation in the last year, then each episode
should be recorded separately)
□ Yes, continue
□ No, GO TO 22

ID of individuals 1 2
Q2. What is your age?
Q3. Gender
Male=1, Female=2
Q4. How many times you had 1 2 3 1 2 3
illness? (give the following
information on each episode)
Q5. What was the illness?
Q6. Did you get treatment?
OP=1, IP=2, No=0
IF NO, GO TO Q19
Q7. How many days were you
hospitalised? (If IP)
Q8.Where did you go for
treatment?
Q9. Why did you go there?

273
Q10. What was the amount
spent for treatment?
Hospital expenses
Medicine
Laboratory/ diagnostic charges
Indirect expenses (pay phones,
lodging, food or drink)
TOTAL
Q 11. What was the result of
the treatment?
Q12. How did you pay the
bills?
Q 13. Did you have money to
pay medical bills? [yes >GO
TO 19; no=0, go to next
question]
Q 14. How did you mobilise
the money?
Q15. How much money did
you borrow?
Q16. What was the interest rate
charged?
Q17. Is you used savings, how
much was used?
Q18. If you sold assets, how
much you got from the sale?
Q19. How many days were
you ill?
Q20.How many days you did
not go for work?
Q21.What was the daily wage
rate(rupees)? (if you are a
labourer)
Instructions: Code for above questions
Q8.
1-Home remedy 2-Clinic
3-Government hospital 3- Small private hospital
4-Large private hospital 5- Pharmacy
6- Ayurveda/Homeopathy 9- Others (specify) _____________

274
Q9.
1-Accessibility □ Yes □ No
2-Lack of improvement □ Yes □ No
3-Lack of money to pay □ Yes □ No
4-Quality of treatment □ Yes □ No
5-Low cost of treatment □ Yes □ No
6-Trust in treatment □ Yes □ No
7-Near to home □ Yes □ No
8-Severity of illness □ Yes □ No
9-Nature of illness □ Yes □ No
Q11.
1-Better 2- Slight improvement 3- No improvement
Q12.
1- From pocket 2- Other health insurance (private)
3- Sampoorna Suraksha 4- Others______________
Q14.[ Tick all relevant answers]
Borrowed □ Yes □ No
Sold crop □ Yes □ No
Used savings □ Yes □ No

For Sampoorna Suraksha members only, Non members GO TO Q25

Q22. How long you have been the member? ______years

Q23. Did you claim any benefit under the scheme so far?
□ Never □ Once □ Twice □ Thrice or more

Q24.What is the amount of benefit availed by your household so far?____________

Q25. How far is the hospital or clinic from your house?

Distance_____km or time taken____minutes/hours
Q26. How far is the Suraksha hospital or clinic from your house?
Distance ____________km or time taken _____________minutes/hours
Q27.What made you join Sampoorna Suraksha Programme scheme? [Tick all relevant
answers]
Benefit package □ Yes □ No
Can go to better hospitals □ Yes □ No
All members in the group have enrolled □ Yes □ No
Need not worry about money □ Yes □ No
Peace of mind □ Yes □ No
Let others get benefitted □ Yes □ No
May need in future □ Yes □ No

275
Q28. Did you take any loan to pay the premium?
□ Yes □ No

Q 29.Do anyone in your family have a permanent illness or suffer from any illness for
longtime?
□ Yes □ No

Q30. If yes, write their age and illness ______,______; ______;______

How satisfied are you with the network hospitals of SS?

Fully Somewhat Neither satisfied Somewhat Fully

satisfied satisfied nor dissatisfied dissatisfied dissatisfied
4 3 2 1
5
Q 31.Overall cleanliness of the hospital 1 2 3 4 5
Q 32.Expertise /experience of doctors 1 2 3 4 5
Q 33.Care given by nurses 1 2 3 4 5
Q 34.Facility (laboratory, X ray, equipments) 1 2 3 4 5
Q 35.Availability of drugs in pharmacy 1 2 3 4 5
Q36. Friendliness and courtesy of the staff 1 2 3 4 5
Q 36.Time spent by doctor in examination 1 2 3 4 5
Socio-economic information
Q 38. Gender of respondent:
□ Male □ Female

Q 39. Marital status of the respondent

□ Married □ Widow(er)
□ Divorced or separated □ Have never been married

Q 40. Relationship to household head:

□ Head of household □ Spouse
□ Brother/Sister □ Son/Daughter
□ Son-in-law/Daughter-in-law □ Parent

Q 41. Religion:
□ Hindu □Muslim
□ Christian

276
Q 42. For eeach househ
hold membeer, please prrovide the fo
ollowing infformation
S.No. 1 Gender 2. Age
1. 3. Maritall 5. 6. Prim
mary 7.In ncome
M
Male=1 (in years) Status Educatio on occuppation (daily/monthlyy)
F
Female=2 ( )

1
2
3
4
5
6

3. Married 1 Widow(er)/divorcee 2
Unmarrieed 3
5. Nursery School 00 Passed Massters degreee or above 14
1 to 12 th standard com
mpleted 01 through
t 12 Passed Bacchelor Level 13
Passed techhnical diplom
ma 15 Illiterate 0

6. Primary occupation:
Unemployeed (includin ng not able to
t work) 01
Unskilled daily
d y workeror beedi)
wages (agriculturaal or factory 02
Presently sttudying 03
Farming (inncluding daairy farmingg) 04
Self-employed (petty-b business) 05
Unskilled monthly
m salaary (housem
maid) 06
Skilled dailly wages (plumber, electrician) 07
Employed in i formal seector (includ
ding govern nment) 08

7. If daily labourer, ho
ow many day
ys do you w
work in a mo
onth?________

Q 43.Area of residencee:
□Urban □ Semi-urb
ban □ Rural

27
77
Q 44.How many off the follow
wing are own
ned by yourr household
d members?

House
Land fo
or house
Land fo
or cropping;; Owned
Mortgaaged
Goats, chicken,
c cowws
Radio
Television
Bicyclees
Scooterrs/mopeds/mmotorcycle
Four orr three wheeeled vehicles
Electricc fans
Mobile phone
Refrigeerator
Any maachines
Others (specify)
(

Q 45.Household inncome from

m itemized so
ources:

Incomee source Amount (iin )

Crops (annual)
Livestoock (milk/eggg) (monthly y)
Interestt on savingss (monthly)
Rent onn land/ builddings (montthly)
Gifts (reeceived fromm relatives and
friends))
Any oth her income

Q46. Do you use grain

g from your
y own fieeld or receivve grain instead of wagges?
□ Yes □ No
N
Q47. Iff yes, what is the money
y equivalen
nt of the graiin? ______
______

278
Q48 . Item wise expen
nditure (Insttructions: m
month or yeaar).

Expenditurre- item wisee Am

mount (in ) Expenditture- item Amount (in )
wise
Food LPG/ firrewood
Clothing Electriciity
Transportaation Moblie
Healthcaree (including
g Educatioon
medicine)
Daily goodds Social activities
Durable gooods Gifts (asssistance to
relativess and friendds)
Loan interrest Tobacco o/alcohol
Entertainm
ment Other iteems
Expenditurre in expensive month (in
( rupees) __________
_ ______
Expenditurre in cheap month
m (in ru
upees) ____
__________ ____
Housing reelated inforrmation:
Q49. Do yo
ou have elecctricity conn
nection?
□Yes □Noo

Q50. What is the typess of flooring

g?
□ Mud □ Cerammic tiles □ Cement-rred oxide □ Marble

Q51. What types of waall material is used?

□ Mud □ Brickk □ Cement blocks
b

Q52. What types of ceeiling is used

d?
□ Thatchedd □ Clay tiles □ Metaal tin □ Conccrete

Q53. How m
many room
ms are there in
i your hou
use?_______
_

Q54. What types of toiilet facility does your family

f use?
□ Open □ Private □ Public
P

Q55. What is the sourcce of water for your fam

mily?
□ Piped into house □ Public tap □ Well
W
□ River/sprring □ Water
W tank

I thank you
u for your co-operatio
on and spending yourr valuable ttime.

