Thesis On Randomized Controlled Trial

EVALUATING BREASTFEEDING SUPPORT:
A RANDOMISED CONTROLLED TRIAL OF SUPPORT FROM

BREASTFEEDING COUNSELLORS.
by
Jonathan Peter Graffy
A thesis submitted to the Faculty of Medicine and Dentistry

of The University of Birmingham
for the degree of
Doctor of Medicine
Department of Primary Care and General Practice,

School of Medicine,
Faculty of Medicine and Dentistry
The University of Birmingham
November 2001
University of Birmingham Research Archive
e-theses repository
This unpublished thesis/dissertation is copyright of the author and/or third

parties. The intellectual property rights of the author or third parties in respect
of this work are as defined by The Copyright Designs and Patents Act 1988 or
as modified by any successor legislation.
Any use made of information contained in this thesis/dissertation must be in

accordance with that legislation and must be properly acknowledged. Further
distribution or reproduction in any format is prohibited without the permission
of the copyright holder.
EVALUATING BREASTFEEDING SUPPORT:
A RANDOMISED CONTROLLED TRIAL OF SUPPORT FROM
BREASTFEEDING COUNSELLORS.
ABSTRACT
Two-thirds of UK mothers begin breastfeeding, but many soon stop. Although breastfeeding
benefits health, infant feeding is influenced by social and attitudinal factors.
Study one prospectively investigated the attitudes and experiences of 514 women. Past
experience predicted which multiparae would stop by six weeks. Manual social class and
considering bottle feeding did so for primiparae. Perceived insufficient milk was the
commonest reason for stopping.
Study two, a randomised trial of support from breastfeeding counsellors, recruited 720
women. At four months, 46.1% (143/310) intervention and 42.3% (131/310) control women
breastfed (Chi2=0.942, P=0.33); 73.9% (229/310) vs 79.4% (246/310) gave bottle feeds
(Chi2=2.60, P=0.11). Survival analysis confirmed that differences between intervention and
control women's partial and full breastfeeding duration were not significant (P=0.45 and 0.15
respectively.) Significantly fewer intervention women felt they had insufficient milk.
Qualitative analysis of women’s comments revealed they wanted better information, practical
help with positioning, effective advice, encouragement and their feelings acknowledged.
Women valued counselling, but their feeding behaviour changed little, which may reflect the
strength of social influences and that not all mothers contacted the counsellors postnatally.
Practical support in the early postnatal period is important. Counselling may increase
women's confidence in breastfeeding and producing enough milk.
ACKNOWLEDGEMENTS
I would like to thank the many people who have contributed to this work and supported me
in undertaking it.
For the Nottingham study, I thank Carol Picksley, Ginny Hayes and the midwives, health
visitors and mothers who took part. Prof Chris McManus gave me invaluable statistical
advice.
In agreeing to participate in the randomised controlled trial, counsellors trusted me with

something very dear to them. I thank them for this and their commitment to the study. I am
also grateful to Rosie Dodds, Gill Gyte, Phyll Buchannan, Maggie Hamand and Elizabeth
Butler for their contributions to the study design.
More than anyone else, Jane Taylor made the RCT a success, with the tireless enthusiasm she
brought to her role as research assistant, liaising with counsellors, practices and mothers. She
also played an important role in the analysis of the support women wanted. Others who
assisted with the study administration included Sophie Dean, Jessie Ricketts, Petra Moxley
and Janet Turner. I am also grateful to Caroline Lee for handling the accounts for the project.
Tamara Hibbert and Moshin al Quarrum piloted questionnaires as a medical student project.
Tony Williams, my supervisor has been a tremendous support. I also thank Sandra Eldridge
for advice on statistical analyses and Stephen Evans for advice on study design. I am grateful
to my partners, colleagues and patients at Statham Grove for giving me the space and support
to undertake this.
I have listed the participating practices in appendix F and thank the receptionists, midwives
and doctors at each. I also thank the women who took part in the study.
I am grateful to the following for financial support: Trent Regional Health Authority (for
study one), the Royal College of General Practitioners, NHS North Thames Responsive
Funding programme, NHS R&D Support Funding through the East London and Essex
Network of Researchers, the Kings Fund and Central and East London Education
Consortium.
Finally, I would like to thank Clare and our children, Simon, Hilary and Ian for their support
and allowing me the time to undertake this.
TABLE OF CONTENTS
Preliminary Listings:
Foreward: Why research breastfeeding? ......................................................................................i
Introduction: Structure and outline of the thesis...........................................................................ii
Glossary: ....................................................................................................................................v
Part One: Literature Review
Chapter 1.1 How great are the benefits of breastfeeding?
1.1.1 Introduction: Women's perspectives on the benefits of breastfeeding..............................2

1.1.2 Scope of this review..........................................................................................................4
1.1.3 Sources of evidence to assess the benefits of breastfeeding .............................................5
1.1.3a Evidence from randomised controlled trials ..............................................................6
1.1.4 Methodological issues in epidemiological studies of infant feeding ................................8
1.1.5 Mechanisms by which breastfeeding may benefit infants' health...................................10
1.1.6 Evidence for associations between infant feeding and infection in infants ....................11
1.1.6a Gastrointestinal illness.............................................................................................12
1.1.6b Respiratory illness....................................................................................................14
1.1.6c Otitis media ..............................................................................................................15
1.1.7 Breastfeeding and neurological development.................................................................16
1.1.8 Other potential benefits for infants .................................................................................19
1.1.8a Allergic disorders.....................................................................................................19
1.1.8b Case control studies for uncommon conditions .......................................................20
1.1.8c Infant nutrition and childhood growth .....................................................................21
1.1.9 Breastfeeding and maternal health..................................................................................22
1.1.10 Disadvantages of breastfeeding and situations in which breastfeeding
is inadvisable ..................................................................................................................23
1.1.11 Conclusions and summary ..............................................................................................24
Chapter 1.2 Epidemiology of breastfeeding:

What factors influence infant feeding?
1.2.1 Introduction: The importance of epidemiological data...................................................26

1.2.2 Breastfeeding in the United Kingdom ............................................................................26
1.2.3 Trends in breastfeeding...................................................................................................28
1.2.4 Giving formula feeds to breastfed babies .......................................................................31
1.2.5 Understanding trends in breastfeeding ...........................................................................33
1.2.5a Maternity services and breastfeeding ......................................................................33
1.2.5b Media coverage of infant feeding.............................................................................36
1.2.6 Socio-demographic factors associated with breastfeeding .............................................37
1.2.7 International comparisons of breastfeeding rates ...........................................................40
1.2.7a World Health Organisation data on breastfeeding rates.........................................41
1.2.7b Breastfeeding in Europe and North America...........................................................42
1.2.8 Summary ...................................................................................................................45
Chapter 1.3 Breastfeeding support in society
1.3.1 Introduction ...................................................................................................................46

1.3.2 Sources of social support and influences on breastfeeding behaviour............................46
1.3.3 Men's perspectives on breastfeeding...............................................................................48
1.3.4 Negotiation and breastfeeding support ...........................................................................49
1.3.5 Support from professionals.............................................................................................51
1.3.6 Summary ...................................................................................................................52
Chapter 1.4 Theories of behaviour change
1.4.1 Introduction ...................................................................................................................53

1.4.2 Theoretical approaches which underpin models.............................................................53
1.4.2a Studies of social influence........................................................................................53
1.4.2b Behavioural models..................................................................................................54
1.4.2c Attitudinal models ....................................................................................................55
1.4.3 The Health Belief Model ................................................................................................56
1.4.4 The Theory of Reasoned Action.....................................................................................58
1.4.4a The applicability of the Theory of Reasoned Action to breastfeeding......................58
1.4.4b Studies applying the Theory of Planned Behaviour to breastfeeding ......................59
1.4.5 The Stages of Change Model..........................................................................................62
1.4.6 The Health Action Model ...............................................................................................63
1.4.7 Self-initiated change .......................................................................................................65
1.4.8 A critique of models in breastfeeding research...............................................................68
1.4.9 The limits to health promotion........................................................................................70
1.4.10 Implications for study design..........................................................................................70
1.4.11 Summary ...................................................................................................................72
Chapter 1.5 Interventions to promote breastfeeding
1.5.1 Introduction ...................................................................................................................73

1.5.2 Observation studies and the move to feeding on demand...............................................73
1.5.3 Implications of changes in feeding behaviour for research ............................................74
1.5.4 The case for intervention studies ....................................................................................75
1.5.5 Assessing intervention studies ........................................................................................76
1.5.6 Meta-analysis of studies of the effectiveness of additional support ...............................78
1.5.7 Implications for study design from published literature .................................................81
1.5.7a Sample size ...............................................................................................................82
1.5.7b Randomisation bias..................................................................................................82
1.5.7c Bias from differential follow-up ...............................................................................82
1.5.7d The implications of a high breastfeeding rate in the control group.........................83
1.5.7.e Independent assessment of outcome.........................................................................83
1.5.7f Generalising from interventions delivered by a single lactation consultant............84
1.5.7g Is the focus on initiation or maintainance of breastfeeding?...................................84
1.5.7h Describing the intervention......................................................................................85
1.5.8 Recent studies of breastfeeding support .........................................................................85
1.5.9 Summary ...................................................................................................................87
Chapter 1.6 Evidence based practice in supporting breastfeeding mothers
1.6,1 Introduction ...................................................................................................................88

1.6.2 Physiology of lactation ...................................................................................................88
1.6.3 Prevalence of common breastfeeding problems .............................................................89
1.6.4 Evidence based management of breastfeeding ...............................................................90
1.6.4a Antenatal preparation ..............................................................................................91
1.6.4b Establishing and maintaining breastfeeding............................................................92
1.6.4c Pain-free effective feeding........................................................................................92
1.6.4d Perceived insufficient milk .......................................................................................93
1.6.4e Other feeding problems............................................................................................95
1.6.5 Delivering evidence based breastfeeding support...........................................................95
1.6.6 Summary ...................................................................................................................96
Chapter 1.7 Counselling and breastfeeding counsellors
1.7.1 Introduction ...................................................................................................................97

1.7.2 A historical perspective on breastfeeding support ..........................................................97
1.7.3 History of NCT Breastfeeding Counsellors....................................................................99
1.7.4 The contribution of breastfeeding counsellors..............................................................101
1.7.5 How women contact a breastfeeding counsellor ..........................................................102
1.7.6 Training and conduct of breastfeeding counsellors ......................................................102
1.7.7 Who becomes a breastfeeding counsellor? ...................................................................104
1.7.8 An examination of the counsellor's role and its implication for research.....................106
1.7.9 Summary .................................................................................................................108
Part two
Study one: a prospective study of mothers' attitudes to and experiences of breastfeeding
Chapter 2.1 Rationale and objectives of study one
2.1.1 Introduction .................................................................................................................110

2.1.2 Specific objectives ........................................................................................................111
Chapter 2.2 Study one: method
2.2.1 Recruitment .................................................................................................................112

2.2.2 Questionnaire design.....................................................................................................112
2.2.3 Statistical analysis.........................................................................................................113
Chapter 2.3 Study one: results
2.3.1 Study population...........................................................................................................114

2.3.2 Feeding plans ................................................................................................................115
2.3.3 Incidence and duration of breast feeding......................................................................116
2.3.4 Factors which might predict early cessation of breastfeeding ......................................117
2.3.5 Reasons for discontinuing breastfeeding ......................................................................120
2.3.6 Common feeding problems...........................................................................................120
2.3.6a Perceived insufficient milk supply..........................................................................121
2.3.6b Sore nipples............................................................................................................122
2.3.7 Summary .................................................................................................................123
Chapter 2.4 Study one: discussion
2.4.1 Strengths and limitations of the study...........................................................................124

2.4.2 Implications for policy and practice .............................................................................126
2.4.2a Identifying women who need most help..................................................................126
2.4.2b Perceived insufficient milk .....................................................................................127
2.4.2c Management of sore nipples ..................................................................................129
2.4.3 Feeding technique – the missing link ...........................................................................130
2.4.4 Implications for study two ............................................................................................130
Part three:
Study two: a randomised controlled trial of support from breastfeeding counsellors
Chapter 3.1 Rationale and objectives of study two
3.1.1 Rationale and selection of intervention for study two ..................................................133

3.1.2 Aims .................................................................................................................134
3.1.3 Primary objective..........................................................................................................135
3.1.4 Additional objectives ....................................................................................................135
Chapter 3.2 Study two: Methods
3.2.1 Outline of study two .....................................................................................................137

3.2.2 Recruitment of women to the study..............................................................................139
3.2.3 Inclusion criteria ...........................................................................................................140
3.2.3a Premature labour ...................................................................................................141
3.2.4 Sample size .................................................................................................................142
3.2.5 Recruitment of practices and counsellors .....................................................................143
3.2.6 The intervention............................................................................................................145
3.2.7 Outcome Measures .......................................................................................................147
3.2.7a Duration of breastfeeding ......................................................................................147
3.2.7b Maternal satisfaction with breastfeeding...............................................................148
3.2.7c Feeding problems...................................................................................................148
3.2.8 Selection of statistical tests ...........................................................................................148
3.2.9 Questionnaire design.....................................................................................................149
3.2.9a Coding social class ................................................................................................150
3.2.10 Study administration, randomisation and blinding .......................................................151
Chapter 3.3 Results of study two: recruitment and follow up
3.3.1 Introduction and overview ............................................................................................156

3.3.2. Study population...........................................................................................................157
3.3.3 Assessment of women to determine their eligibility.....................................................157
3.3.3a Reasons why questionnaires were not returned .....................................................158
3.3.4 Recruitment by practice................................................................................................159
3.3.5 Reasons for exclusion ...................................................................................................161
3.3.6 Characteristics of participants.......................................................................................162
3.3.7 Effectiveness of Randomisation ...................................................................................167
3.3.8 Withdrawals .................................................................................................................171
3.3.9 Postnatal follow up .......................................................................................................171
3.3.10 Analysis and exclusion of premature births..................................................................174
3.3.11 Summary .................................................................................................................176
Chapter 3.4 Results of study two: uptake of counselling and women's perspectives on
support they received
3.4.1 Introduction .................................................................................................................177

3.4.2 Uptake of counselling in the intervention group...........................................................178
3.4.2a Antenatal contact .............................................................................................178
3.4.2b Postnatal contact .............................................................................................179
3.4.2c Combining antenatal and postnatal contact data............................................180
3.4.3 Uptake of counselling and feeding behaviour ..............................................................181
3.4.4 Comparison of the uptake of breastfeeding counselling in the intervention,
control and observation groups.....................................................................................183
3.4.5 Women's explanations of why counsellor was helpful or unhelpful ............................185
3.4.6 Mothers' opinions on the quality of advice they received from different sources ........188
3.4.7 Summary of key findings..............................................................................................192
Chapter 3.5 Results of study two: effect of the intervention
3.5.1 Introduction .................................................................................................................194

3.5.2 Incidence and duration of breastfeeding for women in RCT........................................194
3.5.3 Outcome of RCT: Incidence of breastfeeding ..............................................................195
3.5.4 Outcome of RCT: Prevalence of breastfeeding ............................................................196
3.5.5 Outcome of RCT: Prevalence of bottle feeding............................................................196
3.5.6 Outcome of RCT: Duration of breastfeeding................................................................198
3.5.7 Outcome of RCT: Time to introduction of bottle feeding ............................................199
3.5.8 Sensitivity analyses to assess impact of the imbalance in number of undecided
women .................................................................................................................201
3.5.9 Outcome of RCT: Maternal Satisfaction with breastfeeding........................................204
3.5.10 Common feeding problems...........................................................................................207
3.5.11 Incidence and prevalence of breastfeeding in observation group .................................208
3.5.12 Summary .................................................................................................................209
Chapter 3.6 Results of study two: counsellors' perspectives on the study
3.6.1 Introduction .................................................................................................................211

3.6.2 The content of additional support .................................................................................211
3.6.3 The role and experiences of the counsellors .................................................................212
3.6.4 Summary .................................................................................................................215
Chapter 3.7 Discussion of study two
3.7.1 Evaluating a complex intervention ...............................................................................216

3.7.2 Strengths and limitations of the study...........................................................................217
3.7.2a Design and conduct of the trial........................................................................218
3.7.2b Assessment of outcome ....................................................................................222
3.7.2c The intervention ...............................................................................................223
3.7.2d Summary of factors which may have influenced the result..............................227
3.7.3 The results in context....................................................................................................228
3.7.4 Randomised controlled trials to assess behavioural interventions................................229
Part four:
Further analysis of study two: Qualitative analysis of what information and support women
want with breastfeeding
Chapter 4.1 Rationale and objectives for qualitative analysis of women’s comments
4.1.1 Introduction .................................................................................................................232

4.1.2 Objectives .................................................................................................................233
Chapter 4.2 Qualitative analysis of women’s comments: participants and methods
4.2.1 Recruitment of participants...........................................................................................234

4.2.2 Content of postnatal questionnaire ...............................................................................236
4.2.3 Administration of postnatal questionnaire....................................................................237
4.2.4 Analysis .................................................................................................................237
Chapter 4.3 Results of qualitative analysis
4.3.1 Women's experiences of advice and support with breastfeeding..................................241

4.3.2 Concepts that encompass the information, advice and support women want
with breastfeeding.........................................................................................................243
4.3.2a Information about breastfeeding and what to expect ......................................243
4.3.2b Practical help with positioning the baby to breastfeed ...................................244
4.3.2c Effective advice and suggestions .....................................................................245
4.3.2d Acknowledgement of mothers’ experiences and feelings.................................246
4.3.2e Reassurance and Encouragement....................................................................247
4.3.3 A model of the information, advice and support that women want with breastfeeding248
4.3.4 Comments on support from breastfeeding counsellors.................................................248
4.3.5 Checking the findings with participants .......................................................................249
Chapter 4.4 Discussion of qualitative findings
4.4.1 Introduction .................................................................................................................251

4.4.2 Strengths and limitations .............................................................................................251
4.4.3 The findings in context .................................................................................................253
4.4.3a Ante-natal information.....................................................................................253
4.4.3b Help with positioning.......................................................................................254
4.4.3c Effective advice................................................................................................255
4.4.3d Encouragement and empathy ..........................................................................256
4.4.4 Summary and conclusions ............................................................................................257
Part five: Conclusions and recommendations
Chapter 5 Conclusions and recommendations

5.1 Policy implications of main findings ............................................................................261
5.2 How can we improve breastfeeding support?...............................................................263
5.2.1 A new strategy for postnatal care ....................................................................263
5.2.2 The role of the volunteer..................................................................................264
5.2.3 Training and professional practice..................................................................266
5.2.4 Marketing breastfeeding..................................................................................267
5.3 Implications for future research....................................................................................269
5.3.1 Lessons from previous research ......................................................................269
5.3.2 Maternal Assessment of Breastfeeding Support ..............................................269
5.3.3 Educational research for professional and lay breastfeeding supporters.......270
5.3.4 Perceptions of volunteers ................................................................................271
5.3.5 Ethnicity .................................................................................................................271
5.3.6 Research to inform a review of postnatal care.......................................................271
5.3.7 Breastfeeding and the media ..................................................................................272
5.3.8 The benefits of breastfeeding..................................................................................272
Reference list .........................................................................................................................273
Appendices
Appendix A Study one questionnaires

Appendix B Pilot study to assess validity and reliability of six-week postnatal questionnaire
Appendix C Study two questionnaires, consent form and information for participants
Appendix D Note on generating outcome variables for duration of breastfeeding
Appendix E Additional data tables
Appendix F Participating practices
Appendix G WHO/UNICEF Ten Steps to Successful Breastfeeding
Appendix H Summary tables for intervention studies
Appendix I NCT Breastfeeding Counsellors' Code of Conduct
Appendix J Publications and Presentations of this work
Appendix K Examples of womens’ comments
Appendix L Literature Search Strategy
LIST OF TABLES 1
Part One:
Literature Review
Table 1.1.1 Mother's reasons for choice of infant feeding method.........................................3

Table 1.1.6 Components of human milk which are thought to protect against infection .....11
Table 1.2.2 Whether mothers would have liked to have breastfed longer by duration of
breastfeeding......................................................................................................28
Table 1.2.3 Incidence and prevalence of breastfeeding in England and Wales from 1975
to 1995 ...............................................................................................................30
Table 1.2.4 Proportion of breastfed babies given formula feeds ..........................................32
Table 1.2.5a Changes in reported hospital practice between 1975 and 1995.........................34
Table 1.2.6 Characteristics associated with mothers who initially breastfed .......................39
Table 1.2.7a Exclusive breastfeeding and median duration of breastfeeding:
a global overview, 1996.....................................................................................41
Table 1.2.7b Surveys of breastfeeding in Europe and North America ...................................43
Table 1.3.5 Comments received in a study of maternity care; Analysis by stage of care.....51
Table 1.5.7 Methodological issues raised by previous research on breastfeeding support...81
Table 1.6.3 Feeding problems experienced by breastfeeding mothers .................................90
Part two
Table 2.3.1 Age, parity and marital status of study population, compared with 1983 birth
notifications for Nottingham............................................................................115
Table 2.3.2 Reasons given by women for their choice of feeding method.........................116
Table 2.3.4b Effectiveness of predictors of stopping breastfeeding by six weeks ...............119
Table 2.3.5 Reasons given by women for stopping breastfeeding......................................120
Table 2.3.6a Health professionals consulted by women with perceived insufficient milk...121
Table 2.3.6b Health professionals consulted by women with sore nipples ..........................122
Part three:
Table 3.3.2 Reasons women were no longer receiving maternity care...............................157

Table 3.3.5 Recruitment and exclusions .............................................................................161
Table 3.3.6a Parity of women assessed for eligibility ..........................................................163
Table 3.3.6b Age of women assessed for eligibility.............................................................164
Table 3.3.6c Mean age of primiparae assessed for eligibility ..............................................165
Table 3.3.6d Terminal educational age of women assessed for eligibility ...........................165
Table 3.3.6e Social class of women recruited ......................................................................166
Table 3.3.6f Ethnicity of women recruited ..........................................................................167
Table 3.3.7a Parity of intervention and control groups .......................................................168
Table 3.3.7b Age group of intervention and control groups ................................................168
Table 3.3.7c Ethnicity of intervention and control groups ..................................................168
Table 3.3.7d Social class of intervention and control groups ..............................................169
Table 3.3.7e Terminal educational age of intervention and control groups .........................169
1
Throughout the thesis, tables are numbered according to the section of the thesis that they relate to. Thus table
3.3.7a is found in section 3.3.7.
Table 3.3.7f Intended breastfeeding duration of intervention and control groups ...............169
Table 3.3.7g Intention to return to work of intervention and control groups .......................170
Table 3.3.7h Feeding plan of intervention and control groups .............................................170
Table 3.3.9a Postnatal follow-up for control and intervention groups .................................172
Table 3.3.9b Postnatal follow-up by feeding behaviour at six weeks...................................172
Table 3.3.9c Postnatal follow-up by feeding behaviour at four months...............................173
Table 3.3.10a Contacts with intervention group mothers by gestation at delivery ................174
Table 3.3.10b Contacts with intervention group mothers who reported on six-week
questionnaire that baby was premature............................................................175
Table 3.3.10c Postnatal follow-up for control and intervention groups (excluding prem
births)...............................................................................................................176
Table 3.4.2a Venue for antenatal contacts............................................................................178
Table 3.4.2c(i) Type of antenatal and postnatal contact for intervention group women..........180
Table 3.4.2c(ii) Antenatal and postnatal contacts for intervention group women who were
followed up to six weeks .................................................................................181
Table 3.4.3a Type of postnatal contact and parity................................................................182
Table 3.4.3b Type of postnatal contact and terminal educational age..................................182
Table 3.4.3c Postnatal contact and breastfeeding at six weeks ............................................183
Table 3.4.4a How women in control and observation groups heard of counsellor ..............185
Table 3.4.4b Women's reports of difficulty contacting a counsellor ....................................185
Table 3.4.5a Responses to "Did you find the counsellor helpful?" ......................................186
Table 3.4.5b Multiple response analysis of free text comments intervention group
women made when asked to explain why counsellor was helpful or not ........187
Table 3.4.5c Multiple response analysis of free text comments control and observation
women made when asked to explain why counsellor was helpful or not ........188
Table 3.4.6a Source of advice perceived as least and most helpful by women in
intervention group............................................................................................190
Table 3.4.6b Source of advice perceived as least and most helpful by women in control
group................................................................................................................190
Table 3.4.6c Source of advice perceived as least and most helpful by women in observation
group................................................................................................................191
Table 3.5.8a Sensitivity analysis of prevalence when undecided women excluded.............201
Table 3.5.8b Binary logistic regression to assess the association between treatment
allocation and prevalence of breastfeeding and bottle feeding, when association
with having made decision to breastfeed has been taken into account............203
Table 3.5.9a Satisfaction with breastfeeding at six weeks ...................................................205
Table 3.5.9b Mann-Whitney Test for mean rank of satisfaction scores ...............................206
Table 3.5.10a Common feeding problems at six weeks .........................................................207
Table 3.5.10b Mann-Whitney Test for mean rank of common problem scores .....................208
Table 3.5.11 Comparison of feeding rates between Observation Group and Women in main
trial .................................................................................................................209
Table 3.7.2d Factors which may have inluenced the results.................................................227
Table 3.7.3 Comparison of proportions stopping breastfeeding with other studies............228
Part four:
Table 4.2.1 Characteristics of participants .........................................................................235

Tables in appendix E
Table 3.3.4 Recruitment by practice
Table 3.3.6 Language spoken by non-English speakers
Table 3.4.8a Raw data for table 3.4.8a: Sensitivity analysis of prevalence when undecided
women excluded
LIST OF FIGURES2
Frontispiece Breastfeeding counsellor and client....................................................................vi
Part One:
Literature Review
Fig 1.1.3a Promotion of Breastfeeding Intervention Trial - Main results.............................7

Fig 1.1.4a Methodological standards proposed by Baucher et al .........................................8
Fig 1.1.4b Schema for breastfeeding definition ....................................................................9
Fig 1.1.10 Situations in which breastfeeding may be contra-indicated ..............................23
Fig 1.3.1 Societal influences on breastfeeding behaviour.................................................46
Fig 1.4.3 The Health Belief Model ...................................................................................57
Fig 1.4.4 The Theory of Reasoned Action........................................................................58
Fig 1.4.4b Theory of Planned Behaviour - model for those working more than 20 hours
a week ................................................................................................................61
Fig 1.4.5 The Stages of Change Model.............................................................................62
Fig 1.4.6 The Health Action Model ..................................................................................64
Fig 1.4.7a Pre-requisites for self-initiated change ..............................................................66
Fig 1.7.6 NCT Breastfeeding Counsellors' Code of Conduct .........................................103
Fig 1.7.7 Abilities breastfeeding counsellors should have..............................................105
Part two
Fig 2.3.3 Duration of breastfeeding (Study one).............................................................117

Fig 2.3.4a Antenatal responses which were significantly associated with stopping
breastfeeding before six weeks. .......................................................................118
Part three:
Fig 3.2.1a Flow chart describing progress of participants through trial ...........................138
Fig 3.2.1b Summary of methods.......................................................................................139
Fig 3.3.10 Study administration flowchart........................................................................152
Fig 3.3.1 Overview of recruitment and follow up...........................................................156
Fig 3.3.3 Assessment of eligibility..................................................................................158
Fig 3.3.4 Recruitment by practice...................................................................................160
Fig 3.5.2 Duration of breastfeeding in RCT ...................................................................195
Fig 3.5.6 Percentage of women giving any breastfeeds .................................................199
Fig 3.5.7 Percentage of women fully breastfeeding........................................................200
Fig 3.6.2a Counsellors' perspectives on antenatal support ...............................................211
Fig 3.6.2b Counsellors' perspectives on postnatal support ...............................................212
2
Throughout the thesis, figures are numbered according to the section that they relate to. Thus figure 3.6.2a is
found in section 3.6.2.
Part four:
Fig 4.2.2 Questions from postnatal questionnaire which asked mothers to assess
the advice and support they received...............................................................236
Fig 4.2.4 Questions on the validation response sheet .....................................................240
Fig 4.3.1a Themes identified from the text: - What women found most helpful.............242
Fig 4.3.1b Themes identified from the text: - What women found least helpful .............243
Fig 4.3.3 Information, advice and support women want with breastfeeding ..................248
Part five:
Conclusions and recommendations
Fig 5.2 Ways to improve breastfeeding support ..........................................................263

FOREWORD: WHY RESEARCH BREASTFEEDING?
The decision to research a particular topic is both a personal choice and one made on the
interest and importance of the subject. For me, the personal trigger was a consultation that
went wrong. Early in my postgraduate training for general practice, I watched an
experienced doctor try to persuade a woman who had "successfully" breastfed her previous
child that she should do so again. Although he meant well, she had not enjoyed
breastfeeding; she had felt exhausted and found weaning difficult. In the end, she left in
tears. I learnt that women base their decisions on a range of personal and emotional factors;
medical advice is only one of the influences they consider. Unravelling the complex
psychosocial and practical factors in the decisions women make was to prove a rich vein to
research.
More objectively, breastfeeding merits research because, despite widespread acceptance that
breast milk is the most appropriate first food for the human infant, many women have
difficulties in both initiating and sustaining breastfeeding. Throughout the work reported in
this thesis, my main aim has been to find ways to improve support for women who want to
breastfeed.
A final reason to research breastfeeding has been the almost universally encouraging
response I have had from those I have asked to help with these studies. The enthusiasm of
the counsellors, health professionals and mothers involved has helped me sustain my own
interest.
i
INTRODUCTION: STRUCTURE AND OUTLINE OF THE THESIS
This thesis is concerned with breastfeeding support and includes three studies, which
approach the topic from different perspectives. My aim in this introduction is to identify the
main strands of the thesis and briefly discuss how they connect together.
Part One: Literature review
The main justification for health services being concerned with infant feeding is the evidence
that breastfeeding is beneficial for human health. Although many health promotion studies
take this for granted, I have felt it is important to understand and draw on this evidence in
initiatives to promote breastfeeding. The first chapter reviews the literature on breastfeeding
and human health, focusing on infection rates and neurological development. Chapters two
and three consider the epidemiology of breastfeeding and studies that have asked women
about their experiences of support. Together, they describe the context in which women
make their decisions on infant feeding.
Chapter four takes a step back from the specifics of supporting breastfeeding and considers
the theoretical basis of health promotion and behaviour change, based largely on models
developed from research in social psychology. Models offer a way to consider the impact of
different factors on health behaviour, and this approach is used to consider how interventions
to promote breastfeeding may work.
Chapter five reviews a group of randomised controlled trials, which were included in a recent
meta-analysis and considers the methodological issues involved in evaluating interventions to
ii
promote breastfeeding. The lessons learnt from this had important implications for the design
of the randomised controlled trial of additional support from breastfeeding counsellors which
is reported in part three.
Chapter six considers the practical aspects of helping mothers with breastfeeding. It draws
on both the physisiogy of lactation and the evidence base for the management of particular
breastfeeding problems such as sore nipples and perceived insufficient milk.
Chapter seven focuses on the intervention being tested in the main trial. It begins with an
account of the origins of breastfeeding counselling and the National Childbirth Trust and then
considers the role of the counsellor and some of the issues involved in evaluating counselling.
Part Two: Observation study
In the second part of the thesis I report the study which first engaged me in breastfeeding
research. I conducted this between 1983 and 1984 whilst working as a GP Registrar in
Nottingham, and eventually published the findings in 1992. The study involved recruiting
women antenatally and then observing their experiences of breastfeeding and the support they
received until their babies were six months old. I was able to describe the support women
received for common problems and identify factors that predicted early cessation of
breastfeeding. Despite the time since this was published, I believe that this work has retained
its relevance, because it was conducted within the context of normal primary care.
Inevitably the literature on breastfeeding has since moved on and many of the papers cited in
the literature review were published after this was undertaken.
iii
Part Three: Intervention study
The main focus of this thesis is the randomised controlled trial of additional support from
breastfeeding counsellors. This was designed in the light of experience in the Nottingham
study and set out to find out whether women who were offered additional support breastfed
for longer, had fewer problems and were more satisfied with their experience. Women were
recruited through general practice antenatal clinics and those allocated to receive additional
support were visited antenatally by a breastfeeding counsellor and offered postnatal support,
by telephone or at further home visits if they requested this.
Part Four: Mothers' perspectives
Part four reports a qualitative analysis of women's perspectives on breastfeeding support,
which is based on responses to open questions they were asked as part of the randomised
controlled trial. Participants were asked what advice they found most and least helpful and
were also invited to give fuller accounts of their experiences on the back page of the six-week
postnatal questionnaire. These were analysed thematically to identify five key concepts
which described the information, advice and support that women wanted with breastfeeding.
These have important implications for the way that breastfeeding support is delivered.
Conclusions and summary
The final section draws together the findings and considers their implications. It suggests
ways that breastfeeding support may be improved and makes suggestions for further research.
iv
GLOSSARY
This glossary lists the main definitions of feeding practice used in this thesis. The literature
on defining infant feeding practices is discussed in section 1.1.4.
Feeding practices:
• Exclusive breastfeeding: The infant has received only breast milk (including expressed
breast milk), but no other liquids or solids with the exception of drops or syrups
consisting of vitamins, mineral supplements or medicines.
• Predominant breastfeeding: The infant’s predominant source of nourishment has been
breast milk. However the infant may have received water, or water-based drinks,
including fruit juices, but not infant formula, solids or other foods.
• Full breastfeeding: This encompasses both exclusive and predominant breastfeeding,
but excludes infant formula, solids or other foods.
• Partial breastfeeding: This refers to infants receiving both breast milk and infant
formula, or solids or other foods.
Measures of feeding rates:
• Incidence of breastfeeding: The proportion of infants who are initially breastfed. This
includes all babies put to the breast at all, even if only on one occasion.
• Prevalence of breastfeeding: The proportion of infants still breastfeeding at specific
ages, even if also receiving infant formula or solid food.
• Duration of breastfeeding: The length of time that infants who were initially breastfed
continue to receive breast milk, even if also receiving other foods.
• Prevalence of bottle feeding: The proportion of babies given any formula feeds,
regardless of whether they were also breastfed.
v
Breastfeeding counsellor and client
(Photograph included with consent)
vi
PART ONE
LITERATURE REVIEW
1
Chapter 1.1
HOW GREAT ARE THE BENEFITS OF BREASTFEEDING?
1.1.1 Introduction: Women's perspectives on the benefits of breastfeeding
The majority of mothers in Britain opt to breastfeed and in many ways, their
perspectives on the advantages and disadvantages of breastfeeding are the most
important. It is individual mothers who make the day-to-day decisions on infant
feeding and although they may be influenced by social factors and scientific
knowledge, they make their own decisions. Although this chapter is primarily
concerned with the evidence that breastfeeding is beneficial for human health, it is
appropriate to see that evidence in the context of the factors mothers consider when
they decide how to feed their babies.
In 1995, 66% of mothers in the United Kingdom began breastfeeding, with higher
proportions in England and Wales, (68%) than in Scotland (55%) or Northern Ireland
(45%). As part of the five-yearly national survey1, they were asked their reasons for
choosing the feeding method adopted and these are reported in table 1.1.1.
The most common reason for breastfeeding, given by 83% of mothers, was that it was
best for their baby. Other reasons were that breastfeeding was more convenient,
cheaper or created a closer bond between mother and baby. The commonest reasons
women gave for opting to bottle feed were that other people could help with feeding,
or that they did not like the idea of breastfeeding. Mothers who had previous children
1.1 Literature Review: How great are the benefits of breastfeeding? 2

often based their decisions on whether breastfeeding or bottle feeding had suited them
before.
Table 1.1.1
Mother's reasons for choice of infant feeding method
United Kingdom, 19951
Mothers planning to breastfeed Mothers planning to bottle feed

n = 3,301 n = 1,557
2 2
Mother's reasons % Mother's reasons %
Breastfeeding is best for baby 83 Other people can feed baby 36

Breastfeeding is more convenient 37 Did not like the idea of breastfeeding 27
Breastfeeding is cheaper 21 Mother's own previous experience 30
Closer bond with baby 20 Would be embarrassed to breastfeed 7
Mother's own experience 17 Can see how much baby has taken 6
Breastfeeding is natural 12 Expecting to return to work soon 6
Breastfeeding is good for mother 12 Bottle feeding is less tiring 4
Influenced by medical personnel 3 Medical reasons for not breastfeeding 4
Influenced by friends or relatives 2 Persuaded by other people 3
Other reasons 2 No particular reason 1
Other reasons 6
1 1
Reproduced from: Foster, Lader and Cheesbrough, 1997 (p45)
2
Percentages do not add to 100 because some mothers gave more than one reason.
While the overwhelming majority of mothers accept that "Breast is Best", it is unclear
how many are aware of the evidence on which the statement is based. But if they are
not, they may be unable to sustain their motivation if they encounter difficulties.
Professionals also make judgements about how much emphasis to place on particular
health promotion activities and the priority they give to supporting breastfeeding will
be influenced by their understanding of the health benefits of breastfeeding.

1.1.2 Scope of this review
Recent reviews of the evidence for associations between infant feeding and human
health have been conducted by the Standing Committee on Nutrition of the British
Paediatric Association in 19942, the American Academy of Paediatrics in 19973 and
Heinig and Dewey from the Department of Nutrition at University of California in
19964. Golding and colleagues from Bristol have also reviewed the evidence for
associations with a range of conditions.5 6 7 8 9 10 11 12
Research on potential benefits for infants has included studies on infections, allergic
disorders, neurological development, growth patterns, diabetes mellitus and
inflammatory bowel disease. Aspects of maternal health that have been studied
include lactational amenorrhoea, osteoporosis and pre-menopausal breast cancer. It
would be beyond the scope of this thesis to appraise the evidence on all these
conditions. Instead, I have chosen to consider the methodological issues raised, and to
focus on the two areas of infection and neurological development. Other potential
benefits are reported.
Although there is strong evidence for the benefits of breastfeeding in developing
countries, confirmed in a recent WHO review of the impact of breastfeeding on infant
mortality,13 I have elected to focus on the situation in developed countries such as the
United Kingdom. Similarly, I have not set out to assess in-vitro evidence for the
importance of components of breast milk.

1.1.3 Sources of evidence to assess the benefits of breastfeeding
Evidence for the benefits of breastfeeding is derived from several types of study and
observation.
Mammalian evolution provides evidence of the benefits of breastfeeding, as over the
years, each species' milk has evolved to meet the needs of their young. While this
may seem too obvious to merit attention, it underpins any consideration of the
benefits of breastfeeding. Indeed it is accepted implicitly by the manufacturers of
infant formulae who attempt to adapt their products each time the value of some
component of breast milk is demonstrated. Breast milk is the benchmark by which
infant formula is judged.
In-vitro studies have investigated the composition of human milk and the functions of
its components. These studies both point to potential benefits, but also help explain
how those benefits may be mediated, an essential step in demonstrating connections
between feeding behaviour and health.
Observational cohort and case-control studies have provided evidence of associations
between feeding behaviour and health and are useful for problems which are
relatively uncommon. They are however subject to methodological problems which
can be only partly resolved by improvements in study design, as discussed in section
1.1.4.

1.1.3a Evidence from randomised controlled trials
Although randomised controlled trials are the ideal way to assess associations
between interventions and their consequences, practical and ethical factors dictate that
they cannot usually be used to assess the benefits of breastfeeding. Because of what is
already known about the benefits of breastfeeding, it would not be ethical, or
acceptable to mothers, to randomise those intending to breastfeed to a group who
would give formula feeds. An RCT would therefore need to recruit from those
intending to use infant formula, but then persuade mothers in the group allocated to
breastfeeding to do so.
Despite the constraints, some have successfully used randomised designs. Lucas et al
14 15 16 17 18
recruited 926 babies who were born prematurely. If their mothers chose
not to breastfeed, or did not provide enough milk themselves, the babies were
randomised to receive either additional donor milk, or additional formula. While
these studies have provided invaluable data on associations between breastfeeding and
necrotising enterocolitis, allergic disorders and neurological development, Lucas has
emphasised that these findings may not necessarily apply to healthy babies born at
term.19
Several studies comparing different infant formulae have employed randomisation.
For example, Burr et al20 compared cow's milk formula with soya-based formula for
infants with a family history of allergic disease, and Lucas et al21 supplemented infant
formula with long-chain polyunsaturated fatty acids to investigate their effect on

neurological development. Whilst this approach does not provide evidence relating
directly to the benefits of breastfeeding, it does allow researchers to investigate the
impact of factors identified from in-vitro studies.
In January 2001, Kramer et al reported the results of a large cluster-randomised trial.22
The Promotion of Breastfeeding Intervention Trial (PROBIT) was conducted in the
Republic of Belarus and recruited 17,046 mother-infant pairs under the care of 31
hospitals and polyclinics. The 16 intervention sites received an intervention based on
the WHO/UNICEF Baby Friendly Hospital Initiative23, while control sites continued
normal care. 96.7% of babies completed follow-up to 12 months. The results of this
study are important, as it was large enough to demonstrate an effect not only on
duration of breastfeeding, but infant health. The main results are shown in figure
1.1.3a. While the results do have lessons for health services around the world, it is
worth noting that the researchers chose to conduct the study in Belarus partly because
the country's health services were less developed than in the West.
Fig 1.1.3a
Promotion of Breastfeeding Intervention Trial (PROBIT)22 - Main results
Infants from intervention sites were more likely to:
• receive any breast milk at 12 months (19.7% vs 11.4%; adj OR 0.47; 95% CI, 0.32-0.69)
• be exclusively breastfed at 3 months (43% vs 6.4%; p < 0.001)
• be exclusively breastfed at 6 months (7.9% vs 0.6%; p = 0.01)
• have lower risk of 1 or more gastrointestinal infections
(9.1% vs 13.2%; adj OR, 0.60; 95% CI, 0.40 - 0.91)
• have lower risk of atopic eczema (3.3% vs 6.3%; adj OR, 0.54; 95% CI 0.31 - 0.95).
No differences were found in rates of respiratory infections;

(Intervention group, 39.2%; control group, 39.4%;
adj OR, 0.87; 95% CI, 0.59 - 1.28.)

1.1.4 Methodological issues in epidemiological studies of infant feeding
Writing in 1986, Bauchner et al24 identified methodological flaws, which reduced the
validity of studies investigating links between artificial feeding and infection. As a
result, he proposed four standards, which are reported in figure 1.1.4a
Fig 1.1.4a
Methodological standards proposed by Baucher et al24
Avoidance of detection bias

In retrospective case-control studies, those who breastfeed or bottle feed may
perceive illness differently and therefore report problems more or less frequently.
Adjustment for potential confounding variables

Because factors such as social class and behaviours such as maternal smoking
are associated with the mode of feeding, it is not always clear whether
associations between infant feeding and health are real or due to the
confounding variable.
Definition of outcomes
Bauchner et al concluded that in a number of the studies they reviewed, the
definition of health outcomes was unclear.
Definition of breastfeeding
Bauchner et al found that inadequate definitions of breastfeeding were common,
with some studies failing to report on supplementary feeding or the duration of
breastfeeding. If partial breastfeeders are included with exclusive breastfeeders
in a "breastfed" group, there may little difference between the feeding experience
of the "breastfed" group and those who receive infant formula, reducing the
power of the study.
Similar concerns were also raised in 1988 by Kramer,25 who advocated a more
stringent approach to evaluating studies of atopic disorders and others26. Although
subsequent studies, such as Howie and Forsyth's work in Dundee27 28 29

and the
PROBIT study reported above22 have attempted to control for the problems that
Bauchner et al identified, they remain important factors to consider. As a result of

these concerns, the Interagency Group for Action on Breastfeeding30 proposed a
common approach to defining breastfeeding (fig 1.1.4b).
Figure 1.1.4b
Schema for breastfeeding definition
BREASTFEEDING
TOKEN
FULL PARTIAL
Breastfeeding
episodes have
insignificant caloric
contribution
EXCLUSIVE ALMOST EXCLUSIVE HIGH MEDIUM LOW

No other liquid or Vitamins, water, juice and More than 80% 79% - 20% Less than 20%
solid is given ritual bikost given not of feeds of feeds of feeds
to infant more than once or twice
per day; not more than
1-2 swallows.
Reference: Schema for breastfeeding definition. (Modified from Labbok M, Krasovec K:

Towards consistency in breastfeeding definitions, Stud Fam Plan 21: 226, 1990.
Although this schema has focussed attention on the need for clarity, it has not been
universally adopted, perhaps in part because of the lack of a simple data collection
tool which researchers can incorporate into their surveys. The original paper30
suggests gathering more detail on feeding than may be realistic in most survey-based
studies. It also fails to reflect feeding behaviour when infants are weaned.
A meeting organised by the WHO Division of Diarrhoeal and Acute Respiratory
Disease Control in 199131 considered further the definition of breastfeeding. The

meeting agreed definitions derived from those proposed by the Interagency Group30,
but allowed medicines as drops or syrups within “exclusive breastfeeding” and added
definitions of “timely complimentary feeding” and “continued breastfeeding” to
assess feeding practices beyond the first four to six months of life. The WHO report
renamed “almost exclusive breastfeeding” as “predominant breastfeeding”. “Full
breastfeeding” includes both “exclusive” and “predominant breastfeeding”.
Auerbach et al26 have highlighted inconsistencies in the breastfeeding definitions used
in the literature, which often relate to imprecise use of the term “exclusive
breastfeeding”. In this thesis, I have used “exclusive” when quoting other authors’
findings, but have preferred the term “full breastfeeding” when reporting my own,
unless referring specifically to “exclusive breastfeeding” as defined by the WHO
report. In the discussion, I have used the most appropriate term for the context.
1.1.5 Mechanisms by which breastfeeding may benefit infants' health
Although infant feeding research often focuses on the benefits of breast milk, some
associations, such as the impact on maternal-infant bonding, neurological
development and suggested reductions in Sudden Infant Death Syndrome may be
partly due to the act of breastfeeding, its frequency and associated child-care
practices, rather than the composition of human milk. Other benefits relate to the
circumstances and level of hygiene under which formula feeds are prepared.
Some potential benefits, such as reduced infection rates are likely to involve a dose-
response relationship, because the more a mother breastfeeds, the higher the dose of

components such as secretory IgA, macrophages and other protective substances the
baby will receive. Conversely, the less formula the baby receives, the fewer
contaminating bacteria he or she will encounter. However other potential benefits may
be unrelated to the amount of formula a mother gives. If protection from allergy is
mediated by early sensitisation of the infant to foreign proteins, it may require very
little contact with infant formula for this to happen and as a result, only exclusive
breastfeeding may be effective in preventing the adverse effect. This makes it harder
to demonstrate associations with problems which require abstinence from other feeds.
1.1.6 Evidence for associations between infant feeding and infection in infants
Evidence that breastfeeding protects against infection comes from both laboratory and
epidemiological studies. Breast milk may be seen as a carrier of biochemical
messages from mother to baby, an approach which Bernt and Walker adopted in a
recent review of the function of different components of breast milk.32 For example, it
contains a range of anti-infective components that are thought to protect infants
against infection and mothers against mastitis4 (table 1.1.6).
Table 1.1.6
Components of human milk which are thought to protect against infection
(Adapted from Heinig and Dewey, 1996 4)
• Carbohydrates that inhibit pathogenic bacteria binding to epithelial cells

• Nitrogen-containing sugars that promote growth of beneficial lactobacilli
• Antibodies such as secretory IgA and serum IgG
• Anti-inflammatory agents
• Anti-oxidants
• White blood cells such as neutrophils and macrophages
• Lactoferrin which binds iron, making it unavailable to certain bacteria
• Lysozyme, an enzyme that attacks microbial pathogens
• Anti-viral lipids
• Anti-protozoan factors

The American Academy of Paediatrics Workgroup on Breastfeeding3 describes
evidence that breastfeeding reduces the incidence and/or severity of a number of
infections as strong, echoing Heinig and Dewey's rather fuller review.4 These
conditions include diarrhoea, lower respiratory infection, otitis media, haemophilus
influenzae bacteraemia and meningitis, botulism, urinary tract infection and for
premature babies, necrotising enterocolitis.
1.1.6a Gastrointestinal illness:
One of the most important studies of the impact of infant feeding on human health is
based on observations on 674 babies born in Dundee between September 1983 and
May 1986.27 28 29 The study was designed to address the concerns raised by Bauchner
et al24 and controlled for confounding variables identified by logistic regression. By
controlling for social class, maternal age and parental smoking, the authors were able
to be more confident that associations between feeding behaviour and health
outcomes were real.
In their report on the first year of life, Howie et al27 found that babies breastfed for 13
weeks or more had significantly less gastrointestinal illness at all ages up to 52 weeks,
with reductions of between 6.6% and 16.8% (95% confidence intervals) in the first
three months. The benefit was maintained beyond the period of breastfeeding itself
and fewer infants required hospital admission.
Others have also compared the relative risk of gastroenteritis in breastfed and
formula-fed infants. Duffy33 found a significant reduction in gastroenteritis amongst

US infants who were exclusively breast fed for four months. Dewey34, monitored 87
Californian infants weekly for two years and found that diarrhoeal illness was
approximately twice as common amongst formula-fed infants, but no association with
respiratory disease. However, a prospective study of 500 babies born to largely
middle class parents in Copenhagen35 found no significant relationship between infant
feeding and gastroenteritis, respiratory illness or otitis media during the first year of
life. This finding may be related to the low power of the study, because few of the
infants received formula in the first months. Eighty percent were exclusively breastfed
at one month and almost 60% at four months.
In 1999, Raisler et al36 reported on the health of 7,092 infants in the US National
Maternal and Infant Health Survey. Fully breastfed babies had lower odds ratios for
diarrhoea, vomiting and cough or wheeze. Predominantly breastfed babies had lower
odds ratios for diarrhoea and cough or wheeze, while those fed equal amounts of
breast and formula feeds had lower odds ratios for cough or wheeze. These results
suggest that reductions in infection rates are associated with the amount of
breastfeeding a baby receives. The recent PROBIT trial22 referred to in section
1.1.3a also found a significant reduction in the risk of gastrointestinal infections.
Necrotising enterocolitis is the commonest serious gastrointestinal illness seen in
neonatal intensive care units. In 1990, Lucas and Cole14 reported on a study of 926
premature babies, which has been described in section 1.1.3a. There were 31 cases of
confirmed necrotising enterocolitis, eight of whom died. Babies who were
exclusively fed formula were 6 - 10 times more likely to have confirmed disease than

those fed breast milk alone and 3 times more likely to than those who received
formula plus breast milk.
1.1.6b Respiratory illness:
In 1998, the Dundee study group reported on the health of infants up to the age of
seven.29 They found that the probability of children ever having respiratory illness
was significantly less for those who were exclusively breastfed for at least 15 weeks,
17%, (95% confidence interval 15.9% - 18.1%), than for those who were partially
breastfed, 31.0%, (26.8% - 35.2%), or those who were bottle fed, 32.2%, (30.7% to
33.7%). Other findings at age seven included the observation that children who were
exclusively bottle fed had higher systolic blood pressures than those given any breast
milk. Introducing solids before fifteen weeks was associated with more wheezing and
an increased percentage of body fat.
In a study of more than 1,200 Arizona infants enrolled with a health maintenance
organisation, Wright37 found that at the age of six, 11% of recurrent wheeze in non-
atopic children could be attributed to not having been breast-fed. Other studies,
which investigated possible associations with respiratory illness, have been referred to
in section 1.1.6a above.34 35 36. The PROBIT trial22 referred to in section 1.1.3a found
no association with rates of respiratory infection, which the authors attributed to
relatively high breastfeeding rates in both control and intervention groups and women
in Belarus relying little on child-care facilities.