27
79
APPENDIX II: ENGLISHQUALITATIVE INTERVIEW SCHEDULESUSED FOR
THE SURVEY
AP 3.1: Focus Group Discussion with Insured and Uninsured Groups
We would like to know your thoughts and ideas about Sampoorna Suraksha and
your past illness related actions. Please feel free to share your ideas when relevant. You
may also ask for clarification if a question is not clear. Your names will not be recorded
or associated with any remarks. This discussion group will last approximately half an
hour.
• Do you access health services during illness?
• What are the reasons for not accessing care despite illness?
• (For insured)
1. Did Sampoorna Suraksha remove barriers to access care?
2. If you get admitted, do you stay longer days in the hospital?
3. Are there very rich or very poor members in your group?
4. Are you happy with Sampoorna Suraksha benefits?
5. Do you have any complaints regarding Sampoorna Suraksha?
6. Do you want any changes to be made in Sampoorna Suraksha? If yes, what
features are to be modified? Benefit package, network hospitals or premium
amount?
7. How do field staffs behave with you?
8. Who took the decision to enrol in Sampoorna Suraksha?
9. Did you have trust in Sampoorna Suraksha since you joined self-help group?
• (For uninsured only)
1. Why didn’t you join Sampoorna Suraksha?
2. Did you know the credit facility given by Sampoorna Suraksha to pay the
premium?
I thank you for participating in this discussion.

280
AP 3.2: Interviews with Health Care Providers
Name of the hospital:
Town:
Address:

Position of the respondent:

• What types of health care do you provide?
□ Outpatient (OP) □ Inpatient (IP)
□ Both OP and IP □ Traditional (ayurvedic, unani, homeopathy)
□ Others_____________________
• Is there any protocol or standard treatment guideline?
□ For all patients □ For insured only
• Do you participate in scheme decision making related to cost of care, health care
quality and related issues?
□ Yes □ No □ Don’t know
• Do insured patients have special queues (speedy service)?
□ Yes □ No □ Don’t know
• What is the nature of your interaction with scheme administrators, if any?

_________________________________________________________________

• What is the nature of your interaction with insurance company, if any?

___________________________________________________________________

I thank you for your kind co-operation.

281
AP 3.3: Interviews with Scheme Administrators

• When did your Micro health insurance (MHI) start? Year: ________
• Why was your MHI program created?
__________________________________________________________________
• Who supported your MHI program financially at the beginning?
_______________________________________________________________________
• How is your program funded now? What are its sources of revenues? (Check all that
apply)
□ Member Premiums/Contributions □ Government
□ Donors □ Others _____________________
• Is membership voluntary or mandatory?
□ Voluntary for all □ Mandatory/Compulsory for Self help group members
□ Other_______________________
• What do members of your MHI have in common? (Check all that apply)
□ Region □ Community
□ Professional group □ Other: ______________
• Does your MHI offer reduced contributions for certain members?
□ Yes □ No
• Who was involved in designing benefit package? (Check all that apply)
□ Insurance company managers □ Target community
□ Government □ SKDRDP management
□ Other ____________
• How often is this revised/updated?
□ Every year □ Every two years □ Other _____________
• Are SS members involved in the management of the scheme?
□ Yes □ No

282
• If yes, what kind of roles/decisions members take?
□ Fixing benefit package □ Premium collections
□ Claim management □ Enrolment of new members
□ Creating awareness about SS □ Renewal of membership
□ Assessing quality of care of hospitals □ Others (specify)_______________
• What is the types of provider payment method used by the scheme?
□ Global budget □ Capitation
□ Case -based payment □ Line item budget
□ Per diem (per day) □ -Fee-for-service
• How do you deal with shortage of funds?
• Are your records regarding your operation (on members, claims, premium and
expenditure) computerized [Management Information System]?
• □ Yes □ No
• Was there any incidence of fraud and abuse of the scheme?
• □ Yes □ No
• If yes, what kind of fraud took place?(supplier or insured moral hazard, adverse
selection, impersonation)
• Describe the design and management of the scheme.
• How do people join in the scheme?
• What is the nature of your interaction with beneficiaries and health care providers?
• What is the regulatory framework in which your scheme works?

• Do you use referral system to refer insured to different hospitals?

□Yes □ No
• Do you impose waiting period for newly insured (period during which coverage is not
provided)
□Yes □ No

283
• Do you crosscheck the beneficiaries?
______________________________________________________________
• Does it offer reduced premium to enrollees to encourage them to use restricted choice
of providers?
□Yes □ No
• Does the scheme accept in-kind contributions of premium from insured?
□Yes □ No
• Do you monitor the behaviour of hospitals?
□Yes □ No
• Do you subsidise the premium for the poorest?
□Yes □ No
• Do you accommodate the income-generating patterns of households employed in
agriculture and the informal sector (irregular, often noncash) for contribution
payment?
□Yes □ No
• Do you have any mechanisms of strategic purchasing?
□ Gate keeping □ Drug formularies
□ Selective contracting □ Referral practice
□ Provider financial incentives
□ Financial incentives to encourage insured to use particular providers
□ Others______________
(Human Resource department managers only)
• How the staffs are recruited?
□ Local community □ Internal candidates
□ External sources □ Drawn from members
□ Others_____________________
• How is SSorganised (organisation structure)?
• What is the basic qualification of different cadre of staff?

284
• Do you provide any training to your employees in managerial skills?
□Yes □ No

• Do you provide any training to your employees on health insurance?

□Yes □ No
• (For field staff only) How often do you contact the members?
______________________________________________________
• Do you educate target population regarding health insurance especially Sampoorna
Suraksha?

□ Yes □ No

• How do you assist Sampoorna Suraksha members during admission to hospitals?

_____________________________________________________________

• Do you maintain accurate and up-to-date records of members’ addresses?

□ Yes □ No

• Were there reports of non-sanctioning of the claim? If yes, for what reasons they were
rejected?

________________________________________________________________

I thank you for the support and information you have shared with me for this study.

285
APPENDIX III:PILOT STUDY REPORT

A pilot study was undertaken in December 2010 to test the ideas and measure the
validity and reliability of questionnaire that was used to collect data to answer research
questions. The questionnaire was drafted based on literature keeping the research
objectives in mind. The sample size was 30 and the respondents were selected using
convenience sampling method. The study took place in Mangalore and Bantwal taluk of
Dakshina Kannada district in Karnataka, India. After 15 days, retest was performed on 15
respondents of earlier sample. Statistical testing was done to find out the reliability. The
content validity of the questionnaire was scrutinised by subject experts.
The respondents included renewed insured (12 individuals), newly insured (10
individuals) and uninsured (eight individuals) self-help group members to represent the
target population. Twenty of thirty respondents had an incidence of illness in the family.
One third of respondents was male; either spouse or children of head of household.
Majority of them were from informal sector mainly unskilled labourers.
The respondents understood most of the questions easily. Some of the questions
that were found to be difficult were re-framed. Some of the options that were not
included in the questionnaire but opted by the respondents were later included in the final
questionnaire. Retest found recall and response to be reliable. The questionnaire was well
understood and had clear instructions.
Questionnaire was re-drafted after making changes to wordings of the sentence,
order of questions, range of answers on multiple-choice questions and removal of some
questions that was unnecessary or ambiguous. The reliability was checked by calculating
kappa coefficient and inter class coefficient. Most of the items in the questionnaire had
high value of coefficient (range from 0.634 to 0.99; p<0.05). Cronbach’s alpha for quality
of care questions was 0. 765. The time taken to fill the questionnaire was found to be 20
minutes.

286
 APPENDIX IV: SURVEYED DISTRICTS, TALUKS, VALAYAS AND KARYAKSHETRAS

Dakshina Kannada Uttara Kannada Gadag

Mangalore Puttur Sullia Karwar Sirsi Kumta Honnavar Bhatkal Gadag-Betageri Shirhatti

kaikamba Nalkuru Amdalli SIrsi A Kumta-A Manki Jali Hulkoti Suranagi

Bannur
Kuppepadavu Amarapadnur Sirsi B and B Honnavar Gadag- Shirhatti
Mulki Karnad Chikmudnur
Betageri

Boliyaru Aikala
Kemral Karnad Nekkiladi Kelanja Amdalli A Gandhinagar Bhaggon Kasarkoda Honnagadde KurthakotiA Shigli A Shigli B
Kemmayi Guthigaru Amdalli B Ganeshanagar Kalabhagh Manki B Harnagadde Kurthakoti B Laxmeshwar
Pakshikere
Balya Chendia Indiranagar Kumta A
Malavur Kunjathur Bannur Manki C Okkalageri Basaveshwara Nagar Mundaragi
Nalkuru Kasturbanagar Hanehalli
Kinnigoli Adyar Chikmudnur Karwar A Edagunji Bhatkala city Vivekanandanagar Shirahatti A
Ubaradka Neharunagar
Halengadi Kabaka Holanagadde Manki A LakkundiA Shirahatti B
Amarapadnur Rajivanagr
Permude Kasaba Aranthodu HonnavarA Lakkundi B
Ramanabailu
Kaikamba Yeyyadi Ramakunja Peraje Honnavar B Okkalageri
Kenya Bislakoppa
Edapadavu Haleneranki Belake Shidhalinganagar
Kodimbala Chokkadi Ambedikarnagar
Ganjimath Pilaru
Kumpala Mogaru Kedambadi Hulkoti
Malali Gundalike

107+79+150=334 39+38+30= 47+50+23= 41+18+ 12= 45+25+19= 44+ 22+ 12= 28+55+36= 18+ 37+ 25= 26+33+36=95 19+15+15=49
107 120 71 93 78 119 80

287
APPENDIX V: PROFILE OF KARNATAKA AND SAMPLE DISTRICTS

Karnataka Uttara Kannada Dakshina Kannada Gadag

Population Persons 61130704 1436847 2083625 1065235
Males 31057742 727424 1032577 538477
Females 30072962 709423 1051048 526758
Decadal growth 15.67 6.15 9.8 9.6
rate (2001-11)
Sex ratio 968 975 1018 978
Number of Persons 41029323 1084277 1,666,834 705136
Literates Males 22808468 585127 866331 401560
Females 18220855 499150 800503 303576
Literacy rate Persons 75.6 84.03 88.62 75.18
Males 82.85 89.72 93.31 84.89
Females 68.13 78.21 84.04 65.29
Human Development Index 0.65 0.653 0.722 0.634
Health index 0.712 0.781 0.823 0.628

Source: Census of India, 2011; Karnataka Human DevelopmentReport 2010; Directorate

of Economics and Statistics, Karnataka

288
 REFERENCES

Abel-Smith, B. (1992). “Health insurance in developing countries: Lessons from experience.”

Hth. Poli.Plann.,7(3), 215-226.
Aday, L.A. and Awe, W.E. (1997). “Health services utilisation models.” Handbook of Health
Behaviour Research I: personal and Social determinants, D.S. Gochman, ed., Plenum, New
York, 153-172.
Aday, L.R. and Andersen, R. (1974). “A framework for the study of access to medical care.”
Hlth. Serv. Res., Fall; 9(3), 208-220.
Aggarwal, A. (2010).“Impact evaluation of India’s ‘Yeshasvini’ community-based health
insurance programme.” Hlth. Econ., 19, 5-35. DOI: 10.1002/hec.1605.
Ahmed, S. M., Adams, A.,Chodhary, M.and Bhuiya, A. (2000). “Gender, socio-economic
development and health seeking behaviour in Bangladesh.” Soc. Sci.Med., 51,361-171.
Ahmed,S.M., Adams,A., Chodhary,M. and Bhuiya,A.(2003). “Changing health-seeking
behaviour in Matlab, Bangladesh: do development interventions matter?”Hth. Poli.Plann.,18(3),
306-315.
Ahmed,M..U., Islam, S.K., Quashem, M. and Ahmed, N. (2005).Health Micro Insurance: A
comparative study of the Three Examples in Bangladesh, CGAP Good and Bad Practices Case
Study No. 13. ILO Social Finance Programme, Geneva.
Alberts, J. F., Sanderman, R., Eimers, J. M. and Heuvel, W. (1997). “Socio-economic inequity
in health care: A study of services utilisation in Curacao.”Soc. Sci.Med., 45(2), 213–220.
Alvarez, B., Pellise, L. and Lobo, F. (2000). “Payment systems for health care providers in Latin
America and OECD countries.”Rev. Panam. Salud. Publica., 8, 55–70.
Andersen, R. and Newman, J.F. (1973). “Societal and individual determinants of medical care
utilisation.” Milbank Mem. Fund Quart.:Hlth Society, 51, 95-124.
Arhin-Tenkorang, D. (2001). Health Insurance for the informal sector in Africa: Design
features, Risk protection and resource mobilisation, Commission on Macro economics and
Health, WHO, Geneva.
Arrow, K. J. (1963). “Uncertainty and the welfare economics of medical care.” The Amer.

289
Econ.Rev., 80(5), 941–973.
Asfaw, A., Lamanna,F., Klasen,S. (2010). “Gender gap in parents' financing strategy for
hospitalisation of their children: evidence from India.” Hlth. Econ., 19(3), 265-279.
Atim, C. (1998). Contribution of mutual health organisations to financing delivery and access to
health care: synthesis of research in nine West and Central African countries, Abt Associates
Inc., Bethesda.
Atim C. (1999) “Social movements and health insurance: a critical evaluation of voluntary, non-
profit insurance schemes with case studies from Ghana and Cameroon.” Soc. Sci.Med., 48
(7),881–96.
Auchincloss, A., Nostrand,J. and Ronsaville,D.(2001) “Access to Health Care for Older Persons
in the United States: Personal, Structural, and Neighborhood Characteristics.” J Aging Health
13; 329. DOI: 10.1177/089826430101300302.
Baeza,C., Montenegro,F.andNunez,M.(2002). Extending Social Protection In Health Through
Community Based Health Organisations: Evidence and Challenge, International Labour
Organisation, Geneva.
Basaza,R., Criel,B. and van der Stuyft,P. (2008). “Community health insurance in Uganda: Why
do enrolment remain low? A view from beneath.” Hlth. Poli., 87 (2008), 172–184.
Bennett S, Creese A. and Monasch R. (1998).Health insurance schemes for people outside
formal sector employment, World Health Organisation, Geneva.
Bennett, S. (2004). “The role of community-based health insurance within the health care
financing system: a framework for analysis.” Hlth. Poli.Plann., 19(3),147–158.
Birch, S., McIntyre, D. and Thiede, M. (2007). “Access as a policy-relevant concept in low-and
middle-income countries.” Global forum for health research- Helping correct the 10|90 gap,
Forum 11, Beijing, People’s Republic of China.
Block, E. (1986). “The comprehension strategies of second language readers.” TESOL Qtrly, 20
(3), 463-494.
Bogg, L., Hengjin, D., Keli, W., Wenwei, C., Diwan, V.(1996). “The cost of coverage: rural
health insurance in China.” Hlth. Poli. Plann.,11(3), 238-252.