Although not necessarily applicable to developed countries, the results of a Brazilian
case-control study merit attention, because of the rigour with which it was conducted
and the focus on mortality, rather than morbidity.38 Victora et al identified 127 infants
who died due to respiratory infection and compared them with 254 neighbourhood
controls. While the main risk factor associated with mortality was low socioeconomic
status, breastfeeding was also shown to have a significant protective effect.
1.1.6c Otitis media
A number of studies have investigated whether there is an association between
breastfeeding and otitis media, with mixed results.
Those reporting an association include Duncan et al39 from the Arizona study referred
to above37. Infants who were exclusively breastfed for four months had half as many
episodes of acute otitis media as those who were never breastfed and 40% less than
those who were partly breastfed. Similarly, a community-based study from Goteburg
in Sweden,40 demonstrated a significantly lower incidence of otitis media in the first
year of life. A prospective study of more than 500 infants from Texas41 also found
that exclusive breastfeeding was associated with less otitis media during the first six
months. Dewey,34 in the study referred to above, found the incidence of otitis media
was 19% less in those who were breastfed throughout the first year of life.
Others have however not shown associations between infant feeding and otitis media.
Howie et al27 found no association in the Dundee cohort, although it is worth noting
that otitis media was less common than respiratory or gastrointestinal infections.

Rubin et al's study of 500 Danish infants35 appeared to show a protective effect
against otitis media, but this was no longer significant when the investigators
controlled for age of the baby. It was also interesting to note that the US Maternal
and Infant Health Survey referred to above,36 infants without siblings who were
predominantly breast-fed had lower odds ratios for ear infections, whereas those who
had siblings did not. This may reflect the importance of siblings as a vector for
infection, which may outweigh any protective effect of breastfeeding.
Taken together, these studies suggest that breastfeeding does protect against otitis
media, but that as children get older, this may be overshadowed by other factors such
as nasopharangeal colonisation by pathogenic bacteria.42 40

There is however a case
for a meta-analysis based on the studies which have considered this.
1.1.7 Breastfeeding and neurological development
Over the years, a number of studies43 44 45 46

have found that breastfed children score
higher on tests of cognitive function than those fed formula. However research in this
area is complicated by the fact that breast-fed infants tend to come from more
advantaged family environments and their mothers have higher educational
attainment. One example of such a study was Rodgers' finding that breastfed babies
in a 1946 UK cohort study of 5,362 children had a 1.76 point advantage on non-verbal
IQ score, after adjusting for identifiable confounding factors.43 Reporting data from
771 low birthweight infants included in the trials of preterm infant feeding referred to
in section 1.1.4 above, Morley et al noted that approximately half the advantage

detected in infants' Bayley mental developmental index could be accounted for by
having a mother who opted to breastfeed.
In 1999, Anderson et al47 published a meta-analysis of differences in cognitive
development between breast-fed and formula-fed children and attempted to clarify the
role of potential confounding factors. Before adjusting for key covariates, breastfed
infants had an IQ advantage of 5.32 points (95% CI, 4.51 - 6.14). After adjusting for
covariates, breastfed infants still had a 3.16 point advantage (95% CI 2.35 - 3.98).
The IQ advantage was detectable from age 6 months until 15 years, the last age for
which reliable data were available, and appeared to correlate with the length of time
for which infants had been breastfed. Low-birth weight babies benefited most, with a
cognitive development advantage of 5.18 points, compared with 2.66 points for those
with normal birth weights.
The mechanism for this effect has been questioned and may involve a combination of
both biological factors and the mother-infant interaction involved in breastfeeding. In
their trials of preterm diet, Lucas et al found that tube-fed premature infants given
breast milk had more favourable cognitive outcomes than those who were given
formula. It has been suggested that long chain polyunsaturated fatty acids in breast
milk may play a part in enhancing neurological development, but three trials in term
babies have reported conflicting results. 21 48 49.
Discussing a later follow-up study, based on IQ tests at age 7 to 8 years, which found
that boys born prematurely were more likely to benefit from nutrient-enhanced as

opposed to standard formula, Lucas et al18 hypothesised that suboptimal nutrition at
key stages in early brain development may have long term effects on cognitive
function. They suggested that early nutrition might "program" longer-term outcomes.
However, it may not be appropriate to extrapolate from this work to healthy term
babies, because many preterm babies develop illness in addition to their prematurity
and they are born at an earlier stage of neurological development.
Evidence for the role of early social environment as either a mechanism for, or
confounding factor in the association between breastfeeding and intelligence comes
from the range of factors identified as covariates in Anderson et al's meta-analysis.47
The importance of this is also illustrated by a study of men and women born in
Hertfordshire between 1920 and 1930 which used multivariate analysis to compare
health visitor records with intelligence tests conducted years later as adults.50
Predictors of a higher IQ were having fewer older siblings, older maternal age and
non-manual social class, but use of a dummy (pacifier), perhaps a measure of less
stimulation as a baby, was associated with lower adult intelligence. In this study,
although those who had been exclusively breastfed had slightly higher IQ scores, no
association between feeding and IQ was detected in the multivariate analysis.
In summary, meta-analysis has confirmed that breastfeeding is associated with a small
benefit in IQ which is likely to be partly mediated by the mother-infant interaction
involved in breastfeeding, and partly by differences between the composition of
human milk and infant formula. This benefit is greater for premature, or low-birth
weight babies, but does also apply to term babies.

1.1.8 Other potential benefits for infants
A number of other benefits for infants have been reported, although it would be
beyond the scope of this thesis to consider the evidence for all of these critically.
1.1.8a Allergic disorders
Over the last 30 years, researchers have expended considerable energy investigating
whether breastfeeding protects against allergic disease,51 but despite this effort, until
the publication of the recent PROBIT trial22, it has been hard to draw clear
conclusions. One problem has been that if avoidance of food antigens protects against
atopy, few breastfeeding mothers achieve the avoidance needed to demonstrate an
effect, either because they supplement with formula or early solids, or because of the
presence of food antigens in their own milk.52 53

As a result, many studies have had
inadequate power to detect differences in the prevalence of atopic conditions.
Evidence for an association with wheezing, (but not eczema) was provided by Burr et
al20 in a study of 453 children with a family history of atopic disease. Similarly,
Saarinen,54 Moore et al 55
and Chandra et al 52
all reported lower rates of atopic
disease amongst who were breast-fed. The PROBIT trial22 referred to in section 1.1.3a
found that infants of women delivering at intervention sites were breastfed for longer
and had a lower risk of atopic eczema (3.3% vs 6.3%; adjusted odds ratio 0.54; 95%
CI 0.31 - 0.95). In their trial with preterm babies, Lucas et al15 found that amongst the
subgroup of 160 infants who had a family history of atopic disease, those randomised
to receive infant formula experienced more atopic disorders at 18 months.

However there was no association in Howie et al's Dundee study,27 or a recent study
of 3,856 Japanese children at age three56. Golding et al57 found that those who
breastfed reported more eczema, but suggested that this might be because those with a
family history of atopy may be more committed to breastfeeding.
Until the publication of the PROBIT trial,22 the question of whether breastfeeding
prevents atopic disease had been controversial and it may take a little while for the
scientific community to interpret the results. However it seems likely that many
mothers with a strong family history of atopy will breastfeed in the hope of protecting
their babies against conditions of which they have personal experience.
1.1.8b Case control studies for uncommon conditions
There is evidence that breastfeeding is associated with lower rates of insulin-
dependent diabetes in childhood58 59 60 61.
Case-control studies conducted by a team in Naples have shown lower rates of
breastfeeding amongst those with urinary infection,62 appendicitis,63 intussusception64,
pyloric stenosis65, inguinal hernia66 and tonsillectomy67. However case-control studies
are prone to bias in recall of feeding method and the selection of controls and these
findings need independent confirmation. Although some have shown lower rates of
ulcerative colitis and Crohn's disease68, others have not confirmed these69.

There has been considerable debate about whether breastfeeding protects against
Sudden Infant Death Syndrome after a case-control study from New Zealand found
that significantly fewer cases than controls had been breast-fed.70 71

However it is
noteworthy that even in that study, sleeping prone (relative risk 5.74, 95% CI 3.26 -
10.1) and maternal smoking (relative risk 2.45, 95% CI 1.32 - 4.55) were more
important risk factors than not breastfeeding (relative risk 1.83, 95% CI 1.02 - 3.29).
This was not however confirmed in a UK study and an expert group established by
the UK chief medical officer concluded that published studies had not demonstrated a
link between bottle feeding and Sudden Infant Death Syndrome. 72
Another topic, on which the evidence is equivocal, is a suggested link between infant
feeding and childhood cancer. While some studies have suggested a benefit, others
have not and there is also concern that carcinogenic substances in the environment
may be transmitted to infants by breast milk.
1.1.8c Infant nutrition and childhood growth
Breast-fed and formula-fed infants have different growth patterns73 and there has been
interest in whether this may herald a lifelong difference in body composition and the
prevalence of disease in adult life. A German study74 of 9,357 children who were
examined at school entry found that 4.5% of those who were never breastfed were
obese (body mass index > 97th centile), compared with 2.8% of those who were
breastfed. Similar findings were reported by Wilson et al29 from the Dundee cohort
of children at seven years. These findings suggest a causal link, but an alternative
explanation would be that mothers who breastfeed also give their children a more

healthy diet during the early years. There are however interesting parallels between
these findings and the hypothesis that early nutrition "programmes" the body for later
life, which was also suggested by research on neurological development referred to
previously.
1.1.9 Breastfeeding and maternal health
Although this chapter has focussed on benefits for infants, there are also benefits for
mothers. Lactational amenorrhoea reduces menstrual blood loss and plays a role in
helping women space their children. Research has also confirmed the popular belief
that breastfeeding helps women regain their pre-pregnant weight75 and that they
experience improved bone remineralisation post-partum76 and fewer hip fractures in
later life.77 Breastfeeding has also been associated with a lower incidence of pre-
menopausal breast cancer78 79

, confirmed in two recent reviews 80 81
and a reduced
risk of some ovarian cancers.82
"Bonding" may be seen as an advantage for both mothers and babies and is cited by
around 20% of mothers as a reason to breastfeed.1 It has both physical and
psychological components,83 some of which are mediated by the release of oxytocin
during breastfeeding.84 85
Although the concept encompasses the experiences of
many mothers and babies, it may be conditional on breastfeeding being a success and
for some, as Schmied and Barclay's qualitative work describes,86 breastfeeding is
painful and dysfunctional.

1.1.10 Disadvantages of breastfeeding and situations in which breastfeeding is
inadvisable
As has been shown, there is evidence that breastfeeding confers a range of health
benefits to mothers and babies. There are however some circumstances in which this
may not be the case. These are listed in figure 1.1.10. Lawrence87 discusses the
issues involved in this more fully and also provides detailed information on drugs in
breast milk.
Fig 1.1.10
Situations in which breastfeeding may be contra-indicated.
Mothers with HIV

The UK Chief Medical Officer has recommended that in the UK, mothers who are
HIV positive should not breastfeed319 because of evidence that the HIV virus can
pass from mother to baby.315 316 317 Untested mothers at high risk of HIV need to
be aware of this advice.
Galactosaemia
A rare condition in which Infants are unable to metabolise galactose and need a
special formula. The condition presents with jaundice, vomiting and electrolyte
disturbance.86
Medication
Some drugs pass from mother to baby in breast milk, and although most cause
no harm, prescribers need to check the safety of medications. In the few cases
where medications may cause harm, women should be fully involved in decisions
about the most appropriate approach.318
Illegal drugs
A number of illegal drugs pass from mother to baby in breast milk and the
American Academy of Pediatrics has recommended that mothers using
amphetamine, cocaine, heroin and marijuana should not breastfeed.320
In addition to specific situations where breastfeeding may be inadvisable, it is also
worth referring to some other potential disadvantages of breastfeeding. There is
evidence that breastfed babies are more likely to suffer bleeding due to vitamin K
deficiency - haemorrhagic disease of the newborn. Formula feeds are supplemented

with vitamin K and although all infants are given an initial dose of vitamin K,
breastfed babies receive two further oral doses. 88
Another issue that has caused concern has been the presence of environmental
contaminants, such as polychlorinated biphenyls (PCBs) in human milk. This has
been widely reported with headlines such as "Breast Milk Poison Alert" (The Express,
12th July 199989) and may have alarmed many mothers. There is evidence that these
chemicals may be concentrated in breast milk and that breastfed babies are exposed to
higher levels than those receiving formula.90 91 92

It has been suggested that higher
intakes of PCBs are associated with neurotoxic effects on the developing brain and
higher levels of thyroid stimulating hormone, so these concerns need to be taken
seriously. However a recent review93 concluded that "only extreme levels of
contaminants in breast milk represent more of a hazard than failure to breast feed".
This conclusion was supported by two other reviews,94 95

but does reinforce the
importance of reducing levels of PCBs and other contaminants in the environment.
1.1.11 Conclusions and summary
The relationship between breastfeeding and human health has been subject to
extensive study, as shown by the wide range of literature reviewed. Because of this,
many authors elect to refer to the benefits in passing, but in doing so, there is a danger
of merely reinforcing the mantra that "breast is best", without understanding why,
how, or to what extent this is the case. Indeed, this last point - that researchers need to
quantify the benefits of breastfeeding and explain them in ways that people can
understand is a theme which runs through many of the topics reviewed in this chapter.

Although Anderson's review of breastfeeding and cognitive development47 is an
important exception, few of the other associations with human health have been
investigated by meta-analysis. Doing so would provide clearer information for
mothers on the extent to which they can expect breastfeeding to benefit their health
and that of their babies. It might be helpful to present the results as Numbers Needed
to Treat, which could be expressed as for example, "For every x mothers who
96
breastfeed, one mother could expect her baby to avoid a particular adverse effect."
97
Adopting this approach would do much to make information about the benefits of
breastfeeding more accessible to mothers. There is also a case for further research to
find out whether quantifying the benefits increases mothers' motivation and ability to
breastfeed.
As has been shown, mothers consider a wide range of medical, psychological and
social factors in deciding whether or not to breastfeed. When feeding, they continue
to weigh these factors. In subsequent chapters I consider the ways in which they do
so, how this influences their feeding behaviour and ways in which those concerned to
promote breastfeeding have attempted to influence that behaviour.

Chapter 1.2
EPIDEMIOLOGY OF BREASTFEEDING:
WHAT FACTORS INFLUENCE INFANT FEEDING?
1.2.1 Introduction: The importance of epidemiological data
Understanding the epidemiology of infant feeding behaviour is important for a
number of reasons:
• Data on the prevalence of breastfeeding underpins research on the benefits of
breastfeeding.
• By examining the behaviour of groups, it is possible to learn more about the
factors that influence behaviour.
• It allows interventions to be directed towards those who most need help.
• Comparing trends in feeding behaviour provides a measure of the impact of public
policy and social changes.
It is not a substitute for broader sociological observation, in-depth qualitative studies
or experimental studies to assess specific interventions. But knowing the
characteristics and experiences of women who breastfeed provides an essential
context for other work.
1.2.2 Breastfeeding in the United Kingdom
Since 1975, infant feeding patterns have been the subject of quinquennial national
surveys, conducted by the Office of Population Censuses and Surveys, (OPCS),
recently renamed the Office of National Statistics, (ONS).98 99 100 101 1 These surveys
were established after a DHSS Committee on Medical Aspects of Food Policy
1.2 Literature Review: Epidemiology of breastfeeding 26

(COMA) report in 1974102 recommended monitoring of infant feeding trends because
of concern about the decline in breastfeeding and evidence demonstrating the health
benefits of breastfeeding. Although the initial survey in 1975 looked only at England
and Wales, Scotland was included in 1980, and Northern Ireland in 1990.
In the 1995 survey,1 66% of women in the United Kingdom began breastfeeding, but
by six weeks only 42% were still giving any breastfeeds. By four months, this had
fallen to 27% of women. These figures however include both full and partial
breastfeeding and although the surveys did ask about the introduction of formula
feeds, questions on this related to when the questionnaire was completed, rather than
the baby's age. When mothers completed the first questionnaire, between six and ten
weeks postnatally, 21% were fully breastfeeding, 18% were partially breastfeeding
and 62% exclusively bottle feeding. When compared with the Department of Health's
recommendation that all infants should be exclusively breastfed for a minimum of

102
four months , it is clear that four out of five UK babies do not receive optimum
feeding.
Whilst government recommendations provide one target against which feeding rates
can be measured, women themselves have views on how long they would have liked
to breastfed. Table 1.2.2, reproduced from the 1995 report1 suggests that the majority
of women who discontinue breastfeeding would have liked to have breastfed for
longer, a finding supported by the observation that the majority of women who
stopped before four months did so because of difficulties with breastfeeding.

Table 1.2.2
Whether mothers would have liked to have breastfed longer by duration of
breastfeeding
(% of those breastfeeding mothers who stopped during the study period)
Baby's age when breastfeeding % of mothers who would have liked Number stopping in period
ceased to breastfeed longer
Breastfed for:
Less than 1 week 90 477
1 - 2 weeks 91 163
2 - 6 weeks 89 528
6 weeks - 3 months 74 348
3 - 4 months 59 247
4 - 6 months 48 208
6 - 8 months 36 267
8 months or more 32 117
1
Reproduced from: Foster, Lader and Cheesbrough, 1997 (table 7.11, page 79) .
Whether assessed against government recommendations, or mothers' aspirations,
breastfeeding rates in the UK are too low and represent a significant challenge to
those concerned with health promotion.
1.2.3 Trends in breastfeeding
The five-yearly retrospective surveys provide comparative data on feeding over the
last 25 years, but it is helpful to look back further to understand the development of a
bottle-feeding culture in Britain and other Western countries.
A fuller discussion of the history of infant feeding is beyond the scope of this thesis,
but excellent accounts are provided by Lawrence87 and, from a more overtly political
perspective, Palmer103. They detail how social, commercial and professional pressures
have interfered with relationships between mothers and their babies, resulting in a

collective loss of confidence in breastfeeding. The implications of these trends for
breastfeeding support are also discussed in section 1.7.1.
Until the end of the 19th century, wet-nursing was the only alternative for mothers
who were unable, or unwilling to breastfeed. It was an option only available to the
better off and led Trollope to write; "How is it that poor men's wives, who have no
cold fowl or port wine on which to be coshered up, nurse their children without
difficulty, whereas the wives of rich men, who eat and drink everything that is good,
cannot do so?" Anthony Trollope 1847
Soon however, the development of infant formulae provided an alternative that was
seized upon by commercial interests and often supported by the medical profession.
Science, which offered so much in other areas of life, allowed formula manufacturers
to refine their products and encouraged the view that breastfeeding mothers should
know how much milk their babies were getting. Restrictive feeding regimes and
bottle feeding offered a way for mothers to order their lives, and contribute to a more
industrialised society. In 1940, the British government began subsidising National
Dried Milk, partly to enable women to contribute to the war effort, a policy which has
been continued with the Welfare Foods Scheme. However data from three studies in
the late 1940s suggest that most women continued to breastfeed with 57%, 42% and
48% breastfeeding at three months in Luton104, a national study105 and Newcastle
upon Tyne.106 In 1953, Westropp reported breastfeeding rates of 85% at one month
and 70% at three months from an Oxford child health survey. 107

Over the next ten years the picture appears to have changed dramatically and
breastfeeding became less popular, with only 29% of Nottingham women still
breastfeeding at three months in Newson and Newson's 1959-60 survey.108 109
110 111
Grosvenor, in an MSc thesis quoted by Jones et al reported breastfeeding rates
of 54% at one month and 29% at three months. A fuller account of the results of local
infant feeding surveys is available in a report published by the MRC Environmental
Epidemiology Unit in Southampton which considered the impact of childhood
nutrition on the risk of cardiovascular disease in later life. 112
As more mothers turned to bottle feeding, the disadvantages of the higher solute
concentrations of the early formula feeds became apparent and prompted the
102
Committee on Medical Aspects of Food Policy and World Health
Organisation113to advocate breast feeding in 1974. In the first National survey in
1975, only 51% of women ever breastfed and only 24% continued to six weeks.
Table 1.2.3
Incidence and prevalence of breastfeeding in England and Wales from 1975 to 1995
98 99 100 101 1
(Percentage feeding at time shown)
1975 1980 1985 1990 1995

Initial feed: 51 67 65 64 68
1 week: 42 58 56 54 58
6 weeks: 24 42 40 39 44
4 months: 13 27 26 25 28
Over the next five years, breastfeeding rates increased significantly, but since 1980,
they have remained remarkably static. Indeed, the proportion of mothers both
initiating breastfeeding and continuing to do so fell slightly between 1980 and 1990.

Although the data presented suggest that the falling trend was reversed in 1995, there
has been concern that this may merely reflect changes in the way the survey was
conducted.1 New data protection legislation restricted the survey team's ability to
follow-up non-responders and there was a lower response rate from younger mothers
in manual social class groups. Overall, the response rate to the initial request to take
part in the study fell from 89% in 1990 to 75% in 1995. Although the researchers
weighted the sample to correct for the imbalance in social class groups, if those who
were not committed to breastfeeding responded less frequently, there may have been
additional bias that was not corrected for. Because of this, it will be interesting to see
whether the survey conducted in 2000 shows a further rise in breastfeeding rates.
Taking a closer look at data on the duration of breastfeeding, it is apparent that 47%
of the 51% who initially breastfed in 1975 were still giving any breastfeeds at six
weeks, but that five years later 63% of the 67% who began breastfeeding continued to
six weeks. Over that five-year period, more women began breastfeeding and more
continued to feed, but there has been little change in the proportions giving at least
some breastfeeds in the following twenty years. It is worth noting that the first study
reported in this thesis was conducted in 1983, when the second five-yearly survey
appeared to herald ongoing increases in breastfeeding rates. Instead, rates changed
little and the results of study one may be more representative of present-day feeding
behaviour than one might have anticipated.
1.2.4 Giving formula feeds to breastfed babies
Although there is evidence that many of the benefits of breastfeeding relate to

exclusive, rather than partial breastfeeding, many women combine breast and bottle
feeding. This ranges from giving occasional bottles when breastfeeding is
inconvenient to regular supplementation and sometimes reliance on bottle feeding
with only token breastfeeding.
Table 1.2.4 shows that fewer babies are being given formula in hospital, which
represents progress towards meeting the sixth of the WHO/UNICEF Ten Steps;23 "To
give newborn infants no food or drink other than breastmilk, unless medically
indicated." The importance of this is underlined by the observation that in 1995, 34%
of mothers whose babies had been given a bottle in hospital gave up breastfeeding in
the first two weeks, compared with 11% of those who had not. It is however unclear
how much of the change observed is due to the WHO/UNICEF initiative as the UK
Breastfeeding Initiative was only established in 1994.114
Table 1.2.4
Proportion of breastfed babies given formula feeds
100 101 1
(Percentage of mothers who were breastfeeding at each stage, Great Britain)
1985 1990 1995
Given formula in hospital 50 45 36
Given formula at 6 - 10 weeks 34 39 46
Given formula at 4 - 5 months 24 27 43
Note:
The denominator for this table is the number of women breastfeeding at each stage of the study. Because of
this, useful comparisons can only be made across, rather than down the table. Many of those who gave bottle
feeds in hospital were not breastfeeding at six to ten weeks, and similarly, many of those giving bottles at 6 -
10 weeks were not breastfeeding at 4 - 5 months.
As well as pointing to a reduction in the use of formula in hospital, table 1.2.4 also
demonstrates another trend. It shows that at both six to ten weeks and at four to five

months, increasing numbers of breastfed babies are being given formula feeds. There
appears to have been a move away from exclusive towards partial breastfeeding over
the last ten years
1.2.5 Understanding trends in breastfeeding
Baer,115 (quoted in Lawrence 87

,) identifies two types of interventions to promote
breastfeeding: "supply" and "demand". Supply interventions are those that increase
the availability of breast milk, while decreasing the availability of substitutes.
Making hospital routines more conducive to breastfeeding by enabling early contact
and restricting the availability of substitutes are supply interventions, as are extending
maternity leave or providing time at work for breastfeeding. Demand interventions on
the other hand address mothers' motivation to breastfeed, emphasising health benefits
with messages such as "breast is best"; motivations, such as "breastfeeding is a great
mothering opportunity" or risk messages, such as "using substitutes is a health risk."
This framework is also helpful in considering the reasons for changes in breastfeeding
rates in the last 25 years. Because supply interventions tend to focus on the postnatal
period, they are more likely to influence the duration of breastfeeding than its
initiation, whereas demand interventions have the potential to influence both
incidence and duration.
1.2.5a Maternity services and breastfeeding
Maternity services have provided an important focus for "supply interventions" and
there has been considerable debate about the way that hospital practices have

influenced infant feeding. Whereas health professionals were previously criticised for
supporting rigid approaches to feeding, practice has now changed, as evidenced by
reports of women's experiences in the quinquennial surveys (table 1.2.5a).
Table 1.2.5a
Changes in reported hospital practice between 1975 and 1995
98 99 100 101 1
(Percentage of British mothers who began breastfeeding) )
b
1975 1980 1985 1990 1995
Baby put to breast immediately after birth 3 16 27 26 25

Baby put to breast within an hour of birth 8 30 32 37 43
a
Feeding on demand 36 68 81 90 -
c
Mother and baby together continuously in 4 17 47 63 74
hospital
c
No bottles given in hospital 23 49 48 54 61
Notes:
a Rigid schedules and feeding on demand did not merit a mention in the 1995 report.
b The 1975 data exclude Scotland.
c The 1975 data relate to the first day in hospital. Over the 25-year period, length of hospital stay fell.
These surveys provide a measure of practice in maternity units, but it is harder to
compare changes in the way that midwives and others have supported women. During
the 25-year period, a succession of reports and guidelines from the Department of
Health 102 114, the World Health Organisation and UNICEF 23 and professional bodies
such as the Royal College of Midwives116 have set out to influence practice in
maternity services, and it is helpful to compare changes in reported hospital practice
with the "Ten Steps to Successful Breastfeeding" proposed by WHO/UNICEF
(appendix G). Writing in 1992, Beeken and Waterston337 compared the accounts
that health professionals gave about their adherence to the Ten Steps with the
accounts given by a sample of mothers in Newcastle upon Tyne and identified a
significant gap between what professionals claimed to do and the care that mothers

received. The development of evidence-based practice in supporting breastfeeding
mothers is discussed further in chapter 1.6.
Over the period, there have been other important changes in maternity care, such as
the closure of small maternity units, the move towards early discharge and the
development of team midwifery but it is unclear what impact these have had on
support for breastfeeding. Staffing levels are rarely discussed in academic studies, but
a recent Audit Commission report117 identified them as an increasingly important
factor affecting the delivery of postnatal care.i Similarly, the English National Board
for Nursing, Midwifery and Health Visiting has expressed concerns that midwifery
staffing levels may account for a fall in the number of deliveries attended by
midwives rather than obstetricians.118 Midwives are concerned that they do not have
enough time to spend with mothers, which may help explain why apparent
improvements in maternity unit practice do not appear to be leading to improvements
in the length of time for which women breastfeed.
The fact that duration of breastfeeding has remained roughly constant over the last 20
years, despite changes in ward practice does call into question the effectiveness of
"supply interventions" which form a major component of the approach that
WHO/UNICEF have adopted in the Baby Friendly Hospital Initiative. Dropping
outdated feeding practices may have removed some barriers to breastfeeding between
1975 and 1980, but it does not offer an adequate basis to change the prevalence of
breastfeeding in the 21st century.
i
Further details of the Audit Commission survey are given in section 1.3.5 below.

1.2.5b Media coverage of infant feeding
There is a debate about whether the media lead public opinion, or merely reflect the
values and priorities of the time. But newspapers and magazines do offer a
contemporaneous record of issues and concerns that contribute to the cultural context
in which people make decisions about health. As such they provide a perspective on
the "demand interventions" referred to by Baer.115
119
For example, a quote from Parents magazine in 1938, (reported in Lawrence's
text87) reflects the attitude of women's magazines at the time, paying lip-service to
breastfeeding, but undermining even the staunchest of breastfeeders: "You hope to
nurse him, but there is an alarming number of young mothers today who are unable
to breastfeed their babies and you may be one of them…"
A recent BMJ report120 documented that significant media bias against breastfeeding
persists in both broadsheet and tabloid newspapers and across the spectrum of
television programming. For example in March 1999, the television programmes
monitored showed bottle feeding in 170 scenes, but breastfeeding in only one. When
breastfeeding was referred to, it was normally as something problematical, or worthy
of comment.
Initiatives to promote breastfeeding are also vulnerable to media attack. For example,
The Guardian celebrated National Breastfeeding Awareness week in 1998 with an
article by columnist Julie Burchill which attacked the "Nipple Police" who seek to

promote breastfeeding.121 It appears that efforts to promote breastfeeding are fair
game for those who seek to score easy points off presumed feminists.
Hoddinott has considered the contextual background to the changes in breastfeeding
documented in the five-yearly national surveys.122 She examined newspaper coverage
by comparing references to infant feeding in The Times Index over the period from
1970 to 1995. When she classified entries as appearing to adopt a "pro", "anti" or
"neutral" stance towards breastfeeding, she found that pro-breastfeeding articles
predominated between 1970 and 1980, but that subsequent coverage was more neutral
or focussed on negative aspects of breastfeeding.
Looking in detail, she identified a period around 1976 when a number of articles
reported concerns about hypernatremia and that bottle feeding might contribute to cot
death. These echoed an increase in publications on breastfeeding in the medical press
and a letter sent by the Chief Medical Officer at The Department of Health to all
doctors in 1976 which warned of the dangers of hypernatremia.123 She suggests that
the message conveyed in consultations up and down the country may have changed
from “breast is best” to “formula milk is dangerous”. The five-year period from 1975
to 1980 was marked by the only significant rise in the incidence and duration of
breastfeeding this century.
1.2.6 Socio-demographic factors associated with breastfeeding
Successive infant feeding surveys101 1 have demonstrated strong associations between
mothers' socio-demographic characteristics and their infant feeding behaviour. Of

those who had previous children, mothers who had previously breast-fed were more
likely to both initiate breastfeeding and continue for longer than were those who had
not, or who had stopped in the first six weeks. Amongst mothers of first babies,
higher levels of education achievement and non-manual social class were associated
with both the initiation and maintenance of breastfeeding. Infant feeding behaviour
also varied significantly between different parts of the UK, with 76% of mothers
initially breastfeeding in London and the South East, but only 55% of Scottish
mothers doing so in 1995.1
Because of the complex inter-relationships between many of the characteristics
investigated, the 1995 survey1 included logistic regression analyses of associations
between characteristics of mothers and the incidence of breastfeeding both initially
and two weeks after the birth. As an example of this, table 1.2.6 reports the
characteristics associated with initially breastfeeding in England and Wales.
Although national data from previous years was not analysed in this way, the
associations shown in table 1.2.6 can be compared with those found in Nottingham in
the first study reported in this thesis (Section 2.3.4). Although the national data can
detect weaker associations more accurately, because they are based on a larger
sample, the prospective approach adopted in the Nottingham study was able to
include questions that the midwives asked mothers. Both sets of results show the
importance of past feeding as a predictor of future behaviour, a finding which was
used to focus the intervention study (Study Two) on those women who were more
likely to discontinue early.

Table 1.2.6
Characteristics associated with mothers who initially breastfed
1
2,076 primiparae and 2,522 multiparae in England and Wales
First births Later births
Odds ratios
Percentage who breastfed initially 74% 62%
Mother's age Under 20 1.00 NS

20 - 24 1.56**
25 - 29 2.00***
30 and over 2.20***
Age mother completed full-time 16 or under 1.00 1.00

education 17 or 18 1.73*** 1.34*
Over 18 2.32*** 2.04***
Social class of partner I or II 2.06*** NS

III non-manual 1.17
III manual 1.35
IV or V 1.03
No partner / not classified 1.00
How mother was fed Bottle fed or method not known 1.00 1.00
Breastfed 3.24*** 1.75***
How friends fed their babies Most bottle fed 1.00 1.00
Half breastfed, half bottle fed 1.43* 1.08
Most breastfed 3.31*** 2.21***
Don't know 1.52* 1.61**
Went to classes with talks or Did not go to classes 1.00 1.00

discussion of feeding Classes with no talks on feeding 2.43*** 1.5
Classes with talks on feeding 2.27*** 2.08***
Whether mother smoked during Smoked 1.00 1.00

pregnanacy Did not smoke 1.44** 1.53***
When mother first held the baby Immediately 2.19** 1.67

Within one hour 1.75 2.38
Between one and 12 hours 2.19* 1.15
Over 12 hours 1.00 1.00
Significance of odds ratios:

* P < 0.05
** P < 0.01
*** P < 0.001
1
Reproduced from: Foster, Lader and Cheesbrough, 1997 (pages 54 and 55)
A number of other studies have also shown the importance of socio-demographic
factors in infant feeding behaviour.124 125 126 111

However one gap in our knowledge
on breastfeeding in the UK relates the role of ethnicity. Although individual studies

127 128 129 130
have addressed this and the government commissioned a survey of the

feeding practices of mothers of Indian, Pakistani and Bangladeshi origin in
1994/95,131 the quinquennial national surveys do not include information on ethnic
origin.
1.2.7 International comparisons of breastfeeding rates
Although this chapter has focussed on breastfeeding patterns in the United Kingdom,
it is also helpful to consider patterns of infant feeding behaviour in other countries.
Doing so allows us to learn from each other's experiences and see our own culture and
approach to infant feeding from a different perspective. It is also enables us to
understand the cultural context of research conducted in other countries - which we
might otherwise take for granted. It is easy to assume that the findings of research
done in one country are applicable in others, but this may not be the case. As an
example of this, McInnes and Stone132 challenged the applicability of a Mexican
study that showed significant increases in breastfeeding rates amongst Mexican
women who received peer support,133 on the basis that while this might apply in a
culture where women valued breastfeeding, it did not in Glasgow's disadvantaged
urban estates, where few women wanted to breastfeed.
Health and social policy is influenced by international initiatives such as the
WHO/UNICEF Ten Steps134 and recent controversy over whether the WHO should
recommend extending its advised duration of exclusive breastfeeding from four to six
months135 reflects the way experiences and priorities are different in different
countries. A further reason why international comparisons may be useful is that in a

multicultural society, patterns of feeding in the countries from which people come
may influence their behaviour in the countries they migrate to. 128 129 136
1.2.7a World Health Organisation data on breastfeeding rates
The World Health Organisation has developed a global data bank on breastfeeding,
which collects data on breastfeeding rates in different countries.137 138

Summary data
by region is reported in table 1.2.7a, reproduced from the WHO's 1998 review of
complementary feeding practices.139
Table 1.2.7a
Exclusive breastfeeding and median duration of breastfeeding: a global overview,
1996.1
WHO Region Countries of Infants in the Exclusive Median

the region region breast- duration of
included included feeding to 4 breast-
months feeding
Number of Number of
infants countries % % (months)
(millions)
Africa 23.3 25/46 71 19 21
The Americas 16.0 14/35 38 34 10
SE Asia 42.2 5/10 93 49 25
Europe 11.5 4/50 19 16 11
Eastern Mediterranean 15.5 11/22 84 36 19
Western Pacific 28.7 2/27 7 33 14
World total 137.2 61/190 58 35 18
1 137 139
Source: WHO Global Data Bank on Breast-feeding.
Table 1.2.7a reveals that overall, about 35% of the world's children are exclusively
breastfed at four months, and that the median duration of breastfeeding is eighteen
months. In Africa, the exclusive breastfeeding rate is low because women often give

supplementary water: in Europe, those who supplement are more likely to use infant
formula. African women continue well into the baby's second year, but most
European women discontinue sooner. It is also striking that the WHO was unable to
find comparable data from more than four European countries. Although this may
partly reflect the importance of breastfeeding in developing countries, it suggests that
breastfeeding is not seen as meriting a high priority in public health surveys across
Europe.
In a 1989 review of breastfeeding rates internationally, Williamson140 identified four
factors that helped describe variations in infant feeding behaviour. These were;
• In most countries, women living in rural areas breastfeed longer than those in
urban areas.
• In developed countries, more educated women are more likely to breastfeed,
whereas in developing countries, the reverse is often the case.
• In some countries, there were marked differences between the breastfeeding rates
of different ethnic groups.
• There are regional differences in the age at which women wean, as illustrated in
table 1.2.7a above.
1.2.7b Breastfeeding in Europe and North America
Grouping data on a regional basis illustrates global breastfeeding patterns, but it does
conceal variations between countries and masks the effects of national initiatives.
Although data has not been collected in a systematic way, there are striking

differences between the experiences in different countries in Europe and North
America, as shown in table 1.2.7b.
Table 1.2.7b
Surveys of breastfeeding in Europe and North America
Country and survey: Findings:
USA
141
- Ross Laboratories survey (1995) 59.7% initiated breastfeeding. 21.6% continued to six months.
(This was a significant increase since 1989, when the figures
were 52.2% and 18.1% respectively.)
1
UK - England and Wales (1995) 68% initiated. 44% at six weeks. 22% at six months.
1
UK - Scotland (1995) 55% initiated. 36% at six weeks. 19% at six months.
1
UK - Northern Ireland (1995) 45% initiated. 25% at six weeks. 8% at six months.
136
Republic of Ireland
- National survey(1982) 29% initiated breastfeeding.
- Births at Rotunda Hosp, Dublin (1994) 40% initiated breastfeeding.
France
142
- National hospital data (1995) 52% breastfed at hospital discharge, (42% exclusively.) 76% of
foreign nationals breastfed, whereas only 49% of French
nationals did so.
143
Netherlands (1992) 44% breastfeeding at 12 weeks.
144
Denmark (1992) 70% breastfeeding at 12 weeks.
145
Sweden (1990) 80% at 8 weeks. 70% at 16 weeks.
146
Norway (1985) 80% breastfeeding at 12 weeks. A survey of maternity wards
suggested breastfeeding was almost universal at hospital
discharge.
Iceland
126
- Study in one health district (1990) 83% breastfeeding at 12 weeks in 1990 (70% exclusive).
Increased from 67% and 57% respectively in 1985.
A number of trends emerge from the studies presented. Firstly, breastfeeding rates
across the UK and in Ireland reveal lower rates amongst Celtic (Scots and Irish)
populations. Ineichen, Pierce and Lawrenson have explored the attitudes behind this
in a recent literature review.136 It is also interesting to note that France has relatively

low breastfeeding rates.142 Breastfeeding rates appear to be rising in the United States,
whereas in the UK they are relatively stable.
It is the Scandinavian experience which is perhaps most interesting. In 1968, only 25
- 30% of Norwegian mothers breastfed at 12 weeks, similar figures to those in the
UK. However, since then the rates have diverged, with only small rises in the UK,
but progressive increases in Norway so that now, 80% of Norwegian mothers
breastfeed at 12 weeks.146
Hulme Hunter reviewed the reasons for this in 1996,147 and identified three
interleaving factors that had worked synergistically to produce the changes.
• The Norwegian Breastfeeding Association, Ammehjelphen which both offers
mother-to-mother support and campaigns for measures to support breastfeeding
from a feminist perspective.
• The Norwegian government has been generous in its support for maternity leave.
• Thirdly maternity services in Norway have enthusiastically adopted the Ten Steps
and currently half of the country's 60 maternity units have achieved 'Baby
Friendly' status.134
The importance of adequate paid maternity leave has been emphasised by a number of
authors.148 149
In Norway, mothers may receive 100% pay for 42 weeks, or 80% pay
for 52 weeks, whereas in the United States, the legal entitlement is only 12 weeks
unpaid leave. As a result many American women make considerable personal
sacrifices to continue breastfeeding, they switch to part-time work, they introduce

formula feeds or adopt the "Pump and Go" strategy of relying on expressed milk.150
151
But despite the efforts of individual American mothers, it seems likely that
maternity leave has played a major role in enabling more Scandinavian mothers to
continue breastfeeding.
1.2.8 Summary
Over the last twenty years, there has been little change in UK breastfeeding rates.
About two-thirds of mothers begin breastfeeding, but many stop in the early weeks.
There have however been changes in hospital practice and the use of infant formula.
Fewer babies now receive formula in hospital, while more do so at home in the first
few months.
While professional practice appears to have changed, the social pressures on mothers
appear unchanged. Feeding behaviour remains strongly related to socio-economic
factors and breastfeeding attracts little positive media coverage.
International comparisons suggest ways in which breastfeeding may be successfully
promoted. In particular, Scandinavian experience suggests that the combination of
improved maternity leave provision, changes in professional practice and enthusiastic
voluntary sector support for breastfeeding can work synergistically to create a culture
in which breastfeeding becomes the norm.

Chapter 1.3
BREASTFEEDING SUPPORT IN SOCIETY
1.3.1 Introduction
As has been shown, the rates of breastfeeding initiation and duration vary between
different social groups and between different societies. These differences reflect a
range of societal influences, which although complex, may help us understand ways
to modify breastfeeding behaviour. Dykes and Griffiths have identified a number of
these influences in a recent review,152 which is summarised in figure 1.3.1. They
comment that, "attitudes and behaviour in relation to infant feeding do not occur
within a social vacuum, but are highly influenced by the woman's social and cultural
environment."
Fig 1.3.1
Societal influences on breastfeeding behaviour
150)
(Adapted from: Dykes and Griffiths, 1998
- Perceptions of the female breast

- Socialisation and the observation of breastfeeding
- Influence of significant others, especially male partners
- Influence of health professionals
- Expectations of contented babies
- Return to employment
1.3.2 Sources of social support and influences on breastfeeding behaviour
Mothers receive support from a range of sources, their families, friends, and
professionals, but whether this promotes, or undermines breastfeeding depends on the
attitudes and experiences of the people giving support. Whereas in the past,
1.3 Literature Review: Social influences on infant feeding 46

grandmothers and other female relatives played a key role in this, Oakley,153 in her
study of social support and motherhood argued that the decline in breastfeeding in the
1960's led to a loss of collective knowledge about breastfeeding. This was echoed by
Hoddinott,154 who identified the importance of embodied knowledge, learnt by
observing others feeding, in the decisions first-time mothers make about infant
feeding.
Much of the published research on support for breastfeeding mothers is based on
American work, often from a single institution, which may limit its applicability to
women in the UK. Many of the studies are based on cross-sectional surveys, using
questionnaires to examine associations between support factors and either feeding
intentions or behaviour, but this approach has been criticised155 because the authors
do not adequately distinguish between association and causation. Examples of this
would be two studies that identified strong associations between women's views of
their partners' attitudes to breastfeeding and their own decision to breastfeed.156 157
Whereas partners' attitudes might influence women's decisions, alternative
explanations would be that they influenced their partners, or that people with similar
attitudes might be more likely to set up a family together.
Male partners have a significant influence on infant feeding, as shown by Kessler et
al158, who found that 71% of a stratified sample of Baltimore mothers identified their
partner as the person "whose opinion mattered most" in her infant feeding decision.
However this may not apply in all ethnic groups, as evidenced by a qualitative study
conducted with black women in Sheffield. In this, Higginbottom 128 found that while

older African and Caribbean women perceived their own mothers as their key role
models, younger women were more ambivalent about this, perhaps because more of
them had been brought up in the UK where observing women breastfeeding was not
an everyday occurrence.
In a longitudinal study of 32 predominately breastfeeding mothers in Utah, Isabella
and Isabella 159 noted that during the first month, most women turned to their partners
for emotional support, or help around the house, but to professionals for information.
Similarly, Buckner and Matsubara, working in Alabama, 160 found that during the first
fortnight, husbands played a key role in encouraging mothers through difficulties.
Although the Utilisation of Support Network Questionnaire they developed for the
study appears to be a useful measure of how women access support, the low response
rate, (48% of 126 participants,) does mean the results should be interpreted with
caution.
1.3.3 Men's perspectives on breastfeeding
Although women turn to their partners for support, there is evidence that many men
feel uncertain about this role.161 Freed et al162 found that although partners of women
who planned to breastfeed had positive attitudes towards breastfeeding, 71% regarded
breastfeeding in public as not acceptable and 24% agreed that "breastfeeding
interferes with sex". Similarly Voss et al163 found that some men felt "left out and
envious of the 'special bonding' between mother and infant." Guigliani et al164
assessed fathers' knowledge about breastfeeding, and found that those who had
previous children, or had attended antenatal classes were better informed. Worryingly,

the majority of fathers did not want to learn more about breastfeeding, which raises
questions about whether fathers really want the supporter's role.
Molinari and Speltini165 have shown that people's attitudes to breastfeeding vary
according to their views of gender roles, maternal roles and sexuality from their life
perspective at the time. Thus fathers often conceptualise the "instinct" to breastfeed as
a combination of biology and sacrifice - and leave it to the mother. Jordan and
Wall166 however identify that interactions between parents and their infants are more
complex than this. They write: "The breastfeeding mother has the control over
parenting and must realise that she has the power to invite the father in or exclude
him… Just as the father is viewed as the primary support to the mother-infant
relationship, the mother is the primary support to the father-infant relationship.”
Lewis, in a recent conference presentation,167 argued that health professionals have an
important role to play in these dynamics.
1.3.4 Negotiation and breastfeeding support
Support is not constant, but changes over time, as found by Morse and Harrison 155 in
a well-conducted longitudinal qualitative study based on telephone interviews with 30
breastfeeding mothers. They reported that breastfeeding women faced increasing
pressures to wean after the first six months, with friends, then parents, and finally
partners changing their stance from being supportive to making neutral or questioning
comments.

The concept of support as a measurable entity, which can be counted, or assessed for
its quality, may limit our understanding of the interactions that take place between
mothers and those around them. Implicit in the concept is the idea that support is
something given by one person to another, but mothers are not merely passive
recipients of support; they exercise choices about how to access it. Morse and
155
Harrison noted that when friends withdrew their support for breastfeeding, many
mothers withdrew from those who made negative comments and sought friendship
from mothers who were "still nursing." Further evidence that women access support
in different ways comes from a Canadian study that compared women who breastfed
for short or long durations. Hewat and Ellis168 found that those who breastfeed for
longer tended to feed more often, take a more relaxed approach, interpret their infants'
behaviours more positively and to have more emotional support from their partners.
In a paper published online, Bowes and Domokos130 investigated the experiences of
Pakistani and white women in Glasgow. They considered how women charted a
course through the conflicting social pressures they encountered and characterised
breastfeeding as "a socially-negotiated project". They wrote that, "Successful
breastfeeding projects are most likely for white middle-class women who have
effective stocks of knowledge, and can negotiate concerted action with health
professionals. Women belonging to socially excluded groups have greater difficulty
in the negotiation process, and their breastfeeding projects are less likely to be
successful. Whilst influenced by patterns of constraint, breastfeeding projects also
show marked individuality." It may be that approaches built on concepts of

empowerment will be effective in helping women negotiate their way to successful
breastfeeding.
1.3.5 Support from professionals
Whereas professionals appear to have little direct influence on the decisions women
make on whether or not to breastfeed,169 170

many do turn to professionals for advice
and support171. In 1995, 58% of UK mothers reported discussing breastfeeding with a
professional antenatally.1 Although in that survey, most women who reported
breastfeeding problems did receive support with them, there is evidence that mothers
often receive conflicting advice or feel unsupported in the early postnatal period.172 173
A recent Audit Commission report117 also expressed concern about the fragmentation
of postnatal care, about which the women reported more negative comments than
other aspects of maternity services.
Table 1.3.5
Comments received in a study of maternity care; Analysis by stage of care1
Antenatal Labour Postnatal Neonatal2 All phases
Positive 115 95 86 8 192
Negative 301 269 443 12 38

i
983 of the 2375 respondents in this study commissioned by the Audit Commission made additional comments
on their care. Some commented on more than one aspect of maternity care, but the breakdown shown does
suggest that mothers are particularly concerned about the quality of postnatal care.
2
Although not defined in the report, the "neonatal" stage appears to relate to care on neonatal units.
116
Source: The Audit Commission, 1997 (p81)
It is perhaps surprising that, despite the evidence that many women continue to have
difficulties with breastfeeding and the range of studies of societal influences on infant

feeding behaviour, few have asked women their views on the best way to support
breastfeeding mothers.
1.3.6 Summary
Women are subject to a range of social and cultural influences that influence their
infant feeding behaviour. Some of these provide support, but there is evidence from a
range of sources that many women feel unsupported in the early postnatal period.
Their male partners are often uncertain of their role and may lack the confidence to
support them. These findings underline the importance of understanding women's
needs and finding effective ways to support breastfeeding mothers.