290
Cameron, A.C., Trivedi, P.K., Milne, F. and Piggott, J. (1988). “A micro econometric model of
the demand for health care and health insurance in Australia.” Rev. Econ. Stud., 1, 85-106.
Carrin, G., Ron, A., Hui, Y., Hong, W., Tuohong, Z, Licheng, Z. et al. (1999). “The reform of
rural cooperative medical system in the People’s Republic of China: interim experience in 14
pilot counties.” Soc. Sci. Med., 48(7), 961–72.
Carrin,G., Waelkens, M. and Criel, B.(2005). “Community- based health insurance in
developing countries: a study of its contribution to the performance of health financing
systems.” Trop. Med. Intl. Hlth., 10(8), 799-811.
Census of India,
http://www.censusindia.gov.in/vital_statistics/Publications/Vital_Statistics_Publications.aspx
(Feb.8, 2012).
Chankova, S., Sulzbach,S. andDiop, F. (2008). “Impact of mutual health organisations: evidence
from West Africa.” Hlth Poli Plann., 23(4), 264-276.
Chuma, J., Gilson, L.and Molyneux, C. (2007). “Treatment-seeking behaviour, cost burdens and
coping strategies among rural and urban households in Coastal Kenya: an equity analysis.” Trop.
Med. Intl. Hlth., 12 (5), 673-686.
Criel, B. and Kegels, G. (1997). “A health insurance scheme for hospital care in Bwamanda
district, Zaire: lessons and questions after 10 years of functioning.” Trop. Med. Intl. Hlth.,2, 654-
672.
Criel, B. (1998). District-based Health Insurance in sub-Saharan Africa, Part II, Case Studies.
Studies in Health Services Organisation and Policy, Nos. 9 and 10, Institute of Tropical
Medicine, Antwerp.
Criel, B. and Waelkens, M.P. (2003). “Declining Subscriptions to the Maliando Mutual Health
Organisation in Guinea-Conakry (West Africa): what is going wrong?” Soc. Sci. Med., 57,
1205–19.
Cummins J.D., Weiss M.A. Zi H. (2007) Economies of scope in financial services: A DEA
bootstrapping analysis of the US insurance industry, Working paper, The Wharton School,
Philadelphia

291
Cutler, D. M. and Zeckhauser, R.J. (1999).The Anatomy of Health Insurance, NBER Working
Paper No. W7176, http://ssrn.com/171140 (Jan 12, 2011).
Data Monitor (2005). “Insurance industry in India: Industry profile.” Reference Code: 0102-
2087, http://www.datamonitor.com/Default.aspx (Apr 18, 2010).
Datt, G. (1997). Poverty in India, 1951 – 1994: trends and decompositions, Mimeo, The World
Bank and IFPRI, Washington D.C.
Dave, P. (1993). “Community and self-financing in voluntary health programs in India.” Paying
for India’s health care, P. Berman and M.E. Khan, eds., Sage publications, New Delhi, 297–
321.
Dave, Sen, P. (1997). Community control of health financing in India: A review of local
Experiences, Technical Report, Abt Associates Inc., Bethesda.
De Allegri, M., Kouyate, B., Becher, H., Gbangou, A., Pokhrel, S., Sanon, M. et al. (2006).
“Understanding enrolment in community health insurance in sub-Saharan Africa: a population-
based case-control study in rural Burkina Faso.” Bull. of the World. Hlth. Org., 84, 852–8.
Dekker, M. and Wilms, A.(2009). “Health Insurance and Other Risk-Coping Strategies in
Uganda: The Case of Microcare Insurance Ltd.” World Dev., 38(3), 369-378.
doi:10.1016/j.worlddev.2009.09.004.
Deolalikar, A. (2002). “Access to Health Services by the Poor and the Non-Poor: The Case of
Vietnam.” J. Asi. Afr. Stu., 37(2), 244 -261.
Dercon, S. (2002). “Income Risks, Coping Strategies, and Safety Nets.” World Bank Res. Obs.,
17 (2), 141-66.
Dercon, S., Bold,T. and Calvo,C. (2004). Insurance for the Poor?, QEH Working Paper Series –
QEHWPS125, Working Paper Number 125, University of Oxford,
http://www3.qeh.ox.ac.uk/pdf/qehwp/qehwps125.pdf (Mar 9, 2010).
Devadasan,N., van Damme,W., van der Stuyft, P. and Criel,B.(2005). “Does community health
insurance increase access to health care for the poor? Evidence from India.” Global Health
Forum, Mumbai, India.
Devadasan,N., Ranson,M.K.,Damme,W V., Acharya,A. and Criel, B.(2006) “The landscape of

292
community health insurance in India: An overview based on 10 case studies.” Hlth. Poli.,
78(2006), 224-234.
Devadasan,N. (2006) Planning and implementing health insurance programmes in India: An
operational guide, Institute of Public Health, India.
Devadasan, N., Criel,B., Damme,W.V.,Ranson,K. and Stuyft, P.V.(2007). “Indian community
health insurance schemes provide partial protection against catastrophic health expenditure.”
BMC Hlth Ser. Res., 7:43, http://bmj.com/cgi/content/full/334/7607/1309 (Jun 12, 2009).
Diop,F., Yazbeck,A. and Bitran,R. (1995). “The impact of alternative cost recovery schemes on
access and equity in Niger.” Hlth. Poli.Plan., 10(3), 223-240.
Directorate of Economics and Statistics, Government of Karnataka,
http://www.karunadu.gov.in/des/pages/survey.aspx, (Oct 22, 2011).
Dror, D.M., Soriano, E., Sarol, J., Azcuna, R., Lorenzo, M. and Koren, R.(2005). “Field based
evidence of enhanced healthcare utilisation among persons insured by Micro Health Insurance
Units in Philippines.” Hlth Poli., 73(3), 263-271.
Dror,M., Radermacher,R., Khadilkar, S., Schout, P., Hay, F., Singh, A. and Kore, R.(2009).
“Microinsurance: Innovations In Low-Cost Health Insurance.” Hlth Affr., 28(6), 1788–1798.
Duffy, K. (1995). Social exclusion and human dignity in Europe, Council of Europe, Strasbourg
Economic survey of India, http://indiabudget.nic.in/survey.asp (Dec 14, 2011).
Economic survey of Karnataka, http://planning.kar.nic.in/EconomicSurver2010-11 (Dec 18,
2011).
Ekman,B. (2004). “Community-based health insurance in low-income countries: a systematic
review of the evidence.”Hlth. Poli.Plan., 19(5), 249–70.
Ekman,B.(2007a). “The impact of health insurance on outpatient utilisation and expenditure:
evidence from one middle-income country using national household survey data.”Hlth. Res.
Poli.Sys., 5:6 doi:10.1186/1478-4505-5-6 http://www.health-policy-systems.com/content/5/1/6
(Jul 12 2010).
Ekman,B.(2007b). “Catastrophic health payments and health insurance: Some counterintuitive
evidence from one low-income country.”Hlth Poli., 83 (2007), 304-313.

293
Falkingham, J. (2004). “Poverty, out-of-pocket payments and access to health care: evidence
from Tajikistan.” Soc.Sci.Med., 58 (2), 247–258.
Franco,L.M., Diop,F., Kelley,A.G., Makinena,M. andSimparae,C.H. (2008) “Effects of mutual
health organisations on use of priority health- care services in urban and rural Mali: a case–
control study.” Bull. of the World. Hlth. Org., 86 (2008), 830–838.
Galarraga,O., Sosa-Rubi,S., Salinas-Rodriguez,A.and Sesma-Vazquez,S. (2010). “Health
insurance for the poor: impact on catastrophic and out-of-pocket health expenditures in Mexico.”
The Eur. J.Hlth. Econ.,11(5), 437-447.
Gotsadze,G., Bennett,S.andGzirishvili,D. (2005). ”Health care-seeking behavior and out-of-
pocket payments in Tbilisi, Georgia.” Hlth Poli Plan., 20(4), 232-242.
Gottret,P. and Schieber,G. (2006). Health financing revisited, a practitioner’s guide, The World
Bank, Washington D.C.
Gumber,A. (2001). Hedging the health of the poor, The case for community financing in India,
Health, Nutrition, and Population Family (HNP) discussion Paper Series, The World Bank,
Washington D.C.
Haddad, S., Fournier, P., Potvin, L (1998). “Measuring Lay People’s Perceptions of the Quality
of Primary Health Care Services in Developing Countries: Validation of a 20-item Scale.” Intl J
Qlty Hlth Care, 10, 93-104.
Hammond J.D., Melander E.R., Shilling, N. (1971). “Economies of scale in the property and
liability insurance industry.” The J of Risk Insur, 38 (2),181-191.
Heltberg, R and Lund, N.(2009) “Shocks, Coping, and Outcomes for Pakistan’s Poor: Health
Risks Predominate.” J. Dev. Stud., 45(6), 889-910.
Hensely (1962). Competition, Regulation and the Public interest in Non life insurance,
University of California Press, Los Angeles, California.
Hsiao, W. C.(2004). “Experience of community health financing in the Asian region.” Health
Financing for Poor People: Resource Mobilization and Risk Sharing, A.S. Preker and G. Carrin,
eds., The World Bank, Washington, D.C, Ch.3.

294
Hsiao,W.(2001). Unmet health needs of two billion. Is community financing a solution? Health,
Nutrition, and Population Family (HNP) discussion Paper Series, The World Bank, Washington
D.C.
Insurance Regulatory and Development authority (2005), www.irdaindia.org/regulations/IRDA-
Mirco-Ins_reg-2005.pdf(Oct 12, 2009).
Israel, G.D. (1992). “Determining sample size.” IFAS, University of Florida, United States of
America www.edis.ifas.ufl.edu/PD006 (September 2010)
Jakab,M. and Krishnan, C. (2001). Community Involvement in Health Care Financing: a Survey
of the Literature on the Impacts, Strengths and Weaknesses, Health, Nutrition, and Population
Family (HNP) discussion Paper Series, The World Bank, Washington D.C.
Jakab, M., Preker A.S., Krishnan C, Schneider P, Diop F, Jutting J, et al.(2004). “Analysis of
community financing using household surveys.” Health financing for poor people: resource
mobilisation and risk sharing, A.S. Preker, G. Carrin, G, eds., The World Bank, Washington
D.C., Ch.5.
Jowett, M., Contoyannis, P. and Vinh, ND.(2002). “The impact of public voluntary health
insurance on private health expenditures in Vietnam.”Soc. Sci. Med., 56(2), 333-342.
Jowett, M. (2004). “Do informal risk sharing networks crowd out public voluntary health
insurance? Evidence from Vietnam.” Apld. Econ., 35(10), 1153–1161.
Jutting, J. and Tine, J. (2000).Micro insurance schemes and health care provision in developing
countries: An empirical analysis of the impact of mutual health insurance schemes in rural
Senegal.ILO/ZEF-Project No. 7359 Project report: 5. Centre for Development Research (ZEF),
Germany.
Jütting, J. (2001). The impact of health insurance on the access to health care and financial
protection in rural areas of developing countries: the example of Senegal. Centre for
Development Research, Germany.
Jütting, J. (2003). “Do community-based health insurance schemes improve poor people’s
access to health care? Evidence from rural Senegal.” World Dev., 32, 273–88.
Karnataka Human Development Report

295
(2010),planningcommission.nic.in/plans/stateplan/sdr_pdf/shdr_kar05.pdf (Aug 2 2010)
Kawabata, K, Xu, K. and Carrin, G. (2002). “Preventing impoverishment through protection
against catastrophic health expenditure.” Bull. of the World. Hlth. Org., 80(2002), 612.
Klein T. (2011) Why do India’s urban poor choose to go private? Health policy simulations in
slums of Hyderabad. SAS Institute Prize for Econometric Analysis, University Meets
Microfinance Award, University of Cambridge, Cambridge
Krishna, A., Kapila, M., Porwal M. and Singh, V. (2005). “Why Growth is not Enough:
Household Poverty Dynamics in Northeast Gujarat, India.” J. Dev. Stud., 41(7), 1163-1192.
Kutzin, J. and Barnum, H. (1992). “Institutional features of health insurance programs and their
effects on developing country health systems.” Intl. J. Hlth. Plann Mgmt., 7, 51-72.
Lei, X.and Lin, W.(2009). “The New Cooperative Medical Scheme in rural China: does more
coverage mean more service and better health?” Hlth. Econ., 18(52), S25-S46.
Leive,A. and Xu,K.(2008). “Coping with out-of-pocket health payments: empirical evidence
from 15 African countries.” Bull. of the World Hlth. Org., 86(2008), 849–856.
Leliveld, A. (2006). “Poverty trap or safety net? Dynamics in social security arrangements in
Ugandan rural economies.” The End of Poverty in Africa? African Studies Centre, Leiden,
Holland.
Litvack, J., Bodart, C. (1993). “User fees plus quality equals improved access to health care:
results of a field experiment in Cameroon.” Soc. Sci. Med., 37(3), 369–83
Liu,G.,Zhaob,Z.,Cai,R.,Yamadad,T. and Yamada,T. (2002). “Equity in health care access to:
assessing urban health insurance reform in China.” Soc. Sci.Med., 55 (2002), 1779–1794.
Liu,H.,Zhang, L.,Summerfield,G. and Shi,Y. (2009). “A gendered view of reforming health care
access for farmers in China.” China Agr.Econ. Rev., 1(2),194-211.
Mathiyazhgan M. (2006) Cost efficiency of public and private hospitals: Evidence from
Karnataka state in India. ISAS Working Paper, Institute of South Asian Studies, National
University of Singapore, Singapore
Mayur,A. (2009). “Health,not home, gets the moolah.”The Economic Times, 25 November,p.15
McCord, M. and Osinde,S.(2005). “Reducing vulnerability: the supply of health microinsurance

296
in East Africa.” J. Intl. Dev.,17(3), 327–381.
McPake, B., Hanson, K.and Mills, A. (1993). ”Community financing of health care in Africa:
An evaluation of the Bamako initiative.” Soc. Sci. Med., 36(11), 1383–1395.
McPake, B., Kumaranayake, L., Normand, C. (2002). Health economics: an international
Perspective, Routledge, London.
Messen, B., Zhenzhong, Z., Damme, W. V, Devadasan, N., Criel, B. and Bloom, G. (2003).
“Iatrogenic Poverty.” Trop. Med. Intl. Hlth.,8(7), 581-584.
Mladovsky, P. and Mossialos, E. (2008) “A conceptual framework for community-based health
insurance in low-income countries: social capital and economic development.” World Dev., 36,
590–607.
Morris, S., Devlin, N., Parkin, D. (2007).Economic analysis, John Wiley & Sons, United
Kingdom.
Msuya, J.M., Jütting, J. and Asfaw,A.(2004). Impacts of Community Health Insurance Schemes
on Health Care Provision in Rural Tanzania, Discussion Papers on Development Policy No. 82,
Centre for Development Research, Bonn.
Musau, S. (1999). Community-Based Health Insurance: Experiences and Lessons Learned from
East and Southern Africa, Partner-ships for Health Reform Project, Abt Associates Inc.,
Bethesda.
Mutangadura, G., Mukurazita, D, andJackson, H. (1999). A review of household and community
responses to the HIV/ AIDS epidemic in rural areas of sub-Saharan Africa, UNAIDS best
practice collection, Geneva.
Nanda, P. (1999). “Women’s participation in rural credit programmes in Bangladesh and their
demand for formal health care: is there a positive impact?” Hlth Econ., 8, 415–28.
National Commission on Macroeconomics and Health (2005).Financing and Delivery of Health
Care Services in India, http://mohfw.nic.in/Report%20on%20NCMH/Background%
20Papers%20report.pdf (May 29, 2010).
National Family Health Survey (2005-06),
www.measuredhs.com/pubs/pdf/FRIND3/00FrontMatter00.pdf(June 2010).