Chapter 1.4
THEORIES OF BEHAVIOURAL CHANGE
1.4.1 Introduction:
Understanding human behaviour is important for anyone planning a health promotion
programme and it may be that an inadequate understanding of why women choose to
breast or bottle feed is one of the reasons professionals have had limited success
influencing their choices. Well-constructed models can provide a framework that
allows the factors influencing behaviour to be compared. By using models, health
workers may be able to identify which factors are most important and most amenable
to change. Similarly, models may help focus efforts on those people who are most
open to change. This chapter considers some of the models used to study human
behaviour and examines their relevance to breastfeeding promotion.
1.4.2 Theoretical approaches which underpin models
Most models are derived from research in social psychology and draw on three
different theoretical approaches; the social influence model, behaviourism and
attitudinal models. These approaches have however been criticised from a
sociological perspective as being too focussed on persuading individuals to change,
rather than understanding the socio-cultural factors which influence behaviour within
social groups.
1.4.2a Studies of social influence
The concept of conformity is based on the observation that, when faced with a
1.4 Literature Review: Theories of behaviour change 53

discrepancy between their own actions and those of the group around them, most
people tend to change in the direction of the group norm. However, whilst this
concept may help to explain the behaviour of individuals within groups, it is less
useful in explaining the behaviour of whole groups. This can be illustrated by
reference to the association between infant feeding behaviour and social class, which
has been reported above in section 1.2.6. Social influence theory can explain that an
individual mother in social class IV or V is more likely to bottle feed, to avoid
discrepancy between her own behaviour and that of her friends, but not why mothers
in social class IV or V as a whole are more likely to do so.
As has been discussed in chapter 1.3, how mothers feed their babies has been shown
to be associated with the attitudes of their family and friends.174 158

Indeed, as Morse
found in a qualitative study of weaning, the length of time that mothers continue to
breastfed could be related to the age at which their social contacts felt they should
wean.155
1.4.2b Behavioural models
Behavioural models are based on the premise that behaviour develops in response to
rewards and punishment and have been used in programmes to change a range of
behaviours, from desensitising people to phobias to stopping smoking. Whilst the
pain of sore nipples and the encouragement or discouragement of peers may act as
behaviour modifiers, they may also receive encouragement from professionals or
peers.

Although behavioural approaches have not often been adopted in promoting
breastfeeding, one exception was a small randomised trial conducted in Arizona by
Sciacca et al.175 They offered intervention group mothers and their partners a range of
incentives for both attending antenatal breastfeeding classes and subsequently
breastfeeding. Significantly more of the intervention group mothers breastfed
throughout the three-month follow-up, but the study had methodological flaws which
limit its value. Only 68 women were recruited and eight of the thirty-four in the
intervention group were excluded from analysis because they did not attend the
antenatal classes. Additionally, although those who reported they were still
breastfeeding were eligible for valuable prizes, the investigators appear to have made
no attempt to check the veracity of their responses.
1.4.2c Attitudinal models
Although they vary in their complexity and the extent they consider other factors, all
the models of behaviour change which are of help in understanding breastfeeding
acknowledge the importance of attitudes. To an extent this states the obvious, that
people make choices based on what they think and feel about the subject in question.
But research into attitudes is bedevilled by difficulty defining and measuring them
reliably. This is partly because attitudes are complex hypothetical constructs that
include elements of behaviour, knowledge and feelings. Although hard to define,
they represent a predisposition to act in a certain way.176 As constructs, they cannot be
measured directly, but they should ideally correspond to the verbal statements used to
measure attitudes in questionnaires or interview studies. Whereas epidemiological
studies tell us how many breastfeed, attitudinal studies, (which may range from large

surveys such as Jones's work in Cardiff,177 to small qualitative studies such as
Hoddinott's work with women in East London154), can help us understand why they
do so.
While attitudinal studies are useful in understanding the decisions people make, it
does not necessarily follow that changing someone's attitudes will alter the way they
behave.
1.4.3 The Health Belief Model
The health belief model178 179 was one of the first attempts to understand how people
make decisions about their health. It suggests that, prompted by some trigger to
consider an issue, individuals weigh up the pros and cons of a particular course of
action. The model suggests their decision will depend on their own perception of
both their susceptibility to health problems and the likely severity of those problems,
balanced against the 'costs' to them of taking the course of action. Thus the health
belief model suggests that women balance their perceptions of the likelihood and
severity of the problems their babies might face if they bottle feed against the 'cost' to
themselves of breastfeeding.
One problem with this is that whereas it may apply to people considering how their
behaviour may affect their own health, it may not apply when mothers and babies
have conflicting interests. Women with sore nipples have the “cost” side of the
equation, whereas most of the benefits of them continuing breastfeeding accrue to
their babies. Similarly, there is evidence that most women who choose to bottle feed

accept that breast feeding would be “better” for their babies,171 180 181 170
but decide
not to because they don't feel comfortable with the idea of doing so.
Fig 1.4.3
The Health Belief Model 176 177
Participation in preventative health behaviour can be predicted on the basis of

an individual's perceptions of:
• his or her own susceptibility to a given disorder

• the seriousness or severity of that disorder
• the benefits of taking action
• the barriers to taking action
• the cues to taking action
In a study of 41 low-income American women who were expecting their first babies
and intended to breastfeed, Libbus et al182 identified infant health benefits and
improved bonding as the main perceived advantages of breastfeeding. Disadvantages
included interference with maternal schedule, inability of others to feed infant and
physical discomfort. While health beliefs may differ between different populations,
these findings echo those of other studies such as the quinquennial infant feeding
surveys in the United Kingdom,98 99 100 101 1

and merit attention because it seems
likely that approaches that address women's actual concerns will be most effective in
facilitating breastfeeding.
While health beliefs are undoubtedly a factor in women's decisions, the model ignores
many of the social and cultural factors that influence infant feeding.

1.4.4 The Theory of Reasoned Action
The Theory of Reasoned Action, developed by Ajzen and Fishbein,183 184

regards
attitudinal and normative factors as predictors both of whether people intend to
undertake a particular behaviour and whether they actually do so. Subjective norms
reflect the role of social influences on decision making and include the individual's
own willingness to comply with those pressures. The attitudes referred to are the
individual's positive or negative evaluation of performing the action; their attitude
towards the behaviour. It is one of the few models that has been rigorously tested to
determine its ability to predict infant feeding decisions.
Fig 1.4.4
The Theory of Reasoned Action 181 182
Attitudes towards the behaviour

Intention Behaviour

Subjective norms
(a measure of social pressure)
In later work, Ajzen added a third construct, perceived behavioural control, which
concerns people's perceptions of the ease or difficulty of carrying out the behaviour
and is determined by both past experience and expectations regarding the future. This
is known as the Theory of Planned Behaviour. 185
1.4.4a The applicability of the Theory of Reasoned Action to breastfeeding
Manstead et al 186 conducted a well-designed study with 215 mothers to measure how

closely their attitudes and subjective norms correlated with their feeding intentions
and subsequent behaviour. They found that almost 60% of the variation in mothers'
intentions could be accounted for by the attitudinal and normative variables studied.
For mothers having their first babies, attitudes and subjective norms were almost
equally strongly correlated with feeding intentions, whereas amongst those who had
previous children, attitudes were more strongly correlated with feeding intentions than
subjective norms. This was perhaps not surprising as women who had previously fed
a baby had a real experience on which to base their attitudes. The results suggest that
experiences with a first child influence mothers' subsequent feeding intentions,
mediated by changes in their attitudes, but that the social pressures they experience
are largely unchanged.
Interestingly, they also found that mothers' attitudes exerted an influence on their
feeding behaviour, over and above the influence mediated by intentions; their
attitudes influenced their ability to carry out their intentions.
Overall, the finding that intentions and subjective norms are associated with infant
feeding behaviour support the use of the theory of reasoned action in predicting
women's infant feeding decisions.
1.4.4b Studies applying the Theory of Planned Behaviour to breastfeeding
As has been noted in section 1.4.4, the Theory of Planned Behaviour (TPB) adapted
the Theory of Reasoned Action to include the construct of perceived behavioural
control. A number of studies have tested the applicability of the theory to

breastfeeding. For example, in a study of women who planned to breastfeed,
published in 1994, Janke187 was able to predict 73% of those who weaned before 8
weeks using a "Breastfeeding Attrition Prediction Tool" based on the TPB.
Duckett et al188 further refined the model, adding specific factors known to be
associated with infant feeding behaviour and related the model generated to whether
women planned to return to work postnatally. The result is complex and although it
does provide a better understanding of the relative importance of different factors
associated with infant feeding behaviour, it is too complex for use in clinical practice.
It is also important to remember that the correlations between different factors
reported were derived from a population of first-time mothers delivering at a private
hospital in the American Midwest and are likely to vary for other populations.
Despite this caveat, this work brings together and quantifies a number of strands of
breastfeeding work in a coherent framework and seems likely to have a significant
influence on future breastfeeding research.
The correlations found for women who returned to work for more than 20 hours a
week before their babies were six months old are shown in figure 1.4.4b. A number
of observations can be drawn from this.
• As might be expected, duration was more strongly associated with feeding
intention than any other variable.
• Knowledge about breastfeeding influenced duration directly, without being
mediated by attitudes or intentions.

• Maternal education both altered women's attitudes, and their ability to carry out
their intentions (duration).
• As Manstead et al186 found, attitudes both influenced intentions, and women's
ability to carry them out.
• Perceived insufficient milk was an independent factor influencing duration, which
did not correlate closely with any of the other variables.
Figure 1.4.4b
Theory of Planned Behaviour - model for those working more than 20 hours a week
.13
Breastfeeding
knowledge .13 Insufficient
milk
-.13
Maternal Attitude:
.14
education bottle feeding
-.12
.10 -.16
Beliefs about
outcomes of .30
formula .20 .33
Attitude: Breastfeeding Breastfeeding
breast feeding intention duration
-.20 .28
Beliefs about
outcomes of
breastfeeding .11
.19 -.20
.37
Subjective norm .12 .28

-.15
.26
Referent beliefs Perceived control
Control beliefs .44
Reference: Duckett et al. A theory of planned behaviour-based structural model for breast-feeding.
Nursing Research 47(6): 325-336. 1998
The findings of this study both confirm the applicability of the Theory of Planned
Behaviour, and also the value of including women's perceived behavioural control in
the model. This aspect, of how able women feel to modify their behaviour is also
considered in a number of the models that follow.

1.4.5 The Stages of Change Model
Prochaska and Di Clemente189 developed the Stages of Change model to explain the
stages individuals may go through when changing an addictive behaviour. It is most
useful in individual health promotion work, as it allows health workers to match their
interventions to individuals' needs and identify those who are most receptive to help.
The stages identified are precontemplation, contemplation, preparation, action,
maintenance and relapse. Interestingly, Prochaska and Di Clemente see relapse not as
failure, but as part of the learning process; having learnt from one attempt to change
behaviour such as smoking, people are better equipped to succeed next time.
Figure 1.4.5
The stages of change model
RELAPSE PRECONTEMPLATION
Individual not interested in change
CONTEMPLATION
Individual is thinking about change
MAINTAINING A CHANGE
READY TO CHANGE
MAINTAINING THE MAKING A CHANGE

HEALTHIER LIFESTYLE
Reference: Prochaska JO, DiClemente CC. Stages and processes of self-change in smoking:
towards an integrative model of change. J of Consulting and Clin Psych. 1983: 51:390-5

The model is of some relevance in understanding women's readiness to consider
breastfeeding and how best to support them. However, it does not fully address the
reasons why people undertake a behaviour such as breastfeeding, or why some think
more than others about this decision. Also, the notion of relapsing is rather different
in infant feeding. Whereas smokers who relapse can try stopping again at any stage,
women who stop breastfeeding will need to wait until a future child is born before
they can put into practice what they have learnt from the previous attempt.
1.4.6 The Health Action Model
The Health Action Model190 offers a framework to explain both the reasons why
people make certain choices about their health and the factors that influence whether
they carry out those choices. It incorporates elements of the Health Belief Model, the
Theory of Reasoned Action and Prochaska and Di Clemente's work. But it also
recognises the importance of individuals' self-esteem and the environment in which
they try to implement decisions. As a result, the model is complex and harder to test;
a reflection perhaps of the complexity of human behaviour.

Figure 1.4.6
The Health Action Model
Health Action Routine

Facilitating Factors
Knowledge Skills
Supportive Environment
} Reappraisal
and possible relapse
Decision
Behavioural intention
Self Concept Self Esteem
Belief System Motivation System
Normative System
Health Promotion Input
Reference: Tones BK. Devising strategies for preventing drug misuse: the role of the health action model.
Health Education Research. 1995; 17-19.
Although the model has been used to illustrate a woman's experience of breastfeeding
191
, its applicability to this has not been fully evaluated. The separation of factors
influencing intentions from those influencing their implementation is helpful, but in
reality the same factors may influence both stages, as found by Manstead when
testing the Theory of Reasoned Action .186 That study, discussed more fully in section
1.4.4a suggests that attitudes influence both intentions and their implementation.
Similarly, there is likely to be an overlap between the extent to which a woman's
environment is supportive to breastfeeding and her normative system, which also
reflects her perception of how supportive are those around her.

Despite its complexity, the Health Action Model does make it easier to understand the
way in which interventions may work. Mass media campaigns, or attempts to portray
breastfeeding in a positive light through television, are most likely to affect women's
attitudes and norms, influencing their intentions, whereas interventions that focus on
individuals, such as support from breastfeeding counsellors, aim to help women
implement their decisions. The relevance of the Health Action Model for the
intervention study reported in this thesis is discussed further in section 1.4.10.
1.4.7 Self-initiated change
Addressing behaviour change from a sociological perspective, Backett et al at the
Edinburgh Research Unit in Health and Behavioural Change have challenged the
biomedical and psychological approaches which have dominated much health
promotion work. In “Changing the Public Health”192, they argue that public health
should be primarily concerned with those factors in the social structure which
impinge on health status, rather than focussing narrowly on trying to persuade
individuals to change their behaviour. Although not constructed as a formal model,
the prerequisites for behaviour change that they identify offer a framework for health
promotion programmes.

Fig 1.4.7a
Pre-requisites for self-initiated change
1). The behaviour needs to become salient, allowing the individual to think about it.
2). The way the behaviour becomes salient should not trigger a reaction or denial.
3). The individual needs time to think about the behaviour.
4). Change is more likely when individuals' abilities to cope are not already under
strain.
5). The climate of opinion should be supportive to change.
6). It should be possible to make the change, implying that the individual has the skills
to do so.
7). Change will be harder if the behaviour the individual might change plays a role in
his or her coping strategies.
190
Source: Backett et al, 1989
Backett et al see established behaviour as habitual, so that although people may think
consciously about their actions when they start doing something such as smoking,
they tend not to subsequently. They are only able to change the behaviour if for some
reason they reconsider it.
A number of these prerequisites appear to apply to women's decisions on whether to
breast or bottle feed. The relevance of the first, that women need to consider the
question at all, is borne out by findings presented later in section 2.3.4, that when
asked in pregnancy, many women did not consider an alternative to their intended
feeding method.171 The second also seems relevant, because if the reason most

women choose to bottle feed is because they don't fancy breastfeeding, there is a risk
that some approaches to promoting breastfeeding may trigger a reaction and make
them more likely to bottle feed.
While the idea of attempting to change behaviour at a time when individuals are not
under strain is attractive, (as suggested in Backett et al's fourth point), it is hard to
imagine a more momentous event than childbirth to precede the onset of
breastfeeding. Many women are exhausted, around 17% have had caesarean sections
117
and a number have had so traumatic an experience as to later develop post-
traumatic stress disorder.193 This does however reinforce the importance of the fifth
point, that there should be a supportive environment for mothers in the early postnatal
period. Further evidence of the importance of the climate of opinion for the
decisions women make on feeding is provided by Manstead's findings on the
importance of subjective norms, reported in section 1.4.4a. 186
The seventh point emphasises that it is hard for a mother to implement a behaviour
that may adversely affect her coping strategies. If she is exhausted, her partner may
help her cope by giving a bottle feed at night, even though in the longer term this may
undermine her breastfeeding. However Hoddinott, in her M Phil thesis,122 identified
that switching to bottle feeding may appear to new mothers as the only available
strategy (box 1.4.7b).

Box 1.4.7b
A coping strategy
"For most women who initiated breastfeeding, the post-natal period was one of
unexpected difficulties. For women wanting to breastfeed these were often exacerbated
by the lack of an apprenticeship model for learning how to breastfeed. Women often felt
isolated, had feelings of failure and lost confidence. These women in particular reported
conflicting advice from health professionals and relatives. Everyone seemed to be
experts in breastfeeding apart from themselves. Their expectations that it would be easy
and that professionals hold the recipe for success were shattered. They experienced a
secrecy about the difficulties learning to breastfeed, which was reinforced by the lack of a
visible breastfeeding culture.
For some women a crisis point was reached, where they felt they could no longer cope
and wanted to regain control of their situation. In particular they wanted the settled,
contented baby that every new parent hopes for. Women found it difficult to ask for more
support or a different quality of support to meet their needs. The only factor which was
easily amenable to change was the feeding method - from breast to formula."
Hoddinott, 1998 121
Perhaps the most important conclusion to draw from this is that attempts to promote
breastfeeding need to address factors in women's lives and social environment that
may affect their infant feeding behaviour.
1.4.8 A critique of models in breastfeeding research
Models may make behaviour easier to understand, but the way they have been used in
breastfeeding research has also been questioned. MacLean,194 challenges the
assumption that human behaviour can be described objectively, manipulated,
controlled and studied in the same way as other phenomena. Although it may be
attractive to try to isolate specific factors, (such as the timing of the first feed), which
can be modified, she questions whether they can really provide the key to successful
breastfeeding. For her, the social context in which women make decisions is so

important that a reductionist approach is inevitably flawed. Instead she advocates a
qualitative approach, to understand women's behaviour on their own terms and in the
context of their daily lives.195
Whereas MacLean's argument leads her to rather nihilistic conclusions, understanding
the socio-cultural context in which people make decisions about breastfeeding need
not imply that change is impossible - just that it is most likely to be achieved by
addressing the socio-cultural factors which contribute to the pattern of behaviour.
Bunton, Murphy and Bennett have reviewed the theoretical basis of health
promotion196 and identify a shift from reliance on psychological models to an
increasing emphasis on socio-cultural influences on behaviour change. Because so
much behaviour is collective, they argue that persuasive campaigns focussed on the
individual have only limited effect. They draw on ideas about communication in
society such as Rogers' work on the diffusion of innovations197 to consider the
importance of "opinion leaders" who have good communication channels with others
and are sufficiently established in a community to avoid alienating others by their
adoption of innovation. For them, understanding how subcultures interact in
transmitting information or persuasion is key to how behaviour change spreads - or
does not spread through society. Because social groups draw on their own resources,
they may not welcome behaviours arriving from other groups they do not identify
with. This has implications for promoting breastfeeding because infant feeding
behaviour varies so widely between different social class groups.

Bunton et al argue that the most productive approaches to health promotion will be to
understand the socio-cultural context in which people live and plan approaches which
maximise indirect effects by exploiting socio-cultural processes.
1.4.9 The limits to health promotion
It is possible to build ever more complex models, reflecting more and more factors
which may, or may not be amenable to change, but it is also important to understand
the wider implications of the balance between the personal and societal factors
198
influencing health behaviour. Stott et al put this in context in a thoughtful
editorial, pointing out the realities that those who are committed to health promotion
have to accept.
"People need individual care when they are frightened or ill; they will often support
sensible legislation for environmental improvement; but their willingness to change
cultural and social habits comes in small steps in response to external opportunities
for change and an inner readiness to change." 198
1.4.10 Implications for study design
Five main themes emerge from the consideration of models of behaviour change and
these have proved invaluable in the design and interpretation of the studies reported in
this thesis. These are:
• women's attitudes
• cultural norms and women's readiness to comply with them
• their readiness to change

• self-esteem
• the social context in which women find themselves.
The first study in this thesis, investigated the relative importance of women's
attitudes, cultural norms and readiness to change, in predicting how they would feed
their babies. Because of the exploratory nature of this study, it seemed important to
allow women to express themselves in their own words, using "open" questions.
However, this contrasts with Manstead's approach186, which was based on using
validated questionnaires to measure the strength of associations between women's
attitudes and norms and their feeding intentions and behaviour.
In one sense, study two adopted a logical positivist approach, to assess whether giving
individual women information, advice and support enabled them to carry out their
intention to breastfeed. However, the breastfeeding counsellors who delivered the
intervention may have also addressed women's self-esteem and ability to cope,
included by Tones in the Health Action Model.190
The second study was also designed to look beyond the impact of the intervention and
understand the role of breastfeeding support for women. To investigate this, the
follow-up questionnaires included a number of open questions that asked women
about their experiences and support that they received. These were then analysed
separately using a qualitative approach and are reported in chapter four.

1.4.11 Summary
Models of behaviour change provide a framework to consider the different factors
influencing women's infant feeding behaviour. They make it easier to weigh up the
relative importance of those factors and understand the ways in which interventions
may work. There is evidence that women's attitudes and the social pressures they
experience influence their decisions and that attitudes also affect the ability to
implement decisions to breastfeed.
Women differ in the extent to which they are open to change which suggests that
health workers should match their interventions to the individual. The context in
which people make decisions is important and those who want to promote
breastfeeding need to address factors in society that influence infant feeding
behaviour. However, it is also necessary to be realistic about what health promotion
can achieve, given the way that social pressures and personal beliefs reinforce current
infant feeding practices.

Chapter 1.5
INTERVENTIONS TO PROMOTE BREASTFEEDING
1.5.1 Introduction:
This chapter considers the role of intervention studies in assessing the effectiveness of
different approaches to breastfeeding support. Whereas epidemiology has
documented changes in infant feeding behaviour and socio-economic factors that are
associated with successful breastfeeding, it cannot answer the questions these
observations invite - whether intervening in some way can change that behaviour.
Although a number of studies have set out to evaluate interventions to promote
breastfeeding, many have serious methodological flaws. Some of the more robust
controlled trials have been included in a recent Cochrane review199 and the findings of
this are discussed. The implications for study design are considered.
1.5.2 Observation studies and the move to feeding on demand:
Epidemiological studies have identified factors associated with infant feeding
behaviour and can provide pointers to practices which may be beneficial or harmful.
For example, there is strong evidence of an association between mothers feeding on
demand and continuing to breastfeed and although this does not prove that demand-
feeding leads to prolonged breastfeeding, it suggests this may be the case. In the 1980
national survey, 28% of mothers who said they had to feed at set times had stopped
within two weeks of birth, whereas only 19% of those who fed on demand had done
so.200
1.5 Literature Review: Interventions to promote breastfeeding

73
This reinforced research published in 1952 by Illingworth and Stone201 comparing the
experiences of mothers and babies on postnatal wards that encouraged demand
feeding with those on wards following a four-hourly regimen. Demand-fed babies
were more likely to have regained their birth weight by the time they left hospital and
were more likely to be fully breastfed a month after delivery. It took some time for
these findings to filter into clinical practice, but the five-yearly national surveys do
illustrate the way hospital practices have changed. While 64% of mothers in England
and Wales initially followed a rigid schedule in 1975,98 in 1980, 32% of British
mothers did so.200 By 1985100, this had fallen to 19% and by 1990101, to 10%. Rigid
schedules did not deserve a mention in the 1995 report.1 (See also section 1.2.5a)
1.5.3 Implications of changes in feeding behaviour for research
This account of the demise of rigid regimens demonstrates that non-random studies
can uncover important results, but also shows the extent to which the setting in which
today's clinical trials are conducted has changed. Whereas mothers in Illingworth and
Stone's study stayed in hospital for nine days, they are now discharged much earlier.
Many unhelpful hospital practices, such as delaying the first feed, supplementing in
the nursery at night and offering formula milk samples have gone. Increasingly,
midwives have received training on how to help women establish breastfeeding,
although they do not always have the time they would wish to devote to this. More
women are breastfeeding.

74
These changes have implications for infant feeding research. Whereas in the 1970s
and 1980s, it was enough to change hospital practice and remove barriers which
prevented women establishing breastfeeding, much of that task has been done and
current research needs to address the problems women face today. A second issue is a
statistical one; as more women breastfeed, the sample size required to demonstrate
that an intervention is effective rises steeply.
1.5.4 The case for intervention studies:
While observational studies can identify factors that are associated with a behaviour
such as breastfeeding, they are less useful in clarifying whether modifying those
factors will lead to a change in the behaviour. Well-constructed intervention studies
offer a better guide to how to proceed, but they necessarily reflect the effectiveness of
interventions in the context in which they have been tested.
The importance of intervention studies is illustrated by research on the timing of the
first feed. Observational studies have consistently demonstrated that women who put
their babies to the breast earlier are more likely to continue breastfeeding. In the 1995
UK survey1, a quarter of mothers breastfed immediately and of these, 14% stopped
within the first two weeks. However, 26% of the one third of mothers who took more
than an hour to put the baby to the breast had discontinued by two weeks.
As a result of studies like this, the WHO/UNICEF Baby Friendly Initiative134 initially
included a recommendation to give the first feed within the first half-hour. Other
researchers however investigated this further and when Renfrew reviewed this for the

75
Cochrane Collaboration202 and as part of a longer review of interventions which
support breastfeeding203, she found no evidence to suggest there was a "critical
period" for the first feed. Indeed, in one study, women randomised to feed within 30
minutes had more difficulties204 and as a result, the WHO/UNICEF recommendations
have been revised and now call on maternity service providers to "Help mothers
initiate breastfeeding soon after birth." Randomised controlled trials can play a
crucial role in determining which interventions should or should not be adopted.
1.5.5 Assessing intervention studies
Assessing intervention studies requires an understanding of the method adopted, the
intervention tested, the participants and the setting in which they were recruited and
the outcome measures used. While bias, due to systematic error in the design or
conduct of a study can invalidate its results, it is also important to consider the extent
to which it is reasonable to generalise from the findings in deciding whether they
should be adopted in other settings.
Concerns about the prevalence of breastfeeding have spawned a variety of efforts to
promote it, but many of the studies have methodological flaws which have left them
open to bias, particularly as many mothers and professionals have strongly-held views
about breastfeeding. Although some investigators have succeeded in blinding
observers to subjects' treatment groups, few appear to have concealed this from the
subjects themselves. This raises the possibility that subjects were influenced by
knowing the treatment group to which they were allocated.

76
In designing the randomised controlled trial reported in this thesis, I considered ways
to conceal from subjects which treatment group they had been allocated to. One
option would have been to randomise potential participants before obtaining their
consent and then recruit them to one of two linked studies - receiving either the
intervention or normal care. However this would have required the doctors and
midwives to behave impartially in their approach to recruitment, which could not be
assumed. Because of the risk of recruitment bias, it seemed safer to adopt the
conventional approach of recruitment and then randomisation, even though it meant
that subjects would be aware of their treatment allocation. This issue did not however
arise in two recent cluster-randomised trials conducted in Bangladesh205 and
Belarus.22
206
Tedstone et al conducted a systematic review of studies promoting breastfeeding
and reported a range of concerns about their quality. These included criticism of the
information provided about how subjects were recruited, the possibility of selection
bias and the failure of many studies to take account of potential confounders such as
previous breastfeeding experience or socio-economic factors. They also commented
on the likely "Hawthorne effect"; that participating in a study might modify
participants' behaviour and lead to higher breastfeeding rates in both control and
intervention groups. A third area of concern was that outcome measures used relied
on self-reported behaviour, had rarely been validated, and often did not distinguish
between exclusive and partial breastfeeding. This mirrored concerns expressed by
Auerbach et al26 and the Interagency Group for Action on Breastfeeding30 which
proposed a schema for breastfeeding definitions in 1988 (Figure 1.1.4b).

77
Another useful assessment of research on breastfeeding support is the meta-analysis
which Sikorski and Renfrew199 conducted for the Cochrane Collaboration. They
appraised studies on the effectiveness of extra support for breastfeeding mothers and
criticised the lack of information provided about the content of interventions and the
background care received by comparison groups.
Although both reviews206 199

excluded studies which appeared to have used
inadequate methods of randomisation, both did include quasi-randomised studies with
treatment groups determined by the time of delivery, tossing a coin, drawing a
numbered ticket or whether the consent form had an odd or even number. All the
studies were conducted on single sites and many had small sample sizes and relied on
just one nurse or counsellor to deliver the intervention. However, those involved in
small studies may have a particular enthusiasm that others who are asked to adopt the
intervention may not share. This raises questions about whether it is appropriate to
generalise from the findings.
1.5.6 Meta-analysis of studies of the effectiveness of additional support.
The publication of a systematic review of extra support for breastfeeding mothers199 is
important for other researchers because it provides a systematic and independent
assessment of the research to date. It provides a benchmark against which studies can
be assessed. However, it also makes the work of subsequent researchers harder,
because there is little value in reassessing studies which have been subjected to the
scrutiny of a systematic review if the aim is merely to find errors in the reviewer's

78
work. It is a paradox that systematic reviews, based on the skills of critical appraisal,
may make those very skills redundant for the majority of practitioners.
There are however dangers if the scientific community succumbs to the temptation of
over-reliance on meta-analyses, particularly when considering health promotion and
human behaviour. Firstly, the decisions that reviewers take on which studies to
include not only reflect their own values and professional perspectives, but also serve
to define a topic. Secondly, because much of what individual researchers learn is
discarded, the composite intervention reported may tell us little about how to proceed,
even if it does give a useful indication about whether or not we should be doing
something.
In their review, Sikorski and Renfrew make a distinction between "education" and
"support", which may not be valid, given the paucity of information about the content
of most interventions. Most reports do not reveal how much of the work of the
lactation nurses and counsellors was educational, but as health education is a
significant component of health promotion, it may have been artificial to exclude it.
However, as a consequence of this, they excluded studies such as Bolam's 1998 BMJ
paper207 which reported no benefit from individually-delivered postnatal health
education in Nepal.
It is also worth noting that the review includes studies that assess the impact of
additional support on breastfeeding initiation, as well as maintenance of
breastfeeding. Whilst this provides a larger sample, it is questionable because the

79
processes women go through in deciding whether to breastfeed appear different from
those involved in its maintainance.
Sikorski and Renfrew199 identified 13 studies which met their quality criteria and
were able to include 3,616 women in their meta-analysis. They found that extra
support for breastfeeding mothers was effective in increasing the proportion of
women breastfeeding exclusively to two months, (Relative Risk for stopping
exclusive breastfeeding 0.83, 95% CI: 0.72 to 0.96), and in reducing the proportion
who gave up breastfeeding before two months, (RR for stopping breastfeeding 0.74,
95% CI: 0.65 to 0.86). This implied that nine women would need to be given extra
support to enable one extra mother to breastfeed to two months (95% CI: 6 to 21).
They were unable to detect significant benefit beyond two months, and suggested that
this may have been because most interventions were concentrated in the early post-
partum period, with subsequent contact mainly by telephone. Those strategies that
relied on face-to-face contact appeared more effective than those based primarily on
telephone contact.
A detailed analysis of each of the studies included in Sikorski and Renfrew's review is
reproduced in appendix H. These are based on the summaries prepared by the
reviewers, but also include some additional information from the original reports and
interpretation.

80
1.5.7 Implications for study design from published literature
Reviewing the published literature on interventions to support breastfeeding offers an
opportunity to improve the design and conduct of subsequent trials. Indeed, arranging
the studies in chronological order highlights the lessons which more recent
investigators have learnt from earlier studies. In designing the intervention study
reported in this thesis, I sought to learn as much as possible from the experience of
others, but it is worth noting that this study was designed in 1995, before either
Tedstone et al,206 or Sikorski and Renfrew199 had published their reviews. In
particular, the methodological concerns raised highlight the importance of obtaining
statistical advice in designing a randomised controlled trial.
The main methodological issues raised by previous research on breastfeeding support
are discussed in sections 1.5.7a - 1.5.7h below and are summarised in table 1.5.7.
Table 1.5.7
Methodological issues raised by previous research on breastfeeding support.
Factors relating to the design and conduct of the trial

• Sample size
• Randomisation bias
• Bias from differential follow-up
• The implications of a high breastfeeding rate in the control group.
• Lack of independent assessment of outcome
Factors relating to the intervention

• Generalisability of interventions delivered by a single lactation consultant
• Confusion between promoting initiation, or maintainance of breastfeeding
• Inadequate description of the intervention

81
1.5.7a Sample size:
Few studies included sample size calculations and, apart from Brent et al208, which
had important methodological problems and Haider et al 209, which was conducted in
a hospital setting with sick children, no study with a sample size below 650 reported a
significant result. An adequate sample size was therefore seen as essential. The
calculations used to determine this are reported in section 3.2.6.
1.5.7b Randomisation bias:
It was surprising that five of the thirteen studies included in the Cochrane review did
not rely on secure randomisation procedures. Insecure methods included allocation
by time of day, by alternating recruitment weeks on the postnatal ward, tossing a coin,
numbered consent forms and drawing numbered tickets. Many other studies were
excluded because of less secure methods of obtaining comparison data, such as
describing a retrospective survey as a "control group", or allocating subjects to the
treatment or intervention groups depending on the availability of a counsellor.210
Because of these concerns, secure randomisation procedures were adopted for the
study, as described in section 3.2.9.
1.5.7c Bias from differential follow-up:
Seven of the thirteen studies in the Cochrane review achieved follow-up rates of 90%
or above, four had lower rates and two studies did not provide information on the
proportion lost to follow-up. This may however have been a significant source of bias,
because women who stop breastfeeding may be less likely to provide follow-up data.

82
A number of steps were taken to minimise this risk in the design of the present study.
These included asking women to give their telephone numbers and permission for the
NHS to release details of a new address if they moved. Additionally, those who
planned to move away before their babies were four months old were excluded from
the study. This idea was taken from Redman et al,211 although that study did have
particular difficulties and only managed to follow up 66% of participants.
1.5.7d The implications of a high breastfeeding rate in the control group:
The rate of breastfeeding in the control group is a key factor in determining the
sample size needed to show an effect, and the numbers needed rise as the
breastfeeding rate in the control group rises, because more women need to receive the
intervention to reach adequate numbers of those who might stop without it. A
number of studies in the Cochrane review reported feeding at three months and for the
six studies that did so and recruited women who either intended to breastfeed or had
begun breastfeeding, the rates in the control group were, 66%212, 64%213, 57% 214
,
53%215, 49% 216 and 48%217.
The implications of this are that it is easier to get an adequate sample size in a
population who are less likely to breastfeed. Because of this, particular effort was
made to recruit practices in deprived areas and women who had successfully breastfed
before were excluded from the study.
1.5.7.e Independent assessment of outcome:
In six of the studies in the Cochrane review, outcome was not assessed independently

83
of the intervention. This observation raises the possibility of bias if mothers wanted
to please the person who had delivered the intervention, and also questions about how
data was gathered on women in the control group and what contact they had with the
person delivering the intervention.
To minimise the risk of bias, all follow-up data in the study now reported was taken
from questionnaires completed by women themselves, or by telephone calls from the
researchers. Information provided by counsellors was only used to describe the
content of the intervention delivered.
1.5.7f Generalising from interventions delivered by a single lactation consultant:
Apart from Barros et al's larger study215, in most studies, the interventions were
delivered by a single lactation consultant, which raises questions about the extent to
which it is appropriate to generalise from the findings. This supported the case for a
larger multi-centre trial employing a number of counsellors to deliver the
intervention.
1.5.7g Is the focus on initiation or maintainance of breastfeeding?
Two studies in the review included encouragement of women to initiate
breastfeeding, as well support to maintain breastfeeding. These seem rather different
objectives and it is questionable whether they should be investigated in the same
study. The intervention study in this thesis focussed on maintainance of
breastfeeding.

84
1.5.7h Describing the intervention:
Whilst all studies describe the intervention planned, fewer provide adequate
information about the support women actually received. It is also clear that different
studies offered support at different times, antenatally, in the early postnatal period in
hospital and subsequently at home. To clarify the support that women actually
received, the questionnaires designed for the present study included specific questions
about support they received from a range of sources. Counsellors were also asked to
complete records of each contact with mothers.
1.5.8 Recent studies of breastfeeding support
Since the publication of the Cochrane review, a number of further studies have been
207
published. Bolam et al found that postnatal health education had no impact on
infant feeding behaviour in Nepal. In contrast, Haider et al (2000)205 found that
mothers supported by community-based peer counsellors who visited 10 to 15 times,
were significantly more likely to breastfeed (70%, 202/228 of intervention and 6%,
17/285 of control mothers breastfed exclusively at five months; difference = 64%,
95% CI 57% - 71%, p<0.0001). Working in Mexico City, Morrow et al 133 conducted
a randomised trial of normal care, and either three or six visits by peer counsellors. At
three months, 67% of six-visit, 50% of three-visit and 12% of control mothers
practised exclusive breastfeeding (intervention vs controls, p<0.001; six visit vs three-
visit, p=0.02.) Fewer infants in the intervention group had episodes of diarrhoea.
218
In another study in Mexico City, Langer et al found that the presence of a female
companion (doula) during childbirth increased the prevalence of exclusive, but not

85
full breastfeeding at one month. (Intervention prevalence for exclusive breastfeeding
12.3% vs 7.5% for controls; RR 1.64, 95% CI 1.01-2.64). (37.4% vs 36.3% for full
breastfeeding: RR 1.00, 95% CI 0.82-1.22)
The results of the cluster-randomised PROBIT study22 which recruited 17,046
mothers in Belarus have been reported in section 1.1.3a. Although the study focussed
primarily on intervening systematically, based on the Baby Friendly Hospital
Initiative, rather than offering particular individual support, its findings are important.
The study found significant increases in the prevalence of any breastfeeding at 12
months and of exclusive breastfeeding at three and six months.
In contrast, Morrell et al219 found that intensive postnatal support by a community
midwifery support worker did not influence breastfeeding rates in Sheffield, UK.
How should we interpret these recent studies? Firstly, it is striking that they are larger
and appear to have addressed many of the methodological concerns raised in previous
work. Both Haider et al (2000)205 and Morrow et al 133

evaluated quite intensive
support by peer counsellors and in Morrow's study, the number of visits made a
significant difference to the outcome.
It is however important that the results should be viewed in their cultural context, as
McInnes and Stone132 argued in response to Morrow's results. Whereas in Mexico,
breastfeeding is highly valued and initiation rates are high, they argued that in
Glasgow's socially disadvantaged urban estates the few mothers who wish to

86
breastfeed find success elusive. It may be easier to persuade mothers in a pro-
breastfeeding culture to drop supplementary feeds, than to help mothers in a less
supportive environment to continue at all.
1.5.9 Summary
This chapter considers studies of the effectiveness of breastfeeding support. Whereas
earlier studies had significant methodological problems, recent studies have adopted
more robust designs. A meta-analysis conducted published in 1998 found that
breastfeeding support was effective at increasing the proportion of women
breastfeeding, both exclusively and at all, to the age of two months. However the
results of intervention studies may depend on the level of support provided to control
women and whether the prevailing climate of opinion is favourable to breastfeeding.
Because of this, it may not be appropriate to assume that results found in one context
are necessarily applicable in other settings.

87
CHAPTER 1.6
EVIDENCE BASED PRACTICE IN SUPPORTING BREASTFEEDING MOTHERS
1.6.1 Introduction
Evidence based management of breastfeeding relies on both an understanding of the
physiological of lactation, and evidence of the effectiveness of different approaches to
breastfeeding problems. As has been discussed in chapter 1.4, women’s attitudes,
self-esteem and the social environment in which they live play a key role in
breastfeeding, but breastfeeding is also a practical skill. This chapter focuses on the
practical aspects of helping mothers establish breastfeeding and overcome problems
they may encounter.
1.6.2 Physiology of lactation
An understanding of the physiology of lactation underpins the management of
breastfeeding, but a full account of this would be beyond the scope of this thesis.
Lawrence’s text87 includes a thorough review of the topic and it is also covered in a
number of other publications220;221.
During pregnancy, a number of sex hormones contribute to the development of breast
tissue in preparation for breastfeeding. After the delivery, levels of inhibitory
hormones – oestrogen and progesterone fall and prolactin is released in response to
feeding.
1.6 Literature Review: Evidence based practice in supporting breastfeeding mothers 88

Prolactin plays a key role in initiating milk production, but autocrine control becomes
more important later. This involves milk that remains in the breast after feeds acting
locally to inhibit the production of more milk. This mechanism allows the baby to
regulate milk production according to his or her needs: the more the baby takes from
the breast, the more the mother makes.222;223 Thus effective suckling and drainage of
the breast stimulate lactation.
Oxytocin is released rapidly by the pituitary in response to suckling (and other stimuli
such as the baby’s cry) and mediates the “let down reflex”, releasing milk stored in
the alveoli.
Ineffective suckling may lead to milk production being inhibited and women feeling
they have insufficient milk. Factors that interfere with effective suckling include poor
positioning at the breast,224 225

rigid feeding regimes,201 the use of supplementary
formula feeds23 and women feeling reluctant to feed due to sore nipples. (The nature
and management of the “perceived insufficient milk syndrome” is discussed further in
section 1.6.4d below.)
1.6.3 Prevalence of common breastfeeding problems{128}
National data on the prevalence of breastfeeding problems is available from the
quinquennial infant feeding surveys1 (table 1.6.3). While in hospital, 35% of mothers
experienced feeding problems, the most common of which was difficulty getting the
baby to suck or latch on. During the early weeks at home a similar proportion, (35%
of those who were breastfeeding when they left hospital) experienced feeding

problems, the commonest of which was the baby being hungry or the mother feeling
she had insufficient milk. These results will be compared later with the prevalence of
feeding problems in the two studies reported in this thesis (Sections 2.3.6 and 3.4.10).
Table 1.6.3
Feeding problems experienced by breastfeeding mothers
Percentages are of mothers breastfeeding (Great Britain) 1
Problems in hospital Problems after

% of those initially
breastfeeding
leaving hospital
% of those breastfeeding
when left hospital
Baby would not suck / rejected breast 17% 7%
Mother had sore or cracked nipples 8% 14%
Baby was hungry / insufficient milk 6% 16%
Baby was ill 5% 1%
Baby falling asleep / slow feeder / poor weight gain 4% 4%
Mother found breastfeeding uncomfortable 2% 1%
Baby didn’t like milk 1% 1%
Baby vomiting 1% 2%
Baby had wind <1% 2%
Other problem affecting mother 2% 4%
Other problem affecting baby 1% 1%
1.6.4 Evidence based management of breastfeeding
Evidence based medicine has been defined by Rosenberg and Donald226 as “the
process of systematically finding, appraising and using contemporaneous research
findings as the basis for clinical decisions.” Much of the initial impetus to develop
Evidence Based Medicine came from the work of Archie Cochrane,227 after whom the
Cochrane Collaboration was named. One of the first fields of medicine to adopt
evidence based practice has been maternity care, an approach reflected in the

publication of Effective Care in Pregnancy and Childbirth. This includes chapters on
228
“Supporting parents and promoting attachment”, “Establishing and maintaining
breastfeeding” 229 and “Common breastfeeding problems” 225. The Cochrane Library
includes a number of reviews of relevance to infant feeding practice, but some are
based on only a few studies, which limits the value of the meta-analyses they report.
Renfrew et al221 have reviewed practices which promote or inhibit breastfeeding,
while Vallenas and Savage230 have drawn on a wide range of observation and
intervention studies to review the evidence base for the WHO/UNICEF Ten Steps to
promote breastfeeding (Appendix G).
Although the different reviews have taken different approaches, they have largely
achieved a consensus about the principles of managing breastfeeding. It would be
beyond the scope of this thesis to re-examine the evidence for, or against specific
practices, but some of the main themes identified by these reviews are now
highlighted. They offer a standard against which advice and support reported by
women in the studies reported in this thesis can be compared.
1.6.4a Antenatal preparation
Initiatives to help women prepare for breastfeeding have adopted both educational
approaches and physical techniques to “preparing the nipples” for breastfeeding.
Educational approaches such as groups that focus on teaching positioning may be
effective231, partly by increasing women’s confidence, but it may be harder to engage
women from disadvantaged groups in antenatal classes. (In the 1995 national survey1,

among mothers of first babies, 87% of mothers whose partners were in Social Class I
attended antenatal classes, compared with 62% of those classified to Social Class V.)
Physical approaches involving rolling the nipple, or applying creams have not been
shown to be effective232 233 221 and in one important study of antenatal preparation for
women with inverted nipples, the proposed intervention appeared to deter women
from attempting to breastfeed234. Renfrew et al221 suggested that the ‘Niplette’
device, designed to prepare inverted nipples by suction, should be evaluated in a
randomised controlled trial.
1.6.4b Establishing and maintaining breastfeeding
Evidence from a range of sources221 230 suggests that good practice in establishing and
maintaining breastfeeding includes:
- early skin-to-skin contact
- the avoidance of unnecessary delay to the first feed (see section 1.5.4)
- support for early feeds from a professional or other experienced person
- flexible timing and duration of feeds, in response to the baby’s needs
- avoidance of supplementary feeding.
1.6.4c Pain-free effective feeding
Many women experience sore nipples, commonly due to poor attachment and
ineffective sucking. In 1945, Gunther235 showed that ineffective feeding may result in
the baby exerting high pressure suction on the breast causing trauma. More recent
work using ultrasound to image the breast during feeds has confirmed the importance

of the baby taking enough of the breast into the mouth to protect the nipple236. That
helping mothers achieve this is part of good clinical practice was shown by Righard
and Alade,237whose small randomised trial found that correcting poor positioning
could enable more women to continue breastfeeding.
A variety of topical preparations have been advocated as treatments for sore nipples,
but there is little evidence to support their use. Chlorhexidine sprays have been
evaluated in two randomised trials221 and although Herd and Feeney238 found that
fewer intervention women discontinued during the first four weeks, few women in the
study had evidence of nipple trauma. Sharp239 has advocated the use of waterproof
barriers to keep skin lesions moist and allow granulation and there may be a case for
further studies to evaluate this. Expressed breast milk has also been advocated.
However focussing on applying creams or other treatments may distract from the
importance of good positioning to feed.
1.6.4d Perceived insufficient milk
There remains considerable uncertainty about both the nature and most effective
management of the “perceived insufficient milk syndrome”. Indeed the term
“perceived insufficient milk” may be controversial and others240 have used the term
“lactation failure”. Woolridge241 has suggested it is hard to distinguish between a
physiological inability to produce enough milk, a transient reduction in milk supply,
(or increase in a baby’s demands) and a mother or caregiver’s anxiety that the mother
may not be producing enough milk. If quite different conditions are grouped together

into a single “syndrome” and offered the same intervention, it is perhaps not
surprising that the results of intervention studies are equivocal.

242
The WHO Update, “Not enough milk” advises health workers to assess the
problem in three steps:
• First, decide whether the baby is getting enough milk or not.
• Second, if the baby is not getting enough milk, decide why.
• Third, decide how to help the mother and baby.
Physiology research has adopted two approaches to assessing milk supply; test
weighing to observe changes in the baby’s weight and breast volume measurement,
both before and after feeds223;243. This has the potential to assess whether mothers are
actually producing insufficient milk, with the subsequent option to “treat” those not
producing enough milk, but reassure those who are.
However research from other perspectives may be equally important in understanding
mothers perceptions of insufficient milk. Using qualitative methods, Dykes and
Williams244 identified that feeding behaviour, socio-cultural, physiological and
psychological factors all interact to contribute to mothers’ perceptions of their milk
supply. Professionals need to be aware of these factors and adopt a holistic approach
if they are to engage effectively with mothers’ concerns.
Encouragement, support and appropriate reassurance play an important role in helping
mothers who feel they have insufficient milk. The aim of management should be to
help them feed effectively, ideally emptying the breast fully to maximise the stimulus

to further milk production. Drug treatments, including sulpiride and domperidone, (to
increase prolactin production), and oxytocin have been tested in small studies, but the
results do not justify their widespread adoption.221
1.6.4e Other feeding problems
Mastitis is an inflammatory condition of the breast, which may or may not be
accompanied by infection. A recent WHO review245 summarises the causes and
management of mastitis and offers advice on distinguishing milk stasis from
infections which require antibiotics. Continuing to feed on the affected breast, with
correction of feeding technique and additional expressing if needed is important, but
women with evidence of infection need antibiotics. Supportive counselling and
symptomatic treatment are also important.
A variety of conditions affecting mothers and babies may impact on breastfeeding and
it is important that those affected should have appropriate advice and support. This is
however beyond the scope of this thesis.
1.6.5 Delivering evidence based breastfeeding support
Over the last twenty years, there have been a number of initiatives23 102 114 116
to
promote evidence based support for breastfeeding and the impact of these has been
discussed in section 1.2.5a. However, producing a policy document is not enough to
change practice and this requires a sustained multi-faceted initiative. This has
116
included initiatives such as the distribution of “Successful breastfeeding” to all
midwives and health visitors and training initiatives such as the programme adopted

by NCT breastfeeding counsellors (Section 1.7.6). Another useful source of
information has been the MIDIRS Midwifery Digest, which is published quarterly
and summarises research findings in an accessible wayi.
Rogers,197 writing about the diffusion of innovations, identified the importance of
opinion leaders in spreading good practice and this approach has been adopted by the
WHO/UNICEF Baby Friendly Hospital Initiative23 and the recent U.K. government
Infant Feeding Initiative, in encouraging local ownership of initiatives to promote
breastfeeding. Interestingly, in a number of areas, breastfeeding counsellors have
played a significant role in local breastfeeding working groups.
1.6.6 Summary
Although gaps in our knowledge remain, there is a substantial base of evidence on
which to base practice in supporting breastfeeding mothers. This also provides a
standard against which professional and lay breastfeeding support can be assessed.
There have been a number of initiatives to promote the uptake of evidence based
practice and the impact of these has been considered in chapter 1.2. Finally, although
the results of clinical trials are of great importance in guiding practice, it is important
that approaches to supporting breastfeeding take into account the many cultural and
psycho-social factors which influence infant feeding behaviour. These have been
discussed in chapters 1.2, 1.3 and 1.4 of this thesis and reflect the context in which
mothers make their decisions about infant feeding.
i
MIDIRS Midwifery Digest is published by Midwives Information and Resource Service, 9, Elmdale Road, Bristol,

BS8 1SL.

Chapter 1.7
COUNSELLING AND BREASTFEEDING COUNSELLORS
1.7.1 Introduction
Because the main focus of this thesis is a study of the effectiveness of support
provided by breastfeeding counsellors, it is important to examine the origins and role
of the breastfeeding counselling movement. This chapter begins with an account of
the history of the National Childbirth Trust and breastfeeding counselling in the UK.
It assesses the contribution made by volunteer counsellors and reports on the training
they receive. It then considers the role of the breastfeeding counsellor more critically,
whether the term "counsellor" is the most appropriate for the role and the
implications of this for research to evaluate the contribution counsellors make.
1.7.2 A historical perspective on breastfeeding support
In traditional societies, most support for breastfeeding mothers has come from family
members whose role is to 'mother' the mother. Urbanisation, the rise of the nuclear
family and separation of the generations have all played a part in cutting new mothers
off from support, traditionally provided by female relatives. A generation of
grandmothers, who were persuaded to bottle feed by formula milk companies103 and
the medical profession's support for rigid feeding regimens,246 have found themselves
inadequately equipped to support their daughters, a point illustrated by infant feeding
surveys over the last two generations. As has been reported in section 1.2.3, in the
late 1940s three surveys104 105 106

reported breastfeeding rates of 57%, 42% and 48%
at three months, but by 1959-60, only 29% of Nottingham women were still
1.7 Literature Review: Counselling and Breastfeeding Counsellors 98

breastfeeding at that stage.108 By 1975, only 24% were breastfeeding at six weeks in a
national survey.102 As most of today's mothers were born between 25 and 35 years
ago,1 this suggests that only a quarter of UK-born mothers have mothers who
themselves successfully breastfed. Oakley153 emphasised the importance of this and
the loss of collective knowledge about breastfeeding in her study of social support and
motherhood.
Into this vacuum, three groups have stepped. As has been shown in section 1.3.3,
there is evidence of the importance of support from husbands and partners for women
who want to breastfeed,158 157 247 248

although many men are uncertain and
inadequately prepared for the role.163 164

Professionals too have made considerable
efforts to remove obstacles to breastfeeding imposed by rigid hospital regimens and to
offer women consistent support. Additionally, voluntary movements have played a
key role, often formalising the mother-to-mother support that has always existed.
These include the La Leche League, founded in Illinois in 1956, which now has
around 4,000 groups in 43 countries,249 the Association of Breastfeeding Mothers, the
National Childbirth Trust and more recently the Breastfeeding Network. These groups
combine individual support with a political role to promote breastfeeding and
although there may sometimes be conflicts between these roles, their codes of conduct
emphasise a non-directive approach (appendix I).
Although often arising outside formal health services, voluntary movements received
considerable support from the Baby-Friendly Hospital Initiative, launched in 1991 by
the World Health Organisation and UNICEF134 (appendix G). The 10th step in the

initiative recommends that Health Services should, "Foster the establishment of
mother support groups and refer mothers to them…" Increasingly, health services
have seen peer counselling programmes and voluntary sector contributions as a means
to achieve this.250 The potential for voluntary initiatives to make a real difference to
the uptake of breastfeeding is perhaps best illustrated by the contribution of the
Norwegian Breastfeeding Association, Ammehjelphen which has been referred to in
section 1.2.7c.
1.7.3 History of NCT Breastfeeding Counsellors
The training and recruitment of NCT breastfeeding counsellors had its origins in a
meeting of the then Natural Childbirth Association, (later renamed National
Childbirth Trust) in 1967. Breastfeeding in Britain had reached its nadir and
midwives were losing their traditional skills at helping mothers. The challenge was
clear, and the new subcommittee, later renamed the Breastfeeding Promotion Group,
was established with a grant of £25.00 and a remit of "helping and encouraging
breastfeeding ... in co-operation with the medical profession." 251
The group was encouraged by an American woman, Ruth Wilf who was an active
member of the International Childbirth Education Association and had spent time
helping women at Mile End Hospital in East London.252 Vanessa Redgrave spoke at a
conference in 1968, talking about "Breastfeeding and a Career." Over the next few
years, a number of study days were organised, so that by 1972 there were over 170
trained breastfeeding counsellors. By 1975 this number had doubled and in the

1980s, the NCT Annual Reports253 referred to around 500 counsellors. By 1997,254
this had further increased to 621.
Although it had no direct impact on the studies reported in this thesis, a rift opened
within the National Childbirth Trust in 1997. Writing in the Observer newspaper,
Mills255 reported concerns that the trustees had accepted sponsorship from Sainsbury's
- a supermarket chain that sold its own brand of infant formula. Additionally,
members were concerned that a proposed increase in membership subscriptions255 to
£36.00 annually would exclude the less well-off and "confirm its reputation as a
largely white middle class preserve." 1,760 members signed a letter of protest.
Following a heated debate at the 1997 NCT Annual General Meeting, a number of
breastfeeding counsellors decided to start a new organisation, the Breastfeeding
Network (BfN).
The establishment of the Breastfeeding Network has highlighted a number of issues.
The crisis arose from attempts by the trustees to raise money to support a more
"professional" mode of working, but this ignored the values of the volunteers who had
contributed so much to the organisation. It demonstrated the commitment many
breastfeeding counsellors have to supporting women from all social groups and to a
political approach to promoting breastfeeding. Another interesting observation is that
the Breastfeeding Network dropped the term "counsellor", preferring instead the term
"Breastfeeding Supporter". The issues behind this are discussed further in section
1.7.8.

1.7.4 The contribution of breastfeeding counsellors
The 1996 NCT Annual Report256 lists the activity of breastfeeding counsellors in
more detail. In that year there were 549 registered counsellors, supervised by 64 local
tutors and 380 women were training to become counsellors. They reported 20,032
antenatal contacts, usually with women attending NCT antenatal classes, and 33,411
postnatal contacts. 35,392 women telephoned them and they visited 6,228 at home.
While there may be some double counting in these figures, they do convey the
workload of these volunteer counsellors and suggest that each has around 200
contacts with mothers a year. Individual counsellors are phoned at home about 65
times a year and visit women on average once a month. Comparing the 35,392
women who telephoned a counsellor with the 732,049 babies born in the United
Kingdom in 1995,257 suggests that 4.8% of mothers telephoned a counsellor that year.
Two recent national surveys have also assessed the support provided by voluntary
organisations such as the National Childbirth Trust, Association of Breastfeeding
Mothers and La Leche League. In 1990,258 7.5% of mothers reported that they had
received help from a voluntary organisation, whereas in 1995,1 this figure had risen to
10%.
In 1998, the Breastfeeding Network established a Supporterline, which connects
mothers directly to a local breastfeeding supporter. Although organised nationally, the
service is only promoted in areas where supporters are available, but as the service is
extended to other areas, the numbers making contact may be expected to rise.

1.7.5 How women contact a breastfeeding counsellor
Women make contact with breastfeeding counsellors in a number of ways. For some,
the first contact is through NCT antenatal classes and although these are led by NCT
antenatal teachers, breastfeeding counsellors usually attend the session on
breastfeeding. Others hear of them from professionals involved in their care and many
maternity departments provide counsellors' telephone numbers on leaflets. NCT
headquarters provides details of local counsellors and others make contact via the
Breastfeeding Network Supporterline.
Increasingly, counsellors have set up initiatives to reach women from manual social
class groups who are more likely to have difficulties with breastfeeding. In
Portsmouth, breastfeeding counsellors visit the postnatal wards on a regular basis,
liaising with the midwives who suggest mothers for the counsellor to see.259
Similarly, one counsellor visits women in Holloway Prison.260
1.7.6 Training and conduct of breastfeeding counsellors
The National Childbirth Trust has established a rigorous training programme which
aims to ensure that counsellors are seen as committed volunteers, with a high level of
knowledge, rather than as well-meaning amateurs with an axe to grind.259 261
Counsellors must have themselves breastfed for at least six months and need to be
nominated by their local NCT group. The training usually extends over a two-year
period and includes study days and monthly meetings with an experienced tutor.262
Trainees learn about counselling and the importance of a non-directive approach, as

well as more specific information about breastfeeding and common breastfeeding
problems. Before they qualify, they submit written case studies, which need to be
approved by an independent tutor. In addition to the initial training, established
counsellors are expected to meet regularly and attend three study days every two
years.
A detailed Code of Conduct underpins the counsellor's role and addresses issues such
as confidentiality, the importance of supervision and of maintaining good relations
with health professionals.263 Sections 1.4 and 1.5 of the Code, which emphasise the
importance of taking a non-directive approach and respecting the wishes of mothers
who decide to bottle feed, are reproduced in figure 1.7.6 and the full code is included
in Appendix I.
Fig 1.7.6
235
NCT Breastfeeding Counsellors' Code of Conduct.
"A counsellor should avoid being directive in her approach to mother, but should offer
information and practical suggestions to enable a mother to make an informed choice.
The aim of a counsellor is to increase a mother's confidence in her own mothering abilities
and to enable her to make her own decisions. A counsellor should take care to
differentiate between reasonable support and intrusion." (Section 1.4)
"A counsellor should respect the decision of a mother not to breastfeed and remember
that a mother who turns reluctantly to bottle feeding may need continuing sensitive, caring
support in coming to terms with her decision." (Section 1.5)

1.7.7 Who becomes a breastfeeding counsellor?
Although between five and ten percent of mothers turn to them for support, little has
been published about who becomes a breastfeeding counsellor and there has been no
formal evaluation of their effectiveness. Most counsellors become involved through
a local NCT group, but some are also health visitors or midwives who want to learn
more about supporting breastfeeding mothers. For many women, local NCT groups
offer an opportunity to make new friends, at a time when the arrival of children
triggers a shift in their priorities from work to home. Sharing the expectations and
experiences of caring for a new baby provides an introduction to peer support and
some women go on to become more active in their local group, or train as a
breastfeeding counsellor.
A commitment to breastfeeding, which for many counsellors is derived from their
own breastfeeding experience, appears to be an important factor in the motivation to
become a counsellor, a suggestion which is supported by the results of American
research. In a qualitative study of 17 low income, culturally diverse women who were
supported by peer counsellors, Locklin264 265 266

found that the perception of
successful breastfeeding could empower women. She identified five themes: Making
the Discovery, Seeking a Connection, Comforting Each Other, Becoming Empowered
and Telling the World. This last theme, summed up as including, "women's
expressions of pride about breastfeeding successfully, persevering under sometimes
trying circumstances, and feeling a commitment to tell others to do the same," may
explain why some decide to become breastfeeding counsellors.

The two-year training to become a counsellor selects women who are prepared to
make a significant commitment, but also engages them in a process of change from
being a mother who is keen to help others, into a skilled volunteer, working within a
supportive, but ideally non-directional framework. There are elements of
professionalisation in the process, so that whilst they are still "peer counsellors," in
their role within the NCT, increasingly they are going beyond this and supporting
women in other settings.
87
Lawrence's invaluable text includes a discussion of the characteristics of a good
counsellor and stresses that, "being a sympathetic listener is the most important
quality" (p 642). Although there are differences between the organisation of
breastfeeding support in the United States and Britain, the six abilities she identifies
of a good counsellor are relevant in any setting (fig 1.7.7). She recognises that while
training is important, not everyone has the personal qualities to make a good
counsellor.
Fig 1.7.7
Abilities breastfeeding counsellors should have
- To truly understand
- To avoid judgement
- To understand other life-styles
- To admit it when they do not know
- To seek appropriate help form professionals
- To recognise incompatibility in a given relationship
86
Source: Lawrence, 1994 (p642)

1.7.8 An examination of the counsellor's role and its implication for research.
The range of terms adopted by women who support others with breastfeeding may
reflect some uncertainty about the role being described. When the Natural Childbirth
Association, (the precursor of the NCT) was first established, the first counsellors
were known as "breastfeeding helpers", but later the term "breastfeeding counsellor"
was adopted. As has been mentioned, when the Breastfeeding Network was
established in 1997, the organisation adopted the term "breastfeeding supporters",
suggesting a less neutral stance. The La Leche League uses the term "La Leche
Leader", while in the United States, the title "lactation consultant" reflects the more
commercial approach adopted. Another term in widespread use is "peer counsellor",
adopted when volunteers are trained to support mothers from their own
communities.267
The Concise Oxford Dictionary268 defines the term "counsellor" as "a person who
gives counsel; an adviser," or as "a person trained to give guidance on personal,
social or psychological problems". However as reported in section 1.7.6, the NCT
Breastfeeding Counsellors' Code of Conduct emphasises that counselling should be a
non-directive activity.
One of the most thoughtful assessments of the contribution of breastfeeding
counsellors is Mary Smale's 1996 D Phil thesis.269 This was based on her experience
as a counsellor over a ten-year period and in it she adopts a qualitative approach to
explore the meaning behind the calls she received. She suggests that for many
women, the comments or actions of those who had attempted to support them thus far

had caused as much distress as any problems inherent in breastfeeding. She argues
for a woman-centred approach and sees counselling as enabling women to put their
own principles into practice.
This issue of whether breastfeeding counselling can be non-directive, is more than a
semantic discussion. Whilst this probably does apply when considering psychosocial
contacts, it may not when mothers have concerns of a practical or physical nature,
when mothers may interpret counsellor's suggestions as advice. Also, given the
personal commitment counsellors make to promoting breastfeeding, it is hard to
perceive them as neutral, however much effort they make to adopt a non-directive
stance with individual women.
The explicit objective of offering non-directive counselling to help mothers achieve
their infant feeding goals also raises issues for attempts to evaluate its effectiveness.
Put starkly, is it appropriate to evaluate non-directive counselling using an outcome
measure such as duration of breastfeeding? Surely, the only valid outcome would be
one that reflected mothers' satisfaction and perceptions of their ability of achieve their
goals. But to accept this approach would be to deny the main justification for health
services to be concerned about infant feeding; that it makes a difference to human
health. Indeed, to accept this would also invalidate other evaluations of non-directive
counselling which have accepted that people adopting a non-directive approach can
still be concerned about whether people feel better or worse as a result of their
intervention.270 271 272

A definition suggested by the British Association for Counselling may help take this
discussion forward. They suggest that counselling aims "to provide an opportunity
for the client to work towards living in a way he or she experiences as being more
satisfying or resourceful."273 Inherent in this is the concept that counselling may
empower people to achieve their goals and therefore that it is appropriate to use
duration of breastfeeding as an outcome measure in a study of the effectiveness of
counselling. The complexity of the interaction between counsellors and mothers, and
the range of factors which have been shown to influence infant feeding behaviour do
however require that any study assesses the impact of counselling from a range of
perspectives.
1.7.9 Summary
Since 1967, breastfeeding counsellors have played an increasing role in supporting
mothers. The NCT annual reports suggest that around five percent of mothers receive
support from a counsellor, while a national survey of mothers suggests the figure is
twice that. There are however a number of potential conflicts in the role, with some
taking a "professional" role, while others seek to remain peers. Similarly, there may
be conflicts between promoting breastfeeding and non-directive approaches to
counselling. If however the aim of counselling is to strengthen women's ability to
achieve their goals, this suggests a way that counselling may work if it is effective.
This account of the history, training and role of breastfeeding counsellors provides a
background to the intervention assessed in study two.

PART TWO
STUDY ONE: A PROSPECTIVE STUDY OF MOTHERS' ATTITUDES TO

AND EXPERIENCES OF BREASTFEEDING

STUDY ONE: A PROSPECTIVE STUDY OF MOTHERS' ATTITUDES TO
AND EXPERIENCES OF BREASTFEEDING
Chapter 2.1
RATIONALE AND OBJECTIVES OF STUDY ONE
2.1.1 Introduction
This first study was conducted between 1983 and 1984 and provides information
about the epidemiology of breastfeeding and levels of breastfeeding support available
at the time. However it is noteworthy that patterns of infant feeding have changed
little since then, as shown earlier in table 1.2.3. There may however have been
changes in professional practice, in the light of initiatives such as the Baby Friendly
Initiative23 and other "supply interventions" discussed in section 1.2.5a.
It is clear from the literature reviewed in chapters 1.2 and 1.3 that social influences
play a key role in infant feeding behaviour and one of the main aims of this study was
to explore the extent of this. By doing this in the context of routine primary care, it
was hoped that the study would identify factors of relevance to clinical practice.
This study aimed to investigate why women choose whether to breast or bottle feed,
their experiences and the role of health professionals in advising them. Whereas most
studies had recruited women in hospital, or from birth notifications, this study was
organised prospectively, to avoid the initial experience of infant feeding modifying
their responses. In a study of first-time mothers, Wright and Walker 124 had suggested
2.1 Study one: Rationale and objectives 110

that midwives could predict who was most likely to stop breast feeding and it was
hoped to investigate this in the primary care setting.
2.1.2 Specific objectives
a) To find out women's reasons for choosing to breast or bottle feed.
b) To measure the incidence and duration of breastfeeding in the study
population.
c) To identify factors which could be used to predict early discontinuation of
breastfeeding.
d) To identify the reasons women give for stopping breastfeeding during the first
six months postnatally.
e) To determine the prevalence of common breastfeeding problems.
f) To find out what advice women receive from health professionals for common
breastfeeding problems.
2.1 Study one: Rationale and objectives 111

Chapter 2.2
STUDY ONE: METHOD
2.2.1 Recruitment
Pregnant women who were registered with 24 Nottingham general practitioners
during the study period in 1983-84 were recruited to the study by their community
midwives. This was usually at a routine antenatal home visit at about 25 weeks
gestation and the midwives interviewed women using the first questionnaire at that
time. Subsequently, they were interviewed by their health visitors six weeks and then
six months after the birth of their child. The six participating practices or health
centres were distributed throughout Nottingham, but the main criterion in recruiting
sites was the willingness of both midwives and health visitors to participate in the
study. All pregnant women registered with the practices were eligible for inclusion,
but those who were subsequently found to have moved away before birth were
withdrawn from the study.
2.2.2 Questionnaire design
Following a procedure agreed with the Nottingham Research Ethics Committee,
questionnaires were numbered, but women's names held on a separate register for
each practice or health centre. Structured questionnaires were used which had been
piloted in one of the practices before the start of the study and are included in
appendix A. The questionnaires included closed questions about women's social and
family circumstances and how long they had fed for, but open questions were used to
ask about their choice of feeding method and problems that they encountered. The
2.2 Study one: Method 112

women's social class was coded by her partner's occupation, using the Registrar
General's tables.274 The reasons for adopting this approach are discussed in section
3.2.9a below.
2.2.3 Statistical analysis
The chi square test275 was used to determine which factors were significantly
associated with women stopping breastfeeding before six months. These potential
predictors were then considered further, using logistic regression, within the
SPSS/PC+ package,276 to determine which were most strongly associated with
discontinuing breastfeeding by six weeks. This allowed each predictive variable to be
considered simultaneously and took into account the interactions between them. The
analysis was done in stages and at each stage, the variable least strongly associated
with stopping, (using the Wald criterion), was removed from the analysis. This
process, called backwards elimination, continued until all the remaining variables
were significantly associated with stopping breastfeeding at the 5% level.
2.2 Study one: Method 113

Chapter 2.3
STUDY ONE: RESULTS
2.3.1 Study population
A total of 514 women were recruited to the study and completed the antenatal
interview. 491 (96%) were interviewed again at six weeks and 463 (90%) at six
months. By six months 51 women had moved away or were lost to follow up for
some other reason.
Of the 514 women, 213 (41%) were having their first child and 301 (59%) already
had children. At the time of the antenatal interview, 83% were married, but a further
7% were living with the baby's father. The family's country of origin was given as the
United Kingdom by 92% of the women.
Sixty two per cent of the study population were over 25, compared with 56.5% of the
birth notifications to the Office of Population Censuses and Surveys, (OPCS) for
277
Nottingham. The study population also included fewer unmarried women, and
rather more married women who had already had children, as shown in table 2.3.1.
This may have led to the study population having a slightly higher breastfeeding rate
than was the case for all births in Nottingham in 1983.
2.3 Study One: Results 114

Table 2.3.1
Age, parity and marital status of study population, compared with 1983 Birth
Notifications for Nottinghami
(Figures are shown as percentage of 514 women in study and 7,614 birth notifications)
Married: Unmarried: Total:

Primiparae: Multiparae:
Under 25: Study population 13.8% 12.1% 11.7% 37.5%
OPCS data 14.5% 13.3% 15.9% 43.7%
25 and over: Study population 16.9% 40.5% 5.1% 62.5%
OPCS data 16.0% 34.2% 6.1% 56.3%
All women: Study population 30.7% 52.5% 16.7%
OPCS data 30.5% 47.5% 22.0%
2.3.2 Feeding plans
By the time they were interviewed, most women had already decided how they would
feed their babies. Indeed, 56% (289/514) had made their decision before they became
pregnant. A total of 359 women (70%) intended to breast feed, 120 (23%) intended to
bottle feed and 35 (7%) were undecided. Only 26 women later changed their plans
which suggests that their decisions were relatively fixed.
The reasons women gave for their choice of feeding method are shown in table 2.3.2.
Additionally, all women were asked which type of feeding would be best for their
baby. Of those planning to bottle feed, 86 (72%) said breast feeding would be best,
21 (18%) said bottle feeding and 13 (11%) were uncertain. This suggests that they
i
Figures in italics are percentages of the 7,614 births recorded by OPCS Survey District 52A (Nottingham) in
1983. In the data provided, parity is only available for married women and the study data have therefore been
presented in the same way for comparison. Data provided by the Office of Population Censuses and Surveys,
Titchfield, Fareham, Hants. PO15 5RR

accepted breastfeeding would be best, but chose not to because they disliked the idea,
or anticipated problems with breastfeeding.
Table 2.3.2
Reasons given by women for their choice of feeding method
% of women
Reason: giving reason
Women intending to breastfeed (n = 359)

Better for baby 66
Natural 24
Convenient / easy 20
Emotionally satisfying 16
Breastfeeding satisfactory previously 11
Cheap 9
Better for mother 7
Other 12
Women intending to bottle feed (n = 120)

Disliked the idea of breastfeeding 34
Convenient 30
Expected, or previously had problems with breastfeeding 25
Bottle feeding previously satisfactory 13
Personal preference 11
Other 11
n = total number of women in group. NB some women gave more than one reason.
2.3.3 Incidence and duration of breast feeding:
Although 359 of the 491 women interviewed at six weeks (73%) had begun
breastfeeding, by that stage, only 243 (49%) were still giving any breast feeds. By six
months, only 122 of the 463 women interviewed (26%) were breastfeeding.

Of the 205 mothers of first babies interviewed at six weeks, 165 (80%) had
commenced breastfeeding, compared with 193 of the 286 who already had children
(67%). However, at six months 42 of the 189 mothers of first babies interviewed
(22%) were still breastfeeding, compared with 73 of the 272 who had previously had
children (27%). These figures are illustrated by figure 2.3.3 which shows that first-
time mothers were more likely to attempt breastfeeding, but less likely to succeed.
F ig 2 .3 . 3 D u r a t io n o f b r e a s t f e e d in g
100
90
80
70
Percentage of women
60
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
W e e k s f r o m b ir t h
P r im ip a r a e M u ltip a r a e
2.3.4 Factors which might predict early cessation of breastfeeding
Women who intended to breast feed, or were undecided, were studied further to see if
it was possible to predict who would stop. In the first stage of the analysis, their
responses at the antenatal interviews were cross-tabulated with whether they were
giving any breast feeds at six weeks. Those responses that, on chi squared testing

(24), were found to be significantly associated with stopping breastfeeding before six
weeks are listed in figures 2.3.4a. Details of the wording of these questions and the
full range of questions asked at the antenatal interview are available by reference to
the antenatal questionnaire, provided in appendix A.
Fig 2.3.4a
Antenatal responses which were significantly associated with stopping
χ2 > 3.84, p < 0.01).
breastfeeding before six weeks (χ
Mothers of first babies: (n = 178)

Age 25 or under
Social Class IIIM, IV, V, Unemployed
Age of leaving school 15 or 16
When decided to breastfeed "During pregnancy", rather than "before"
Considered bottle feeding? Yes
Intended duration Less than 4 months
Own feeding as an infant "Bottle"
Mothers with previous children: (n= 216)

Age 25 or under
Social Class IIIM, IV, V, Unemployed
Age of leaving school 15 or 16
When decided to breastfeed "During pregnancy", rather than "before"
Considered bottle feeding? Yes
Intended duration Less than 4 months
How family usually feed "Bottle"
Partner's preference "Bottle" or "uncertain"
Confidence "Fairly" or "rather anxious"
Previous child's feeding "Bottle" or "breast for less than 6 weeks"
The results of the second stage of the analysis, in which the potential predictors were
considered together using logistic regression, are shown in table 2.3.4b. For
primiparae, two variables remained after removal of those that did not significantly
contribute to predicting who would stop. Women in manual social class groups were
more than three times more likely to stop than those in non-manual groups. Similarly,

those who had considered bottle feeding were more than twice as likely to stop as
those who had not. For multiparae, the most effective predictor was how women had
fed their previous child followed by the age of leaving school and considering bottle
feeding.
Table 2.3.4b
Effectiveness of predictors of stopping breastfeeding by six weeks
Odds ratioa Sensititivity Specificity

Predictor 95% CI % %
Mothers of first babies (n = 178)

Social class IIIM, IV, V or unemployed 3.68 (1.44 to 6.20) 77 46
Considering bottle feeding 2.40 (1.23 to 4.70) 60 62
Mothers with previous children

(n = 216)
Previous child bottle fed, or breastfed
for less than 6 weeks 5.15 (2.53 to 10.66) 70 75
Age of leaving school 15 or 16 years 3.80 (1.60 to 8.99) 84 41
Considering bottle feeding 3.38 (1.62 to 7.02) 63 64
a
CI = confidence interval, n = total number of women considering breastfeeding. Odds ratio is the probability that
women giving the response will stop breastfeeding by six weeks, compared with that of women not giving that
response.
Other measures of the effectiveness of a predictive tool are its sensitivity, that is the
percentage of those who will stop that the test detects, and its specificity, the
percentage of those who will continue to breastfeed that the test excludes. These are
shown in table 2.3.4b for the predictors that the logistic regression analysis suggested
were the most useful.

2.3.5 Reasons for discontinuing breastfeeding
Women who discontinued breastfeeding were asked why they stopped and it was
noteworthy that the vast majority had done so because they had difficulties feeding.
The commonest reason they gave, both during the first six weeks and subsequently,
was that they had insufficient milk. Sore nipples, engorgement and other breast
problems were common reasons for stopping in the first six weeks, but less common
later. It is also of interest that more women stopped because they themselves were ill
than because their baby was unwell.
Table 2.3.5
Reasons given by women for stopping breastfeeding
% of women giving reason

Stopped before Stopped between
six weeks (n = six weeks and six
Reason 109)a months (n = 109)b
Not enough milk 51 64

Breast problems 22 10
Feeding problems 15 14
Disliked breastfeeding 14 1
Maternal illness 10 14
Baby ill 6 5
Social factors or planned to stop 6 34
a b
N = total number of women in group. Seven women did not give a reason. Eleven women did not
give a reason. NB: Some women gave more than one reason
2.3.6 Common feeding problems
All women who had given any breast feeds were asked about common feeding
problems to determine their incidence and both the source and nature of the advice
they received.

2.3.6a Perceived insufficient milk supply
Of the 359 women who had begun breastfeeding, 166 (46%) had felt they did not
have enough milk at some stage during the first six weeks and 56 gave this as a reason
for stopping. When asked why they felt they had insufficient milk, 144 women (87%)
said it was because their baby was unsettled, while 18 (11%) said their breasts had felt
less firm and 17 (10%) were concerned because their baby's weight gain was poor.
Those who were concerned about their milk supply were asked why they thought this
had happened to them. Seventy six (46%) of the 166 women said they did not know
why they did not have enough milk, but 24% blamed it on being tired or overworked.
Some felt they were incapable of making enough milk, while others blamed their diet
or illness. A few thought it was because the baby had been ill or refused to feed.
Ninety six (58%) of the women who felt they had insufficient milk sought advice
from a health professional (table 2.3.6a).
Table 2.3.6a
Health professionals consulted by women with perceived insufficient milk
(n = 166 women)
Number of women % of women who felt
consulting they did not have
enough milk
Community midwives 56 34%
Health visitors 53 32%
Hospital staff 16 10%
General practitioners 5 3%
n = Number of women with perceived insufficient milk. NB some women consulted more than one
health professional.

The advice they received was: to persevere or feed more often (47 women, 49% of
those seeking advice), to give a bottle as a supplement or substitute (40, 42%) and to
rest or drink more (21, 22%). Eighteen women (19%) received reassurance or support
and eight reported that they had received advice on breastfeeding technique (8%).
Eighty five of the 166 women who felt they had insufficient milk (51%) gave bottle
feeds in response to the problem and of these, only 38 (45%) were still breastfeeding
at six weeks. In contrast, 67 (85%) of the 81 who did not give bottle feeds when they
felt they had insufficient milk were still breastfeeding at six weeks.ii
2.3.6b Sore nipples
One hundred and fifty eight women, (44% of all those who began breastfeeding) had
sore nipples during the first six weeks and 21 gave this as a reason for stopping. Of
the 158 women, 134 (85%) sought professional advice, the source of which is listed in
table 2.3.6b.
Table 2.3.6b
Health professionals consulted by women with sore nipples
(n = 158 women)
Number of women % of women who
consulting had sore nipples
Community midwives 77 49%
Health visitors 13 8%
Hospital staff 46 29%
General practitioners 30 19%
n = Number of women with sore nipples. NB some women consulted more than one health
professional.
ii
Some caution is necessary in interpreting this because giving a bottle feed is not only a response to a problem
such as perceived insufficient milk, but also a measure of outcome as stopping breastfeeding almost inevitably
involves switching to bottle feeding.

The advice they received was largely physical in approach. Ninety two of the 134
women who sought advice from a health professional (69%) were advised to use
creams, 38 (28%) to use a disinfectant spray and 16 (12%) to use a breast shield.
Other breast care was recommended to 15 women (11%), nine women were told to
persevere or feed more often and seven (5%) reported that they had been given advice
on feeding technique.
2.3.7 Summary
• Although 73% of women began breastfeeding, this had fallen to 49% by six weeks
and 26% by six months.
• Logistic regression showed that mothers who had previous children who had
bottle fed, or stopped breastfeeding their previous child during the first six weeks
were 5.15 times more likely to stop breastfeeding in the first six months than those
who had previously breastfed for six weeks or more.
• First-time mothers in manual social class groups were 3.68 times more likely to
stop than women in higher social classes, while those who said antenatally that
they had considered bottle feeding were 2.40 times more likely to stop.
• Most women gave an inadequate milk supply as a reason for stopping, but sore
nipples were also a common problem in the first six weeks.

Chapter 2.4
STUDY ONE: DISCUSSION
2.4.1 Strengths and limitations of the study
This study set out to investigate women's choices of infant feeding method, their
experiences and the role of health professionals in advising them. Although some of
the findings, such as the incidence and duration of breastfeeding, echo those of the
national infant feeding surveys,200 100

the study was also able to assess the potential
for primary care professionals to engage with women on infant feeding. The
prospective design reflected the real situation midwives might encounter if they
wished to identify antenatally those who might need more support. Additionally,
because women's antenatal responses to questions about their feeding intentions were
not clouded by their subsequent experiences, they are likely to reflect their decisions
more accurately than could have been obtained using a retrospective questionnaire
after birth. The primary care setting also contributed to the high follow-up rates of
96% at six weeks and 90% at six months.
A number of methodological issues merit consideration in interpreting the results.
Whilst the prospective, primary care-based design had several advantages, women
who were not visited in the antenatal period were not included and this may explain
why the study included fewer younger and unmarried women than expected from the
birth notifications for Nottingham.277 This may have led to an overestimate of the
incidence and duration of breastfeeding amongst mothers of first babies, but is less
2.4 Study One: Discussion 124

likely to have influenced the multivariate analyses to identify predictors of early
breastfeeding cessation.
The questionnaires were adapted from that used in the national infant feeding
surveys,98 99 and although they were piloted before use, no attempt was made to assess
their reliability or validity. The mode of completion also deserves mention, as it is
possible that the fact that women completed questionnaires during routine
consultations with midwives and health visitors may have influenced their responses.
With hindsight, it was apparent that the questionnaire had focussed on "advice", rather
than other forms of support and also that it did not enquire about mothers' contacts
with breastfeeding counsellors. It therefore seems likely that women were more likely
to report contacts with professional than lay supporters. Despite these reservations,
the fact that the questionnaire relied largely on "open" questions makes it less likely
that the questionnaire design influenced women's responses.
A further issue that merits consideration is that the study was conducted in 1983.
Since then, a number of initiatives, such as the World Health Organisation's "Baby
23
Friendly Hospital Initiative" and publication of "Successful Breastfeeding" by the
Royal College of Midwives116 have set out to modify professional practice. Whilst
there has been little change in infant feeding rates between 1980 and 1995, there is
some evidence, reported in section 1.2.5a of changes in professional practice since the
study was conducted.

2.4.2 Implications for policy and practice
2.4.2a Identifying women who need most help
The results of study one are relevant to health professionals working in both primary
care and hospital settings. The finding that many women had decided how they would
feed before becoming pregnant suggests that during the antenatal period,
professionals have limited influence on women's choice. It would therefore seem
sensible for midwives to concentrate on helping those who want to breastfeed.
Women who chose to bottle feed did so largely because they disliked the idea of
breastfeeding. However, of those planning to bottle feed, 72% believed that
breastfeeding would be best for their baby. This suggests that professionals who want
to overcome women's objections to breastfeeding should promote it as natural and
convenient, rather than just as better for the baby.
It is possible to identify women who are more likely to stop breastfeeding. Mothers of
previous children who were considering breastfeeding, but had bottle fed, or stopped
breast feeding their last child before six weeks, were five times more likely to stop.
First-time mothers from manual social class groups were nearly four times more
likely to stop than those from non-manual groups, echoing the findings of the
quinquennial national surveys reported in section 1.2.61. The simple question; "Have
you considered bottle feeding?" was also shown to be valuable and by asking it, 60%
of the first-time mothers who would stop could be identified and 62% of those who
would continue to breastfeed could be excluded. It may not be possible for midwives
to assign women to social class groups without using the Registrar General's tables274

278
and therefore asking women whether they have considered bottle feeding may be
more useful in routine antenatal care. However, the observation that social class was
related to stopping breast feeding does have important implications for the allocation
of resources and suggests that those organising antenatal classes should target the less
well-off.
Identifying women who are likely to need more help would enable members of the
primary care team to make a particular effort to establish a supportive relationship
with them during the antenatal period. The purpose of this would be to offer
education and encouragement, whether by giving time to discuss breastfeeding,
providing literature, or introducing them to other mothers through antenatal classes or
a support group.
The rapid decrease in the numbers of women breastfeeding in the first few weeks after
birth and the observation that most women who discontinue do so because of feeding
problems suggest that women who want to breast feed need more help.
2.4.2b Perceived insufficient milk
Problems such as not having enough milk or sore nipples were strikingly common.
Concern about inadequate milk supply was the commonest reason given for stopping
breastfeeding, but there has been some debate about what this really means.109 279
Women appear to become concerned because their babies are unsettled, but that might
not necessarily mean they are hungry. More controversially, Newson & Newson
suggested that women sometimes gave physical problems as a justification for

stopping when the underlying reason was their ambivalence about breastfeeding.109
However in more recent work, Duckett et al188 found that when they evaluated a
model based on the Theory of Planned Behaviour, perceived insufficient milk was not
related to other attitudinal or normative factors tested (Section 1.4.4b). This suggests
that perceived insufficient milk is a phenomenon mothers experience, rather than just
a proxy for other attitudes. In 1996, Hill and Humenick280 reported the validity and
reliability of an instrument to assess perceptions of insufficient milk supply
Although it can be measured, the perception of insufficient milk has complex origins,
as Dykes and Williams discussed in qualitative work on this.244 They concluded it
was underpinned by a complex and synergistic interaction between socio-cultural
influences, feeding management, the baby's behaviour, lactation physiology and the
woman's psychological state.
In physiological terms, (as has been discussed in section 1.6.2), lactation is stimulated
by the infant sucking at the breast; if the baby is hungry and sucks for longer, the
breast will be emptied more effectively and produce more milk. "Not enough milk"
can therefore also been seen as a stage in the feedback between mother and baby. For
this mechanism to be effective the baby needs to be able to take milk from the breast,
which is dependent on effective positioning at the breast.224 In a small, but important
study, Righard and Alade237 demonstrated that correcting poor positioning ("nipple
sucking") led to a significant increase in the numbers of women breastfeeding at four
months.

The concept that effective suckling stimulates lactation may be one of the most
important things for professionals to convey. However the finding that 42% said
they had been advised to either supplement or switch to bottle feeding would seem to
undermine this, even if some were asking for endorsement of a change they had
already decided on. This also highlights the danger of women getting conflicting
advice, which could be avoided if team members were to discuss their approach to
common problems more often.
It was also striking that few women who felt they did not have enough milk had any
idea why this had happened. This suggests that few were aware of the physiological
processes that regulate lactation and points to a gap which health education should
seek to address.
2.4.2c Management of sore nipples
Women’s reports of the advice they received for sore nipples convey a remarkably
consistent picture of the advice professionals in Nottingham gave in 1983 – 84. The
overwhelming majority were advised to use creams, sprays or other topical breast care
and only five percent reported that they had been given advice on feeding technique.
The focus was on responding to skin trauma, rather tackling the cause of the problem.
This can be contrasted with the evidence presented in section 1.6.4 a and c, that the
main focus of the management of sore nipples should be to help the baby take enough
of the breast into his or her mouth, thereby feeding effectively and avoiding high
pressure suction and trauma. Writing in the Nursing Times in 1991, Hulme258

described how “unnecessary annointing” failed to address the cause of sore nipples
and in recent years professional advice has emphasised positioning, rather than topical
treatments.
2.4.3 Feeding technique – the missing link
Perhaps the most striking omission from the advice reported by mothers was advice
on how to breastfeed. Women who felt they did not have enough milk were either
encouraged to persevere and feed more often, or given permission to supplement with
bottle feeds, while those with sore nipples were offered creams, sprays or other breast
treatments. But very few mothers said they had been shown how to breastfeed
effectively.
This conclusion can be contrasted with the results of qualitative analyses of the
information, advice and support women value most, which are presented later in this
thesis. It also underlines the importance of the initiatives to promote evidence based
practice23;114;116 which have been developed since this study was conducted.
2.4.4 Implications for study two
This study demonstrated how those women who are most likely to have difficulties
with breastfeeding can be targeted during the antenatal period and offered more
support in the first few weeks after birth. This finding was used to increase the power
of the subsequent intervention study by excluding mothers who had previously
breastfed for six weeks or more.

Study one also showed that excellent follow-up rates could be achieved in primary
care and this approach was also adopted in the randomised trial. The questionnaires
and coding frames used to code open questions were shown to work reasonably well
and they were adapted for the subsequent study.
Knowing who is most likely to benefit from health promotion interventions is
valuable, but it is also essential that the interventions offered should be shown to be
effective and meet women's perceived needs. These issues were considered further in
the two further studies reported in this thesis.

PART THREE
STUDY TWO: A RANDOMISED CONTROLLED TRIAL OF SUPPORT FROM

BREASTFEEDING COUNSELLORS
132
STUDY TWO: A RANDOMISED CONTROLLED TRIAL OF SUPPORT FROM
BREASTFEEDING COUNSELLORS
Chapter 3.1
RATIONALE AND OBJECTIVES OF STUDY TWO
3.1.1 Rationale and selection of intervention for study two
Study two was conceived some years after the completion of study one, but it was
designed to build on the experience gained in it and evaluate an intervention which
might address the rapid decline in breastfeeding rates identified in that and other
studies. Whereas observation studies, such as those reported in chapter 1.2 can help
explain who breastfeeds and why, they are less good at assessing the effectiveness of
interventions. As Stephenson and Imrie281 argued, random assignment of individuals
to different treatment groups is the best way to avoid false conclusions, whether for
behavioural interventions or drug therapies.
Selecting the intervention for study two involved some thought and discussion.
Although initially I had considered evaluating a video to promote breastfeeding, a
small pilot study revealed that the best available film was one which had been had
been sponsored by a formula milk manufacturer, undermining its message. In
subsequent discussions, it became clear that volunteer counsellors were playing a
major role in providing breastfeeding support to between 5% and 10% of mothers, but
that this had not been evaluated. It was thought that if the intervention were shown to
be effective, this might encourage more women to train as counsellors and encourage
the NHS to promote their work. Making counselling more widely available might
enable more women to breastfeed.
3.1 Study two: Rationale and objectives 133

In setting up the trial, it became clear that it was not possible to evaluate the work
counsellors were already undertaking - because it would not have been ethical to ask
women not to approach a counsellor. The trial could only evaluate extending their
role to women who were not already approaching them for support. Because of this,
when women were recruited, they were asked if they planned to contact a counsellor
and those who did plan to do so were recruited to a parallel observation study.
Study two was designed between 1994 and 1995, with women recruited between
April 1995 and August 1998i. The initial stages of study design were supported by a
Research Training Fellowship from the Royal College of General Practitioners and
the study itself was funded by a grant from the NHS North Thames Responsive
Funding Scheme. A part-time research assistant, Jane Taylor, was employed from
April 1996 and undertook many of the practice visits from then on. Additionally, a
part-time data entry clerk was employed from June 1997 to handle the large numbers
of questionnaires being returned. The study office was based at Statham Grove
Surgery, which received funding as a research practice, initially from the RCGP and
subsequently from the NHS R&D Programme.
3.1.2 Aims
The main aims of study two were to find out whether women offered additional
support from a breastfeeding counsellor breastfed for longer and were more satisfied
i
Although 32 practices eventually took part in the study, they were recruited in phases. During 1995, only
Statham Grove Surgery recruited, which provided an opportunity to refine the study administration before
involving other practices.

with their experiences. Additional aims were to find out more about women who
contacted breastfeeding counsellors and the support they received.
3.1.3 Primary objective
To determine whether women who were offered additional support from a
breastfeeding counsellor breastfed for longer than those who were not (Sections 3.5.3,
3.5.4, 3.5.6 and 3.5.8).
3.1.4 Additional objectives
The randomised controlled trial
a) To determine whether women who were offered additional support from a
breastfeeding counsellor introduce bottle feeding later than those who were
not (Sections 3.5.5, 3.5.7 and 3.5.8).
b) To determine whether women who were offered additional support from a
breastfeeding counsellor were more satisfied with their experiences than those
who were not (Sections 3.5.9).

c) To determine whether women who were offered additional support from a
breastfeeding counsellor had fewer feeding difficulties than those who were
not (Section 3.5.10).
Comparisons with the observation group

d) To compare socioeconomic and other characteristics of women who decide
antenatally to see a breastfeeding counsellor with women who do not decide
this (Section 3.3.6).
e) To compare the feeding behaviour of women who decide antenatally to contact
a counsellor with women who do not decide this (Section 3.5.11).

Contacting a counsellor
f) To determine characteristics of women who contact breastfeeding counsellors
postnatally (Section 3.4.3).
g) To determine whether women had any difficulty contacting breastfeeding
counsellors postnatally (Section 3.4.4).
h) To determine whether women who contact breastfeeding counsellors
postnatally find them helpful (Section 3.4.5 and 3.4.6).

Chapter 3.2
STUDY TWO: METHODS
3.2.1 Outline of study two
The study was conducted in primary care and women were recruited when they
attended for antenatal care between 28 and 36 weeks of pregnancy. Women who
planned to breastfeed or were undecided, who were having their first child, or had not
previously breastfed for more than six weeks were eligible. Those who intended to
contact a breastfeeding counsellor were excluded from the randomised trial, but were
followed up to observe their experience of breastfeeding. These formed the
'observation' group.
Eligible women were allocated to the control or intervention group according to
numbered randomisation codes kept in sealed envelopes at the study office and
notified to women in writing. Women in the intervention group were visited
antenatally by one of the breastfeeding counsellors, who also offered support
postnatally, either by telephone or at further home visits if appropriate.
The main outcome measures were duration of breastfeeding, maternal satisfaction
with breastfeeding and the incidence of feeding problems. Women were asked to keep
a weekly diary card for the first four months and complete questionnaires when they
attended for check-ups or immunisations at six weeks and three and four months. The
study design is summarised in Figures 3.2.1a and 3.2.1b.
3.2 Study two: Methods 137

Fig 3.2.1a
Flow chart describing progress of participants through trial
Early Women registered for

pregnancy maternity care
Potential recruits
28+ wks Assessed for eligibility
Recruited to
Recruited to RCT
Observation study
Randomisation
Observation Control Intervention
Counsellor
-Birth- Support
Observation -6 wks- Control Intervention
Observation -3 mths- Control Intervention
Observation -4 mths- Control Intervention

Fig 3.2.1b
Summary of methods
Setting: 32 practices in & around London

Recruitment in antenatal clinics
Inclusion criteria: Intention to breastfeed, or undecided

Not previously breastfed 6 weeks
English speaker
No plan to move away
No plan to contact breastfeeding counsellor anyway
Safe for counsellor to visit
Not previously included in study
Randomisation: Sealed envelopes in study office
Intervention: Antenatal home visit

Postnatal support if requested
Follow up: Questionnaires at 6 weeks, 3 months, 4 months

Diary cards
Calls to non-responders
Outcome measures: Duration of breastfeeding, (Survival analysis)

Maternal satisfaction with breastfeeding
Feeding problems
Satisfaction with support & advice
Qualitative analysis: Mothers' comments

Content of counselling
3.2.2 Recruitment of women to the study
Women recruited to the study were all registered with one of 32 participating general
practices and identified from the records that practices used to claim for maternity
care. Each month, practices were asked for the names and expected date of delivery of
expectant mothers and these were used to prepare the baseline antenatal
questionnaires. The researcher then returned to place these in the medical records
when each woman was around 28 weeks pregnant, checking she had not left the
practice, miscarried or had a termination of pregnancy. If she had, she was excluded
at this stage.