297
National Sample Survey Organisation,
http://www.mospi.gov.in/sarvekshana_publication_main.htm (June 21, 2010).
National Health Policy
2002,www.mohfw.nic.in/NRHM/.../National_Health_policy_2002.pdf(July 7, 2010).
Newbold, K. B., Eyles, J. andBirch, S. (1995). “Equity in health care: Methodological
contributions to the analysis of hospital utilisation within Canada.” Soc. Sci. Med., 40(9), 1181–
1192.
Nyman, J. A. (2003). The theory of demand for health insurance, Stanford University Press:
Stanford.
Ouimet, M., Fournier,P., Diop,I. and Haddad,S. (2007) “Solidarity or Financial Sustainability
An Analysis of the Values of Community-based Health Insurance Subscribers and Promoters in
Senegal.” Cand. J. Pub.Hlth., 98(4), 341-346.
Pag´an,J., Ross,S., Yau, J. and Polsky, D. (2006). “Self-medication and health insurance
coverage in Mexico.” Hlth. Poli., 75 (2006): 170–177.
Patel, R.K.,Trivedi, K.N., Nayak, S.N. and Patel, P. (2010). “Treatment seeking behaviour of
peri-urban community of Chandkheda.” Natl. J. Comm. Med.1 (1), 35-36.
Pauly, M. V. (1968). “The economics of moral hazard: comment.” Amer.Econ.Rev., 58, 531-
537.
Pauly, M. and Nicholson, S. (1999). “Adverse consequences of adverse selection.” J. Hlth. Pol.
Poli. Law, 24(5), 921–930.
Pauly, M. V., Zweifel, P., Scheffler, R. M., Preker, A. S.and Bassett, M. (2006). “Private health
insurance in developing countries.”Hlth Affrs,25(2):369–379.
Perez, D., Ang, A., Wega W.A (2006). “Effects of health insurance on perceived quality of care
among Latinos in the United States.” J Gen Intrnl Medi, 24 (Suppl 3), 555-60.
Peters, D., Yazbeck.A., Sharma, R., Ramana,G.,Pritchett,L. andWagstaff,A., (2002). India,
Raising the Sights: Better Health Systems for India’s Poor, Poor: Findings, Analysis, and
Options, Health, Nutrition, and Population Series, World Bank, Washington, D.C.
Planning Commission of Government of India,

298
http://planningcommission.nic.in/reports/genrep/index.php?repts=b_repgen.htm (June 14, 2010).
Polonsky,J., Balabanova,D., McPake,B., et al. (2008). “Equity in community health insurance
schemes: evidence and lessons from Armenia.” Hlth. Poli.Plan., 24(3), 209-216.
Pradhan,M.and Prescott,N. (2002). “Social riskmanagement options for medical care in
Indonesia.”J. Hlth. Econ., 11, 431–446.
Prasad, K.V.E. (1998). “Health insurance for tribals: ACCORD’s experience in Gudalur, Tamil
Nadu.” Social security for all Indians, W. van Ginneken, ed., Oxford University, New Delhi,
77–97.
Preker, A.S., Carrin, G., Dror, D.M., Jakab, M., Hsiao, W. and Arhin-Tenkorang, D.(2002).
“Effectiveness of community health financing in meeting the cost of illness.” Bull. of the World
Hlth. Org., 80(2), 143-150.
Preker, A.S., Carrin,G., Dror, D., Jakab, M., Hsiao,W. and Arhin-Tenkorang,D. (2004). “Rich-
Poor Differences in Health Care Financing.” Health financing for poor people: resource
mobilisation and risk sharing, A.S. Preker. and G. Carrin, eds., The World Bank, Washington,
D.C.
Radwan L (2005). India-Private health services for the poor: A policy note, Discussion Paper,
The World Bank, Washington D.C.
Ramachandran, R. and Rajalakshmi, T. K. (2009). “Unhealthy trend.” Frontline, 26(7),114-117.
Ranson, K.M. (2001). The impact of SEWA’s medical insurance fund on hospital utilisation and
expenditure: Results of a household survey, Health, Nutrition, and Population Series, The World
Bank, Washington, D.C.
Ranson, K.M. (2002). “Reduction of catastrophic health care expenditures by a community-
based health insurance scheme in Gujarat, India: current experiences and challenges.” Bull. of
the World Hlth. Org., 80(8), 613-621.
Ranson, K.M. (2003). “Community-based health insurance schemes in India: a review.” Ntn.l
Med. J.Ind., 16(2), 79–89.
Rao, K.,Waters,H., Steinhardt, L.,Alam, S.,Hansen, P. and Naeem A. (2009). “An experiment
with community health funds in Afghanistan.” Hlth. Poli. Plan, 1-11.

299
doi:10.1093/heapol/czp018.
Rothschild, M.and Stiglitz, J. (1976).“Equilibrium in competitive insurance markets.” Qtrly J.
Econ., 90, 629–649.
Russell, S. (1996). “Ability to pay for health care: concepts and evidence.” Hlth.
Poli.Plan.,11(3), 219–37.
Sauerborn, R., Adams, A., and Hien, M.(1996). “Household strategies to cope with the
economic costs of illness.” Soc. Sci.Med., 43(3):291–301
Savage, E. and Wright, D.J. (2003). “Moral hazard and adverse selection in Australian private
hospitals: 1989–1990.” J. Hlth. Econ., 22, 331–359.
Scheil-Adlung, X., Carrin, G., Jutting, J. and Xu, K. (2006).What is the impact of social health
protection on access to health care health expenditure and impoverishment? A comparative
analysis of three African countries, Extension of Social Security, ESS Paper No. 24.
International Labour Organisation, Geneva.
Schneider, P. and Diop, F. (2001). Synopsis of Results on the Impact of Community-Based
Health Insurance on Financial Accessibility to Health Care in Rwanda, Health, Nutrition, and
Population Series, The World Bank, Washington, D.C.
Schneider P. (2005) “Trust in micro-health insurance: an exploratory study in Rwanda.” Soc.
Sci. Med., 61, 1430–38.
Sebstad, J. and Cohen, M. (2002) Microfinance, Risk Management and Poverty, AIMS synthesis
study commissioned for the World Bank’s World Development Report 2000/01 (WDR), The
World Bank, Washington D.C.
Shetty, N. and Veerashekarappa (2009). “Institutional Innovations and Access to Micro-Health
Insurance for the Poor: Evidence from Karnataka.”The IUP J. of Risk & Insu., 6 (1), 50-68.
Sinha,T., Ranson,M., Chatterjee,M., Mills,A.(2005) “Management initiatives in a community
based health insurance scheme.” Intl J of Hlth Plan Mgmt ,22, 289–300. DOI: 10.1002/hpm.898
Sinha,T., Ranson,M. and Mills,A.(2007). “Protecting the Poor? The Distributional Impact of a
Bundled Insurance Scheme.” World Dev., 35 (8), 1404–1421.
Sodani ,P.R. (1999). “Determinants of Demand for Healthcare in Surveyed Tribal Households