The first sheet of the baseline antenatal questionnaire, (reproduced in appendix C),
enabled the doctors and midwives to check whether women were eligibile for the
study. Relevant response boxes had letters beside them and women ticking any box
labelled with the letter “c” were ineligible for the randomised controlled trial.
However, those who had only ticked box “3c”, indicating that they planned to contact
a breastfeeding counsellor, were eligible for the observation study. The second sheet
was designed to gather demographic data on all women, whether recruited or not. A
second envelope was attached to the antenatal questionnaire for the doctor or midwife
to recruit eligible women. This contained instructions for the professional, the
mother’s information leaflet and consent form (Appendix C).
The envelopes were left sticking out of the medical notes to prompt the receptionists
to give women the questionnaires when they attended the antenatal clinic. They were
asked to complete these while waiting and to give them to the doctor or midwife to
check.
3.2.3 Inclusion criteria:
Women who were considering breastfeeding and met the inclusion criteria were
eligible for the study (Fig 3.2.1b). Those who had breastfed their previous child for
more than six weeks were excluded, because study one171 had shown they were more
than five times more likely to breastfeed to six weeks than those who had not.
Excluding them focussed the study on those who were more likely to discontinue

early and reduced the sample size required by reducing the expected breastfeeding
rate in the control group.
It would have been preferable to have included non-English speakers but both
counsellors and women would have faced difficulties in communicating by telephone.
Additionally, the effort involved in translating questionnaires and arranging
interpreters for the range of languages needed in different parts of London precluded
this.
Women who planned to move away before four months after the birth were excluded
to avoid difficulties following them up.i Additionally, the practice staff were asked
not to recruit women if they felt it might be unsafe for a counsellor to visit them
alone. Although this criterion was rarely used, it provided some protection for the
counsellors. Women who had been in the study during a previous pregnancy were not
eligible for recruitment a second time.
Women who planned to contact a breastfeeding counsellor were excluded from the
controlled trial because it was not considered ethical to ask them not to do so, but
were invited to join the observation study and report their experiences.
3.2.3a Premature labour
Premature labour posed a problem that had not been anticipated when the study was
designed. When about 50 women had been recruited, it became apparent intervention
i 209
Redman et al excluded women who planned to move away in their Australian trial and although they had
other difficulties with follow up, it seemed sensible to not recruit women who would soon move away.

group women who delivered prematurely might do so before they met the counsellor.
If they were included and an “intention-to-treat analysis” performed, this might
underestimate the benefit of counselling, but to withdraw only intervention women
might have left more women who delivered prematurely in the control group and
biased the results. Because of this, it was decided to include those who delivered
prematurely and ask the counsellors to contact them postnatally if they had not met
before the birth.
However, it was recognised that it might not be appropriate to analyse data from very
premature women who could not have received the intervention and that a decision to
exclude both intervention and control women might need to be taken at the analysis
stage. The gestation selected for this could then be based on the proportion of
intervention women who were not seen antenatally at particular weeks of pregnancy.
3.2.4 Sample Size:
In study one, only 50% of women who would have been eligible for inclusion in this
study were still giving breast feeds at six weeks. The sample size was calculated
using the Statcalc programme in Epi info282, assuming 95% confidence limits and
80% power for an unmatched case control study. This assumed that women were
allocated equally to control and intervention groups and that, as in the previous study,
5% were lost to follow-up by six weeks.
The sample size calculation suggested that to demonstrate an increase in the
proportion breastfeeding at six weeks to 60%, would require 854 women, 427 of

whom would receive counselling. If twenty counsellors were to take part, they would
need to see around 21 women each.
In the initial sample size calculations, it was estimated that 50% of women who
completed the antenatal questionnaire would be eligible for the study, suggesting that
practices would need to screen 1,708 women. This was based on the results of the
previous study,171 which were similar to those of the 1990 national infant feeding
survey101. The main assumptions made were that 23% would be ineligible because
they intended to bottle feed; 55% of mothers who were considering breastfeeding
would have had previous children and of these, 40% would have not previously
breastfed for six weeks, excluding 22% of women. Additionally, it was assumed that
7% of women would be excluded for other reasons, an estimate that was to prove
optimistic.
3.2.5 Recruitment of Practices and Counsellors:
Setting up this study involved recruiting both practices and counsellors in and around
London and required considerable time and effort. Although most of the counsellors
approached were keen to participate, they needed to have completed their training and
to be able to arrange cover if they were on holiday or otherwise unavailable.
The choice of practice was important, but proved to be problematic because the
practice population would be a major factor in determining the breastfeeding rate of
the control group. As noted in section 1.5.7d, the higher the breastfeeding rate, the
larger the sample size needed to show any benefit from the intervention. Because of

this, most of the practices approached were in mixed or deprived, rather than affluent
areas. Practice organisation was a further factor, because they needed to see most of
their mothers for antenatal care and the baby’s six-week check if they were to recruit
and follow up women. Ideally, practices would be conveniently sited for counsellors
to do home visits and should not have given breastfeeding support a high priority in
the past.
Potential practices were identified either by personal contacts or at the suggestion of
local counsellors. When contacting new practices, the best approach appeared to be
to telephone at the end of morning surgery and speak to one of the doctors. They were
then sent information about the study and if interested, a meeting to explain the study
was arranged with the doctors, midwives and receptionists. Although practice
managers were very helpful if contacted first, the doctors seemed less likely to agree
to participate if they had not been contacted personally at the outset.
During the course of the study, it became clear that fewer women than expected were
being recruited and as a result, further practices were approached. This meant that
practices were recruited in phases. Around forty practices were approached and the
majority agreed to participate. However a few declined, usually because they were
too busy, but in one case because they felt a counsellor had been over-zealous in
encouraging a woman whose baby was not gaining weight in the past. Several
practices were already making significant efforts to promote breastfeeding, which
would have obscured the effect of the intervention and others had high breastfeeding
rates. Sometimes the way maternity care was organised would have made it difficult

for practices to follow the study protocol, particularly when midwifery teams working
on a geographical, rather than practice-attached basis provided most maternity care.
Decisions not to participate were usually made informally, although on three
occasions potential difficulties only emerged when the practice was visited.
3.2.6 The intervention:
Women in the intervention group were to be visited antenatally and offered postnatal
support, either by telephone or at further home visits if needed. Although the
intervention was intended to reflect the way breastfeeding counsellors normally
support women, some differences were inevitable because of the context of a research
project. This raised a number of issues, which were discussed with counsellors and
the NCT Breastfeeding Promotion Group nationally.
Normally, counsellors see women who have sought them out, either through NCT
antenatal classes, or postnatally for help with feeding problems, whereas in this study,
those who planned to contact a counsellor were excluded. This provoked discussion
because some counsellors were unhappy that the study should only assess the
effectiveness of counselling for women who would not normally have contacted
them. This did however reflect the study’s focus on those who were more likely to
discontinue early.
A few counsellors had concerns about visiting women antenatally, rather than
meeting them in an antenatal class. This was partly unfamiliarity, partly because
women would not get the support of the class and partly because it implied the study

might not assess their normal way of working. However, they recognised that home
visits were essential to establish relationships with women and that it would not be
enough to meet only those who were motivated to attend an antenatal class.
The arrangements for women to contact counsellors postnatally were also discussed in
detail. Although initially, it had been planned that counsellors would visit routinely,
the NCT Breastfeeding Promotion Group felt counsellors should not impose a
postnatal visit on women, or contact them uninvited. In the event, it was agreed that
counsellors would ask women at the antenatal visit whether they wanted to be
contacted, or left to call themselves. However as the study progressed, it became
apparent that some were reluctant to ask for help and the counsellors were asked to
encourage women to contact them postnatally, to say how they were getting on, rather
than only if they had problems.
Counsellors were asked to offer women the same support they would have normally
done and to follow their usual code of conduct (Appendix I). Although they were not
given guidelines on managing particular problems, they were asked to complete a
record form that had been prepared with the help of two counsellors in Hackney. This
was designed to reflect their normal practice and captured descriptive data on the
contacts, from the counsellors’ perspective.
At the antenatal visit, counsellors gave women a calling card, an introductory letter
and two leaflets; "Breastfeeding; a Good Start", published by the NCT283 and

"Breastfeeding - Your questions answered", published by the Health Education
Authority.284
3.2.7 Outcome Measures:
The main outcome measures were duration of both full and partial breastfeeding,
mothers’ satisfaction with the experience of breastfeeding and the incidence of
feeding problems. Additionally, women contacting the counsellor postnatally were
asked to score whether or not she had been helpful, and all women were asked
whether advice they received was helpful.
3.2.7a Duration of breastfeeding
Mothers were asked to complete a diary card each Saturday and questionnaires at 6
weeks, 3 months and 4 months. The diary card was included to provide more accurate
information about when mothers changed how they fed, to allow the use of survival
analysis. Previous work by Launer et al285 had demonstrated that maternal recall of
infant feeding events was accurate, but Freer286 had found that health diaries were an
effective way of gathering contemporaneous information. (See appendix C for the
questionnaires and diary cards.)
Women were followed up for only four months for three reasons. Firstly, the
counselling focused mainly on the antenatal period and helping women establish
breastfeeding so any effect was most likely to be during the first few months.
Secondly, the most rapid decline in breastfeeding is during the early weeks. The third

reason was that four months coincides with children's third immunisations, which was
a convenient time to issue follow-up questionnaires.
3.2.7b Maternal satisfaction with breastfeeding.
If the aim of counselling is to support women who want to breastfeed, then it is
important to evaluate not only the duration but also the quality of their experiences of
breastfeeding. Because of this, the six-week questionnaire included questions about
satisfaction with breastfeeding, details of which are given in section 3.5.9.
3.2.7c Feeding problems
Mothers who stopped breastfeeding were asked about their reasons for doing so and at
six weeks all mothers were asked if they had experienced specific feeding problems.
This built on the results of study one but also allowed comparison of whether women
who received additional support responded differently from those who did not.
Feeding problems and advice women received were coded using the coding system
devised for study one.
3.2.8 Selection of statistical tests
The statistical tests employed were discussed with the statistical advisers, and chosen
to compare two independent samples - the control and intervention groups. The main
tests employed were:
• Incidence and prevalence of breastfeeding were compared using Chi2 tests.
• Kaplan-Meier survival analysis was used to compare the duration of
breastfeeding.

• Mann-Whitney U tests were used to compare the non-parametric data on
satisfaction and whether women considered they had feeding problems.
The SPSS package287 was used for data handling and analysis, apart from calculations
of confidence intervals for relative risk performed on the comparisons of incidence
and prevalence of breastfeeding. STATA288 was used for these.
3.2.9 Questionnaire Design:
The questionnaires designed for this trial used questions from a range of sources.
These included the first study in this thesis, the first two national infant feeding
surveys98 99
and the questionnaires that Jones used in her controlled trial of the
effectiveness of a lactation nurse.177 216

A further influence was Rajan's use of open
questions from a survey of pain relief in labour to ascertain women's experiences of
postnatal care.172 This formed the basis of the qualitative study of women's
experiences of breastfeeding support reported in study three. Texts by Sudman and
Bradburn289, Streiner and Norman290 and Fallowfield291 292

were invaluable in
providing a broader perspective, particularly in relation to assessing satisfaction.
Leff et al's work on Maternal Satisfaction with Breastfeeding293 294 deserves particular
mention, because of the systematic approach they adopted, first conducting a
qualitative study, and then basing the questionnaire developed on the themes
identified in the study. Although the 42 item questionnaire was too long to include in
the six-week questionnaire, the satisfaction questions used were drawn from that
work. One potential problem about the applicability of that work was however that

the original instrument was developed in a middle class community in New England,
relatively few of whom had difficulties with breastfeeding.
A range of professional and lay advisors, including breastfeeding counsellors and
members of the National Childbirth Trust Breastfeeding Promotion Group provided
valuable comments and the questionnaire was also piloted with 45 mothers attending
the child health clinic at Statham Grove Surgery in East London. (See appendix B for
further details.)
3.2.9a Coding social class
Social class was coded using the Registrar General's Classification, a system which
classifies the social class of households, based on the occupation of a “household
reference person”. Where the woman had a partner, the partner's occupation was used
to determine her social class group, but where she did not, her own occupation was
used. The issues involved in coding by partner's occupation are discussed in the 1991
OPCS coding guide (Vol 3, page 21).295
This research is essentially concerned with empowering women in their breastfeeding
endeavours and with listening to their perspectives, so it may seem incongruous to use
their partners’ occupations to assign their socio-economic status. There are however
important reasons for doing so. Most expectant mothers do have partners, but many
take a career break after the birth of their first child, which may make it harder to code
their occupational group. The successive national infant feeding studies have also

coded social class by partner's occupation, which allows comparisons with the results
of those studies to be made.
3.2.10 Study Administration, randomisation and blinding
A computerised study management database was written in Ability Plus,296 an MS
Dos programme,ii to automatically generate letters and prompts when questionnaires
were overdue. To ensure confidentiality, the study management database was kept
separate from the SPSS records prepared when questionnaires were returned.
Although the database necessarily included details of the allocation group, this was
not included on printouts the researchers used to follow up non-responders and was
not available when returned questionnaires were coded prior to entry into SPSS. This
ensured that whenever possible the researchers were blind to the allocation group, but
inevitably they sometimes discovered this from women's responses on questionnaires,
or from correspondence with the counsellors.
The study administration is summarised in Figure 3.2.10.
ii
Ability Plus combines spreadsheet , database and word processor functions and although no longer available, I
had previously written a general practice program using it and was familiar with it's macro functions.

Fig 3.2.10
Study Administration flowchart
Lists of eligible women obtained from practices.
Early Women registered for
pregnancy maternity care
Questionnaires placed in patient notes. Researcher checks
woman is still pregnant & registered.
Potential recruits
Receptionist gives woman antenatal questionnaire in
antenatal clinic. Midwife or doctor checks forms and obtains
28+ wks Assessed for eligibility consent if eligible.
Questionnaire and consent form returned to study office &
checked.
Recruited to
Recruited to RCT
Observation study
Randomisation code obtained from sealed envelopes.
(Stratified by practice and parity.) Form issued to counsellor.
Mother informed of randomisation and counsellor’s details if
Randomisation
intervention.
Reminder to counsellor if contact record not returned in 4
weeks
Observation Control Intervention
Practice notifies study office of delivery. (Calls to check if

Counsellor
-Birth- overdue.) Diary card posted to mother. 6 week form and
Support diary 2 inserted into notes. Counsellor informed if
intervention.
Observation -6 wks- Control Intervention 8 week postal reminder. 10 week postal reminder and
phone call if needed.
3 month form and diary 3 inserted in notes for 1st
Observation -3 mths- Control Intervention immunisation if breastfeeding at 6 weeks.
4 month form inserted in notes for 2nd immunisation if
Observation -4 mths- Control Intervention breastfeeding at 3 months or no response. 20 week postal
reminder with telephone follow-up if needed.
Randomisation
If a woman was eligible for inclusion, the doctor or midwife explained the study and
if she agreed, asked her to complete the consent form. The doctor or midwife could
then either telephone the research assistant or reception staff at Statham Grove for
randomisation, or forward the questionnaires for postal randomisation.
Women were individually allocated to the control or intervention group according to
numbered randomisation codes prepared by Prof. Stephen Evans, statistical advisor.
Separate lists of random numbers were kept for primiparae and multiparae within
each practice. The random numbers were computer-generated in balanced permutated

blocks of 6, 8, or 10 numbers to ensure the control and intervention groups had a
similar proportion of mothers of first and subsequent babies from each practice. They
were then kept in sealed envelopes, which were numbered to ensure they were used in
sequence. Each time a woman was randomised, her details were added to the
randomisation slip to avoid duplication.
Although the original intention had been for the midwives or doctors obtaining
consent to telephone for women's treatment allocations, this procedure was dropped
after about 100 women had been recruited. Telephone randomisation had been
adopted to ensure the research assistant could not be influenced by women's responses
to the questionnaire when she opened the randomisation envelopes, but it became
apparent from conversations with professionals recruiting women that this might be a
source of bias. The professionals felt disappointed when women were randomised to
the control group and some wanted to offer them more support, to compensate for
them not receiving counselling. Because of this, the procedure was modified so
women were notified in writing of their treatment allocation. Professionals were not
routinely informed.
Follow up
Counsellors were informed when women were recruited and if they did not return the
antenatal visit record within four weeks, they were sent a reminder. After a woman
was 38 weeks pregnant, the computer prompted the research assistant to contact her
practice to find out whether she had delivered. Following concerns raised by the
counsellors that they were not being informed of births soon enough, if the practice

had not been notified of the delivery within a few days of the due date, the researchers
telephoned the woman directly to find out whether she had delivered. In making these
calls, the researchers were careful not to discuss breastfeeding or be drawn into
conversation. When her date of delivery was known, she was sent her first feeding
diary and, if in the intervention group, her counsellor was informed.
The six-week questionnaire and second feeding diary were placed in the baby's
medical records, but if the questionnaire had not been returned by eight weeks, a
postal reminder was sent to the mother. At nine weeks, the research assistant was
prompted to telephone and ask her to return the forms.iii
Unless the six-week form confirmed that the woman had stopped breastfeeding, the
three-month questionnaire and third diary were issued at 11 weeks. Similarly at 15
weeks four-month questionnaires were issued for women who were still
breastfeeding. The computer also sent postal reminders and prompted the research
assistant to contact women who had not returned the four-month questionnaires.
Even without reminders, the computer routinely generated seven letters for women in
the control and observation groups and nine for those allocated a counsellor.
Although mailing these was a significant daily task, it was visiting the practices, and
iii
The telephone interviews, which were often conducted with women who had discontinued breastfeeding, are
discussed further in section 4.3.1.

following up the non-responders, which were to require most commitment and time
from the research team.

Chapter 3.3
RESULTS OF STUDY TWO: RECRUITMENT AND FOLLOW UP
3.3.1 Introduction and overview:
This chapter describes the recruitment and follow-up of study participants in detail. It
is however also helpful to have an overview of the numbers recruited and followed up
as part of the study. This is provided in figure 3.3.1.i
Figure 3.3.1
Overview of recruitment and follow up
Early 5193 women registered for 829 women no longer pregnant

pregnancy maternity care or had moved away(table 3.3.2)
1925 not assessed for eligibility

4364 potential recruits
(See section 3.3.3)
28+ wks 2439 assessed for eligibility 1420 excluded, of whom:
636 breastfed 6wks previously

370 planned bottle
148 recruited to 221 non-English speakers
720 recruited to RCT
Observation study 153 planned to move
40 other reasons
Randomisation 151 not recruited

(See section 3.3.4)
i
148 obs 357 control 363 interven 5 withdrawals
4 neonatal deaths
(See section 3.3.8)
Counsellor
Support
-Birth- i
12 prem births excluded
(See section 3.3.10)
147 remain 350 remain 350 remain
after birth after birth after birth
i
31 lost to follow-up
144 obs -6 wks- 336 control 336 interven
i
68 lost to follow-up
128 obs -4 mths- 310 control 310 interven
i
In figure 3.3.1, the numbers of women shown as being withdrawn, excluded or lost to follow-up relate to all three
arms of the study – the intervention, control and observation groups.
3.3 Study Two Results: Recruitment and follow-up 156

3.3.2. Study population:
During the course of the study, 5,193 baseline antenatal questionnaires were prepared
for distribution to women, whose details were provided by participating practices.
Because this was based on maternity services claims made early in pregnancy, by the
time questionnaires were issued at 28 weeks, some women had moved and others
were no longer pregnant, as shown in table 3.3.2. Thus, 829 women were no longer
receiving maternity care from the practices, leaving 4,364 potential recruits to study.
The remaining data relate to these women.
Table 3.3.2
Reasons women were no longer receiving maternity care
(% shown is % of 5,193 women who had an initial maternity claim)
Miscarriage or termination 345 (6.6%)
No longer registered at 28 weeks 78 (1.5%)
Later found to have moved away 392 (7.5%)
Expected Date of Delivery incorrect 14 (0.2%)
Total 829 (16%)
3.3.3 Assessment of women to determine their eligibility
Baseline antenatal questionnaires were prepared for the 4,364 potential recruits and
returned by 2,439 of them.ii 720 (29.5%) of these were recruited into the main
controlled trial and 148 (6.1%) into the observation study, because they planned to
ii T
he figure of 2,439 women "returning the questionnaire" refers to those about whom enough was known to
determine their eligibility. It includes 221 women who were excluded because they did not speak English, only
16 of whom actually filled out the questionnaire. Additionally, 100 questionnaires were returned with the reason
for exclusion written on, but otherwise blank, presumably because they had either been completed in
consultations or practice staff felt they should not be approached about the project. They are included within the
2,439 reported as completing the questionnaire and some further information is available in footnote iv below.

contact a breastfeeding counsellor anyway. More information on the reasons for
excluding women who returned the questionnaire is given in section 3.3.5 below.
Fig 3.3.3 - Assessment of eligibility

(4,364 potential recruits)
Questionnaire not issued

by research team
152 (3%) In RCT
720 (17%)
Questionnaire not given

to mother, or mislaid by In observation study
practice 148 (3%)
1375 (32%) No consent
152 (3%)
Premature, hospital care Excluded

or not seen (see table 3.3.4)
397 (9%) 1420 (33%)
3.3.3a Reasons why questionnaires were not returned
Of those who did not return antenatal questionnaires, 397 had not attended their
practice for antenatal care. 78 of these delivered prematurely and were not seen at
the practice between 28 weeks and the baby's birth, 116 received only hospital care
but 203 had no apparent reason for their non-attendance.
Administrative factors at the practices accounted for 1,375 of the questionnaires not
being returned. On checking, the research team found 433 blank forms still in
women's notes, even though they had been seen for antenatal care and a further 766

forms were untraceable, so it was not possible to determine whether or not women
had been asked to complete them. At the end of the study, a decision was taken not to
revisit the practices to retrieve the last 176 forms and it is likely that the majority of
these were not recruited for an administrative reason.
The research team did not issue antenatal forms for 152 women, many of whom had
registered with a study practice at a late stage of pregnancy. Thirty of these were due
to deliver during the summer of 1997 when several counsellors in South and East
London planned holidays simultaneously. Because it would it would not be possible
to offer them counselling, these women were not asked to complete questionnaires.
Only one woman is known to have declined to complete the ante-natal form, although
it is possible that some of those whose forms were mislaid had in fact declined to
complete them.
The number of women who were not assessed for inclusion reflected the difficulties
involved in conducting a multi-centre trial, and although this led to inconvenience and
expense, it is unlikely to have affected the study results. These issues are discussed
more fully in section 3.7.2a, but the practice staff's commitment and understanding of
the study seemed very important for successful recruitment. In particular, recruitment
suffered whenever there was a change of receptionist or attached midwife, or if the
usual midwife was on holiday.

3.3.4 Recruitment by practice
Because factors related to the practice staff were thought to be a major factor in the
recruitment rate, this was examined for each of the practices. The proportion of
women recruited varied from 0 - 31%, with a mean of 19.9%. Inevitably, this
analysis conceals changes over time, as recruitment at a number of sites went through
both good and bad phases, but it does allow some comparisons between practices to
be drawn.
Figure 3.3.4
Recruitment by practice
500
400
300
200
100
Recruited Potential recruits
Throughout the study, each practice’s recruitment rate was monitored and when there
appeared to be problems, the research staff made efforts to discuss them with practice
staff and visiting midwives. From this, it became apparent that the staff at one
practice were unlikely to recruit, partly because they were unable to follow the study
protocol and wanted the midwife to complete the questionnaire with women on home
visits. As a result, we agreed they would withdraw from the study. We also
considered this for two computerised practices, where staff appeared less likely to

respond to on-screen prompts than to forms left in paper records, although in the
event they remained in the study.
The reasons for exclusion varied widely between practices and reflected the
demography of the practice populations. Fuller details are shown in appendix E, table
3.3.4. It was also interesting to note that the non-English speakers were concentrated
in relatively few practices, mainly in East London. Eight practices, where more than
5% of potential recruits were excluded on language grounds, accounted for 159 of the
221 non-English speakers. Similarly, the proportion who were ineligible because they
intended to bottle feed varied widely, but was particularly high in the two South Essex
practices where more than 40% of women completing the questionnaire were
excluded for this reason.
3.3.5 Reasons for exclusion
The reasons why women were excluded are reported in table 3.3.5.
Table 3.3.5
Recruitment and exclusions
Frequency %
Breastfed last child more than 6 weeks 636 26%
Plans bottle 370 15%
Non-English speaker 221 9%
Moving away 153 6%
Inappropriate to visit 32 1%
Previously in study 4 <0.1%
Known fetal abnormality 4 <0.1%
Total Exclusions 1420
Recruited to RCT 720 30%

Not recruited to RCT iii 122 5%
Recruited to observation study 148 6%
Not recruited to observation study 29 1%
Total 2439

Where women were excluded for more than one reason, only one is listed. Thus
although 36 of the women who had previously breast fed for more than six weeks and
22 of those who planned to bottle feed also planned to move away, they are included
in the first two categories, rather than the latter. It is also likely that a number of
those excluded because they did not speak enough English would have still been
excluded on other grounds, had they been able to complete the questionnaire.
Because only multiparae could have been excluded on the grounds of having
previously breastfed, those multiparae who remained eligible for inclusion were a
selected group. Previous research suggests that they were less likely to breastfeed than
the primiparae who were more representative of the population as a whole.
During the course of the study, a number of antenatal questionnaires were returned
without consent forms for women who were eligible for either the RCT or observation
study. Although some had comments indicating that the women had declined to
participate, it seems likely that the majority had not been invited to do so, a suggestion
supported by the finding that when we checked this with a few of these women, they
were happy to participate.
3.3.6 Characteristics of participants:
The next section provides information on the socio-demographic characteristics of

women who returned the questionnaires and allows a comparison between those who
were, or were not recruited to the randomised controlled trial. This descriptive data on
the study population may also be used for comparison with other studies and in assessing
the extent to which the results may be extrapolated to other populations.
iii
The women who were "not recruited" include both those who were not invited to participate, and those who
declined to do so. See last paragraph of section 3.3.4 for further information on this.

Tables 3.3.6 (a - f) provide data on those who actually completed the questionnaire,
but not all those who were excluded did so. Because of this, further demographic data
is unavailable for 205 women who did not speak enough English to complete the
questionnaire, or the 100 whose forms were returned blank, apart for the reason they
were not eligible for the studyiv. The total number of women recorded as returning
the baseline antenatal questionnaires differs slightly between the tables, because a few
women missed out questions when completing their forms.
Table 3.3.6a
Parity of women assessed for eligibility
Observation Not recruited

RCT to RCT Not iv All women
Excluded
study recruited to assessed for
observation eligibility
study
Had 181 (26%) 23 (16%) 45 (37%) 6 (21%) 889 (80%) 1144 (54%)
previous
children
No previous 539 (75%) 125 (84%) 77 (63%) 23 (80%) 226 (20%) 990 (46%)
children
Total: 720 148 122 29 1115 2134
Table 3.3.6a shows that three-quarters of those recruited to the RCT were first time
mothers and that even more of those recruited to the observation study, because they
planned to see a counsellor, were first time mothers.
iv
Because 205 of the 221 women who were excluded because they did not speak enough English did not
complete the antenatal questionnaire, ethnic minority women are under represented in the excluded column. As
noted in footnote i above, 100 other questionnaires were blank, apart from the reason for exclusion. (20 had
previously breastfed, 23 planned to bottle feed, 30 planned to move away, 29 were excluded because the doctor
or midwife felt it inappropriate for a counsellor to visit, four because of a known fetal abnormality and four
because they had previously participated. Further demographic details are unavailable on these women.

Table 3.3.6b
Age of women assessed for eligibility

Excluded
study recruited to assessed
observation for eligibility
study
Under 20 44 (6%) 5 (3%) 12 (10%) 0 60 (6%) 121 (6%)
20 – 24 117 (16%) 12 (8%) 28 (23%) 4 (14%) 151 (14%) 312 (15%)
25 – 29 230 (32%) 35 (24%) 28 (23%) 6 (21%) 292 (27%) 591 (28%)
30 – 34 225 (32%) 63 (43%) 43 (36%) 10 (36%) 358 (33%) 699 (33%)
Over 35 98 (14%) 32 (22%) 10 (8%) 8 (29%) 222 (20%) 370 (18%)
Total: 714 147 121 28 1083 2093
Mean age: 28 yrs 8 mths 30 yrs 9 mths 27 yrs 6 mths 31 yrs 4 mths 29yrs 9 mths 29 yrs 4 mths
Tables 3.3.6b shows the distribution of participants by age and the mean ages
respectively for the different study groups. The mean age of women recruited to the
RCT was 28 yrs 8 months.
The age breakdown does however need to be interpreted with caution, because many
of the differences revealed by the comparison reflect the recruitment criteria for the
study. For example, the commonest reason for women to be excluded was that they
had breastfed their previous child and as a result, more multiparae than primiparae
were excluded. This influenced the average age of women in the study, because on
average, the multiparae were older. To prevent parity distorting the age distribution,
it is more useful to limit comparison of age to first time mothers, as shown in Table
3.3.6c. This shows that those recruited to the RCT were older than those excluded.

Similarly, women who were recruited to the observation study because they wanted to
see a breastfeeding counsellor anyway were older than those in the RCT.
Table 3.3.6c
Mean age of primiparae assessed for eligibility
RCT to RCT Not iv All primips
Excluded
study
Mean age: 28 yrs 3 mths 30 yrs 10mth 26 yrs 2 mths 32 yrs 4 mths 25 yrs 2 mths 27 yrs 10mth
Total: 533 124 76 22 218 973
Terminal educational age is both a measure of educational achievement and a proxy
for social class, both of which are known to be associated with infant feeding
behaviour. (See sections 1.2.6 and 2.3.4 ) Table 3.3.6d shows that women recruited to
the RCT had similar levels of educational attainment to the population of women
assessed for eligibility. Those who were recruited to the observation study were
however more likely to have left school at 19 or over, demonstrating an association
between pre-existing intention to contact a counsellor and educational attainment.
Table 3.3.6d
Terminal educational age of women assessed for eligibilityv

Excluded
observation for
study eligibility
Under 16 51 (7%) 6 (4%) 14 (12%) 2 (7%) 113 (11%) 186 (9%)
16 174 (25%) 22 (15%) 36 (32%) 3 (11%) 300 (29%) 535 (26%)
17 103 (15%) 15 (10%) 17 (15%) 3 (11%) 138 (13%) 276 (14%)
v
Whereas it is helpful to compare the mean ages of each study group, this would be inappropriate for Terminal
Educational Age, because the data is skewed.

18 109 (15%) 15 (10%) 12 (10%) 4 (14%) 134 (13%) 274 (13%)
19 and over 266 (38%) 89 (61%) 35 (31%) 16 (57%) 359 (34%) 765 (38%)
Total 703 147 114 28 1044 2036
Table 3.3.6e reports the social class, coded using the Registrar General's tables.94 vi
Because of the work involved in coding, a decision was taken to only do so for
women recruited to either the RCT or observation study. Comparing these two
groups revealed that women in the observation study were more likely to be from
social class I or II than women in the main trial.
Table 3.3.6e
Social Class of women recruited
RCT Observation Study All women recruited
I 69 (10%) 26 (19%) 95 (12%)

II 179 (26%) 59 (43%) 238 (29%)
IIINM 124 (18%) 20 (15%) 144 (18%)
IIIM 178 (26%) 21 (15%) 199 (24%)
IV 76 (11%) 7 (5%) 83 (10%)
V 22 (3%) 3 (2%) 25 (3%)
Other 35 (6%) 1 (1%) 36 (4%)
Total 683 137 820
The ethnicity of those who returned the baseline antenatal questionnaire is shown in
table 3.3.6f. Although only 16 of the non-English speakers returned the
questionnaires, the ethnicity of women recruited to the RCT reflected that of the
population of women assessed for eligibility. The languages spoken by the non-
English speakers are given in appendix E, table 3.3.6.
vi
Social class was coded by husband or partner's occupation when the woman had a partner and by her own if
she did not. The reasons for this approach are given in section 3.2.9a.

Table 3.3.6f
Ethnicity of women assessed for eligibility

Excluded
study
White (UK) 417 (59%) 105 (71%) 74 (63%) 21 (75%) 677 (63%) 1294 (63%)
White 74 (11%) 21 (14%) 6 (5%) 69 (6%) 170 (8%)

(Other)
African / 109 (15%) 12 (8%) 13 (11%) 5 (18%) 211 (20%) 350 (17%)
Caribbean
Indian 55 (8%) 4 (3%) 15 (13%) 68 (6%) 142 (7%)
subcontinent
Other 49 (7%) 5 (3%) 9 (8%) 2 (7%) 46 (4%) 111 (5%)
Total: 704 147 117 28 1071 2067
3.3.7 Effectiveness of Randomisation
Participants recruited to the RCT were allocated to the intervention or control group
according to randomisation codes, which were stored in sealed envelopes, as
described in section 3.2.10.
As a result of the randomisation, 363 (50.4%) women were allocated to receive
additional support from a breastfeeding counsellor and 357 (49.6%) to normal care.
The socio-demographic characteristics of the two groups are reported in tables 3.3.7a
- 3.3.7g, but no statistically significant differences were observed. Specifically, the
groups were approximately similar in terms of their parity, age, ethnicity, social class
and terminal educational age. Similarly there was little difference between the
intended duration of breastfeeding or intention to return to work of the two groups.

Table 3.3.7a
Parity of intervention and control groups:
Intervention Control Total

Yes 94 (26%) 87 (24%) 181
No 269 (74%) 270 (76%) 539
Total 363 357 720
Chi-Square = 0.22 (1df) p = 0.637, NS
Table 3.3.7b
Age group of intervention and control groups:

Under 20 20 (5%) 24 (7%) 44
20 – 24 63 (18%) 54 (15%) 117
25 – 29 119 (33%) 111 (31%) 230
30 – 34 106 (29%) 119 (34%) 225
35 and over 53 (15%) 45 (13% 98
Total 361 353 714
Chi-Square = 2.65 (4df) p = 0.618, NS
Table 3.3.7c
Ethnicity of intervention and control groups:

White (UK) 212 (59%) 205 (59%) 417
White (Other) 37 (10%) 37 (11%) 74
African / Carribean 61 (17%) 48 (14%) 109
Indian Subcontinent 24 (7%) 31 (9%) 55
Other 23 (6%) 26 (7%) 49
Total 357 347 704
Chi-Square = 2.60 (4df) p = 0.627, NS

Table 3.3.7d
Social class of intervention and control groups vi

I 38 (11%) 31 (9%) 69
II 81 (23%) 98 (29%) 179
IIINM 68 (20%) 56 (17%) 124
IIIM 90 (26%) 88 (26%) 178
IV 40 (12%) 36 (11%) 76
V 7 (2%) 15 (4%) 22
Other 22 (6%) 13 (4%) 35
Total 346 337 683
Chi-Square = 8.82 (6df) p = 0.184, NS
Table 3.3.7e
Terminal educational age of intervention and control groups

Under 16 25 (7%) 26 (7%) 51
16 86 (24%) 88 (25%) 174
17 51 (14%) 52 (15%) 103
18 50 (14%) 59 (17%) 109
19 & over 142 (40%) 124 (36%) 266
Total 354 349 703
Chi-Square = 1.98 (4df) p = 0.740, NS
Table 3.3.7f
Intended breastfeeding duration of intervention and control groups

Under 6 weeks 22 (7%) 28 (8%) 50
6 weeks - 3 months 75 (23%) 77 (23%) 152
3 - 6 months 150 (45%) 152 (45%) 302
6 - 9 months 51 (15%) 36 (11%) 87
9 - 12 months 25 (8%) 30 (9%) 55
Over 1 year 8 (2%) 15 (4%) 23

Total 331 338 669
Chi-Square = 5.86 (5df) p = 0.320, NS

Table 3.3.7g
Intention to return to work of intervention and control groups

No plan to return 85 (26%) 91 (29%) 176
Within 6 months 117 (36%) 118 (38%) 235
After 6 months 122 (38%) 104 (33%) 226
Total 324 313 637
Chi-Square = 1.45 (2df) p = 0.484, NS
However, on one variable there was a difference between the control and intervention
groups, which although not large enough to reach statistical significance, merits
attention. The intervention group included ten more women who said they were
undecided about how to feed when they completed the antenatal screening
questionnaire than did the control group. The details of this are shown in table 3.3.7.
Because no differences were detected on any of the other criteria assessed, it seems
likely that this arose by chance. Also, the research team were not aware of any
situation in which the random allocation might have been prejudiced. However,
because feeding intention was known to be an important predictor of behaviour, a
sensitivity analysis was performed to assess the impact of this. This is reported in
section 3.5.8.
Table 3.3.7h
Feeding plan of intervention and control groups
Breast 240 (67%) 244 (67%) 484
Both breast and bottle 104 (29%) 101 (29%) 205
Undecided 16 (4%) 6 (2%) 22

Total 360 351 711
(Chi-Square = 4.51 (2df) p = 0.105 NS)
3.3.8 Withdrawals
Eight of the 720 women recruited to the RCT withdrew, or were withdrawn after
randomisation. In three cases, (two intervention and one control), this was because of
a stillbirth or neonatal death. Four intervention group mothers withdrew their consent
after randomisation and one who was recruited to the intervention group was neither
offered the intervention, nor followed up, due to an administrative error in the study
office.
Although all the women who withdrew their consent were in the intervention group,
this is not surprising, because they needed to make contact with a counsellor
antenatally, whereas the control group had no prompt which might have led them to
reconsider their participation. By chance, the same number of women - 356, were left
in both the control and intervention groups, after the eight withdrew.
One woman was withdrawn from the observation study following a neonatal death.
3.3.9 Postnatal follow up
Six-week questionnaires were returned by 684 (96%) of the 712 women remaining in
the RCT and by 144 (98%) of the 147 in the Observation Study. Information was
available on the duration of breastfeeding up to four months on 634 (89%) of those in
the RCT and 128 (87%) of those in the observation study

As shown by table 3.3.9a, there was no difference between the follow up rates for the
control and intervention groups in the RCT.
Table 3.3.9a
Postnatal follow-up for control and intervention groups:
Intervention: Control: Total RCT:

In study at birth: 356 356 712
Follow up at six weeks: 344 (97%) 342 (96%) 686 (96%)
Follow up at four months: 318 (89%) 316 (89%) 634 (89%)
The high response rates at both six weeks and four months, which were similar for
both the intervention and control groups, reduce the likelihood that any differences
observed between the two groups might reflect, or be masked by, differential response
rates, rather than genuine treatment effects. To assess the extent to which this might
have mattered, had there been a lower response rate, the proportions of women who
were contacted by telephone because they had not returned questionnaires themselves,
were examined for those fully breastfeeding, partially breastfeeding and exclusively
bottle feeding. (Tables 3.3.9b and 3.3.9c)
Table 3.3.9b
Postnatal follow-up by feeding behaviour at six weeks:
Breastfeeding Both breast Only bottle: All (vii)

fully: and bottle: respondents:
Responded at six weeks: 342 209 277 828
6 week form done by phone: 28 (8%) 37 (18%) 79 (29%) 144 (17%)
Diary 1 returned: 137 (40%) 66 (32%) 47 (17%) 250 (30%)
2
Chi for 6 week form done by phone = 44.1 (2df), P < 0.01
2
Chi for Diary 1 returned = 39.0 (2df), P < 0.01
vii
Tables 3.3.8b and 3.3.8c include women recruited to both the RCT and Observation Studies.

Table 3.3.9c
Postnatal follow-up by feeding behaviour at four months:
Breastfeeding Both breast Only All (vii)

fully: and bottle: bottle: respondents:
Information available on 198 158 404 760
feeding to 4 months:
Completed four month 198 158 90 446
questionnaire:
4 month form done by 39 (20%) 40 (25%) 25 (28%) 104 (23%)
phone:
Diary 3 returned: 78 (39%) 52 (33%) 22 (24%) 152 (34%)
Notes:
1. The percentages shown in 3.3.9c are of the numbers completing the four-month questionnaire. This
includes those who did so by telephone.
2. In interpreting the response rates at four months, it is important to note that women who had previously
reported that they had stopped breastfeeding were assumed to be still bottle feeding. Thus, although
404 women were recorded as bottle-feeding at four months, only 90 of these responded at or after four
months. It may be more appropriate to compare the numbers who were interviewed by telephone, or
who returned the third diary, with this number. Because of this, statistical analysis of the responses at
four months would be inappropriate.
At six weeks, those who had discontinued breastfeeding were significantly less likely
to have returned either the feeding diaries, or the postnatal questionnaires, as shown in
table 3.3.9b. Twenty nine percent of those who were known to be exclusively bottle
feeding at six weeks needed to be contacted by telephone, whereas only 8 % of those
breastfeeding fully needed this prompt. Similar trends are also apparent in table
3.3.9c, which shows the response rates at four months. This suggests that had the
non-responders not been contacted by telephone, those who discontinued
breastfeeding would have been under-represented, potentially biasing the results.viii It
is also worth noting that those who discontinued were less likely to return the feeding
diaries. Although some useful information was obtained from the diaries, this meant
viii
A descriptive account of the telephone calls to non-responders is given in section 4.3.1 below.

they were a less effective way of finding out about the behaviour of those who
discontinued than the questionnaires mailed to participants.
3.3.10 Analysis and exclusion of premature births
The issue of whether to include women delivering prematurely is discussed in section
3.2.3a. Early in the conduct of the trial, it became apparent that it might be necessary
to exclude from some of the analyses those women who delivered before they were
able to see the counsellor. The numbers of women in the Intervention Group who did
not see the counsellor at each completed week of pregnancy are shown in table
3.3.10a.
Table 3.3.10a
Contacts with intervention group mothers by gestation at delivery:
Gestation at Seen before birth: Seen after birth: Not seen: Total:
delivery:
30 1 1
33 1 1
34 1 1 2
35 2 3 5
36 1 3 2 6
37 6 1 3 13
38 25 1 13 39
39 38 16 54
40 74 1 13 88
41 86 9 95
42+ 46 5 51
279 6 70 355
On closer inspection it became apparent that for a few women, the original Expected
Date of Confinement, which was compared with the actual date of birth to calculate
the gestation at birth, was incorrect. The six-week postnatal questionnaire asked
women if their babies were premature and the results revealed that fewer women had
actually delivered early (Table 3.3.10.b).

Table 3.3.10b
Contacts with intervention group mothers who reported on six-week questionnaire
that baby was premature.
(Nine mothers whose original expected date of confinement suggested the baby had been born before
37 weeks)
Gestation at Seen before birth: Seen after birth: Not seen: Total:
delivery:
33 1 1
34 1 1
35 1 3 4
36 2 1 3
Although the numbers delivering early are small, tables 3.3.10a and 3.3.10b show that
most women delivering before 36 weeks did not see the counsellor, whereas the
majority of those delivering at 37 weeks or later were able to do so. Although three of
the women delivering at 36 weeks only saw the counsellor after birth, because the
majority did see the counsellor at some stage, it was decided they should remain in the
study.
Twelve women recruited to the RCT delivered before 36 weeks and confirmed that
their babies were premature on their six-week postnatal questionnaires. Six of these
were controls and six from the intervention group. Because prematurity was shown
to directly affect the counsellors' ability to deliver the intervention, these twelve
women were excluded from all the analyses which compared the control and
intervention groups in section 3.5.
Table 3.3.10c reports the numbers of intervention and control women followed up to
four months, when the twelve who delivered prematurely had been excluded.

Table 3.3.10c
Postnatal follow-up for control and intervention groups (excluding prem births)
Intervention: Control: Total RCT:

In study at birth: 350 350 700
Follow up at six weeks: 336 (96%) 336 (96%) 672 (96%)
Follow up at four months: 310 (89%) 310 (89%) 620 (89%)
3.3.11 Summary
This chapter has reported the recruitment and follow up of participants in the trial.
Recruitment proved to be very much harder than expected, partly because the
participating practices either did not give women their forms to complete, or did not
subsequently return the forms, for about a third of the potential recruits. Additionally,
more women were ineligible for the study than had been anticipated. Despite these
difficulties, when the control and intervention groups were compared using a range of
socio-demographic indicators, they were shown to be remarkably similar, suggesting
that the random assignment to treatment groups had been successful. One potentially
important difference was however noted, in that more women in the intervention
group were initially undecided how to feed their babies. Women in both control and
intervention groups were successfully followed up, with 96% of women providing
data on their feeding at six weeks and 89% doing so at four months.

Chapter 3.4
RESULTS OF STUDY TWO: UPTAKE OF COUNSELLING AND WOMEN’S

PERSPECTIVES ON THE SUPPORT THEY RECEIVED
3.4.1 Introduction
This chapter reports on the uptake of breastfeeding counselling by women recruited to
the study. This is important because the comparisons of feeding behaviour reported
later in chapter 3.5 are analysed on the basis of “intention to treat”, but not all women
in the study received the "treatment" to which they were allocated. Those allocated to
receive normal care were still able to contact a counsellor themselves and some of
those allocated to receive support from a counsellor as part of the study did not do so.
The implications of this are discussed later in chapter 3.7.
Secondly, this chapter reports the content of the support that women received from
counsellors and whether they found it helpful. Also, because mothers were asked
about advice from all sources it was possible to compare the extent to which different
groups of people, (professionals, breastfeeding counsellors, family and others) gave
helpful or unhelpful advice
As previously stated in section 3.3.8 and 3.3.10, the data presented on the uptake of
counselling exclude eight women who were withdrawn and twelve who delivered
before 36 weeks, which was before they were likely to have seen a counsellor.
3.4 Study two results: Uptake of counselling 177

3.4.2 Uptake of counselling in the intervention group
3.4.2a Antenatal contact:
Information on the uptake of counselling in the intervention group is available from
the record forms that counsellors completed when they were in contact with women.
Counsellors were sent the antenatal forms when they were informed that women had
been recruited to the study and postnatal ones when they were notified of births.
Counsellors also had spare postnatal forms to use if contacted by mothers before the
forms arrived and received reminders if they did not return forms within four weeks
of them being issued.
During the study, the counsellors returned completed 279 antenatal forms for the 350
women in the intervention group. They reported no difficulty in contacting 222
(80%) of the 279 women, but needed to make repeated telephone calls or visits to see
42 (15%) of them. Fifteen (5%) of the antenatal contacts were solely by telephone.
The venue of the antenatal contacts is shown in table 3.4.9a and shows that although
the majority of contacts took place at women's homes, the counsellors were prepared
to see mothers in a range of settings.
Table 3.4.2a
Venue for antenatal contacts

(Percentages shown are of the total number of intervention women.)
Number of women %
Contacted before birth, of which: 279 79.7
Woman's home 253 72.3
Counsellor's home 7 2.0
Doctor's surgery 2 0.6
Hospital 1 0.3
Woman's workplace 1 0.3

Telephone 15 4.2
Not contacted before birth 71 20.3
Total intervention women 350
The counsellors reported that they had been unable to contact 26 women before the
birth and no antenatal form was returned for a further 45 women. Assuming that none
of these were seen antenatally, this suggests that 71 (20%) of the 350 women
allocated to the intervention group did not receive the antenatal component of the
intervention. The implications of this are discussed in section 3.7.2c.
3.4.2b Postnatal contact
Whereas the study protocol intended that women would see the counsellor
antenatally, it was left to them and the counsellors to agree how much contact they
had after the birth. In the event, 215 (61%) of the 350 women had contact with a
counsellor postnatally. Sixty-eight women (19% of 350) met the counsellor face-to-
face, while a further 147 (42%) were only in contact by telephone. Whilst most face-
to-face contacts occurred when counsellors visited women at home, they also visited
five women in hospital. They also saw two women at their own homes, one at the
health centre where the counsellor worked and one at a breastfeeding centre drop-in
session.
Many women had more than one contact with the counsellor and up to eight contacts
were recorded for individual women. The counsellors returned completed postnatal
record forms for a total of 81 face-to-face contacts and 302 telephone conversations.

When they completed the postnatal record forms, counsellors were asked to record
who initiated each contact. This showed that 105 (49%) of the 215 first contacts were
initiated by mothers or their partners, while 108 (50%) were initiated by counsellors.
Two first contacts were reported as being initiated by a health professional. When all
the 383 postnatal contacts were taken together, 205 (54%) were initiated by mothers
or their partners and 175 (46%) by counsellors. Three of the contacts were initiated by
a professional.
3.4.2c Combining antenatal and postnatal contact data
Comparing the records of antenatal and postnatal contacts provides a fuller account of
the support women received, as shown in table 3.4.2c. This shows that only 8% of
women allocated to the intervention group had no contact with a counsellor and that
the majority of those who were not seen antenatally did have some postnatal contact.
Table 3.4.2c (i)

Type of antenatal and postnatal contact for intervention group women i
(Percentages shown are of the total number of intervention women)
Antenatal contacts:
Face-to-face Telephone No contact Total
Postnatal contacts:
Face- to-face 50 (14%) 2 (1%) 16 (5%) 68 (19%)
Telephone 115 (33%) 6 (2%) 26 (7%) 147 (42%)
No contact 99 (28%) 7 (2%) 29 (8%) 135 (39%)
Total 264 (75%) 15 (4%) 71 (20%) 350 (100%)
i Data in this table relate to the 350 intervention group women who delivered at, or after, 36 weeks gestation.
Of the 12 women who were excluded because they delivered before 36 weeks, five did see a counsellor
postnatally and one spoke to a counsellor on the telephone.
To assess the impact of the level of antenatal and postnatal support that women
received on the trial's power to measure changes in the intervention group, it is

necessary to narrow the comparison made in table 3.4.2c (i) to those women who
responded to the six-week postnatal questionnaire. This is illustrated in the next
table, 3.4.2c (ii). Nine of the fourteen intervention women who were lost to follow-up
had no postnatal contact, four were in contact with the counsellor by telephone and
only one was visited. This showed that 272 (81%) of the women followed up to six
weeks met a counsellor at some stage, 38 (11%) only had telephone contact and 26
(8%) had no contact with a counsellor. Postnatally, 63% of women who were
successfully followed up had some contact with a counsellor.
Table 3.4.2c (ii)

Antenatal and postnatal contacts for intervention group women who were followed
up to six weeksI
(Percentages shown are of the total number of intervention women followed up)
Antenatal contacts:
Face-to-face Telephone No contact Total
Postnatal contacts:
Face-to-face 49 (15%) 2 (1%) 16 (5%) 67 (20%)
Telephone 112 (33%) 6 (2%) 25 (7%) 143 (43%)
No contact 93 (28%) 7 (2%) 26 (8%) 126 (37%)
Total 254 (76%) 15 (4%) 67 (20%) 336 (100%)
i Data in this table relate to the 336 intervention women who delivered at, or after, 36 weeks gestation and who
provided follow-up data on their feeding behaviour by responding to the six-week questionnaire.
3.4.3 Uptake of counselling and feeding behaviour
The uptake of counselling in the intervention group was investigated further, to find
out more about the women who engaged with the counsellor. There were no
associations of note between antenatal contact and either socio-economic variables or
feeding behaviour, which suggested that practical problems making contact were the
main reasons women were not seen before the birth.

The situation for postnatal contact was however different and suggests that some
mothers are more likely than others to contact counsellors during the postnatal period.
Tables 3.4.3a and 3.4.3b demonstrate that mothers of first babies and those who left
school at an older age were more likely to have face to face contact with a counsellor.
Table 3.4.3a
Type of postnatal contact and parity
Type of postnatal contact
Face to face Telephone No contact Total

Previous Children
Yes 8 (9.2%) 41 (47.1%) 38 (43.7%) 87
No 59 (23.7%) 102 (41.0%) 88 (35.3%) 249
Total 67 (19.9%) 143 (42.6%) 126 (37.5%) 336
2
Chi = 8.57, (2df), p = 0.014
Table 3.4.3b
Type of postnatal contact and terminal educational age
Type of postnatal contact
Terminal educational Face to face Telephone No contact Total

age:
Under 16 1 (4.3%) 7 (30.4%) 15 (65.2%) 23
16 9 (11.7%) 43 (55.8%) 25 (32.5%) 77
17 9 (19.6%) 16 (34.8%) 21 (45.&%) 46
18 12 (27.3%) 15 (34.1%) 17 (38.6%) 44
19 & over 35 (25.5%) 58 (42.3%) 44 (32.1%) 137
Total 66 (20.2%) 139 (42.5%) 122 (37.3%) 327
2
Chi = 21.21, (8df), p = 0.007

The next chapter reports the prevalence and duration of breastfeeding amongst
women recruited to the trial. However, it is worth noting at this stage that postnatal
contact with a counsellor could be related to feeding behaviour at six weeks; those
who met the counsellor during the postnatal period were more likely to continue
breastfeeding than those who did not (table 3.4.3c). Despite this observation, evidence
of an association does not necessarily imply causation, and it is noteworthy that
women who had no contact with a counsellor had lower breastfeeding rates than the
control group, 63.4% of whom were breastfeeding at six weeks. Instead the data
suggest that those who contacted a counsellor were more motivated to breastfeed.
Table 3.4.3c
Postnatal contact and breastfeeding at six weeks
Breastfeeding at six weeks

Any breast Only bottle Total
Postnatal contact No % No %
Face-to-face 51 76% 16 24% 67
Telephone 92 64% 51 36% 143
No contact 75 60% 51 40% 126
Total 218 65% 118 35% 336

2
Chi for trend (linear by linear association) = 4.89, (1df), P= 0.027
3.4.4 Comparison of the uptake of breastfeeding counselling in the intervention,

control and observation groups.
Although contemporaneous records of the contacts between counsellors and women
are only available for those in the intervention group, the six-week postnatal
questionnaire asked all women whether they had contacted a counsellor after the

birth. This gives more information about the support normally provided by
counsellors and offers a measure of the uptake of counselling in the control group.

In interpreting this data, it is worth bearing in mind that women in the control and
intervention groups had not planned to "discuss breastfeeding with a breastfeeding
counsellor" when they completed the antenatal questionnaire. Those in the
observation group had planned to do so.
When asked in the six-week questionnaire, "Have you tried to contact a breastfeeding
counsellor since your baby was born?" 179 (53%) of the 336 intervention group
mothers answered "Yes". Similarly, 48 (14%) of the 336 mothers in the control group
and 47 (33%) of the 144 in the observation group had tried to do so.
The proportion of intervention group mothers reporting that they tried to contact a
counsellor is less than the 63% of mothers with whom the counsellors reported having
postnatal contact. However this may be due to the counsellors and mothers being
asked different questions and mothers not reporting all the contacts that counsellors
initiated.
Whereas the overwhelming majority (95%) of women in the intervention group
reported that they heard of the counsellor via the study, (159 of the 168 who
responded to the question,) women in the control and observation groups had heard of
counsellors from a wide range of sources (table 3.4.4a). Reading the detail of
women's responses revealed that their definition of a "breastfeeding counsellor"
included a range of sources of help. Whilst most of the contacts appeared to be with
NCT breastfeeding counsellors, women also sought help from the La Leche League,
hospital breastfeeding specialists, breast pump agents and telephone helplines.
3.4 Study two: Uptake of counselling 184

Table 3.4.4a
How women in control and observation groups heard of counsellor
(Responses from women who had tried to contact a counsellor)
Control group Observation group

(n = 44) (n = 45)
n % n %
NHS staff 18 (41%) 12 (27%)
NCT & NCT antenatal class 13 (30%) 26 (58%)
Friends & others 5 (11%) 4 (9%)
Books & leaflets 7 (16%) 2 (4%)
Through research study 1 (2%) 1 (2%)
Women who tried to contact a counsellor were also asked whether they had any
difficulty doing so. Although women in the intervention group had little difficulty
contacting a counsellor, one in five of women in the control and observation groups,
who needed to identify the counsellor they contacted themselves, reported that they
had difficulty making contact (table 3.4.4b).
Table 3.4.4b
Women's reports of difficulty contacting a counsellor
Intervention group Control group Observation group

(n = 171) (n = 45) (n = 44)
n % n % n %
No difficulty 146 (85%) 36 (80%) 34 (77%)
Yes, had difficulty 25 (15%) 9 (20%) 10 (23%)
3.4.5 Women's explanations of why counsellor was helpful or unhelpful
Table 3.4.5a reports women's responses to the question, "Did you find the counsellor
helpful?" It demonstrates high levels of satisfaction with the counselling women

received as part of the study, but lower scores from those in the control and
observation groups who approached counsellors independently.
Table 3.4.5a
Responses to "Did you find the counsellor helpful?"
I
Intervention group Control group Observation group
(n = 169) (n = 40) (n = 44)
n % n % n %
Very Helpful 123 (73%) 25 (62.5%) 22 (50%)
Fairly Helpful 28 (17%) 8 (20%) 12 (27%)
A little Helpful 12 (7%) 4 (10%) 4 (9%)
Not Helpful 6 (4%) 3 (7.5%) 6 (14%)
Note:
i Five control group women who had difficulty contacting a counsellor did not reply to the question on whether
they found the counsellor helpful. As a result, these figures may underestimate dissatisfaction in the control
group.
After the question on whether they found the counsellor helpful, women were asked
to "Please explain" their response in a free text section. This was coded using the
coding scheme for advice and response to problems adapted from study one. Up to
three concepts could be coded for each woman and the responses were analysed as a
multiple response variable.
Comments made by the 161 intervention women who responded show that they
valued the relationship with the counsellor, what they learnt and practical advice they
received for problems (table 3.4.5b). Only one in ten women commented on receiving
practical demonstrations, or advice on positioning. For simplicity, the thirteen
negative comments are grouped together, but the commonest negative comment,

made by six women, was that they had received poor or insufficient advice on
problems they encountered.
Table 3.4.5b
Multiple response analysis of free text comments made when asked to
explain why counsellor was helpful of not.
(161 Intervention group women who responded)
% of
Coded response: Count respondents
Gave time, good relationship, listened, discussed 73 45.3

feelings
Explanation, conveying knowledge 71 44.1
Practical advice on feeding problems 63 39.1
Reassurance 19 11.8
Practical demonstration/advice on positioning 17 10.6
Persevere, encouragement, keep trying 6 3.7
Discuss expectations of breastfeeding 3 1.9
Explain benefits of breastfeeding 2 1.2
Give bottle, (complementary or supplementary feed) 1 .6
Rest, slow down, relax 1 .6
Drink more fluids 1 .6
Eat better 1 .6
Paracetamol or other painkiller 1 .6
Delay feed 1 .6
Do not give bottle feeds 1 .6
Cream for thrush/Nystatin 1 .6
Poor or insufficient advice 13 8.1
-------
Total responses: 275
Women in the observation study and the control group of the RCT also reported their
experiences and these are grouped together in table 3.4.5c because of the smaller
numbers involved. Whilst these women often referred to the relationship with the
counsellor, fewer women referred to practical advice for problems, or that they valued
explanations.

Table 3.4.5 c
Multiple response analysis of free text comments made when asked to
explain why counsellor was helpful of not.
(56 Control and Observation group women who responded)
% of
Coded response: Count respondents
Gave time, good relationship, listened, discussed 29 51.8

feelings
Practical advice on feeding problems 16 28.6
Explanations, conveying knowledge 12 21.4
Reassurance 10 17.9
Practical demonstration/advice on positioning 9 16.1
Persevere, encouragement, keep trying 5 8.9
Poor or insufficient advice 9 16.1
-------
Total responses: 90
3.4.6 Mothers' opinions on the quality of advice they received from different sources
Towards the back of the questionnaire they completed six weeks after the birth,
mothers were asked to write in the advicei they found "most helpful" and "least
helpful", and who had given this advice. Many gave more than one example, and it
was clear that any attempt to code their responses ran the risk of reducing the
meanings that women expressed in their comments to categories which did not do
justice to the strength of their feelings. Because of this, it was decided to analyse their
comments separately, using a qualitative method and this forms the basis of part four
of the thesis. It was relatively straightforward to compare the source of the advice
which women had described as "most helpful"or "least helpful".
i
The term "advice" adopted in this question may be controversial, because of the way it implies a directive, rather
than non-directive approach. The issues involved in this are discussed more fully in section 1.6.8. While it might
have been better to use the words "information, advice and support" in questions 20 and 21 of the six-week
questionnaire, the term "advice" is commonly used for the content of discussions about how to resolve
breastfeeding problems.

Before considering the results presented in tables 3.4.6a - c, it is appropriate to
mention some caveats, which may influence its interpretation.
• Firstly, because women were able to report more than one piece of advice as being
"most helpful" or "least helpful", it would be inappropriate to apply statistical tests
to the data. The percentages shown are the percentage of women who had begun
breastfeeding that reported helpful or unhelpful advice from a particular source.
• Response rates were higher in the observation group, perhaps because of the
different socio-economic composition of that group, and lower in the control
group, who may have felt less commitment to provide detailed responses on the
questionnaire when they did not receive the intervention.
• The availability of counselling to the intervention group - and to a lesser extent the
observation group who had indicated their intention to seek it out - meant that
more women in these groups could have received their "most helpful" advice from
a counsellor, because more were in contact with one. Because of this, it is
inappropriate to compare satisfaction with counsellors' advice between groups.
Additionally, because so many women in the intervention group said their "most
helpful" advice had come from counsellors, fewer reported advice from other
sources as "most helpful". This does not necessarily mean that women who saw a
counsellor were less satisfied with the advice they received from other sources.

Table 3.4.6a
Source of advice perceived as least and most helpful by women in intervention
group
(320 women who initiated breastfeeding)
Respondents reporting Respondents reporting
"Least helpful" advice "Most helpful" advice
from source. from source.
No % No %
Midwife 54 17% 75 23%
Health visitor 10 3% 17 5%
Hospital nurse 17 5% 3 1%
Other health professional 6 2% 8 3%
Breastfeeding counsellor 10 3% 141 44%
Husband or partner 4 1% 7 2%
Other family 34 11% 26 8%
Friends 20 6% 15 5%
Other people 5 2% 2 1%
Books, leaflets, magazines 8 3% 6 2%
Total respondents reporting helpful 133 42% 250 78%

or unhelpful advice:
Table 3.4.6b
Source of advice perceived as least and most helpful by women in control group
No % No %
Midwife 43 13% 118 36%
Friends 14 4% 30 9%


Table 3.4.6c
Source of advice perceived as least and most helpful by women in observation
group
No % No %
Midwife 24 17% 48 34%
Friends 8 6% 20 14%

Despite the caveats above, tables 3.4.6a - c reveal some striking findings.
• Across the three study groups, approximately twice as many women reported
advice they described as "most helpful" than "least helpful".
• In the intervention group, more women identified advice from breastfeeding
counsellors as "most helpful" than from any other source. Counsellors were the
second most commonly mentioned source of "most helpful" advice in the
observation group and across the three groups very few women said that their
"least helpful" advice had come from counsellors.
• The findings reveal significant levels of dissatisfaction with the advice women
receive from some sources. Advice from hospital nurses was seen as particularly
unhelpful, but women's families and friends were also mentioned as a common
source of unhelpful advice.

• Women regarded more of the advice they received from midwives as helpful than
unhelpful, but despite this, the questions do reveal significant levels of
dissatisfaction with advice from midwives.
• Relatively few women mentioned their husbands or partners as a source of
advice. This may partly reflect the wording of the question - asking about
"advice" rather than "information, advice and support" as might have been more
useful. It may also reflect women taking their partners for granted in this, or
perhaps a real finding that partners are not a source of advice.
3.4.7 Summary of key findings
This chapter reports on the uptake of counselling and women's perspectives on the
support they received. Key findings for the intervention group were:
• 80% of women were in contact with counsellors during the antenatal period and
almost all of these met face-to-face.
• Postnatally, 19% of women met a counsellor face-to-face and a further 42% were
in contact by telephone. Contacting a counsellor postnatally appeared to be
related to motivation to breastfeed.
• Overall, 81% met a counsellor at some stage, 11% were only in contact by
telephone and 8% had no contact.
• 73% of women who contacted a counsellor regarded her as "very helpful". They
said the counsellor was helpful because she gave them time, listened and
discussed feelings (45%), because she explained about breastfeeding (44%) and
because she gave practical advice on feeding problems (39%).

• More intervention group women regarded the advice they received from
breastfeeding counsellors as "most helpful", than from any other source.

Chapter 3.5
RESULTS OF STUDY TWO: EFFECT OF THE INTERVENTION
3.5.1. Introduction:
This chapter reports on comparisons between the control and intervention groups and
assesses the effect of the breastfeeding counselling intervention. It includes
comparisons of the incidence and duration of breastfeeding, as well as participants'
satisfaction with their breastfeeding experience. Additionally, data are provided on
the uptake of counselling and women's comments on the support they received from
the counsellor.
3.5.2. Incidence and duration of breastfeeding for women in RCT
As has been described, 712 women were included in the RCT. 655, (95.8%) of the
686 who responded to the first postnatal questionnaire reported putting the baby to
breast to feed at any stage and by six weeks, 437, (63.9%) were still giving any
breastfeeds. By four months, this figure had fallen to 277. (43.7% of the 634 whose
feeding behaviour was known.)
When the 12 women who delivered before 36 weeks were excluded, 644 (95.8% of
the 672 who responded to the six-week questionnaire) ever put the baby to the breast
to feed. At six weeks, 431 (64.1%) were giving any breastfeeds and by four months,
3.5 Study Two Results: Effect of the intervention 194

this figure had fallen to 274, (44.2% of the 620 respondents) (Fig 3.5.2). The rest of
the data in chapter 3.5 relate to these 700 women.i
These figures may be compared with the 1995 Infant Feeding Survey data for
England and Wales.1 This found that 65% of mothers who had initially breastfed
were still doing so at six weeks, a figure which had fallen to 42% by four months.
Fig 3.5.2 Duration of breastfeeding in RCT (n = 700)
100
90
80
70
Percentage of women
60
Any breastfeeding
50
Exclusive breastfeeding
40
30
20
10
0
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98 105 112 119
Days from birth
3.5.3 Outcome of RCT: Incidence of breastfeeding
Incidence of breastfeeding is defined as the proportion of babies who were breastfed
initially. This includes all babies who were put to the breast at all, even if this was on
i
As discussed in sections 3.2.3a and 3.3.10, women delivering before 36 weeks were excluded from analyses
because they delivered before they were able to see the counsellors and could not therefore receive the
intervention. The numbers involved are small, so including them would not have significantly altered the results.

one occasion only and is the definition used in the quinquennial national infant
feeding surveys.1
In the intervention group, the incidence of breastfeeding was 95.2% (320/336),
compared with 96.4% (324/336) in the control group, (relative riskii 0.99; 95%
confidence interval 0.84 to 1.16; chi2 = 0.596, (1df); P = 0.44).
3.5.4 Outcome of RCT: Prevalence of breastfeeding
Prevalence of breastfeeding refers to the proportion of babies still breastfeeding at
specific ages, even if the babies were also receiving infant formula or solid food.1
At six weeks, the prevalence of breastfeeding in the intervention group was 64.9%
(218/336), compared with 63.4% (213/336) in the control group, (relative risk, 1.02;
95% confidence interval, 0.84 to 1.24; chi2 = 0.162, (1df); P = 0.69).
By four months, the prevalence of breastfeeding had fallen to 46.1% (143/310) in the
intervention group and 42.3% (131/310) in the control group, (relative risk, 1.09; 95%
confidence interval, 0.86 to 1.39; chi2 = 0.942, (1df); P = 0.33).
3.5.5 Outcome of RCT: Prevalence of bottle feeding
The prevalence of bottle feeding refers to the proportion of babies given any formula
ii
Relative Risk, is the ratio of the proportions of women breastfeeding (or bottle feeding)in the control and
intervention groups at the time under consideration. Where the Relative Risk is greater than one, the event is
more likely in the intervention group. Where it is less than one, it is more likely in the control group. The 95%
confidence intervals reflect the range within which the Relative Risk could be expected to have been found in
95% of the occasions a similar trial was conducted. If both the 95% confidence limits are less than, or both
greater than one, it is unlikely that any difference observed between the likelihood of the event happening in the

feeds, regardless of whether they were also breastfed. The prevalence at seven days
was calculated from women's responses to questions about when they introduced
formula feeds on the six-week questionnaire and first feeding diary. The prevalence
calculated included those who gave just one or two bottle feeds during the first seven
days.
At seven days, the prevalence of bottle feeding in the intervention group was 34.5%
(116/336), compared with 38.2% (128/335) in the control group, (relative risk, 0.90;
95% confidence interval, 0.70 to 1.17; chi2 = 0.984, (1df); P = 0.32).
By six weeks, the prevalence of bottle feeding had risen to 60.7% (204/336) in the
At four months, the prevalence of bottle feeding was 73.9% (229/310) in the
These figures do not show statistically significant differences and could have arisen
by chance. However, even if the sample size had been larger and the differences
noted had been statistically significant, they were smaller than those considered
clinically important when the sample size was calculated. As discussed in section
3.2.4, it was anticipated that the prevalence of breastfeeding in the control group
intervention and control groups, arose by chance. The relative risk and exact confidence intervals reported were
288
calculated using the epitab function in the STATA statistical package.

would be 50% at six weeks, and the sample size was calculated to detect an increase
to 60%.
3.5.6 Outcome of RCT: Duration of breastfeeding
Duration of breastfeeding refers to the length of time that mothers who breastfed
initially continue to do so, even if they were also giving their baby other foods. In
this study, as in many investigations that assess participants' survival, or the time until
they experience a particular event, the data available for analysis are incomplete, or
"censored". This arose because some mothers fed for longer than four months, when
data collection stopped, while others were lost to follow-up. Survival analysis297
allows comparisons that take into account the number of events that have occurred
amongst those remaining in both the control and intervention groups, throughout the
period being studied. Because it is influenced by how long each participant fed, it
provides a more sensitive measure of the effect of the intervention than comparisons
between the prevalence of breastfeeding at a single point in time which are only based
on whether participants have stopped or not.
In order to conduct the survival analysis, new variables were calculated to express the
length of time women were known to have breastfed for, and whether or not the data
on each woman were "censored".iii (See appendix D)
The proportions continuing to breastfeed in both the intervention and control groups
were obtained using the Kaplan-Meier survival function in SPSS and are shown in
iii
"Censored" means that the final date to which a woman fed was not known as she was either lost to follow-up,
or she continued to breastfeed beyond the four-month recording period.

figure 3.5.6. The median duration of breastfeeding in the intervention group was 110
days in the intervention group, compared with 96 days in the control group. When the
survival distributions were compared, the differences between them were not
statistically significant. (Log Rank statistic 0.58 (1df); P = 0.445)
Fig 3.5.6
Percentage of women giving any breastfeeds
100
90
80
Randomisation
70
Control
% 60 Control-censored
Intervention
50
Intervention
40 -censored
0 20 40 60 80 100 120 140
Days until last breast feed
3.5.7 Outcome of RCT: Time to introduction of bottle feeding
The time to introduction of bottle feeding, (or duration of full breastfeeding,) for
intervention and control groups is shown in figure 3.5.7. The median time to
introduction of bottle feeding in both intervention and control groups was 28 days.

When the survival distributions were compared, the differences between them were
not statistically significant. (Log Rank statistic 2.03 (1df); p = 0.154).
Fig 3.5.7
Percentage of women fully breastfeeding
(ie those who had not introduced bottle feeds)
100
80
Randomisation
60 Control
% Control-censored
40 Intervention
Intervention
20 -censored
0 20 40 60 80 100 120 140
Days until first bottle feed
Although the differences between the survival distributions were not significant, the
survival graph suggests that there may have been a small, though clinically
unimportant difference in the time to introduction of bottle feeding between women
allocated to counselling and those who were not.

3.5.8 Sensitivity analyses to assess impact of the imbalance in number of undecided
women
As identified in section 3.3.7, sixteen of the women in the intervention group had not
decided whether or not to breastfeed when they completed the antenatal questionnaire,
whereas only six of those in the control group were undecided. To assess the impact
of this, four sensitivity analyses were performed.
Firstly, the prevalence of breastfeeding and bottle feeding was examined for those
who had made a decision to breastfeed, after excluding the undecided women. (Table
3.5.8a). As expected, there was a slightly greater difference between the percentages
in all cases, than when the undecided women were included. However, this did not
alter the conclusions and no P-values reached statistical significance.
Table 3.5.8a
Sensitivity analysis of prevalence when undecided women excluded
All women Only those who had decided to

1
breastfeed
Control Intervention P-value Control Intervention P-value
Breastfed 96.4% 95.2% 0.60 96.7% 96.9% 0.88

initially
Any breast 63.4% 64.9% 0.69 63.6% 67.5% 0.30
at 6 weeks
Any breast 42.3% 46.1% 0.33 43.0% 48.3% 0.19
at 4 months
Any bottle 38.2% 34.5% 0.32 38.0% 31.9% 0.10

at 7 days
Any bottle 64.3% 60.7% 0.34 63.9% 58.8% 0.17
at 6 weeks
Any bottle 79.4% 73.9% 0.11 79.0% 72.4% 0.06
at 4 months
1
Raw data for the numbers who had decided to breastfeed are provided in appendix E. Raw Data for all women
are given in the text of sections 5.4.3, 5.4.4 and 5.4.5.

Secondly, the survival analyses reported in sections 3.5.6 and 3.5.7 were repeated,
excluding the women who were uncertain. The log rank statistic for the comparison
of the time to stopping breastfeeding was 1.66 (1df), (P = 0.197), compared with 0.58
(1df), (P = 0.445) when women who were uncertain were included. The difference in
survival durations for full breastfeeding was just significant at the 95% level when the
uncertain women were excluded, (Log rank statistic 3.86 (1df); P = 0.0496). This
compared with a value of 2.03 (1df), (P = 0.154) when uncertain women were
included.
Thirdly, binary logistic regression was used to compare the strength of the association
between firstly, whether or not women had decided to breastfeed and secondly, the
treatment group to which women were allocated and the prevalence of breastfeeding
and bottle feeding. The results of this are shown in table 3.5.8b. Although the results
do not reach statistical significance at the 5% level, the results of binary logistic
regression also suggest that the intervention was a little more effective than it
appeared when the impact of the imbalance in numbers of undecided women was not
taken into account.

Table 3.5.8b
Binary logistic regression to assess the association between treatment allocation
and prevalence of breastfeeding and bottle feeding, when association with having
made decision to breastfeed has been taken into account.
Crosstabulation Binary Logistic regression

statistics, when association
with decision has been taken
into account.
1 2 3 4
Control Inter- P-value Odds Ratio P-value Odds Ratio
vention (95% Confidence (95% Confidence
intervals) intervals)
Breastfed 96.4% 95.2% 0.60 0.74 0.82 1.1

initially (0.32 - 1.68) (0.49 - 2.45)
Any breast 63.4% 64.9% 0.69 1.03 0.45 1.13
at 6 weeks (0.88 - 1.21) (0.82 - 1.56)
Any breast 42.3% 46.1% 0.33 1.17 0.18 1.25
at 4 months (0.92 - 1.27) (0.90 - 1.72)
Any bottle 38.2% 34.5% 0.32 0.85 0.18 0.80

at 7 days (0.61 - 1.18) (0.58 - 1.10)
Any bottle 64.3% 60.7% 0.34 0.86 0.22 0.82
at 6 weeks (0.62 - 1.19) (0.60 - 1.12)
Any bottle 79.4% 73.9% 0.11 0.74 0.06 0.70
at 4 months (0.50 - 1.09) (0.48 - 1.02)
1 2
P is calculated for the Chi statistic in the first analysis.
2
Although Relative Risk is preferred for comparisons of the feeding rates between the control and intervention
groups, in this table the Odds Ratio is shown to allow comparison with the Odds Ratios derived from the binary
282
logistic regression. (The Odds ratios were calculated using Epi info software. )
3
In the binary logistic regression, P is calculated for the "B" statistic. "B" is an estimate of the change in the
dependent variable (ie the feeding behaviour) that can be attributed to a change of one unit in the independent
variable (the treatment group allocation).
4
The Odds Ratio is 1/exp(B). This is a measure of the association between treatment allocation and feeding
behaviour, when the association with whether or not the woman had decided to breastfeed had been taken into
account.
Lastly, Cox regression was used to make a similar comparison with the duration of
any breastfeeding and full breastfeeding. This technique is similar to the survival
analysis performed in section 3.5.7, but allows the relative contribution of different
factors to be assessed. Using Cox regression, the probability that there was no
association between treatment allocation and duration of any breastfeeding, when the
effect of having decided to breastfeed or not was taken into account, was 0.282. This
compared with a probability of no effect of 0.58 for the log rank statistic calculated in
section 3.5.6. Similarly, the probability of no association with the duration of full

breastfeeding, using Cox regression to take into account the effect of decision was
0.105, compared with 0.154 for the log rank statistic. These results suggest that the
effect on duration was greater than it had originally appeared.
There are dangers in retrospectively excluding women after randomisation and it may
be better to rely on the regression analyses than those which excluded women who
were uncertain. Indeed, Roberts and Torgerson298 caution against adjusting for
baseline imbalance, but advocate that those designing trials should identify potentially
prognostic factors in advance and then fit them into an analysis of covariance, (as in
the binary logistic regression and Cox regression analyses above).
Taken together, these sensitivity analyses suggest the intervention was more effective
than appeared from the results when no allowance was made for the imbalance in the
distribution of undecided women. However, on only one of the analyses - the survival
analysis of duration of full breastfeeding - did this reach statistical significance. The
analyses also suggest that the intervention was more likely to have had an effect on
full than partial breastfeeding.
3.5.9 Outcome of RCT: Maternal Satisfaction with breastfeeding
Maternal satisfaction with breastfeeding was assessed by women's responses to six
questions in the six-week postnatal questionnaire. (See section 3.2.9 and appendix B
for the development and reliability of these.)

Women's responses to questions on satisfaction are reported as percentages in table
3.5.9a, but numerical data are available in table 3.5.9b. Women who delivered before
36 weeks were excluded and it is worth noting that those who never breastfed were
not asked to complete these questions.
Table 3.5.9a
Satisfaction with breastfeeding at six weeks
Responses by group (%)
Intervention Control
8) How have you found breastfeeding?
Much easier than I expected 21.0% 17.8%

A little easier than I expected 14.0% 14.9%
As I expected 17.5% 16.5%
A little harder than I expected 23.6% 27.6%
Much harder than I expected 23.9% 23.2%
9) Have you felt confident or unsure about your ability to breastfeed?
I have felt very confident 26.0% 23.2%

I have felt fairly confident 46.0% 42.8%
I have felt fairly unsure 17.5% 23.5%
I have felt very unsure 10.5% 10.6%
10) Have you found breastfeeding stressful?
Most of the time 15.4% 14.2%

Some of the time 32.7% 36.8%
A little of the time 21.8% 22.3%
Very little of the time 30.1% 26.8%
11) Have you enjoyed breastfeeding?
A lot 38.6% 38.3%

A fair amount 36.0% 36.3%
A little 17.4% 17.4%
Not at all 8.0% 8.0%
13) Have you felt you would be embarrassed about breastfeeding in front of people you don't know?
A lot 26.5% 28.7%

A little 37.1% 31.0%
Not at all 23.5% 20.6%
To test whether there was any difference between the satisfaction scores reported by
women in the control and intervention groups, a non-parametric test for the

equivalence of two independent samples, the Mann-Whitney U test was performed for
each question.iv Table 3.5.9b reports this and confirms that there were no significant
differences between the responses of control and intervention women to questions
about their satisfaction with breastfeeding. The differences between control and
intervention women's responses appeared greater for the questions on embarrassment
about feeding in front of others and confidence in the ability to breastfeed than for the
other questions.
Table 3.5.9b
Mann-Whitney Test for mean rank of satisfaction scores1
2
(644 women who ever put the baby to the breast to feed)
Randomisation Number of Mean Rank Significance of

group women Mann-Whitney
test
How have you found breastfeeding? Intervention 314 310.42
Control 315 319.56
Total 629
P = 0.516
Have you felt confident or unsure Intervention 315 304.14
about your ability to breastfeed? Control 311 322.98
Total 626
P = 0.167
Have you found breastfeeding Intervention 312 315.75
stressful? Control 310 307.22
Total 622 P = 0.537
Have you enjoyed breastfeeding? Intervention 311 311.13
Control 311 311.87
Total 622
P = 0.956
Have you felt you would be Intervention 310 321.61
embarrassed about breastfeeding in Control 310 299.39
front of people you don't know?
Total 620
P = 0.108
1
Women's responses were ranked from 1 to 4 (or 5) as shown in table 3.5.9a. Thus a low score
(and resultant rank) for the question on confidence implies a woman is more satisfied, while a low
score (and resultant rank) for the question on finding breastfeeding stressful implies a woman is
less satisfied.
2
644 women who initiated breastfeeding were included in the analysis, but data were missing for
15 to 24 women who did not answer these questions.
iv
The Mann-Whitney test is used to test whether two independent sampled populations are equivalent in location.
The observations from both groups are combined and ranked, with the average rank assigned in the case of ties.
If the populations are identical in location, the ranks should be randomly mixed between the two samples.

3.5.10 Common feeding problems
Women were asked about common feeding problems in a similar format to the
questions on satisfaction. (Table 3.5.10a) The non-parametric tests reported in table
3.5.10b show that women in the intervention group were significantly less likely to be
concerned that they weren't making enough milk for their babies. The other
comparisons showed no significant difference at the 5% level.
Table 3.5.10a
Common feeding problems at six weeks
Responses by group (%)
Intervention Control
12) Have you worried that your baby may not be gaining enough weight?
A lot 10.2% 8.4%

A little 20.7% 19.0%
Not at all 62.1% 61.7%
15) Have you had difficulty getting the baby to take the breast?

Not at all 55.4% 54.5%
16) Have you felt you felt you weren't making enough milk for baby?

Not at all 49.8% 40.8%
17) Have sore nipples been a problem for you?
Yes, a severe problem 20.0% 17.7%

Yes, a moderate problem 14.8% 18.4%
Yes, a minor problem 26.5% 23.9%
Not a problem for me 38.7% 40.0%

Table 3.5.10b
Mann-Whitney Test for mean rank of common problem scores1
2
(644 women who ever put the baby to the breast to feed)
Randomisation Number of Mean Rank Significance of

group women Mann-Whitney
test
Have you worried that your baby Intervention 314 313.88
may not be gaining enough weight? Control 311 312.11
Total 625
p = 0.887
Have you had difficulty getting the Intervention 307 312.13
baby to take the breast? Control 308 303.88
Total 615
p = 0.526
Have you felt you felt you weren't Intervention 309 321.91
making enough milk for baby? Control 306 293.95
Total 615 p = 0.038
Have sore nipples been a problem Intervention 310 308.80
for you? Control 310 312.20
Total 620
p = 0.805
1
Women's responses were ranked from 1 to 4 as shown in table 3.5.10a. Thus a low score (and
resultant rank) indicates that a woman reported a more severe problem.
2
644 women who initiated breastfeeding were included in the analysis, but data were missing for
19 to 29 women who did not answer these questions.
3.5.11 Incidence and prevalence of breastfeeding in observation group
Women in the "Observation group" were excluded from the main study because they
planned to contact a counsellor when originally recruited to the study. As has been
reported in section 3.3.6, they differed from those recruited to the main trial in terms
of educational attainment and social class. Table 3.5.11 compares the incidence and
prevalence of breastfeeding in the observation group with those recruited to the main
trial. It shows that women in the observation group were significantly more likely to
be giving any breast milk at both six weeks and four months, and were more likely to
be breastfeeding fully at all ages from 7 days to four months.

Table 3.5.11
Comparison of feeding rates between Observation Group and Women in main trial
RCT Observation Group Relative Risk

(95% confidence
interval)
Ever put baby to breast 95.7% (645/674) 99.3% (143/144) 0.96 (0.80 - 1.16)
Breastfeeding at 6 weeks 63.9% (431/674) 79.2% (114/144) 0.81 (0.66 - 1.00)
Breastfeeding at 4 months 44.1% (274/622) 61.7% (79/128) 0.71 (0.55 - 0.93)
Any bottle at 7 days 36.6% (246/673) 22.2% (32/144) 1.64 (1.14 - 2.46)
Any bottle at 6 weeks 62.6% (422/674) 38.9% (56/144) 1.61 (1.22 - 2.17)
Any bottle at 4 months 76.7% (477/622) 60.2% (77/128) 1.27 (1.00 - 1.64)
3.5.12 Summary
Although slightly more intervention than control women breastfed to four months,
these differences were not statistically significant. (46.1% vs 42.3%; RR 1.09; 95% CI
0.86 - 1.39). Similarly, although the differences in the proportion giving any bottle
feeds at four months were a little larger, these were also not significant. (73.9% vs
79.4%; RR 0.93; 95% CI 0.77 - 1.12).
There was a baseline imbalance between the treatment allocation groups in that more
of the intervention group were uncertain about whether to breastfeed or not.
Sensitivity analyses using binary regression and Cox regression suggested that the
intervention was more effective, particularly for full breastfeeding, than appeared
when no allowance was made for the imbalance in the distribution of undecided
women. However, the differences noted were not significant at the 5% level.

The intervention did not significantly influence women's satisfaction with breast-
feeding, but those in the intervention group were significantly less likely to feel they
were not making enough milk. (Mann-Whitney U test, P=0.038).
Comparisons with the Observation Group, who were excluded because they planned
to see a counsellor anyway, showed that they were significantly more likely to
breastfeed, both partially and fully.

Chapter 3.6
RESULTS OF STUDY TWO: COUNSELLORS' PERSPECTIVES ON THE STUDY
3.6.1 Introduction
Reviewing the individual contacts between counsellors and study women provided
valuable information about the delivery and content of the intervention, but it was
also possible to ask the counsellors about their experiences in the study and how they
saw the intervention they delivered. To do this, three review meetings were held
during the course of the study and one at the end to discuss the findings with
counsellors.
3.6.2 The content of additional support
In the discussions, the counsellors were asked to describe the content of their contacts
with women. Their comments were summarised into brief descriptions of the
antenatal and postnatal support, which are shown in figures 3.6.2a and 3.6.2b. These
were then sent to counsellors for further comment.
Fig 3.6.2a
Counsellors' perspectives on antenatal support
The aim of antenatal support was to enable women to make informed decisions. It
involved giving women time to discuss their experiences of, and feelings about
breastfeeding. Women's concerns often included the following topics:
• Expectations
• The benefits of breastfeeding
• Explanation of how breasts make milk and babies feed
• How to do it - learning the skills
• What may be difficult and how to overcome the more common problems
• Support from other sources
• Role of the counsellor and her availability
The antenatal support was usually in the woman's own home, at a time convenient to
her, lasting on average 40 minutes.

3.6 Study Two Results: Counsellors' perspectives on the study 211
Fig 3.6.2b
Counsellors' perspectives on postnatal support
Postnatal support differed from antenatal support in that it usually involved more
listening and responding to the woman's needs as she perceived them. It included:
• Time to discuss concerns, worries and difficulties

• Reassurance and encouragement
• Practical help and suggestions
• Sources of postnatal support
• Exploration of the choices open to women.
In some cases, the support was over the telephone, while in others it involved a home
visit or a combination of the two.
The main difference between the role of counsellor / supporter within the NCT/BfN work
and the study is that the NCT/BfN support is usually mother-led; i.e. the woman calls
the counsellor and seeks support. In the study, the support was often counsellor-led;
the counsellor contacted the mother.
3.6.3 The role and experiences of the counsellors
The counsellors welcomed the opportunity to discuss their experiences as part of the
study and made wide-ranging comments, which may be of interest to others
evaluating "human interventions". These were taken from notes at the meetings, but
not analysed more formally or checked with participants.
The clientele:
The counsellors commented that they saw women from a wider range of social groups
as part of the study than they did in their routine counselling contacts.
While they felt that some women saw them as volunteer counsellors, they felt that
others saw them in a more "professional" role and that the study played a role in
"professionalising" them. When the issues about being a volunteer were discussed,
some counsellors felt that some women were reticent about troubling a volunteer with

their concerns. Although they tried to adopt a non-directive stance, they felt they
were sometimes seen as being there to promote breastfeeding.
Relations with professionals:
A few of the counsellors also worked as midwives or health visitors and were able to
see the counsellor's role from both perspectives. They perceived the key difference
between their input and that of the community midwives to be that they had more
time. As one counsellor who was also a midwife said, "I'd love to be able to give
women an hour." Because of this, women were able to raise a range of issues they
did not have time to discuss with their midwives. They were however also aware that
as counsellors, they were approached less often during the first few days than were
the hospital staff or community midwives.
Talking more generally about relations between breastfeeding counsellors and
midwives, the counsellors were concerned that increasing their input into hospital
settings might de-skill hospital staff. When they had established visiting
arrangements, they were aware of ward staff saving up work for the counsellor and
several counsellors had received calls from women during the first ten days who
reported that "the midwife said you'd come and help me."
The counsellors were keen not to undermine the role of midwives and although they
had little direct contact with them, were not aware of problems.

Comments on the study organisation:
When asked for feedback on the organisation of the study, the counsellors commented
that they had not been notified early enough when women had delivered. They saw
this as important, because many women did not contact them postnatally and
suggested that it might have been worth giving women a stamped addressed postcard
which they could use to report the delivery.
Being a counsellor:
Although the counsellors had enjoyed taking part in the study, and the postnatal visits
had not been too much of a burden, they did discuss some of the disadvantages of
being a breastfeeding counsellor. Most used an answerphone so they did not have to
take calls at an inconvenient time, and liked the "Supporterline", established by the
Breastfeeding Network during the course of the study. This automatically passes
calls down a list if the first counsellor does not answer. One said, "If I feel I can't
answer my phone in case it's a breastfeeding call, I'd have to stop being a
counsellor." Another commented that she limited the numbers of antenatal classes
she was prepared to speak at, as a means to limit the numbers of postnatal calls she
received.
Balancing their role as a mother with the commitment to counselling was sometimes
difficult, and one counsellor gave a graphic description of an occasion when she took
her own two-year-old on a breastfeeding visit.

Future of breastfeeding support:
Although this was not discussed at length, the counsellors did have views on how
breastfeeding support should develop. They identified a mismatch between the
numbers of women needing help and what it was possible for volunteer counsellors to
deliver. Several supported the idea of breastfeeding support / counselling being paid
as a more formal job. They recognised however that this was at variance with the
alternative strategy of training peer counsellors, who would offer mother-to-mother
support in a less "professional" role.
3.6.4 Summary
The discussions with counsellors provide insight into their experiences in the study
and how they perceived their role. It was interesting to note that the counsellors felt
that the study had to an extent "professionalised" them. Although similar discussions
were not held with participants, these observations concur with the quantitative
findings on the uptake of counselling reported in section 3.4.2.

Chapter 3.7
DISCUSSION OF STUDY TWO
3.7.1 Evaluating a complex intervention
It is important to take into account the complexity of the intervention and the context
in which it was evaluated when interpreting the results of this study. Whereas in a trial
of a medication, the dose, mode of action and bio-availability are known and can be
standardised, "human interventions" vary widely. The relationship between
counsellors and women was central to the intervention and depended on the attitudes
of both. In particular, how women saw the counsellors was an important factor in
whether they were open to their suggestions and support.
This intervention was not delivered in a vacuum, but rather it was an attempt to
change behaviour in a maelstrom of conflicting pressures. This context has been
considered from epidemiological, sociological and theoretical perspectives in chapters
1.2, 1.3 and 1.4.
The trial's main aims were to find out whether women offered additional support
breastfed for longer and whether they were more satisfied with their experiences. In
the event, women allocated counselling fed slightly longer but these differences were
not statistically significant. (See section 3.5.7.) There were no significant differences
between the satisfaction scores recorded by control and intervention group women,
although there was a suggestion that women offered counselling might feel less
3.7 Study Two: Discussion 216

embarrassed about breastfeeding and more confident in their ability to do so. They
were also significantly less likely to feel they had insufficient milk.
After this study was conducted, the Medical Research Council published a discussion
document299 which set out a framework for the development of randomised controlled
trials of complex interventions. This emphasises that it is as important to find out
“how” a complex intervention works, as whether it is effective. Knowing what are the
“active ingredients” of the intervention may be the most useful thing in deciding
whether and how to implement an intervention in a different setting. The way study
two was designed has made it possible to address some of these questions for
breastfeeding counselling. For example, gathering information about mothers’
contacts with counsellors and their perspectives on those contacts proved invaluable.
It was also useful to understand what breastfeeding advice and support women valued
most, as described in part four. This meant that the role of breastfeeding counsellors
could be considered in the context of postnatal care as a whole (Section 5.2.1).
3.7.2 Strengths and limitations of the study
A key question about these findings is whether they arose because of methodological
problems, which caused the study to miss a real treatment effect, or whether they are a
genuine reflection of a non-effectiveness of breastfeeding support. As has been
discussed in chapter 1.5, previous research has been bedevilled by methodological
problems and the issues identified in table 1.5.7 may be of relevance in assessing this
study.

3.7.2a Design and conduct of the trial
Chapter 3.3 describes the recruitment and follow-up of participants and allows an
examination of whether bias could have arisen, whether in the randomisation, follow-
up or assessment of outcome.
Potential sources of bias
The randomisation procedure was designed to provide a balanced sample of women in
both control and intervention groups and employed separate sets of randomisation
codes in sealed envelopes for mothers of first and subsequent babies in each practice.
Despite this, there was a baseline imbalance in the numbers of women who were
uncertain about whether to breastfeed, with sixteen in the intervention, as opposed to
only six in the control group.
Sensitivity analyses were performed using logistic regression techniques to assess the
importance of this. Controlling for whether women had decided to breastfeed, or
were uncertain when recruited antenatally increased the correlation between both full
and partial breastfeeding and treatment allocation, although the differences were still
not statistically significant. However, when those who were uncertain were excluded,
survival analysis detected an association that was just significant at the 5% level
between being allocated to counselling and the duration of full breastfeeding. Taken
as a whole, these results suggest that the counselling probably did have a small effect
on the length of time for which women breastfed fully.

It is unclear how the baseline imbalance arose. In particular, there was no evidence of
women who were uncertain and recruited to the control group being withdrawn or lost
to follow-up. When the research team discussed the issue, we were unable to identify
any way that the integrity of the randomisation was breached and it seems likely that
this arose by chance.
Ideally, randomised controlled trials should be "double-blind", with the treatment
allocation concealed from both the subject and the person assessing the outcome.300
Stephenson and Imrie281 however argue that blinding may not be appropriate for trials
of behavioural interventions, but that ensuring blind assessment of outcome is
essential. Almost all studies of breastfeeding support have not concealed treatment
allocations from subjects, apart from two cluster-randomised trials in which the issue
did not arise.205 22 In study two, blind assessment of outcome was largely achieved by
the use of standardised forms and the counsellors playing no role in assessing
outcome, but no attempt was made to conceal treatment allocations from subjects.
(See section 3.2.10)
One question raised by this discussion on blinding is whether knowing they were
allocated to the control group influenced women to behave differently. Perhaps some
were spurred to breastfeed with greater commitment. It is impossible to assess this
from the available data, but it was interesting that, having started out with no plan to
contact a counsellor, 14% of the control group then did attempt to do so. (See section
3.5.4).

As reported in section 3.3.9, there were no differences between the follow-up of
control and intervention women, so it seems very unlikely that differential follow-up
could have affected the results.
Sample size and recruitment
Recruitment to the trial proved to be a Herculean task, taking a year longer than
expected and requiring the recruitment of additional practices to boost numbers. In
the event, although the intended sample size was 854, only 720 participants were
recruited. However, when compared with previous studies, this was one of the largest
randomised controlled trials of breastfeeding support conducted anywhere in the
world.
Looking back, it is worth asking whether the task could have been made easier. The
main difficulties faced in recruitment were due to problems in the practices.
Successful recruitment depended on too many people - receptionists, midwives and
doctors all understanding and following the study protocol. Despite repeated visits
with bagels, prize draws for champagne and truffles, progress reports, Christmas cards
and smiles in the face of adversity, it was not possible to keep all those who needed to
know about the study engaged. In particular, when new receptionists or midwives
began working in the practices, the research team often did not hear about their arrival
for some time. It would have been better to concentrate on fewer practices, to have
visited more often, ideally during antenatal clinics and to have monitored their
performance more closely.

One alternative would have been to recruit through hospital antenatal clinics, but
because so much antenatal care is now delivered in the community, that might have
biased the sample towards those at higher risk of obstetric complications and made
the findings less applicable to primary care settings. Basing the study in hospital
would have also lost the contribution that practices made to follow-up.
Many of the difficulties encountered have been reported by others who have
attempted randomised controlled trials in primary care.301 For example, Wilson et
al302 reported on factors that influenced recruitment to a trial of dyspepsia
management. They identified a need to compromise between including sufficient
practitioners to recruit a representative cohort of patients and the effort involved in
recruiting and motivating those practitioners. Another issue is that practitioners may
be reluctant to randomise their patients to receive care they perceive to be less good,
as Fairhurst and Dowrick303 found when they attempted to compare counselling with
normal care. One approach to recruitment, which should perhaps be considered more
304
often, is to pay practitioners financial incentives. Foy et al have advocated this
and although the practice is common in commercial research it has yet to gain
widespread acceptance in health services research. Although in this study each
practice was paid £100 to cover their administrative expenses, this was not linked to
recruitment and the amount was not large enough to influence their behaviour. Also,
because the money was paid to the practices, it was unlikely to reach the midwives
and receptionists on whom the study most depended.