300
of Selected Three Districts of Rajasthan: Demography India.”J. Hlth. Mgmt., 28 (2), 257-71.
Sommerfeld,J., Sanon,M.,Kouyate,B. and Sauerborn,R.(2002). “Informal risk-sharing
arrangements (IRSAs) in rural Burkina Faso: lessons for the development of community-based
insurance (CBI).” Intl. J. Hlth. Plan. Mgmt., 17, 147–163.
Soucat, A., Gandaho, T., Levy-Bruhl, D., de Bethune, X., Alihonou, E., Ortiz, C., et al. (1997).
“Health seeking behaviour and household expenditures in Benin and Guinea: the equity
implications of the Bamako Initiative.” Intl. J. Hlth. Plan. Mgmt., 12(Suppl 1), 137-63.
Subramanian, S.V., Jones, K. and Duncan, C. (2003). “Multilevel methods for public health
research.” Neighborhoods and health, I. Kawachi and L.F. Berkman, eds., Oxford University
Press, New York, 65–111.
Supakankunti, S. (2004). “Impact of the Thailand Health Card. In Health Financing for Poor
People.” Health financing for poor people: resource mobilisation and risk sharing, A.S. Preker,
G. Carrin, eds., The World Bank, Washington, D.C, Ch.10.
Tediosi,F., Aye,R., Ibodova,S., Thopson,R. and Wyss,K.(2008). “Access to medicines and out
of pocket payments for primary care: Evidence from family medicine users in rural Tajikistan.”
BMC Hlth. Ser. Res., 8, 109 doi: 10.1186/1472-6963-8-109.
http://www.biomedcentral.com/1472-6963/8/109 (Feb 3, 2010).
Tellis, W. (1997). “Introduction to Case Study.” (68 paragraphs), The Qult.Rep., [On-line serial],
3 (2), http://www.nova.edu/ssss/QR/QR3-2/tellis1.html (Mar 22, 2009).
Tipping, G., and Segall, M. (1995).Health care seeking behaviour in developing countries: an
annotated bibliography and literature review.IDS Development bibliography, Institute of
Development Studies, Sussex
Wagstaff, A. (2006). “The economic consequences of health shocks: Evidence from Vietnam.”
J. Hlth. Econ., 26 (1), 82–100.
Wagstaff, A.(2007). Health Insurance for the Poor: Initial Impacts of Vietnam’s Health Care
Fund for the Poor.Impact Evaluation Series # 11.Policy Research Working Paper #WPS 4134.
The World Bank ,Washington D.C http://www-
wds.worldbank.org/servlet/WDSContentServer/WDSP/IB/2007/02/05/000016406_2007020511

301
0453/Rendered/PDF/wps4134.pdf (Sept 15, 2010).
Wagstaff,A., Lidelow,M., Jun,G. and Juncheng,Q. (2008a). “Extending health insurance to rural
population: An impact evaluation of China’s new cooperative medical scheme.” J. Hlth. Econ.,
28 (2009),1–19.
Wagstaff, A. and Lindelow, M.(2008b). “Can insurance increase financial risk? The curious case
of health insurance in China.” J. Hlth. Econ., 27 (4), 990–1005.
Wang, H., Yip, W., Zhang, L., Wang, L. and Hsiao, W. (2005). “Community-based health
insurance in poor rural China: The distribution of net benefits.” Hlth. Poli.Plan., 20(6), 366–374.
Wang, H., Zhang, L., Yip, W. and Hsiao, W. (2006). “Adverse selection in voluntary Rural
Mutual Health Care health insurance scheme in China.” Soc. Sci. Med., 63, 1236-42.
Ward, H., Mertens, T.E. and Thomas, C. (1997). “Health-seeking behaviour and the control of
sexually transmitted disease.” Hlth. Poli.Plan., 12, 19–28.
Waters, H., Morlock, L. and Hatt, L.(2004). “Quality-based purchasing in health care.” Intl. J.
of Hlth Plan Mgmt.,2004; 19:365–381.
Weich, S. Lewis, G. and Jenkins, S.P. (2002). “Income inequality and self rated health in
Britain.” J. epidemiol. Comm. Hlth., 56, 436-441. doi:10.1136/jech.56.6.436.
Wiesmann,D. and Jutting,J.(2001). “Determinants of viable health insurance schemes in rural
sub Saharan Africa.” Qtly J.Intnl Agri., 50(4), 361-378.
Wilms, A.(2006). “The financial impact of formal health insurance schemes: Evidence from
Uganda.” Master Thesis, Vrije Universiteit Amsterdam, Amsterdam
http://microfinancegateway.org/content/article/ detail/36532 (May 6, 2010).
Wong,I., Chan,W., Choi,S., Lo,S.and Leung,G. (2006). ”Moral hazard or realised access to care?
Empirical observations in Hong Kong.” Hlth. Poli., 75 (2006), 251–261.
World Bank (2004).The Millennium development goals for health: Rising to the challenges,The
World Bank, Washington D.C.
World Health Organisation (2000).The World Health Report: Health Systems-Improving
Performance, WHO, Geneva.
World Health Organisation (2004).The World Health Report 2004: Changing History, WHO,

302
Geneva.
World Health Organisation (2009).World Health Statistics, WHO, Geneva.
Wouters A. (1999). Alternative Provider Payment Methods: Incentives for Improving Health
Care Delivery.Primer for Policymakers Series 1, Partnerships for Health Reform, Abt
Associates Inc., Bethesda.
Xu, K., Evans, D.and Kawabata, K. (2003). ”Household catastrophic health expenditure: a
multicountry analysis.” The Lancet 362, 111–7.
Yin, R.K., (1984).Case Study Research: Design and Methods. Sage Publications, Calif.
Yip, W. and Berman,P. (2001). “Targeted health insurance in a low income country and its
impact on access and equity in access: Egypt’s school health insurance.” J. Hlth. Econ., 10, 207–
220. DOI: 10.1002:hec.589.
Yip, W., Wang,H. and Hsiao,W. (2007) The Impact of Rural Mutual Health Care on Access to
Care: Evaluation of a Social Experiment in Rural China. 6th World Congress of Health
Economics. Copenhagen, Denmark. July 8-11, 2007.
Young,P., Kiyaga,E. and Mukwana,P. (2006). Micro insurance: Exploring ways to assess its
impact, Microfinance Opportunities, Washington D.C.
Zhang, L., Cheng,X., Liu,X., Tang,K., Bogg,L.,Dobberschutez,K. and Tolhurst,R. (2009).
“Balancing the funds in the New Cooperative Medical Scheme in rural China: determinants and
influencing factors in two provinces.” Intl. J. Hlth. Plan. Mgmt., 25(2):96-118.
Zhang, L., Cheng,X., Tolhurst,R., Tang,S. and Liu,X. (2010). “How effectively can the New
Cooperative Medical Scheme reduce catastrophic health expenditure for the poor and non- poor
in rural China?” Trop. Med. Intl. Hlth., 15(4), 468-475.
Zweifel, P. (2004). “Private health insurance in developing countries: Supply.” Wharton impact
conference on Voluntary health insurance in developing countries, University of Pennsylvania,
USA.

303
304
RESEARCH PUBLICATIONS

Research Papers Published in Journals

Savitha and Kiran, K.B. (2013a). “Health seeking behaviour in Karnataka: Does micro
health insurance matter?” Indian Journal of Community Medicine, 38 (4), 217-222.
Savitha and Kiran, K.B. (2013b). “Microhealth Insurance and the risk coping strategies
for the management of illness in Karnataka: A case study” International Journal of
Health Planning and Management, doi: 10.1002/hpm.2216.
Savitha and Kiran, K.B. (2013c).“Barriers to enrolment in micro health insurance - An
empirical study of Sampoorna Suraksha Programme in Karnataka.”RVIM Journal of
Management Research, December issue, accepted for publication.
Savitha and Kiran, K.B. (2012a) “Health financing mechanisms in India: An appraisal of
its performance”, Tatva, 9(2), 1-11.
Savitha and Kiran, K.B. (2012b). “Awareness and knowledge of micro health insurance:
A case study.” Journal of Health Management, 14(4), 481-494.
Savitha and Kiran, K.B. (2011). “The role of micro health insurance in realizing universal
coverage of health services: A reappraisal.” Asia Pacific Journal of Research in
Business Management, 2(4), 187-211.
Research Papers Published in Conference Proceedings
Savitha and Kiran, K.B. (2013). “Micro Health Insurance and Utilisation of Health
Services in Karnataka: A Case Study.”International Conference on Contemporary
Debates in Public Policy and Management (CD-ROM), Indian Institute of Management
Calcutta, Kolkata.
Savitha and Kiran, K.B. (2010). “Health for all: The role of health financing in mitigating
iatrogenic poverty in India”, International conference on Health care Market and
Emerging consumers- Innovation, Efficiency and Effectiveness (CD-ROM), Manipal
Institute of Management, Manipal University.