It is also worth asking how much it mattered that the numbers recruited fell short of
the intended sample size. As a result, the study was less powerful than planned, with
a power of 74% to detect an increase in the proportion breastfeeding from 50% to
60%, compared with an intended power of 80%. However, the main conclusion, that
if counselling made a difference it was quite small, would probably remain. One final
observation is that evidence-based healthcare increasingly draws on the results of a
range of studies. Whether or not an individual trial is large enough to achieve
statistical significance may matter less when the results are included in meta-analyses.
3.7.2b Assessment of outcome
Considerable effort went into designing the various questionnaires employed in the
study, so it is worth considering how well they performed. The high response rate -
albeit boosted by telephone follow-up suggests they were acceptable to the women
who were asked to complete them (Section 3.3.9). In contrast, the feeding diaries
were returned by less than 40% of women and contributed little additional
information.
One omission was that three and four month questionnaires did not ask women
whether had introduced other foods, although this was included on the diaries.
Over the last 20 years, a number of authors have reported on questionnaires to assess
aspects of breastfeeding, including knowledge305, health beliefs,306 attitudes,307 308 309
310 311
social support312 159 158
and satisfaction.313 280 294
A number of these187 188
have
been based on work in social psychology such as the Theory of Planned Behaviour185

which has been discussed in section 1.4.4b. It was possible to incorporate questions
on maternal satisfaction with breastfeeding developed by Leff et al 293 294 into the six-
week questionnaire, but there is a question whether it would have been worth drawing
more on other questionnaires in the design of the instruments to measure outcomes in
this trial (Section 3.2.9). Although the validity, internal consistency and reliability of
most of these have been assessed, much of this work was done in the United States
and may not necessarily be valid in the United Kingdom. Additionally some was not
published until after 1994 - 1995 when I designed the questionnaires.
It might have been helpful to have included more validated questions on the type of
support women received and their perspectives on this, in place of some of the open
questions on advice they received for individual problems. However, the six-week
questionnaire was already quite long and it was clear from the pilot study that it
needed to be as easy as possible to complete (Appendix B).
3.7.2c The intervention
The intervention tested in this study was based on breastfeeding counsellors carrying
out an antenatal home visit and offering postnatal support if requested. As discussed
in section 1.7, the aim of counselling was to help mothers who wanted to breastfeed to
achieve their infant feeding goals. Details of the intended contact were provided in
section 3.2.6, while section 3.4.2 reported the actual contacts that took place. In the
event, 76% of women met the counsellor before the birth, with a further 4% having
contact by telephone. After the birth 19% saw the counsellor and a further 42% were

in contact by telephone only. Although all women in the intervention group were
offered support, 8% of them had no contact at any stage.
Those who were in contact with the counsellor were very positive about their
experiences. They liked the way the counsellors gave them time and listened, gave
them useful explanations and practical advice on feeding problems (Section 3.4.3). In
response to a separate question on who gave them the advice they considered to be
most useful, more referred to the counsellors than any other source (Section 3.4.4).
Examining the uptake of counselling highlights a contrast between those who did and
those who did not establish effective contact with the counsellor, because it appears
that those who did so were more motivated to breastfeed than those who did not. This
echoed findings reported in section 3.5.11, that women in the observation group, who
planned antenatally to see a counsellor, breastfed for longer than those who had not.
The effectiveness of telephone support merits attention, not least because of the
proposal to expand the coverage of the BfN "Supporterline". In this study, only 19%
of women were visited, but twice as many women were supported by telephone.
However more than a third of those supported by telephone stopped breastfeeding by
six weeks. This also relates to findings reported later in study three that women saw
having someone to spend time and show them how to position the baby at the breast
as very important. Comparing data from a number of studies, Sikorski and Renfrew's
review199 concluded that strategies that relied on telephone contact were less effective
than those based primarily on face-to-face contact. In the present study, the relatively

low visiting rate does suggest that counsellors were less effective than they might
have been.
In the focus groups, some counsellors said that they did not hear soon enough that
women had delivered. This meant that they were often unable to offer women
support in the crucial first days, but only heard when problems such as sore nipples
had become established. To overcome this, they encouraged women to contact them
as soon as the baby was born - "to let them know how the birth had gone." These
delays reflected difficulties organising a trial in parallel with existing services and it
would have been easier if the counsellors had been routinely notified of deliveries, as
midwives are. An alternative suggestion made was to give mothers stamped
postcards to inform the counsellors when their babies were born.
Although it is tempting to assume that had the counsellors seen more women
postnatally, they would have been more effective, this does not necessarily follow.
Data from two sources suggest that, given the constraints of the study design, they
were able to support the majority of those who wanted help. Firstly, those who tried
to contact a counsellor reported little difficulty doing so and secondly, the counsellors
said they felt women were sometimes reluctant to ask for help. Thus the uptake of
counselling may offer a realistic estimate of the proportion of women who might
contact a counsellor postnatally if such a service were more widely available.
This prompts questions about how postnatal support might have been made more
acceptable to women. One issue, alluded to in the focus groups with counsellors, was

that some mothers did not want to trouble a volunteer with their problems. Although
further clarification of the counsellor's role might have helped, this suggests that
perhaps some people have a more fundamental ambivalence about being helped by
volunteers. One strategy to "normalise" help from the counsellor would have been to
strengthen links between the counsellors and midwives, so that counsellors were more
closely integrated into postnatal care, rather than working in isolation as in this study.
It is also worth reviewing the decision to make postnatal support an optional, rather
than universal component of the intervention, as had been originally envisaged. When
this issue was discussed with counsellors and with the NCT Breastfeeding Promotion
Group in 1994, they felt strongly that counselling should be responsive, but never
imposed on women. While the antenatal home visit was necessary to establish
contact, they saw contacting women routinely, (unless individually agreed in advance)
as inappropriate for a counsellor. While it seems possible that a routine postnatal
contact might have enabled more women to continue breastfeeding, the results would
have revealed less about the likely impact of breastfeeding counselling outside the
setting of a randomised controlled trial.
Some studies of postnatal support with breastfeeding have involved a programme of
regular postnatal visits, which appears to have been effective in promoting the
maintenance of breastfeeding.205;215;314 Writing about the results of her 1981 study of
support by midwifery sisters, Renfrew emphasised that giving mothers reliable
fortnightly appointments encouraged them to persevere until the next postnatal visit,
even if they encountered difficulties.314 Such an approach may be more effective, but

also requires more resources than offering just one postnatal visit, or support only
when requested. It also moves further from the non-directive but responsive approach
currently taken by breastfeeding counsellors.
3.7.2d Summary of factors which may have influenced the result
The preceding sections have considered a range of factors that may have influenced
the result and these are summarised in table 3.7.2d under three headings:
• Factors in the conduct of the study that may have reduced its ability to detect an
effect.
• Aspects of this intervention which may have reduced its effectiveness.
• Factors which reduce the ability of any health promotion intervention to alter
behaviour.
Table 3.7.2d
Factors which may have influenced the results
Potential influences Likely impact

Factors in the conduct Randomisation bias Sensitivity analyses suggest not a large
of the study that may factor.
have reduced its ability Hawthorne effect on control 14% of controls tried to contact a
to detect an effect. group motivation counsellor.
Sample size Reduced power, but unlikely to have
influenced result
Delays notifying Did not stop women calling counsellor, but
counsellors of delivery. reduced potential for counsellors to contact
women proactively if previously asked to do
so.
Aspects of this Postnatal care “on request” Hard to assess as it depends on reasons
intervention which may rather than as routine women did not ask for help. See section
have reduced its contact 3.7.2c.
effectiveness. Socio-cultural barriers May have had some impact. Section 3.4.3,
between counsellors and (discussed below in 3.7.3), indicated that
clients. disadvantaged groups were less likely to
contact a counsellor. See also section
3.7.2c on volunteer role. However women’s
comments on counsellors were positive.
Factors which reduce Social and attitudinal Interventions focussed on the individual can
the ability of any health factors which exert strong only influence some of the factors that
promotion intervention influences on infant feeding influence feeding behaviour. Chapters 1.3
to alter behaviour. behaviour. and 1.4 discuss this.

3.7.3 The results in context
The results of study two may be compared with those in Sikorski and Renfrew's
review,199 which have been reported in sections 1.5.6 and 1.5.7 and appendix H. Table
3.7.3 reports this comparison and shows that in my study, more women stopped by
four months. The difference between the cessation rates for intervention and control
women was less than in the meta-analysis.
Table 3.7.3
Comparison of proportions stopping breastfeeding with other studies
Results of study two Meta-analysis by Sikorski and

199
Renfrew
Control Intervention Control Intervention
Six weeks 36.6% (123/336) 35.1% (118/336) - -
Two months - - 41.2% (275/667) 30.3% (203/671)
Four months 57.7% (179/310) 53.9% (167/310) 50.8% (416/819) 35.4% (295/834)
It is worth noting that different studies gathered data at different stages and that the
Cochrane review does not report on feeding at six weeks. More studies reported on
feeding at four than two months, which explains the larger numbers included at that
stage. Because only three studies reported exclusive breastfeeding rates, it is less
useful to make comparisons of this (Appendix H).
It is also worth picking up a separate topic, the issue of engaging women from
disadvantaged groups. This study did succeed in recruiting disadvantaged women, as
reported in section 3.3.6. They were however less likely to take up the offer of
postnatal support, as reported in section 3.4.3. Reid and Glazener315 also found that

disadvantaged mothers were reluctant to attend postnatal support groups and these
findings emphasise the difficulty providing social support to those who might benefit
most from it.
3.7.4 Randomised controlled trials to assess behavioural interventions
This study was designed as a randomised controlled trial, but is the RCT really the
best way to evaluate a behavioural intervention? As has been reported in section
1.4.8, MacLean194 has challenged this view, arguing that the social context is so
important that cannot be objectively manipulated like other phenomena. Stephenson
and Imrie281 however argue that randomised controlled trials are just as valuable, but
harder to conduct for behavioural interventions. As Oakley316 recounts in her
historical account of the role of experimental methods in the development of social
interventions, the RCT has been both in and out of favour as a means to evaluate
social interventions over the last 120 years.
The RCT had its origins in the work of the American psychologist, C. S. Pierce in the
1880s and educationalists such as Winch and McCall in the early years of the 20th
century. Between the 1960s and early 1980s, RCTs were conducted in the United
States to evaluate a range of social interventions, such as support for disadvantaged
workers or prisoners after release from custody. The approach is also reflected in the
evaluation of a number of breastfeeding support initiatives through the WIC (Women,
Infants and Children) programme, which provides health care to disadvantaged
mothers and their children. 217 208 317 318 319

In the 1980s, the RCT has however declined in popularity as a means to evaluate
social interventions. Oakley highlights three reasons for this; first, disenchantment
with the negative results that many of the studies reported; second, policy makers
were reluctant to accept delays while schemes they favoured were evaluated and third,
a reluctance to face the extent of social change that might be required if the findings
from social research were to be acted on. It is also worth remembering that at the
time, the political agenda was being set by Reagan and Thatcher - conviction
politicians who were reluctant to see their ideas for social change bogged down in
lengthy evaluations.

PART FOUR
FURTHER ANALYSIS OF STUDY TWO:

QUALITATIVE ANALYSIS OF WHAT INFORMATION AND SUPPORT WOMEN
WANT WITH BREASTFEEDING
4.1 Qualitative analysis: Rationale and objectives 231

FURTHER ANALYSIS OF STUDY TWO:
QUALITATIVE ANALYSIS OF WHAT INFORMATION AND SUPPORT WOMEN
WANT WITH BREASTFEEDING
Chapter 4.1
RATIONALE AND OBJECTIVES
4.1.1 Introduction:
Evidence from the studies reported in sections two and three of this thesis, as well as a
range of other reports, confirm that many women do seek advice and support with
breastfeeding. Antenatally, 58% of UK mothers in the 1995 national survey1 reported
discussing breastfeeding with a professional. Although in that survey, most women
who had breastfeeding problems did receive support with these, there is evidence that
mothers often receive conflicting advice or feel unsupported in the early postnatal
period.172 173
The Audit Commission117 has also expressed concern about the
fragmentation of postnatal care, about which the women reported more negative
comments than other aspects of maternity services.
It is perhaps surprising that, despite the evidence that many women continue to have
difficulties with breastfeeding and the number of studies which have documented
societal influences on infant feeding behaviour152, few have asked women their views
on the best way to support those who want to breastfeed. This is a particularly
striking omission, given the importance which women's attitudes have been shown to
have in determining their infant feeding behaviour and the limited effectiveness of
interventions to support breastfeeding.

Conducting the randomised controlled trial presented an ideal opportunity to ask
women about the information, advice and support they received and their perspectives
on that support. The value of this was recognised in the initial proposal for the trial
and open questions about what advice they saw as “most helpful” and “least helpful”
were included in the six week postnatal questionnaire, along with a page at the back
for them to “add anything else they felt was important”.
When the first completed six-week questionnaires were returned to the study office, it
became clear that women had used the open questions to describe not just the support
they received, but also how they felt about it. The richness of their comments could
not easily be coded using the coding frames developed in study one because they
focussed on counting the number of times women received particular advice, rather
than understanding their perspectives on that advice and support. Because of this, it
seemed most appropriate to employ qualitative techniques to conduct a thematic
analysis, keeping women’s comments in mind throughout the analysis.
The qualitative analyses were funded by additional grants from the King’s Fund, the
NHS Central and East London Education Consortium (CELEC) and further support
from the NHS R&D Support Funding to Statham Grove Surgery as a Research
Practice.
4.1.2 Objectives:
To examine women’s experiences of breastfeeding and their views on the advice,
information and support they receive.

Chapter 4.2
PARTICIPANTS AND METHODS
4.2.1 Recruitment of participants
Participants were recruited as part of the randomised controlled trial of support from
breastfeeding counsellors reported in study two and therefore needed to meet the
inclusion criteria for the trial. (See section 3.2.3) Additionally, participants needed to
return the six-week postnatal questionnaire and report that they had put the baby to
the breast to feed at any stage.
As described in section 3.2.2, women booked for antenatal care were identified from
the practices’ maternity claims and questionnaires were left in their notes for them to
complete at their first appointment with a midwife or general practitioner after 28
weeks. Antenatal forms were completed by 2,439 women, 720 (29.5%) of whom were
eligible to take part and were recruited. Of these, 685 responded to the six-week
postnatal questionnaire and 654 reported that they had ever breastfed the baby. Table
4.2.1 provides demographic information on these women.
4.2 Qualitative analysis: Participants and methods 234

Table 4.2.1
Characteristics of participating women:
Number %
1
Age: n = 649
Under 20 36 5.5%
20 – 24 101 15.6%
25 – 29 214 33.0%
30 – 34 207 31.9%
Over 35 91 14.0%
Previous children: n = 654
No previous children 492 75.2%
Had previous children 162 24.8%
Age Completed Education:

n = 639
16 or under 192 30.1%
17 – 18 188 29.4%
19 and over 259 40.5%

2
Social class: n = 626
I & II 240 38.7%
III non-manual 116 18.7%
III manual 155 25.0%
IV & V 82 13.2%
Other 27 4.4%
Ethnicity: n = 640
UK & other white 440 68.8%
African & Caribbean 103 16.1%
Indian Subcontinent 50 7.8%
Other 47 7.3%
1
Mean age of respondents was 28 years, 10 months.
2
Social class was coded using the Registrar General’s classification, based on the partner’s occupation, or
if no partner’s occupation was listed, the woman’s own. Coding primarily by partner’s occupation enabled
significantly more women to be categorised than was possible using women’s own occupations.
Although all the women whose comments were included in the qualitative
analyses had begun breastfeeding, by six weeks most had introduced at least

some formula feeds. Two hundred and forty nine women (38%) were fully
breastfeeding, 183 (28%) were giving both breast and bottle feeds, while 222
(33.9%) were exclusively bottle feeding.
4.2.2 Content of postnatal questionnaire
The postnatal questionnaire was designed to gather information on a range of topics,
including women's experience of birth, feeding behaviour, satisfaction with
breastfeeding and advice they received for common problems. Quantitative data from
this are reported in Part three of this thesis.
The findings reported here are based on women's responses to two open questions
about the advice they received and comments they wrote when asked to add anything
else they felt was important on the last page of the questionnaire (fig 4.2.2). Rajan172
had successfully used this approach in a study of women's experiences of postnatal
care and it was felt that a similar approach might elucidate their experiences of
breastfeeding support. Examples of women’s unedited responses are given in
appendix K to illustrate the nature of the data available for analysis.
Figure 4.2.2
Questions from postnatal questionnaire which asked mothers to assess the
advice and support they received.
Of all the advice you received about breastfeeding, which was most helpful?
Of all the advice you have received about breastfeeding which was least helpful?
Please add anything else you feel is important:-

4.2.3 Administration of postnatal questionnaire
Post-natal questionnaires were left in each new baby's medical notes for participants
to complete at their baby's six-week check-up as described in section 3.2.9. If they
had not returned this by 8 weeks, they were sent the first of two postal reminders.
Seventy four per cent of women returned completed questionnaires by post, but it
became apparent that those who had switched to formula were less likely to do so. As
these women's opinions on the support that they had received were particularly
important, non-responders were asked to complete the questionnaire by telephone,
which a further 20% did. Only 6% of women were lost to follow-up and the majority
of these had moved away.
4.2.4 Analysis
The analysis of women’s comments on the care they had received was based on the
techniques for developing “Grounded Theory” described by Strauss and Corbin.320
This conceptual approach involves description, conceptual ordering and then
theorising. They identify description as the first stage in qualitative analysis –
depicting, telling a story, without stepping back to interpret events. Conceptual
ordering then involves classifying events and objects along various explicitly stated
dimensions without necessarily relating the classifications to each other to form an
overarching explanatory scheme. Theorising is the act of constructing from data an
explanatory scheme that systematically integrates various concepts through
statements of relationship.

Bryman and Burgess321 have however argued that few qualitative researchers who cite
“Grounded Theory” either remain grounded in their data – returning to data collection
during iterative stages of analysis, or truly develop theory. Whilst the “static” nature
of the data, gathered from women’s responses to questionnaires, limited the
opportunity to check emerging interpretations with participants, as might be possible
in an interview-based study, the analysis did follow the stages Strauss and Corbin320
advocated.
All the women's responses were first transcribed, printed and cut out for sorting prior
to conducting the analysis. Three researchersxxiv began the familiarisation process by
reading the transcripts independently to identify initial themes from the text. Then,
with a basic framework of themes, they worked together, searching for patterns, and
comparing the experiences, feelings and perceptions within women's accounts until a
consistent thematic framework developed. This meant that each individual response
could fit into a particular category with no new themes emerging.
Initially, women's responses to the three open questions were analysed separately,
with each question generating around ten to fifteen different topics. At this stage, the
paper slips with women’s comments were stuck down onto separate cards for each
topic. The three lists of topics were then considered as a whole and regrouped into
summary themes for the advice and support women found helpful or unhelpful (Fig
xxiv
Jonathan Graffy, Jane Taylor and Janet Turner undertook this work. JG is a general practitioner and director
of a primary care research network. Jane Taylor has a background in youth work and had worked in market
research after undertaking an MSc in Social Research Methods. She worked part-time on the RCT and this study
for three and a half years. Janet Turner had an MA in women’s studies and a background in women’s health
research. She assisted with managing and analysisng the qualitative data for about four months.

4.3.1a and b). From these, five key constructs were identified, reflecting the main
priorities mothers expressed for advice and support with breastfeeding.
Validity
A number of approaches were adopted to enhance the validity of the findings. Using
triangulation, a technique described by Denzin322 to compare interpretations made by
considering the same phenomenon from different perspectives, the analyses of what
women found most and least helpful were compared with conclusions drawn from
their free text comments. Similarly, the researchers compared their individual
interpretations of the data throughout the analysis.
During the analysis, the researchers approached the data from their personal and
disciplinary perspectives. Bringing different, but complementary perspectives to the
discussion meant that on a number of occasions it was important to go back to
women’s words and agree what they meant before agreeing an interpretation. Two
particular debates deserve mention. Firstly, during the first stages of the analysis, it
seemed that “time” was an important factor. This arose in a number of ways – when
people asked for help, time spent feeding and mothers valuing it when someone spent
time with them during early feeds. However in discussion, it was agreed that time was
a factor which qualified each of the main themes identified, rather than a theme in its
own right. Secondly, at one stage, Jane Taylor was concerned that women’s need for
“replenishment” and to “look after themselves” was not fully reflected in the model
being developed. When women’s responses were reviewed, it appeared that this was
only rarely mentioned, but it had been identified as important by other

researchers.168;323;324 It was agreed that this point would be identified in comparisons
of between the findings and other literature, but that general advice to look after their
own needs would be included within the theme of “effective advice and suggestions”.
This also reflected the way the questions women were asked focussed on advice and
support as transactions, rather than conceptualising support as being looked after in a
supportive environment.
In order to check the findings accurately reflected women's views, the last eighty
participants were sent a two-page summary, accompanied by a structured response
sheet (fig 4.2.4). Additionally, the findings were compared with other literature on
breastfeeding support and this is examined in detail in section 4.4.3 of the discussion.
Figure 4.2.4
Questions on the validation response sheet:
Do you agree with the report? Is there anything important that we have missed out?
Is there anything we should change?
How did you feel about taking part in this research? Is there anything we should have
done differently?

Chapter 4.3
RESULTS OF QUALITATIVE ANALYSIS
4.3.1 Women's experiences of advice and support with breastfeeding
The responses women gave to the three open questions conveyed a rich account of
their experiences and how they felt about the support they had received. In particular,
they often used the invitation to "add anything else they felt was important" on the
back page to tell the story of what happened to them. Many offered suggestions on
how breastfeeding support could be improved. This narrative text provided fuller
personal accounts and helped show what the advice and support meant to them. They
often expressed their need for time and support from those around them, so that they
felt cared for and able to meet the demands of life with a baby. Those who had
stopped breastfeeding often expressed that they felt saddened and disappointed by
their experience.
One in five of the postnatal questionnaires were completed by telephone and these
proved particularly poignant. Mothers who had switched to exclusive bottle feeding
often began the interviews with some reticence, and although a few remained
defensive, most opened up and spoke freely about what happened to them. As the
interviews progressed and they were given the opportunity to talk about their
experiences and feelings, many identified factors around the birth, in their postnatal
care, or some aspect of their home life which had contributed to the difficulties they
had in sustaining breastfeeding. By the end of the interviews, many apologised for not
4.3 Results of qualitative analysis 241

returning their questionnaire and expressed regret that they had not been able to
breastfeed for as long as they had intended.xxv
Whilst many women commented positively on the advice and support they received
with breastfeeding, many felt unsupported, particularly in the first few days after
birth. It was also striking that more of the experiences women identified as "least
helpful" were about the way others had related to them, than the content of advice or
support received. These included feeling that they had not being listened to, or that
they had been pressurised to either give or not give bottle feeds to their babies.
Figure 4.3.1a
Themes identified from the text:
- What women found most helpful
Effective advice for specific concerns:

- Positioning
- Timing and duration of feeds
- Engorgement
- Expressing
- Treatments for sore nipples
Encouragement to keep going

Reassurance that what they were going through was normal
Knowing the benefits of breastfeeding
Encouragement to look after themselves, rest, relax, eat and drink well
"Permission" to give an occasional bottle, supplement or switch to formula feeding
Being able to ask questions about breastfeeding
The main themes identified from women's comments on what was most and least
helpful are listed in figures 4.3.1a and 4.3.1b.
xxv
Further information on the response rate and feeding behaviour of those who responded to the postal
questionnaire or by telephone is given in section 3.3.9.

Figure 4.3.1b
Themes identified from the text:
- What women found least helpful
Advice that didn't fit with women's own feelings or experiences on:
- Timing and duration of feeds
- Supplementing or changing to formula
- Positioning
- Treatments for sore nipples, colic and other problems
Not enough time or help with feeding

Not feeling listened to
Pressure to breastfeed. Being made to feel guilty for bottle feeding
Negative attitudes towards breastfeeding
Conflicting advice
4.3.2 Concepts that encompass the information, advice and support women want with
breastfeeding.
When the themes which emerged from the analysis of what advice and support
women found most and least helpful, and the free-text comments from the back page
were considered together, it was possible to identify five key concepts which
encompassed the information, advice and support that women want with
breastfeeding. These are discussed with illustrative accounts of their experiences and
views.
4.3.2a Information about breastfeeding and what to expect
Some women described feeling unprepared for the "realities of breastfeeding" and ill-
equipped to cope with the difficulties they faced. They wanted more information
about how to handle problems such as sore nipples, engorgement, frequent feeding

and tiredness, before their baby was born. It would have helped them to know what
they might experience in terms of pain, or the time they might need to spend
breastfeeding.
For some, it was important to have learnt about the benefits of breastfeeding. They
found that this motivated them to keep going in times of pain and difficulty and also
helped them to explain why they were breastfeeding to others who viewed it
negatively (box 4.3.2a).
Box 4.3.2a
" I could have prepared myself a bit more"
"I don't think that women are aware of just how painful breastfeeding can be. In the
leaflets it says all the encouraging things like it's good for the baby. It would be more
helpful if they were realistic and also pointed out that you have to be dedicated to keep
it up. You are tied to your baby and get little space for yourself, which can be very
exhausting for the first few weeks. If I had been more aware of this in advance, I could
have prepared myself a bit more. Giving women a full picture may discourage
breastfeeding but it's up to us to make the decision based on 'true' information."
(Respondent 1114)
"I am surprised to find that I hardly know any people who breastfed their babies, so it
was difficult to have a role model. I feel that more should be done to encourage
mothers to breastfeed at parentcraft classes." (Respondent 1892)
4.3.2b Practical help with positioning the baby to breastfeed
Women consistently reported that the most helpful advice they received was when a
midwife, breastfeeding counsellor, friend or relative had shown them how to position
the baby at the breast in the early postnatal period. The words "positioning", "latching

on" and descriptions of how to get the baby to open the mouth wide enough to get
enough of the breast into the mouth recurred in many of their accounts.
Although women detailed the care and support they received from health
professionals and others, all too often they commented that they had not received the
help they needed with breastfeeding. Staff on the postnatal ward seemed too busy and
sometimes unwilling, or unable to help mothers breastfeed. One mother commented
that although the slogan “Breast is Best” was everywhere, there was “nothing to back
it up”.
Women want practical help with positioning the baby at the breast in the early
postnatal period until they feel confident in their own ability.
Box 4.3.2b
"Someone to sit down and show me what to do"
"When my son was born he would not latch on - There was a midwife on the night shift
that sat with me very patiently for about two hours (in the) early hours of the morning -
she was the only midwife that gave me any help while in hospital. I never saw her again.
But thanks to her help and kind words, I continued." (Respondent 4920)
"I wanted someone to sit down with me and show me what to do and help me when it
wasn't working. It was all sort of, "do it like this" and then off." (Respondent 1067)
"I don't think the midwives give you enough information. They should have enough time
to sit with you. There was no help at all. It's enjoyable and it's very sad and I got
depressed when it didn't work out with me." (Respondent 5024)
4.3.2c Effective advice and suggestions
Practical tips or explanations that helped with particular issues such as breast
engorgement, sore nipples, the timing of feeds, or how to express milk were the most

helpful advice for some women. Similarly, encouragement to rest more, relax, get
comfortable when feeding and to look after themselves was welcome.
But mothers felt that conflicting or inappropriate advice was unhelpful, whether from
family, friends or professionals. Topics which caused particular confusion included
how long and often to feed for, when to switch breasts if trying to give both fore and
hind milk, and whether using nipple shields or giving supplementary bottle feeds
would undermine breastfeeding.
Box 4.3.2c
Suggestions which worked - and conflicting advice
"Express when breasts are very hard and uncomfortable and enable the baby to latch on
more easily" (Respondent 5358)
"Make sure you are relaxed, with 'facilities' at hand; i.e. cushions, drink, snack,
telephone, TV, etc. (Respondent 494)
"When I had sore nipples - nipple shields - I would highly recommend their use rather
than struggling or giving up breastfeeding." (Respondent 4804)
"In hospital she fed for over an hour. One midwife said, "keep her on there, she'll take
what she wants". Then another one said, "Oh she shouldn't be on for that long!"
Everyone was telling me different things and they were contradicting and I was very
unsure really. I left hospital very confused." (Respondent 1683)
4.3.2d Acknowledgement of mothers’ experiences and feelings
It was important for women that those giving advice should be concerned about their
feelings and respect their views. Some reported that they felt pressurised to continue
breastfeeding, or made to feel guilty or inadequate for introducing formula feeds.
Conversely, others who were struggling to sustain breastfeeding felt they had been
undermined by pressure to 'give a bottle'.

Box 4.3.2d
Pressure from all sides
"I have felt under pressure from various people to switch to bottle as they thought a baby
who sucks so much must be permanently hungry... I have found it quite hard to resist
this and explain why breastfeeding is better. It is easy to see how people might cave in
under that sort of pressure - I nearly did." (Respondent 882)
"Breastfeeding is not easy for everyone. When trying to feed my first baby in the
hospital I had great difficulty getting him to latch on or suck and I very much felt the
midwives blamed me for this. When I said to one, "It isn't easy," she replied, "of course
it's easy - all the other mothers can do it! My feeling is that the most important thing is
not to make a new mother feel inadequate or guilty in the first few days."
(Respondent 1576)
4.3.2e Reassurance and Encouragement
Women commonly reported feeling that they were not producing enough milk, or that
they were overwhelmed by the time they spent feeding, but valued reassurance that
what they were going through was normal - that they were not failing by finding it
hard going. Praise and encouragement to persevere through difficult times made a
difference to how they felt about continuing.
Box 4.3.2e
"Telling me how well I was doing"
"Helen especially went out of her way to assist me when I almost gave up breastfeeding
due to painful breasts and feeling that Harry was simply not getting enough milk. She
made me feel very comfortable and confident that I could continue breastfeeding with
bottle feeding, to assure me that Harry was getting enough." (Respondent 2110)
"I was tempted to give up when I felt too tired to feed. I received encouragement from my
partner, telling me how well I was doing. "
"I felt a bit let down. I didn't really get any help. I'd had a difficult birth. I didn't seem to
have any milk, I was really tired and I didn't wake up when he was crying and they gave
him a bottle when I was asleep. When I tried the next day, he just wouldn't take it and
when I was trying they said, "You can give him a bottle". I'd rather they had encouraged
and helped me to breastfeed." (Respondent 4127)

4.3.3 A model of the information, advice and support that women want with
breastfeeding
The five concepts reported in section 4.3.2 were considered together and displayed as
a model of the information, advice and support women want with breastfeeding. This
is shown in figure 4.3.3.
Fig 4.3.3
Information, advice and support women want with breastfeeding
Information about
Acknowledgement of breastfeeding and what to Practical help with
mother’s experiences and expect positioning
feelings
Effective Breastfeeding Support
Encouragement and
reassurance Effective advice &
suggestions
4.3.4 Comments on support from breastfeeding counsellors
Although the aim of the qualitative analysis was to consider the information and
support women wanted, based on their experience of support from all sources, half of
the women had been offered additional support by a breastfeeding counsellor.
Whether or not they continued to breastfeed, women often commented positively on

the support they received from counsellors. They liked the fact that they had
themselves breastfed and were knowledgeable, reassuring, non-judgmental and
prepared to listen to their feelings.
4.3.5 Checking the findings with participants
A two-page summary of the findings was prepared and sent, with a structured
response sheet, to the last eighty respondents for comment (figure 4.2.4). Fifty-one
(64%) of the eighty who were invited to comment returned the response sheet. As
non-responders proved difficult to contact by telephone, it is likely that many of them
had moved.
All agreed with the report, but they often emphasised how a particular issue, such as
the need for realistic information, or practical help getting the baby to feed had
mattered most to them. Although the majority did not want to see any changes, some
did make suggestions. One woman commented that the study focussed on the first
few weeks of parenthood, but that women needed different advice later when they
returned to work, wanted to express milk, or were considering weaning. Another had
found nipple shields particularly helpful and another wanted a more explicit reference
to the role that husbands and partners can play in supporting breastfeeding.
The women who responded were positive about taking part in the study, and several
commented that they were pleased to be able to help others by doing so. One had
however found it a nuisance to complete the forms when she had a new baby to look

after, but emphasised that she hoped the results of the research would benefit other
women.

Chapter 4.4
DISCUSSION OF QUALITATIVE FINDINGS
4.4.1 Introduction
Although many women in this study reported positive experiences with breastfeeding
support, many felt let down, judged, or neglected when they needed help. This
mattered to them, as shown by the strength of feeling conveyed in the narrative
accounts they wrote on the back pages of the questionnaires. By exploring their
comments on their experiences, we were able to identify both what advice and support
they wanted with breastfeeding, and also how they wanted that advice and support to
be given.
4.4.2 Strengths and limitations
Whereas most qualitative studies rely on exploring issues in depth with a small
number of participants, this draws on the experiences of a much larger number. There
are both benefits and disadvantages to the approach adopted. The demographic data
on participants and the 94% response rate to the postnatal questionnaire suggest that
the study includes a wide range of women's experiences, with the exception of those
who did not have enough English to participate and were excluded because of the
requirements of the study design. Similarly, the finding that 34% of participants had
stopped breastfeeding and that 28% were giving both breast and bottle feeds suggests
that the study did capture the views of women who might most need breastfeeding
support.

The context in which women were asked their comments deserves mention. Basing
the findings on women's experiences of advice and support for the problems they
actually faced, rather than what they thought might be helpful in a hypothetical
situation, also makes it more likely that the conclusions drawn reflect their perceived
needs in breastfeeding a new baby. Because they were recruited as part of a
randomised controlled trial, half of the women had been seen antenatally and offered
postnatal support by a breastfeeding counsellor. Although the counsellors' advice or
support was often reported as helpful, the analysis deliberately focussed on the
content, rather than source of advice. Additionally, the women's accounts revealed
that counsellors were only one of a number of sources of advice for those in the
intervention group.
One limitation of the approach adopted was the static nature of the data, drawn from
responses to questionnaires, which meant it was not possible to test interpretations
with participants. Instead, it was necessary to rely on discussions within the research
team and a more formal approach to checking our findings when the analysis was
largely complete. The endorsement received in that validation survey does however
provide reassurance that the conclusions are valid for the support women want in the
first six weeks. As has been noted, issues such as expressing, weaning and returning
to work were not commonly raised, but might have been identified as more important
had the study been repeated later in the baby's first year.
Respondent validation has been recommended by Reason and Rowan,325 because it
allows researchers to refine their conclusions in the light of subjects’ reactions to

tentative results. On the other hand, Silverman326 has advocated caution in the use of
the technique, partly because of Abrams’327 argument that it is “only possible if the
results are compatible with the self-image of the respondents”. In this exploration of
women’s perspectives on breastfeeding support, the respondent validation did
however both endorse and contribute to the findings.
4.4.3 The findings in context:
Comparing the findings of this study with those of other researchers who have
adopted different perspectives also provides an opportunity to test the validity of the
conclusions drawn.
4.4.3a Ante-natal information
The request for more information about breastfeeding reflects the reality that many
new mothers have had little contact with breastfeeding. As Hoddinott154 identified,
there is a collective loss of embodied knowledge of breastfeeding as fewer mothers
have witnessed others breastfeeding. Women often feel unprepared and although they
search for information to prepare themselves, media accounts and health promotion
literature often portray breastfeeding as a mechanism to produce the best milk, rather
than a lived experience. Britton328 argued that neglecting the variability and personal
significance of women's experiences leads to conflicts between their expectations and
the reality of breastfeeding. In their detailed account of women's experiences,
Schmied and Barclay86 reported that for some, breastfeeding is painful, disruptive and
unpleasant, but this is not reflected in the images of contented mothers and babies
found in books and leaflets about parenting. O'Connor329 suggested that glossing over

the distress some women experience with breastfeeding may encourage more to try,
but leave those who have difficulties less able to cope.
The case for offering more information about breastfeeding during the antenatal
period is also strengthened by Duckett’s finding188 that knowledge about
breastfeeding was independently associated with duration of breastfeeding for women
who planned to do so (Section 1.4.4b).
In 1995, 68% of first time mothers who planned to breastfeed attended antenatal
classes that included a discussion on infant feeding, suggesting that antenatal classes
do offer a means to provide those first time mothers who plan to breastfeed with more
information.1 However classes have been criticised for focussing too much on the
experience of labour, and not enough on breastfeeding and "parenting" a new baby.330
331
These findings concur with the emphasis that women who took part in this study
placed on receiving more information about what to expect before birth.
4.4.3b Help with positioning
The ten steps to successful breastfeeding adopted by WHO and UNICEF as part of the
baby-friendly initiative230 stress the importance of showing mothers how to breastfeed
in the early postnatal period, and this was reflected in the comments that mothers in
this study made about the care they had received. Many stressed the importance of
postnatal ward staff being prepared to spend time with women who needed support,
echoing concerns raised by the Audit Commission about the quality of postnatal care
and staffing levels on postnatal wards.117

Help with positioning includes both spending time with women while they feed and
specific guidance to help them position the baby at the breast. Good positioning
allows the baby to take a good mouthful of breast tissue into the mouth, leaving the
nipple at the back of the baby's mouth where it is protected from damage and ensuring
an unrestricted milk flow. Woolridge332 has summarised the physiological evidence
for this, and the value of correcting "nipple sucking" has been demonstrated in a small
controlled trial in Sweden (Righard and Alade 1992)237. Most intervention studies
have however combined practical guidance with other support and encouragement, so
it is difficult to separate the relative importance of the different components of
postnatal support. In our study, it was striking how often mothers used the word
"positioning" to describe the most helpful advice they had received, underlining the
extent to which breastfeeding is a practical skill that mothers may need help to learn.
4.4.3c Effective advice
The main reasons women give for discontinuing breastfeeding have been documented
in successive National Surveys and include perceived insufficient milk supply, the
baby not sucking or rejecting the breast, painful breasts or nipples, and mothers
feelings that feeding was taking too long.1 The majority of women with feeding
problems do ask for help, but concern has been expressed about the quality and
consistency of the advice they receive. In a study asking women about their
experiences of maternity care, Rajan77 identified conflicting advice as a significant
problem, which undermined mothers' confidence in their ability to breastfeed, a
finding also noted in this study. It was interesting that although conflicting advice

sometimes reflected pressure to switch to bottle feeding, much of it occurred when
professionals attempted to give "up-to-date" advice to feed on demand, empty one
breast before offering the other, or be cautious about the use of breast shields, which
have been shown to reduce milk flow.332 333
In 1991 the Royal College of Midwives sent all UK midwives a book "Successful
Breastfeeding",116 to help them offer consistent, effective responses when approached
about breastfeeding problems and there have been a number of other initiatives to
promote evidence-based practice in maternity care.225 203 230
As well as advice they received in response to feeding problems, mothers also
reported general advice to look after their own needs, to rest, eat well and get
comfortable when feeding. During the analysis, there was a debate about whether
this should be considered as a separate theme, covering women's need to be looked
after when breastfeeding. Although not often made explicit in this study, this need for
replenishment has been identified by Hewat and Ellis,168 and Bottorff. 323 In a review
of anthropological reports on breastfeeding from around the world, Raphael324 noted
that "one element in most cultures seemed to emerge which facilitated success - the
presence of someone who cares for the mother." (p 13)
4.4.3d Encouragement and empathy
Some of the most disturbing comments women made about their experiences with
breastfeeding support related to misguided attempts by professionals to encourage
them to continue in the face of difficulties, highlighting the fine line between

encouragement and inappropriate pressure which may make mothers feel guilty.
Similarly, advice from family members to supplement with bottle feeds was often
seen as undermining. In both cases, what women found difficult was being told what
to do, when they wanted to be listened to and have their feelings acknowledged. This
echoes the findings of a review of women's experience of support in breastfeeding by
Cronenwett and Reinhardt334 which identified being listened to, feeling cared for and
being given time as being especially important.
The importance of social approval in the maintenance of breastfeeding has been
widely documented, with reports by Morse and Harrison,155 Dykes and Griffiths,152
Tarkka et al,335 and Matthews et al,336 and reinforces the emphasis women in the
study placed on reassurance and encouragement. Reassurance that their experiences
were normal also helped mothers who had been uncertain what to expect in the first
few weeks.
Smale337 has argued that the psychosocial aspects of supporting breastfeeding mothers
have been neglected in professional training and that more emphasis should be placed
on the development of counselling skills and ways to help women develop their own
strategies than just advising solutions to problems.
4.4.4 Summary and conclusions
These results give a clearer view of the support women want with breastfeeding. They
deserve attention because of the large number of women whose perspectives were
included, because they are based on women's views on specific experiences, rather

than what they might want in theory and because of the rigour with which the analysis
was conducted. They are also consistent with, and provide a framework for previous
work on breastfeeding support. They have wide-ranging implications for policy and
practice.
Women want to see changes in the way breastfeeding support is delivered. They want
more realism antenatally, more practical help in the first few days and effective advice
if they have problems. Throughout this, they want to be listened to and to be
encouraged, without feeling pressurised. This represents a challenge to all those
involved in supporting breastfeeding mothers to consider their own practice.
While there are a number of initiatives to improve the quality of breastfeeding support
offered by health professionals, most focus on women themselves and do not engage
with fathers and others within their informal support networks. But because many
fathers, friends and family members feel unprepared for the support role they are
asked to adopt, we need to find effective ways to address their needs and enable them
to offer appropriate support.
Improving breastfeeding support however needs more than individual responses by
health professionals and resources should to be made available to support this.
Staffing levels on postnatal wards need to be improved to allow midwives and others
time to support women: services need to take a more strategic approach to supporting
breastfeeding.

These results could also help health services consider how well they meet women's
needs. This could involve the development of an audit tool based on the priorities
expressed by women. The framework identified also has implications for the training
of health professionals and emphasises the importance of non-directive counselling
skills, practical training in helping women position the baby at the breast and an
understanding of the evidence which underpins advice for specific problems. Ways in
which breastfeeding support could be improved are considered further in the next
chapter.
This study adopted an RCT methodology to evaluate a social intervention -
breastfeeding counsellors providing support for mothers - and like many of the
American social experiments of the 1970s, the comparison between the feeding
behaviour of the control and intervention groups does not provide clear evidence on
how to proceed. But the design included a range of quantitative and qualitative
components, to assess both what support women wanted and to understand the
intervention the counsellors delivered. From this, it has been possible to go beyond
asking whether the intervention was effective, confirmed or refuted by a single "P"
value, to the lessons we can learn about how to support women with breastfeeding.

PART FIVE
CONCLUSIONS AND RECOMMENDATIONS
260
Chapter 5
CONCLUSIONS AND RECOMMENDATIONS
5.1 Policy implications of main findings
The findings of these studies have implications for practice which go beyond the
specific research questions asked. In particular, they point to ways that breastfeeding
support, from whatever source, may be made more effective. Implications for practice
include:
• The negative results of the randomised controlled trial do not justify extending
this model of individual breastfeeding counselling to all mothers.
• The low uptake of postnatal support suggests that services that rely on women to
ask for help will miss those who are less committed to breastfeeding and most
need support. Because of this, it is better to integrate breastfeeding support into
routine postnatal care than provide it as an alternative which women have to
request.
• The qualitative research on women's experiences of support emphasised the need
for practical help with positioning in the first few days. This requires time,
patience and practical skills, but women's comments revealed that for many of
them, postnatal care was inadequate.
• Although telephone support is relatively easy to provide, the findings suggest it is
unlikely to be effective and do not justify investment in this. It is more likely to
be used by those who are more motivated and does not offer an opportunity to
observe a mother feed or help with positioning. Sikorski and Renfrew's meta-
analysis of studies employing telephone support also endorses this conclusion.199
5. Conclusions and Recommendations 261

• Women valued the skills and expertise of the counsellors. Those who contacted
them were very satisfied with their support and women in the intervention group
regarded their advice as more helpful than that they received from other sources.
Other disciplines providing postnatal care have not been subjected to the scrutiny
of a randomised controlled trial and these perspectives from women need to be
taken into account in considering how best to provide postnatal care.
• Counselling approaches may increase women's confidence in their ability to
breastfeed and reduced the proportion who felt they did not have enough milk.
Models of behaviour change, such as the Theory of Planned Behaviour identify
the ability to carry out intentions as a discrete factor in behaviour and the results
of the trial suggest that counselling enhanced this.
• Study one showed that it is possible to predict which mothers are most likely to
stop in the early weeks. They are mothers of first babies who have considered the
option of bottle feeding, or come from manual social class groups and those who
had previous children, but had not breastfed for more than six weeks.
• The qualitative study of women's views on breastfeeding support describes how
they feel it could be improved. This has implications for both professional and lay
supporters, which are discussed further below.

5.2 How can we improve breastfeeding support?
Fig 5.2
Ways to improve breastfeeding support
• A new strategy for postnatal care
• Clarifying the role of volunteers
• Improving training and professional practice
• Marketing breastfeeding
5.2.1 A new strategy for postnatal care
Evidence for a range of sources, the static breastfeeding rates,99 100 101 1
the Audit
Commission,117 midwifery staffing levels118 and the findings of this study suggests
there is a crisis in postnatal care. While the first step is to recognise there is a
problem, there sometimes seems to be a sense of despair - that nothing can be done to
improve the situation. Government initiatives, such as the Joint Breastfeeding
Initiative and National Breastfeeding Working Group114 and more recently, the Infant
Feeding Initiative have attempted to bring people together, at both national and local
levels to improve breastfeeding support, but too often this is seen in isolation from
other postnatal care.
There is a case for a fundamental review of postnatal care, following on from the
ideas developed in "Changing Childbirth" in 1993.338 This should bring together those
involved and consider how best to support women, both during the first few hours in
hospital and then at home. Whereas in the past, postnatal wards were the focus of
breastfeeding support, now women are discharged much earlier and neither hospital

support for a token first feed, nor brief home visits by overworked midwives
adequately meet their needs. As Bick recently argued in MIDIRs Midwifery Digest,339
the pattern of postnatal care dictated by the Midwives acts of 1902 and 1936 no
longer meets the needs of women today. This review needs to consider both the
resources needed to make seamless postnatal support effective and the professional
issues of who should be doing this. Study two demonstrated high levels of
satisfaction with support from breastfeeding counsellors and there is a case for
including breastfeeding counsellors, supporters or lactation consultants within plans
for a new model of postnatal care. The observation that fewer women from
disadvantaged groups asked for help suggests that routine early visits are important if
services are to engage those who are most likely to need help.
5.2.2 The role of the volunteer
These studies have raised, but not fully answered a number of questions about the role
of volunteers in breastfeeding support. Many women were very positive about their
support, but the counsellors sensed that some were uncertain about what they could
reasonably ask of a volunteer.
To an extent the study "professionalised" the counsellors, but is this what they
wanted? Section 1.7.7 considered the motivation to become a counsellor, which is
often a desire to assist others with something they have valued. But do they want to
turn their role into a paid job, with the additional responsibility and lack of flexibility
that might entail? Professionalisation might also prevent those with skills and
commitment, but few academic qualifications from becoming counsellors. At the end

of the day, volunteers retain a veto on how much of a commitment to make and
breastfeeding counsellors need to be able to decide for themselves how their role
should develop.
To what extent are volunteers able to cross the cultural barriers between social
groups? Most of the volunteer counsellors in this study had become involved through
the National Childbirth Trust and while that may equip them to support women within
that social milieu, their contribution may not be as well received in other settings.
Brunton et al196 have highlighted the importance of working with cultural processes
within society and it may be essential that volunteers are drawn from within social
groups with whom they will work. In line with this, the Department of Health Infant
Feeding Initiative340 is funding 14 local peer support groups, which have recruited
mothers from disadvantaged groups to help others. Many of these are however small
projects with only short-term funding, which may make it hard to evaluate their
effectiveness. McInnes341 compared breastfeeding rates in a deprived area of
Glasgow where mothers were supported by seven peer counsellors with rates in a
similar area without support. She found that when other differences between the areas
were controlled for, peer support increased the proportion intending to breastfeed
(odds ratio 1.95; 95% CI 1.22 - 3.41), but that the differences noted at six weeks were
not significant (odds ratio 1.80; 95% CI 0.96 - 3.41). These findings suggest peer
support may be an effective way to promote breastfeeding, but that this merits further
research.

5.2.3 Training and professional practice
The qualitative analysis of women’s comments suggested ways in which professional
practice should be improved. Both initial and in-service training should help
professionals improve their inter-personal and counselling skills, perhaps employing
video techniques as are increasingly used for GP training. Professionals should ask
women ante-natally about their contact with breastfeeding and allow time to discuss
how they expect it to be. They should discuss how women will access support and
the role her partner may play in this. Research presented in section 1.3.3 suggests that
many men feel uncertain about supporting their partners with breastfeeding and
professionals and counsellors need to help men develop understanding and confidence
in this.
Mothers’ comments revealed wide variations in the quality of help they received with
positioning, which suggests that professionals need much more specific training on
this. Helping mothers with positioning is a practical skill and may be best learnt from
mothers, or in joint consultations with experienced midwives.
Although consistent, evidence-based advice is important for specific problems,
reliance on a “medical” problem-solving approach is not enough. Professionals need
to help women negotiate the difficulties they encounter.
Improving professional practice requires a number of parallel initiatives, including
organisational change, educational programmes, audit and the provision of adequate
resources. Examples of different approaches include the distribution of "Successful

Breastfeeding" in 1991,116 the MIDIRS Midwifery Digestxxvi and local projects such as
the South Thames Evidence Based Practice (STEP) project at King's Healthcare.342
The qualitative work reported in part four has particular relevance to audit, because
the five key priorities identified by women could be used to develop an audit tool to
assess how effectively support meets mothers' needs.
5.2.4 Marketing breastfeeding
Evidence from a number of sources suggests that "supply interventions" which
Baer115 described as those designed to increase the supply of milk, can have only
limited impact on breastfeeding behaviour. As discussed in section 1.2.5, there have
been significant changes in hospital practice over the last 20 years, without apparent
rises in breastfeeding rates. Section 1.2.6 and chapter 1.3 detail some of the social
pressures which influence mothers' behaviour and chapter 1.4 considers how models
can help us understand the relative influence of attitudinal factors, the social
environment and support from professional and lay sources on infant feeding. While
meta-analysis of previous research suggests that support can increase the proportion
of women breastfeeding to two months,199 my trial suggests that, at least for the
intervention tested, any benefit is limited.
If supply interventions have only limited impact on breastfeeding rates, would
"demand interventions", to encourage more women to breastfeed and promote a more
positive view of breastfeeding in society be more effective? For this to be successful,
the campaign would need to redress the current media bias against breastfeeding120,
xxvi
MIDIRS Midwifery Digest is published quarterly by Midwives Information and Resource Service, 9, Elmdale
Rd, Bristol, BS8 1SL.

but as Hoddinott has warned, merely slipping images of breastfeeding into
programmes may not be enough if they provoke adverse reactions from viewers. Two
presentations to national conferences in 1995343 and 2000344 by people working in
marketing have explored the importance of developing a "brand image" for
breastfeeding. The current national campaign344 targets women and their male
partners from lower socioeconomic groups, and deliberately emphasised positive male
attitudes towards breastfeeding. The key messages of the campaign were:
• Breastfeeding is about giving your child a healthy start in life
• Most women can breastfeed if they get the right support.
Despite the thought that has gone into the campaign, it may need significantly more
resources to have an impact. The total funding for the Government's Infant Feeding
Initiative, including national co-ordinators, local projects and promotional resources is
£900,000,340 whereas in 1995, Durdle et al 343

estimated that an effective media
campaign would cost between £700,000 and £1.5 million a year.
Evidence from Norway,147 and the evaluation of the Ten Steps to Successful
Breastfeeding230 suggests that it may be most effective to combine a number of
approaches to promoting breastfeeding - that combined approaches may work
synergistically, whereas isolated interventions, (as was the one evaluated in this
thesis,) may not overcome the many barriers to change.