305
Research Papers Published as a Book Chapter
Savitha and Kiran, K.B. (2012). “Impact of micro health insurance on treatment seeking
behaviour: Evidence from Karnataka” Brig. Rajiv Divekar and Londhe B.R. eds.,
Inclusiveness and Innovation – Challenges for Sustainable Growth of Emerging
Economies, SIMS and Excel India Publishers, New Delhi, ch.49,377-383.
Savitha and Kiran, K.B. (2011). “Impact of micro health insurance: A review of
literature”, M R Shollapur and K N Shetty, eds., Microfinance and Sustainable
Livelihood Promotions in India, NABARD and SIT, New Delhi, Ch.33,369-391.
Research Awards
Awarded ‘Best paper’ in early proposal category titled “Impact of micro health insurance;
the case study of Sampporna Suraksha and Karuna schemes in Karnataka” at the 4th
Doctoral Colloquium, Indian Institute of Management Ahmedabad on 3rd and 4th of
January, 2011 at Ahmedabad

306
 BIO-DATA

Name : Ms. Savitha

Address for communication : B1 304, Ashoka Paradise, Hoigebail,

Ashoknagar, Mangalore-575006,

Karnataka, India

Educational Qualification : M.A (Economics), M.B.A. (Finance), H.G.Dipl.(UK),

M.Phil (Management)

Teaching Experience : 8 ½ years

2 years in Madhava Pai Memorial College, Manipal (July

1999 to December 2000)

1 year in Aloysius Institute of Business Administration,

Mangalore (May 2008 to June 2009)

5 ½ years in SDM Post Graduation Centre for

Management Studies and Research, Mangalore (from July
2007 to April 2008 and from July 2009 onwards)

Research Publications

a) Research Papers Published in Journals

1. Savitha and Devaraj, K. (2010).“Corporate social responsibility in micro, small
and medium enterprises- a study on the awareness and practice in
Mangalore.”Tatva,8(2), 56-63.
2. Savitha. (2011a). “Corporate social responsibility: charity or strategy?
Comparison of small, medium and large industries.”Asia Pacific Journal of
Research in Business Management, 2(2),110-118.
3. Savitha and Kiran,K.B. (2011b). “The role of micro health insurance in realizing
universal coverage of health services: A reappraisal.” Asia Pacific Journal of
Research in Business Management, 2(4),187-211.

307
 4. Savitha and Kiran, K.B. (2012a) “Health financing mechanisms in India: An
appraisal of its performance”, Tatva, 9(2), 1-11.
5. Savitha and Kiran, K.B. (2012b). “Awareness and knowledge of micro health
insurance: A case study.” Journal of Health Management, 14(4), 481-494.
6. Savitha and Kiran, K.B. (2013a). “Health seeking behaviour in Karnataka: Does
micro health insurance matter?” Indian Journal of Community Medicine, 38 (4),
217-222.
7. Savitha and Kiran, K.B. (2013b). “Microhealth Insurance and the risk coping
strategies for teh management of illness in Karnataka: A case study” International
Journal of Health Planning and Management, doi: 10.1002/hpm.2216.
8. Savitha and Kiran, K.B. (2013c).“Barriers to enrolment in micro health insurance
- An empirical study of Sampoorna Suraksha Programme in Karnataka.”RVIM
Journal of Management Research, December issue, accepted for publication.
b) Research Papers Published in National and International Conference
Proceedings
1. Savitha and Santhosh, S. (2009). “Marketing for microfinance
institutions.”National seminar on Strategies for Emerging Environment, Cochin
University, Cochin.
2. Savitha and Kiran,K.B. (2010). “Health for all: The role of health financing in
mitigating iatrogenic poverty in India.”International conference on Health Care
Market and Emerging consumers- Innovation, Efficiency and Effectiveness (CD-
ROM), Manipal Institute of Management, Manipal University.
3. Savitha and Kiran, K.B. (2012). “Impact of Micro Health Insurance on Treatment
Seeking Behaviour: Evidence from Karnataka” International Conference on
Inclusiveness and Innovation – Challenges for Sustainable Growth of Emerging
Economies, Symbiosis Institute of Management Studies, Pune.
4. Savitha and Kiran, K.B. (2013). “Micro Health Insurance and Utilisation of
Health Services in Karnataka: A Case Study.”International Conference on

308
 Contemporary Debates in Public Policy and Management (CD-ROM), Indian
Institute of Management Calcutta, Kolkata.
c) Research Papers Published as a Book Chapter
1. Savitha and Kiran, K.B. (2012). “Impact of Micro Health Insurance on Treatment
Seeking Behaviour: Evidence from Karnataka”, Brig. Rajiv Divekar and Londhe
B.R. eds., Inclusiveness and Innovation – Challenges for Sustainable Growth of
Emerging Economies, SIMS and Excel India Publishers, New Delhi, ch.49,377-
383.
2. Savitha and Kiran, K.B. (2011b). “Impact of micro health insurance: A review of
literature”, M R Shollapur and K N Shetty, eds., Microfinance and Sustainable
Livelihood Promotions in India, NABARD and SIT, New Delhi, Ch.33,369-391.
d) Research Papers Presented at the Conferences /Workshops Attended
1. Savitha (2008) AICTE sponsored one week Faculty Development Programme on
“Econometrics for Management Research” organised by Indian Institute of
Management Kozhikode during October 20-25, Kozhikode, Kerala.
2. Savitha (2009). “Marketing for microfinance institutions” in a National Seminar
on Strategies for Emerging Environment, held on 20-21st February 2009 at
Cochin University, Cochin, Kerala.
3. Savitha (2009). “The role of venture capital financing for the development of
entrepreneurship in India” in a National Conference on ‘Entrepreneurship; “ ray
of hope’ held on 4th April 2009 at SDM Postgraduate Centre for management
education and research, Mangalore, Karnataka.
4. Savitha and Kiran K.B. (2010). “Health for all: The role of health financing in
mitigating iatrogenic poverty in India” at the International Conference on Health
Care Market and Emerging Consumers- Innovation, Efficiency and Effectiveness,
held at Manipal Institute of Management, Manipal between January 21-23, 2010.
5. Savitha and Kiran K.B. (2010). “Impact of micro health insurance: A review of
literature” at the national Conference on Microfinance and Sustainable Livelihood
Promotion in India, organised by SIT, Tumkur on December 15 and 16th, 2010.
309
 6. Savitha and Kiran K.B. (2011).“Impact of Micro Health Insurance; the Case
Study of Sampoorna Suraksha and Karuna Schemes in Karnataka” at the 4th
Doctoral Colloquiumorganised by Indian Institute of Management Ahmedabad on
3rd and 4th of January, 2011 at Ahmedabad.
7. Savitha and Kiran, K.B. (2011). “Barriers to enrolment in micro health insurance -
An Empirical Study of Sampoorna Suraksha Programme in Karnataka” at the
International Conference on Innovative Strategies for Global Competitiveness,
RVIM Institute of Management, Bangalore, 8-10 December, 2011.
8. Savitha and Kiran, K.B. (2012). “Impact of Micro Health Insurance on Treatment
Seeking Behaviour: Evidence from Karnataka” at the International Conference
on Inclusiveness and Innovation – Challenges for Sustainable Growth of
Emerging Economies, Symbiosis Institute of Management Studies, Pune, 12-14
December, 2012.
9. Savitha and Kiran, K.B. (2013). “Micro Health Insurance and Utilisation of
Health Services in Karnataka: A Case Study” at the International Conference on
Contemporary Debates in Public Policy and Management, Indian Institute of
Management Calcutta, Kolkata, India, 7-9 February, 2013.
10. Savitha and Kiran, K.B. (2013).“How effectively can Micro health Insurance
provide financial protection?” at First Pan- IIM World Management Conference
on Emerging Issue in Management jointly organised by all IIMs and Ministry of
Human Resource Development, Government of India at Goa Marriott Resort and
Spa, Goa, India, 30th May-1st June, 2013.
e) Books

Savitha (2012). Corporate Social Responsibility in Small, Medium and Large Industries:
Evidence from India, Lambert Academic Publishing, Germany.

310