5.3 Implications for future research
5.3.1 Lessons from previous research
Reviewing the literature revealed a great deal of research on breastfeeding, but much
of it is of poor quality. As has been discussed, many studies of both the benefits of
breastfeeding and interventions to promote it have problems with confounding
factors, inadequate assessment of outcome and small sample size. These issues are
considered further in chapters 1.1 and 1.5, but future studies need to address the
methodological concerns raised more effectively. There is also a case for basing
future interventions on qualitative work, such as that reported in this thesis, to ensure
that interventions meet mothers' needs.
The number of small studies on breastfeeding may reflect the personal commitment of
clinicians and researchers to breastfeeding. This enthusiasm is a tremendous asset, but
if it is to be harnessed effectively, there may be a case for experienced researchers
supporting novices and also helping them identify genuinely new questions, rather
than repeating local epidemiological studies. Often research is conceived as an
adjunct to a service and in this case, those providing the service might be best
supported by being provided with an "off-the-shelf" evaluation package to enable
them to evaluate their service, without having to design a whole study.
5.3.2 Maternal Assessment of Breastfeeding Support
During the course of the qualitative analysis, it became clear that the five themes
identified could form the basis of an instrument to evaluate the quality of
breastfeeding support women receive. This work would derive directly from the

priorities women gave and could then be used to audit services. At present the project
is only at the conceptual stage, but carrying this out might involve:
• Reviewing the data to identify phrases that relate to particular themes and could
be incorporated in a questionnaire.
• Designing the questionnaire, taking into account literature on breastfeeding
research instruments.
• Testing the validity and reliability of the questionnaire in a number of maternity
settings.
5.3.3 Educational research for professional and lay breastfeeding supporters
The qualitative findings showed that how people give support with breastfeeding is of
great importance for women. Many of them wanted practical help with positioning -
getting the baby to feed. This suggests that practical training in this is important and
there is a case for educational research to identify what works best in teaching how to
convey this skill.
Women were very positive about the support they received from counsellors and felt
it was very important that their feelings were acknowledged. There may be a case for
further research on the applicability of counselling approaches to midwifery practice.
There is also a case for studies to investigate ways of engaging more effectively with
fathers. The importance of this is illustrated by the observation that previous research
on fathers' perspectives has been confined to observation studies, with Bar-Yam and
Darby identifying no intervention studies in their review of the topic.161

5.3.4 Perceptions of volunteers
The randomised controlled trial unearthed questions about women's perceptions of
support from volunteers which merit further study. The importance of evaluating
services delivered by voluntary organisations was highlighted in a recent BMJ
editorial345 and acceptability to those receiving help is an important part of this. (The
fact that neither the word "patient" nor "client" satisfactorily describes the recipient's
role illustrates the issues that may arise if statutory services promote services provided
by volunteers.)
5.3.5 Ethnicity
There has been remarkably little research on the feeding behaviour and needs of
different ethnic groups in the United Kingdom. Because of this, there is a strong case
for including questions on ethnicity in the next national infant feeding survey.
However, research on ethnicity should not just focus on differences between different
groups - instead it would be more useful to know how well services meet the needs of
different groups and to learn about the cultural processes which sustain, or undermine
breastfeeding in different ethnic groups.
5.3.6 Research to inform a review of postnatal care
The need for a fundamental review of postnatal care has been identified and there may
well be a need for further research to inform this process. Initially this should involve
reviewing the literature, (which was not included in the topics I reviewed for this
thesis,) but there may be a need for further studies of different approaches before

change is implemented widely. However there is some urgency about improving
postnatal care and further research should not be an alternative to change.
5.3.7 Breastfeeding and the media
Many accounts of coverage of breastfeeding in the media identify problems, but
because the media is outside the health service - and largely in private ownership, few
studies have considered how to engage with the media, and how to best get positive
messages across on television, radio, newspapers or magazines. There is however the
possibility of fascinating qualitative work on the adoption of health promotion
messages by the media.
5.3.8 The benefits of breastfeeding
The review of research on the benefits of breastfeeding identified a number of gaps in
the evidence, but also the difficulties of answering some questions, such as whether
breastfeeding protects against atopic disorders. There is a need for some further work
on the health benefits, which should include further meta-analyses to draw together
research on topics such as infections. Doing this will enable researchers to quantify
the benefits of breastfeeding and give mothers more specific information. It might
then be worth investigating whether giving women a clearer picture of the actual
benefits alters their motivation to both initiate and continue breastfeeding.

REFERENCE LIST
1. Foster, K., Lader, D., and Cheesbrough, S. Infant Feeding 1995. 1997. London, The
Stationery Office.
2. Standing Committee on Nutrition of the British Paediatric Association. Is breastfeeding

beneficial in the UK? Arch Dis Child 1994;71:376-80.
3. Work Group on Breastfeeding. Breastfeeding and the Use of Human Milk. Pediatrics
1997;100:1035-9.
4. Heinig MJ,.Dewey KG. Health advantages of breast feeding for infants: A critical review.
Nutrition Research Reviews 1996;9:89-110.
5. Golding, J., Rogers, I. S, and Emmett, P. M. Breast feeding: benefits and hazards.
Methodology and summary of results. Early Hum Dev 49 Suppl, S1-S6. 29-10-1997.
6. Golding, J., Emmett, P. M., and Rogers, I. S. Gastroenteritis, diarrhoea and breastfeeding.
Early Hum Dev 49 Suppl, S83-S103. 29-10-1979.
7. Golding, J., Emmett, P. M., and Rogers, I. S. Does breast feeding protect against non-gastric
infections? Early Hum Dev 49 Suppl, S105-S120. 29-10-1997.
8. Golding, J., Emmett, P. M., and Rogers, I. S. Eczema, asthma and allergy. Early Hum Dev 49
Suppl, S121-S130. 29-10-1997.
9. Golding, J., Emmett, P. M., and Rogers, I. S. Does breast feeding have any impact on non-
infectious, non-allergic disorders? Early Hum Dev 49 Suppl, S131-S142. 29-10-1997.
10. Golding, J., Emmett, P. M., and Rogers, I. S. Breast feeding and infant mortality. Early Hum
Dev 49 Suppl, 143-155. 29-10-1997.
11. Rogers, I. S, Emmett, P. M., and Golding, J. The growth and nutritional status of the breast-
fed infant. Early Hum Dev 49 Suppl, S157-S174. 29-10-1997.
12. Golding, J., Rogers, I. S, and Emmett, P. M. Association between breast feeding, child
development and behaviour. Early Hum Dev 49 Suppl, S175-S184. 29-10-1997.
13. WHO Collaborative Study Team on the Role of Breastfeeding on the Prevention of Infant
Mortality. Effect of breastfeeding on infant and child mortality due to infectious diseases in
less developed countries: a pooled analysis. Lancet 2000;355:451-5.
14. Lucas A,.Cole TJ. Breast milk and neonatal necrotising enterocolitis. Lancet 1990;336:1519-
22.
15. Lucas A, Brooke OG, Morley R, Cole TJ, Bamford MF. Early diet of preterm infants and
development of allergic or atopic disease: randomised prospective study. BMJ 1990;300:837-
40.
16. Lucas A, Morley R, Cole TJ, Lister G, Leeson-Payne C. Breast milk and subsequent
intelligence quotient in children born preterm. Lancet 1992;339:261-4.
Reference list 273

17. Morley R, Cole TJ, Powell R, Lucas A. Mother's choice to provide breast milk and
developmental outcome. Arch Dis Child 1988;63:1382-5.
18. Lucas A, Morley R, Cole TJ. Randomised trial of early diet in preterm babies and later
intelligence quotient. BMJ 1998;317:1481-7.
19. Lucas A, Cole T, Morley R. Early diet in preterm babies and later intelligence quotient -
Authors reply. BMJ 1999;318:1625.
20. Burr ML, Limb ES, Maguire MJ, Amarah L, Eldridge BA, Layzell JC et al. Infant feeding,
wheezing, and allergy: a prospective study. Arch Dis Child 1993;68:724-8.
21. Lucas A, Stafford M, Morley R, Abbott R, Stephenson T, MacFadyen U et al. Efficacy and
safety of long-chain polyunsaturated fatty acid supplementation of infant-formula milk: a
randomised trial. Lancet 1999;354:1948-54.
22. Kramer, M. S., Chalmers, B., Hodnett, E. D., Sevkovskaya, Z., Dzikovich, I., Shapiro, S.,
Collet, J.-P., Vanilovich, I., Mezen, I., Ducruet, T., Shishko, G., Zubovich, V., Mknuik, D.,
Gluchanina, E., Dombrovskiy, V., Ustinovitch, A., Kot, T., Bogdanovich, N., Ovchinikova,
L., Helsing, E., and for the PROBIT Study Group. Promotion of Breastfeeding Intervention
Trial (PROBIT). JAMA 285(4), 413-420. 24-1-2001.
23. World Health Organisation / UNICEF. Protecting, Promoting and Supporting Breast-feeding:
the Special Role of Maternity Services. 1989. Geneva, W.H.O.
24. Bauchner H, Leventhal JM, Shapiro ED. Studies of breastfeeding and infections. JAMA
1986;256:887-92.
25. Kramer MS. Does breastfeeding help protect against atopic disease? Biology, methodology,
and a golden jubilee of controversy. The Journal of Pediatrics 1988;112:181-9.
26. Auerbach KG, Renfrew MJ, Minchin M. Infant feeding comparisons: a hazard to infant
health? J Hum Lact 1991;7:63-8.
27. Howie PW, Forsyth JS, Ogston SA, Clark A , du V Florey C. Protective effects of
breastfeeding against infection. BMJ 1990;300:11-6.
28. Forsyth JS, Ogston SA, Clark A, du V Florey C, Howie PW. Relation between early
introduction of solid food to infants and their weight and illnesses during the first two years of
life. BMJ 1993;306:1572-6.
29. Wilson AC, Forsyth JS, Greene SA, Irvine L, Hau C, Howie PW. Relation of infant diet to
childhood health: seven year follow up of cohort of children in Dundee infant feeding study.
BMJ 1998;316:21-5.
30. Labbok M,.Krasovec K. Toward consistency in breastfeeding definitions. Studies in Family

Planning 1990;21:226-30.
31. Division of Diarrhoeal and Acute Respiratory Disease Control. Indicators for assessing
breast-feeding practices. 1991. Geneva, World Health Organisation.
32. Bernt, K. M. and Walker, W. A. Human milk as a carrier of biochemical messages. Acta
Paediatr Scand 430, 27-41. 1999.
Reference list 274

33. Duffy LC, Byers TE, Riepenhoff-Taltry M, La Scolea LJ, Zielezny M, Ogra PL. The effects
of infant feeding on rotavirus-induced gastroenteritis; a prospective study. Am J Public Health
1986;76:259-63.
34. Dewey KG, Heinig MJ, Nommsen-Rivers LA. Differences in morbidity between breast-fed
and formula-fed infants. J Pediatr 1995;126:Pt 1):696-702.
35. Rubin, David H., Leventhal, John M., Krasilnikoff, Peter A., and et al. Relationship between
infant feeding and infectious illness: A prospective study of infants during the first year of
life. Pediatrics 1990; 85:464-471.
36. Raisler J, Alexander GR, O'Campo P. Breast-feeding and infant illness: a dose response
relationship? Am J Public Health 1999;89:25-30.
37. Wright AL, Holberg CJ, Taussig LM, Martinez FD. Relationship of infant feeding to
recurrent wheezing at age 6 years. Arch Pediatr Adolesc Med 1995;149:758-63.
38. Victora CG, Smith PG, Barros FC, Fuchs SC. Risk factors for deaths due to respiratory
infections among Brazilian infants. International Journal of Epidemiology 1989;18:918-25.
39. Duncan B, Ey J, Holberg CJ, Wright AL , Martinez FD, Taussig LM. Exclusive breastfeeding
for at least 4 months protects against otitis media. Pediatrics 1993;91:867-72.
40. Aniansson G, Alm B, Andersson B, Hakansson A, Larsson P, Nylen O et al. A prospective

cohort study on breast-feeding and otitis media in Swedish infants. Pediatr Infect Dis J
1994;13:183-8.
41. Owen MJ, Baldwin CD, Swank PR, Pannu AK, Johnson DL, Howie VM. Relation of infant
feeding practices, cigarette smoke exposure and group child care to the onset and duration of
otitis media with effusion in the first two years of life. Journal of Pediatrics 1993;123:702-
11.
42. Harabuchi Y, Faden H, Yamanaka N, Duffy L, Wolf J, Krystofik D. Nasopharyngeal

colonization with nontypeable Haemophilus influenzae and recurrent otitis media.
Tonawanda/Williamsville Pediatrics. J Infect Dis 1994;170:862-6.
43. Rodgers B. Feeding in infancy and later ability and attainment: a longitudinal study.
Developmental Medicine and Child Neurology 1978;20:421-6.
44. Uauy R,.De Andraca I. Human milk and breast feeding for optimal mental development.
[Review]. J Nutr 1995;125:Suppl):2278S-80S.
45. Lanting CI, Fidler V, Huisman M, Touwen BC, Boersma ER. Neurological differences
between 9-year-old children fed breastmilk or formula-milk as babies. Lancet
1994;344:1319-22.
46. Horwood LJ,.Fergusson DM. Breastfeeding and later cognitive and academic outcomes.
Pediatrics 1998;101:E9.
47. Anderson JW, Johnstone BM, Remley DT. Breast-feeding and cognitive development: a
meta-analysis. Am J Clin Nutr 1999;70:525-35.
Reference list 275

48. Scott DT, Janowsky JS, Carroll RE, Taylor JA, Auestad N, Montalto MB. Formula
supplementation with long-chain polyunsaturated fatty acids; are there developmental
benefits. Pediatrics 1998;102:E59.
49. Birch, E. E., Hoffman, D. R., Uauy, R., Birch, D. G., and Prestidge, C. Visual acuity and the
essentiality of docosahexaenoic and arachidonic acid in the diet of term infants. Pediatric
Research 44(2), 201-209. 1998.
50. Gale CR,.Martyn CN. Breastfeeding, dummy use, and adult intelligence. Lancet
1996;347:1072-5.
51. Vandenplas, Y. Myths and facts about breastfeeding: does it prevent later atopic disease?
Acta Paediatr 86, 1283-1287. 1997.
52. Chandra RK, Puri S, Hamed A. Influence of maternal diet during lactation and use of formula
feeds on development of atopic eczema in high risk infants. BMJ 1989;299:228-30.
53. Lovegrove JA, Hampton SM, Morgan JB. The immunological and long-term atopic outcome
of infants born to women following a milk-free diet during late pregnancy and lactation: a
pilot study. Br J Nutr 1994;71:223-38.
54. Saarinen UM,.Kajosaari M. Breastfeeding as prophylaxis against atopic disease: prospective

follow-up study until 17 years old. Lancet 1995;346:1065-9.
55. Moore WJ, Midwinter RE, Morris AF, Colley JRT, Soothill JF. Infant feeding and subsequent
risk of atopic eczema. Arch Dis Child 1985;60:722-6.
56. Nakamura Y, Oki I, Tanihara S, Ojima T, Ito Y, Yamazaki O et al. Relationship between
breast milk feeding and atopic dermatitis in children. Journal of Epidemiology 2000;10:74-8.
57. Golding J, Butler NR, Taylor B. Breastfeeding and eczema/asthma. Lancet 1982;1:623.
58. Mayer EJ, Hamman RF, Gay EC, Lezotte DC, Savitz DA, Klingensmith GJ. Reduced risk of
IDDM among breast fed children. Diabetes 1988;37:1625-32.
59. Soltesz G, Jeges S, Dahlquist G. Non-genetic risk determinants for type 1 (insulin-dependent)
diabetes mellitus in childhood. Hungarian Childhood Diabetes Epidemiology Study Group.
Acta Paediatr 1994;83:730-5.
60. Virtanen SM, Rasanen L, Aro A, Ylonen K, Lounamaa R, Tuomilehto J et al. Feeding in
infancy and the risk of type 1 diabetes mellitus in Finnish children. The 'Childhood Diabetes
in Finland' Study Group. Diabet Med 1992;9 :815-9.
61. Gerstein HC. Cow's milk exposure and type I diabetes mellitus. A critical overview of the
clinical literature. Diabetes Care 1994;17:13-9.
62. Pisacane A, Graziano L, Mazzarella G. Breast-feeding and urinary tract infection. J Pediatr
1992;120:87-9.
63. Bittencourt AL. Possible risk factors for vertical transmission of Chagas' disease. [Review].
Rev Inst Med Trop Sao Paulo 1992; 34:403-8.
Reference list 276

64. Pisacane A, Caracciolo G, de Luca U, Grillo G, Simeone C, Impagliazzo N et al. Infant
feeding and idiopathic intussusception. J Pediatr 1993;123:593-5.
65. Pisacane A, de Luca U, Crisculolo L. Breastfeeding and hypertrophic pyloric stenosis:

population based case-control study. BMJ 1996;312:745-6.
66. Pisacane A, de Luca U, Vaccaro F, Valiante A, Impagliazzo N, Caracciolo G. Breast-feeding

and inguinal hernia. J Pediatr 1995;127:109-11.
67. Pisacane A, Impagliazzo N, De Caprio C. Breastfeeding and tonsillectomy. BMJ

1996;312:746-7.
68. Whorwell PJ, Holdstock G, Whorell GM, Wright R. Bottle feeding, early gastroenteritis and
inflammatory bowel disease. BMJ 1979;i:382.
69. Koletzko S, Griffiths A, Corey M, Smith C, Sherman P. Infant feeding practices and
ulcerative colitis in childhood. BMJ 1991;302:1580-1.
70. Mitchell EA, Taylor BJ, Ford RP. Four modifiable and other major risk factors for cot death:
the New Zealand study. J Paediatr Child Health 1992;28 (suppl 1):S3-S8.
71. Taylor BJ. A review of epidemiological studies of sudden infant death syndrome in southern
New Zealand. [Review]. J Paediatr Child Health 1991;27:344-8.
72. Golding, J. Breast-feeding and sudden infant death syndrome. Report of the chief medical
officer's expert group on the sleeping pattern of infants and cot death. 77-82. 1993. London,
HMSO.
73. Dewey KG, Heinig MJ, Nommsen LA, Peerson JM, Lonnerdal B. Breast-fed infants are
leaner than formula-fed infants at 1 y of age: the DARLING study. Am J Clin Nutr
1993;57:140-5.
74. von Kries R, Koletzko B, Sauerwald T, von Mutius E, Barnert D, Grunert V et al. Breast
feeding and obesity;cross sectional study. BMJ 1999;319 :147-50.
75. Dewey KG, Heinig MJ, Nommsen LA. Maternal weight-loss patterns during prolonged
lactation. Am J Clin Nutr 1993;58:162-6.
76. Melton LJ, Bryant SC, Wahner HW. Influence of breastfeeding and other reproductive factors
on bone mass later in life. Osteoporos Int 1993;3:76-83.
77. Cumming RG,.Klineberg RJ. Breastfeeding and other reproductive factors and the risk of hip
fractures in elderly women [published erratum appears in Int J Epidemiol 1993
Oct;22(5):962]. Int J Epidemiol 1993;22:684-91.
78. United Kingdom National Case-Control Study Group. Breast feeding and risk of breast cancer
in young women. BMJ 1993;307:17-20.
79. Newcombe PA, Storer BE, Longnecker MP. Lactation and a reduced risk of premenopausal
breast cancer. New England Journal of Medicine 1994;330:81-7.
80. Newcombe PA. Lactation and breast cancer risk. Journal of Mammary Gland Biology &
Neoplasia 1997;2:311-8.
Reference list 277

81. Lipworth L, Bailey LR, Trichopoulos. History of breast-feeding in relation to breast cancer
risk; a review of the epidemiological literature. Journal of the National Cancer Institute
2000;92:302-12.
82. Rosenblatt KA,.Thomas DB. Lactation and the risk of epithelial ovarian cancer. The WHO
Collaborative Study of Neoplasia and Steroid Contraceptives. Int J Epidemiol 1993;22:192-7.
83. Newton N. Psychologic differences between breast and bottle feeding. The American Journal
of Clinical Nutrition. 1971;24:993-1004.
84. Uvnas-Moberg, K. Oxytocin may mediate the benefits of positive social interaction and
emotions. Psychoneuroendocrinology 23(8), 819-835. 1998.
85. Klaus, M. Mother and Infant: Early Emotional Ties. Pediatrics 102(5 (Supplement)), 1244-
1246. 1998.
86. Schmied V,.Barclay L. Connection and Pleasure, Disruption and Distress: Women's
Experience of Breastfeeding. Journal of Human Lactation 1999;15:325-33.
87. Lawrence RA. Breastfeeding: A guide for the medical profession. St Louis, Missouri: Mosby-
Year Book Inc, 1994.
88. Hull D. Vitamin K and childhood cancer [editorial; comment]. BMJ 1992;305:326-7.
89. McVeigh, T. Breast Milk Poison Alert. The Express , 1-4. 12-7-1999. London, Express
Newspapers.
90. Abraham K, Hille A, Ende M, Helge H. Intake and fecal excretion of PCDDs, PCDFs, HCB
and PCBs (138, 153, 180) in a breast-fed and a formula-fed infant. Chemosphere
1994;29:2279-86.
91. Lonnerdal B,.Hernell O. Iron, zinc, copper and selenium status of breast-fed infants and
infants fed trace element fortified milk-based infant formula. Acta Paediatr 1994;83:367-73.
92. Huisman M, Koopman-Esseboom C, Fidler V, Hadders-Algra M, van der Paauw CG,

Tuinstra LG et al. Perinatal exposure to polychlorinated biphenyls and dioxins and its effect
on neonatal neurological development. Early Hum Dev 1995;41:111-27.
93. Rogan WJ, Blanton PJ, Portier CJ, Stallard E. Should the presence of carcinogens in breast
milk discourage breast feeding? Regul Toxicol Pharmacol 1991;13:228-40.
94. Frank JW,.Newman J. Breast-feeding in a polluted world: uncertain risks, clear benefits. Can
Med Assoc J 1993;149:33-7.
95. Somogyi A,.Beck H. Nurturing and breast-feeding: exposure to chemicals in breast milk.
Environ Health Perspect 1993;101:Suppl 2:45-52.
96. McQuay HJ,.Moore RA. Using numerical results from systematic review in clinical practice.
Annals of Internal Medicine 1997;126:712-20.
97. Number needed to treat or harm. (NNT/NNH). Bandolier 1998;55-6.
Reference list 278

98. DHSS Committee on Medical Aspects of Food Policy: Panel on Child Nutrition. Breast
feeding. 1978. London, Her Majesty's Stationery Office.
99. Working Party of the Panel on Child Nutrition, Committee on Medical Aspects of Food
Policy. Present Day Practice in Infant Feeding: 1980. 20. 1980. London, Her Majesty's
Stationery Office. Reports on Health and Social Subjects.
100. Working Party on Child Nutrition, Committee on Medical Aspects of Food Policy.
Present day practice in Infant Feeding: Third Report. 32. 1988. London, Her
Majesty's Stationery Office. Reports on Health and Social Subjects.
101. White A, Freeth S, O'Brien M. Infant feeding 1990. London: HMSO, 1992.
102. Department of Health and Social Security. Present day practice in infant feeding. Reports on
health and social subjects. 1974. London, HMSO.
103. Palmer G. The Politics of Breastfeeding. London: Pandora Press, 1988.
104. Dykes RM. Illness in infancy: a comparative study of infant sickness and infant mortality in
Luton. Luton: Leagrave Press, 1950.
105. Joint Committee of Royal College of Obstetricians and Gynaecologists and Population
Investigation Committee. Maternity in Great Britain. London : Oxford University Press,
1948.
106. Spence J. 1000 families in Newcastle upon Tyne. London: Oxford University Press, 1954.
107. Westropp C. Breastfeeding in the Oxford Child Health Survey. BMJ 1953;1:138.
108. Newson LJ,.Newson E. Breastfeeding in decline. BMJ 1962;1744-5.
109. Newson J, Newson E. Food and comfort: breast, bottle and dummy. In Newson J, Newson E,
eds. Infant care in an urban community., pp 43-68. London: George Allen and Unwin LTD,
1995.
110. Grosvenor, P. Influences Affecting Mothers' Decisions about Infant Feeding. 1969.
Edinburgh. (Thesis)
111. Jones DA, West RR, Newcombe RG. Maternal characteristics associated with duration of
breastfeeding. Midwifery 1986;2:141-6.
112. Whitehead RG, Paul AA. Changes in infant feeding in Britain during the last century. Infant
nutrition and cardiovascular disease.(Scientific Report; No 8), Southampton: MRC
Environmental Epidemiology Unit, 1987.
113. Twenty seventh world health assembly. Part 1. Infant nutrition and breastfeeding. Official
records of the World Health Organisation. 217, 20. 1974. Geneva, WHO.
114. National Breastfeeding Working Group. Breastfeeding: Good practice guidance to the NHS.
1995. Department of Health.
115. Baer EC. Promoting breastfeeding; a national responsibility. Studies in Family Planning
1981;12:198.
Reference list 279

116. Royal College of Midwives. Successful breastfeeding. London: Churchill Livingstone, 1991.
117. Audit Commission. First Class Delivery - Improving Maternity Services in England and
Wales. 1997. London, Audit Commission.
118. Boseley, S. Midwife crisis over hours and workload. The Guardian , 8. 9-10-2000. London.
119. Carroll, E. G. C. Home from the hospital. Parents 13(22), 52. 1938.
120. Henderson, L., Kitzinger, J, and Green, J. Representing infant feeding:
content analysis of British media portrayals of bottle feeding and breast feeding. BMJ
321, 1196-1198. 11-11-2000.
121. Julie Birchall. Breast-feeding? It sucks! The Guardian, Guide Weds May 12th, 8-9. 1999.
London.
122. Hoddinott, P. Why don't some women want to breastfeed and how might we change their
attitudes? A qualitative study. 1998. M Phil thesis, University of Wales College of
Medicine.
123. Yellowlees, H. Baby milk and infant feeding. CMO(76)3. 1976. London, HMSO.
124. Wright HJ,.Walker PC. Prediction of duration of breastfeeding in primiparas. Journal of

Epidemiology and Community Health 1983;37:89-94.
125. Piper S,.Parks PL. Predicting the duration of lactation: evidence from a national survey.
BIRTH 1996;23:7-12.
126. Freysteinsson H,.Sigurdsson JA. Breast-feeding in Iceland. Predictive factors and effects of
interventive measures. Scand J Soc Med 1996;24:62-6.
127. Fitzpatrick CC, Fitzpatrick PE, Darling MR. Factors associated with the decision to breast-
feed among Irish women. Ir Med J 1994;87:145-6.
128. Higginbottom G. Breastfeeding and black women: a UK investigation. Health Visitor

1998;71:12-5.
129. Gatrad AR. Attitudes and beliefs of Muslim mothers towards pregnancy and infancy. Arch
Dis Child 1994;71:170-4.
130. Bowes A,.Domokos TM. Negotiating breast-feeding: Pakistani women, white women and
their experiences in hospital and at home. Sociological Research Online 1998;3:1-21.
131. Thomas, M. and Avery, V. Infant feeding in Asian Families. 1997. London, The Stationery
Office .
132. McInnes R,.Stone DH. Promotion of exclusive breastfeeding. Lancet 1999;354:161.
133. Morrow AL, Guerrero ML, Shults J, Calva JJ, Lutter C, Bravo J et al. Efficacy of home-based
peer counselling to promote exclusive breastfeeding: a randomised controlled trial. Lancet
1999;353:1226-331.
Reference list 280

134. UNICEF. Take the Baby-Friendly Initiative! 1992. New York, UNICEF.
135. Ferriman A. WHO accused of stifling debate about infant feeding. BMJ 2000;320:1362.
136. Ineichen B, Pierce M, Lawrenson R. Jewish and Celtic attitudes to breast feeding compared.
Midwifery 1997;13:40-3.
137. World Health Organisation (WHO) Global Data Bank on Breast-feeding. Breast-feeding: the
best starrt in life. WHO/NUT/96.1. 1996. Geneva, World Health Organisation.
138. Saadeh, R. H. et al. Breast-feeding. The technical basis and recommendations for action.
1993. Geneva, World Health Organisation.
139. World Health Organisation (WHO) . Complementary feeding of young children in developing
countries. WHO/NUT/98.1. 1998. Geneva, World Health Organisation.
140. Williamson NE. Breastfeeding trends and patterns. Int J Gynaecol Obstet 1989;Suppl. 1:145-
52.
141. Ryan AS. The resurgence of breastfeeding in the United States. Pediatrics 1997;99:E12.
142. Crost M,.Kaminski M. Breastfeeding at maternity hospitals in France in 1995. National

Perinatal Survey. Archives de Pediatrie 1998;5:1316-26.
143. Geurts, J. J. M. Gezondheitsequetes - Borstvoeding in Nederland
(Health interview surveys. Breast feeding in the Netherlands.). Maandbericht van de
Gezondheidsstastik ((CBS 93/9)), 4-24. 1993.
144. Mose Hansen, A. and Rostgaard, K. Amningen bor styrkes (Breastfeeding should be
strengthened). Journal No 1. 1994. Copenhagen, Danish board of Health.
145. Hillervik-Lindquist C. Studies on perceived breast-milk insufficiency: relation to attitude and

practice. J Biosoc Sci 1992;24:413-25.
146. Heiberg Endresen E,.Helsing E. Changes in breastfeeding practices in Norwegian maternity

wards: national surveys 1973, 1982 and 1991. Acta Paediatr 1995;84:719-24.
147. Hunter HH. Breastfeeding UTOPIA? New Generation 1996;26-7.
148. McLeish J. Combining breast feeding and work: the legal rights. Professional Care of Mother
& Child 1998;8:2-3.
149. Galtry J. Breastfeeding, Feminism and Employment. Breastfeeding Abstracts 1998;3.
150. Hills-Bonczyk SG, Avery MD, Savik K, Potter S, Duckett LJ. Women's experiences with
combining breast-feeding and employment. J Nurse Midwifery 1993;38:257-66.
151. Frank E. Breastfeeding and maternal employment: two rights don't make a wrong. Lancet
1998;352:1083-4.
Reference list 281

152. Dykes F,.Griffiths H. Societal influences upon initiation and continuation of breastfeeding.
British Journal of Midwifery 1998;6:76-80.
153. Oakley A. Social Support and Motherhood. Oxford: Blackwell, 1992.
154. Hoddinott P,.Pill R. Qualitative study of decisions about infant feeding among women in east
end of London. BMJ 1999;318:30-4.
155. Morse JM, Harrison MJ. Social coercion for weaning. In Morse JM, ed. Qualitative health
research, pp 363-75. London: SAGE Publications, 1992.
156. Littman H, Medendorp SV, Goldfarb J. The decision to breastfeed. The importance of fathers'
approval. Clinical Pediatrics 1994;33:214-9.
157. Giugliani ER, Caiaffa WT, Vogelhut J, Witter FR, Perman JA. Effect of breastfeeding
support from different sources on mothers' decisions to breastfeed . J Hum Lact 1994;10:157-
61.
158. Kessler LA, Gielen AC, Diener-West M, Paige DM. The effect of a woman's significant other
on her breastfeeding decision. J Hum Lact 1995;11:103-9.
159. Isabella PH,.Isabella RA. Correlates of successful breastfeeding: a study of social and
personal factors. J Hum Lact 1994;10:257-64.
160. Buckner E,.Matsubara M. Support network utilization by breastfeeding mothers. J Hum Lact
1993;9:231-5.
161. Bar-Yam NB,.Darby L. Fathers and breastfeeding: a review of the literature. J Hum Lact
1997;13:45-50.
162. Freed GL, Fraley JK, Schanler RJ. Attitudes of expectant fathers regarding breast-feeding.
Pediatrics 1992;90:Pt 1):224-7.
163. Voss S, Finnis L, Manners J. Fathers and breast feeding: a pilot observational study. J R Soc
Health 1993;113:176-80.
164. Giugliani ER, Bronner Y, Caiaffa WT, Vogelhut J, Witter FR, Perman JA. Are fathers
prepared to encourage their partners to breast feed? A study about fathers' knowledge of
breast feeding. Acta Paediatr 1994;83:1127-31.
165. Molinari, L. and Speltini, G. Breastfeeding practices and values: a study of social
representations. Swiss Journal of Psychology. 57(4), 235-247. 1998.
166. Jordan, P. L. and Wall, V. R. Breastfeeding and fathers: illuminating the darker side. Birth
Issues in Perinatal Care 7(4), 210-213. 1990.
167. Lewis, C. Men: Supporters or supplanters? Fyle, J., Dodds, R., and Allanach, V. 2-4. 2000.
Royal College of Midwives, National Childbirth Trust, Community Practitioners and Health
Visitors Association and Royal College of Nursing. Barriers to Breastfeeding; time for a new
approach. 15-5-2000. (Conference Proceeding)
168. Hewat RJ,.Ellis DJ. Similarities and differences between women who breastfeed for short and
long duration. Midwifery 1986;2:37-43.
Reference list 282

169. Wylie J,.Verber I. Why women fail to breast feed:a prospective study from booking to 28
days post partum. Journal of Nutrition and Dietetics 1994;7:115-20.
170. Hughes P,.Rees C. Artificial feeding: choosing to bottle feed. British Journal of Midwifery
1997;5:137-42.
171. Graffy JP. Mothers' attitudes to and experience of breastfeeding: a primary care study. British
Journal of General Practice 1992;42:61-4.
172. Rajan L. The contribution of professional support, information and consistent correct advice
to successful breast feeding. Midwifery 1993;9:197-209.
173. Garforth S,.Garcia J. Breastfeeding policies in practice - "No wonder they get confused".
Midwifery 1989;5:75-83.
174. Woollett A. Who breastfeeds? The family and cultural context. Journal of Reproductive and
Infant Psychology 1988;5:127-31.
175. Sciacca JP, Phipps BL, Dube DA, Ratliff MI. Influences on breast-feeding by lower-income
women: an incentive- based, partner-supported educational program. J Am Diet Assoc
1995;95:323-8.
176. Proctor M. Measuring attitudes. In Gilbert N, ed. Researching Social Life, pp 116-34. London
: Sage, 1993.
177. Jones D. Attitudes of breastfeeding mothers: a survey of 649 mothers. Social Science and
Medicine 1986;23:1151-6.
178. Rosenstock IM. Historical origins of the health belief model. Health Education Monographs
1974;2:328-35.
179. Becker MHe. The Health Belief Model and Personal Health Behavior. New Jersey:
Thorofare, 1994.
180. Dix DN. Why women decide not to breast feed. BIRTH 1991;18:222-5.
181. Thomson AM. Why don't women breast feed? In Robinson S, Thomson AM, eds. Midwives,
Research and Childbirth, pp 215-40. London: Chapman & Hall, 1989.
182. Libbus MK, Bush TA, Hockman NM. Breastfeeding beliefs of low-income primiparae.
Int.J.Nurs.Stud. 1997;34:144-50.
183. Ajzen I,.Fishbein M. The prediction of behavioural intentions in a choice situation. Journal of
Experimental Social Psychology 1969;5:400-16.
184. Ajzen I,.Fishbein M. The prediction of behavior from attitudinal and normative variables.
Journal of Experimental and Social Psychology 1970;6:466-87.
185. Ajzen I. From intentions to actions: A theory of planned behaviour. In Kuhl J, Beckmann J,
eds. Action Control, pp 11-39. Berlin: Springer-Verlag, 1985.
Reference list 283

186. Manstead ASR, Proffitt C, Smart JL. Predicting and understanding mothers' infant-feeding
intentions and behavior: testing the theory of reasoned action. Journal of Personality and
Social Psychology 1983;44:657-71.
187. Janke JR. Development of the Breast-Feeding Attrition Prediction Tool. Nurs Res
1994;43:100-4.
188. Duckett L, Henly S, Avery M, Potter S, Hills-Bonczyk S, Hulden R et al. A theory of planned
behavior-based structural model for breast-feeding. Nursing Research 1998;47:325-6.
189. Prochaska JO,.DiClemente CC. Stages and processes of self-change in smoking: towards an
integrative model of change. Journal of Consulting and Clinical Psychology 1983;51:390-5.
190. Tones BK. Devising strategies for preventing drug misuse: the role of the health action
model. Health Education Research 1988;2:305-17.
191. Tones BK. Making a change for the better. Healthlines 1995;17-9.
192. Backett K, Currie C, Hunt S, McQueen D, Martin C, Ziglio E. Health Related Behavioural
Change. In Research Unit in Health and Behavioural Change UoE, ed. Changing the Public
Health, pp 67-86. Chichester: John Wiley & Sons, 1989.
193. Lyons S. A prospective study of post traumatic stress symptoms 1 month following childbirth
in a group of 42 first-time mothers. Journal of Reproductive and Infant Psychology
1998;16:91-105.
194. Maclean HM. Implications of a health promotion framework for research on breastfeeding.
Health Promotion 1989;3:355-60.
195. Maclean HM. Women's experience of breastfeeding: a much needed perspective. Health
Promotion 1989;3:361-70.
196. Bunton R, Murphy S, Bennett P. Theories of behavioural change and their use in health
promotion: some neglected areas. Health Education Research 1991;6:153-62.
197. Rogers E. The diffusion of Innovations. New York: Free press, 1983.
198. Stott NCH, Kinnersley P, Rollnick S. The limits to health promotion. BMJ 1994;309:971-2.
199. Sikorski, J. and Renfrew, M. J. Support for breastfeeding mothers. The Cochrane Database of
Systematic Reviews. Issue 1. 1999. Oxford, Update Software.
200. Martin J, Monk J. Infant feeding 1980. London: Office of Population Censuses and Surveys,
1982.
201. Illingworth RS,.Stone DG. Self-demand feeding in a maternity unit. Lancet 1952;1:683-7.
202. Renfrew MJ, Lang S. Pregnancy and Childbirth Module. In Enkin MW, Keirse MJNC,
Renfrew MJ, Neilsen JP, eds. The Cochrane Database of Systematic Reviews., Oxford:
Update Software, 1996.
Reference list 284

203. Renfrew, M. J. and Ross McGill, H. Enabling women to breastfeed; Interventions which
support or inhibit breastfeeding. A structured review of the evidence. 1996. Leeds, University
of Leeds.
204. Widstrom AM, Wahlberg V, Matthiesen AS, et al. Short-term effects of early sucking and
touch of the nipple on maternal behaviour. Early Human Development 1990;21:153-63.
205. Haider, R., Ashworth, A., Kabir, I., and Huttly, S. R. A. Effect of community-based peer
counsellors on exclusive breastfeeding practices in Dhaka, Bangladesh: a randomised
controlled trial. Lancet 356, 1643-1647. 11-11-2000.
206. Tedstone A, Dunce N, Aviles M, Shetty P , Daniels L. Effectiveness of interventions to

promote healthy feeding in infants under one year of age: a review. London: Health
Education Authority, 1998.
207. Bolam A, Manandhar DS, Shrestha P, Ellis M, M de Le Costello A. The effects of postnatal
health education for mothers on infant care and family planning practices in Nepal: a
randomised controlled trial. BMJ 1998;316:805-11.
208. Brent NB, Redd B, Dworetz A, D'Amico F, Greenberg JJ. Breast-feeding in a low-income
population: program to increase incidence and duration. Archives of Pediatrics & Adolescent
Medicine. 1995;149:798-803.
209. Haider R, Islam A, Hamadani J, Amin NJ, Kabir I, Malek MA et al. Breast-feeding
counselling in a diarrhoeal disease hospital. Bulletin of the World Health Organization
1996;74:173-9.
210. Kistin N, Abramson R, Dublin P. Effect of peer counselors on breastfeeding initiation,

exclusivity, and duration among low-income urban women. Journal of Human Lactation.
1994;10:11-5.
211. Redman S, Watkins J, Evans L. Evaluation of an Australian intervention to encourage

breastfeeding in primaparous women. Health Promotion International 1995;10:101-3.
212. Sjolin S, Hofvander Y, Hillervik C. A prospective study of individual courses of

breastfeeding. Acta Paediatrica Scandinavia 1979;68:521-9.
213. Lynch SA, Koch AM, Hislop TG, Coldman AJ. Evaluating the effect of a breastfeeding
consultant on the duration of breastfeeding. Canadian Journal of Public Health 1986;77:190-
5.
214. Mongeon M,.Allard R. Controlled study of a regular telephone support program given by
volunteers on the establishment of breastfeeding (French) . Canadian Journal of Public
Health 1995;86:124-7.
215. Barros FC, Halpern R, Victora CG, Teixeira ABM, Beria JU. A randomised intervention trial
to increase breast-feeding prevalence in southern Brazil. Rev.Saude Publica 1994;28:277-83.
216. Jones DA,.West RR. Effect of a lactation nurse on the success of breastfeeding: a randomised
controlled trial. Journal of Epidemiology and Community Health 1986;40:45-9.
217. Grossman LK, Harter C, Sachs L, Kay A. The effect of postpartum lactation counselling on
the duration of breast-feeding in low-income women. AJDC 1990;144:471-4.
Reference list 285

218. Langer A, Campero L, Garcia C, Reynoso S. Effects of psychosocial support during labour
and childbirth on breastfeeding, medical interventions, and mothers' wellbeing in a Mexican
public hospital: a randomised clinical trial. British Journal of Obstetrics & Gynaecology
1998;105:1056-63.
219. Morrell CJ, Spilby H, Stewart P, Walters S, Morgan A. Costs and benefits of communiity
postnatal support workers: a randomised controlled trial. Health Technology Assessment
2000;4:1-100.
220. Infant feeding: the physiological basis. Geneva: World Health Organisation, 1989.
221. Renfrew MJ, Woolridge MW, Ross McGill H. Enabling women to breastfeed. A review of
practices which promote or inhibit breastfeeding - with evidence-based guidance for practice.
London: The Stationery Office, 2000.
222. Daly SEJ, Kent JC, Huynh DQ, et al. The determination of short-term breast volume changes
and the rate of synthesis of human milk using computerised breast measurement.
Experimental Physiology 1992;77 :79-87.
223. Daly SEJ, Owens RA, Hartmann PE. The short-term synthesis and infant-regulated removal
of milk in lactating women. Experimental Physiology 1993;78 :209-20.
224. Minchin MK. Positioning for breastfeeding. BIRTH 1989;16:67-73.
225. Inch S, Renfrew MJ. Common breastfeeding problems. In Chalmers I, Enkin M, Keirse
MJNC, eds. Effective care in pregnancy an childbirth., pp 1375-89. Oxford: Oxford
University Press, 1989.
226. Rosenberg W,.Donald A. Evidence based medicine: an approach to clinical problem-solving.

BMJ 1995;310:1122-6.
227. Cochrane A. Effectiveness and efficiency. Random reflections on health services. London:
Nuffield Provincial Hospitals Trust, 1972.
228. Barrera ME, Rosenbaum PL. Supporting parents and promoting attachment. In Chalmers I,
Enkin M , Keirse MJNC, eds. Effective care in pregnancy and childbirth., pp 221-46. Oxford:
Oxford University Press, 1989.
229. Inch S, Garforth S. Establishing and maintaining breastfeeding. In Chalmers I, Enkin M,

Keirse MJNC, eds. Effective care in pregnancy and childbirth., pp 1359-73. Oxford: Oxford
University Press, 1989.
230. Vallenas C, Savage F. Evidence for the ten steps to successful breastfeeding. Geneva: World
Health Organization, 1999.
231. Duffy EP, Percival P, Kershaw E. Positive effects of an antenatal group teaching session on
postnatal nipple pain, nipple trauma and breast feeding rates. Midwifery 1997;13:189-96.
232. Hewat RJ,.Ellis DJ. A comparison of two methods of nipple care. Birth: Issues in Perinatal
Care & Education 1987;14:41-5.
233. Storr GB. Prevention of nipple tenderness and breast engorgement in the postnatal period. J
Obstet Gynecol Neonatal Nurs 1988;17:203-9.
Reference list 286

234. Alexander JM, Grant AM, Campbell MJ. Randomised controlled trial of breast shells and
Hoffman's exercises for inverted and non-protractile nipples. BMJ 1992;304:1030-2.
235. Gunther M. Sore nipples: causes and prevention. Lancet 1945;ii:590-3.
236. Woolridge MW. The 'anatomy' of infant sucking. Midwifery 21986;2:164-71.
237. Righard L,.Alade MO. Suckling technique and its effects on success of breastfeeding. BIRTH
1992;19:185-9.
238. Herd B,.Feeney JG. Two aerosol sprays in nipple trauma. Practitioner 1986;230:31-8.
239. Sharp D. Moist wound healing for cracked nipples. Breastfeeding Abstracts 1992;12:19.
240. Mathur GP, Chitranshi S, Mathur S, Singh SB, Bhalla M. Lactation failure. Indian Pediatrics
1992;29:1541-4.
241. Woolridge MW. Breastfeeding: physiology into practice. In Davies DP, ed. Proceedings of a
conference jointly organised by the Royal College of Physicians of London and the British
Paediatric Association, pp 13-31. RCPL Press, 1995.
242. Division of Child Health and Development, World Health Organisation. "Not enough milk".
1996. Geneva, World Health Organisation.
243. Thomsen AC, Esperen T, Maigaard S. Course and treatment of milk stasis, non-infectious
inflammation of the breast, and infectious mastitis in nursing women. Am J Obstet Gynecol
1984.
244. Dykes F,.Williams C. Falling by the wayside: a phenomenological exploration of perceived

breast-milk inadequacy in lactating women. Midwifery 1999;15:232-45.
245. Inch, S. and von Xylander. Mastitis; Causes and Management. Savage, F.
WHO/FCH/CAH/00.13. 2000. Geneva, World Health Organisation.
246. Apple RD. Mothers & Medicine; a social history of infant feeding 1890 - 1950. Madison: The
University of Wisconsin Press., 1987.
247. Freed GL,.Fraley JK. Effect of expectant mothers' feeding plan on prediction of fathers'
attitudes regarding breast-feeding. Am J Perinatol 1993;10:300-3.
248. Matich JR,.Sims LS. A comparison of social support variables between women who intend to
breast or bottle feed. Soc Sci Med 1992;34:919-27.
249. La Leche League. The womanly art of breastfeeding. Franklin Park, Illinois: La Leche League
International, 1992.
250. La Leche League. Mother-to-Mother Support Resources: Selected Review. 1998. Washington
DC, LINKAGES, Academy for Educational Development.
251. Hunter, H. "Surely other organisations should do this work?". 1998. Personal
Communication. (Letter circulated when she was considering writing a history of the NCT
Breastfeeding Promotion Group).
Reference list 287

252. Rankin G. The Birth of BPG. In Generations. 1998. London, National Childbirth Trust.
253. NCT. National Childbirth Trust. Annual report and accounts 1993. Need, Philippa. 1-36.
1993.
254. National Childbirth Trust. Annual Report and Accounts, 1997. London, National Childbirth
Trust.
255. Mills, H. Breastfeeding gurus caught in cash storm in a baby bottle. The Observer . 2-3-1997.
London.
256. National Childbirth Trust. Annual Report and Accounts, 1996. London, National Childbirth
Trust.
257. Office for National Statistics Information Service. 1998. (Personal Communication)
258. Hulme H. Unnecessary anointing. Nursing Times 1991;87:68.
259. Jones W. NCT counsellors prove their worth. Midwife, Health Visitor and Community Nurse
1991;27:106-8.
260. Hamand, M. Breastfeeding support at Holloway Prison. 1998. (Personal Communication)
261. Seel S. A Mutual Trust. Community Outlook 1987;8-9.
262. National Childbirth Trust. NCT Breastfeeding Counsellors. 1995. London, National
Childbirth Trust.
263. Breastfeeding Promotion Group. Code of Conduct for Breastfeeding Counsellors and
Trainees. 2. 1992. London, National Childbirth Trust.
264. Locklin MP. Telling the world: low income women and their breastfeeding experiences.
Journal of Human Lactation 1995;11:285-91.
265. Locklin MP,.Naber SJ. Does breastfeeding empower women? Insights from a select group of
educated, low-income minority women. BIRTH 1993;20:30-5.
266. Naber SJ,.Locklin MP. Are peer counselors the key to achieving breastfeeding success with
low income women? Breastfeeding Abstracts 1994;13:19-20.
267. Wright J. Breastfeeding and deprivation - the Nottingham peer counsellor programme.
MIDIRS Midwifery Digest 1996;6:212-5.
268. The Concise Oxford Dictionary. Oxford: Oxford University Press , 1995.
269. Smale, O. M. Women's Breastfeeding. 1996. D Phil thesis. University of Bradford.
270. Ward, E., King, M., Lloyd, M., Bower, P., Sibbald, B., Farrelly, S. , Gabbay, M., Tarrier, N.,
and Addington-Hall, J. Randomised controlled trial of non-directive counselling, cognitive
behaviour therapy and usual general practitioner care for patients with depression. I: Clinical
effectiveness. BMJ 2000;321:1383-1388.
271. Counselling in General Practice . Drug and Therapeutics Bulletin 38(7), 49-52. 2000.
Reference list 288

272. King M, Broster G, LLoyd M, Horder J. Controlled trials in the evaluation of counselling in
general practice. British Journal of General Practice 1994;44:229-32.
273. British Association for Counselling. Code of ethics and practice for counsellors. 1998.
Rugby, British Association for Counselling.
274. Office of Population Censuses and Surveys. Classification of occupations 1980. London:
HMSO, 1980.
275. Matthews DE, Farewell VT. Using and understanding medical statistics. Basel: Karger, 1988.
276. Advanced statistics: SPSS/PC + for the IBM PC/ XT/ AT. 1986. Chicago (IL), SPSS.
277. Office of Population Censuses and Surveys. Birth notifications for Nottingham, 1983 and
1984. 1984. Titchfield, Hampshire, OPCS.
278. Office of Population Censuses and Surveys. Standard Occupational Classification - Structure
and Definition of Major, Minor and Unit Groups. London: Her Majesty's Stationery Office,
1990.
279. Martin J. Infant feeding 1975: attitudes and practice in England and Wales. London: HMSO,
1978.
280. Hill PD,.Humenick SS. Development of the H & H Lactation Scale. Nursing Research
1996;45:136-40.
281. Stephenson J,.Imrie J. Why do we need randomised controlled trials to assess behavioural
interventions? BMJ 1998;316:611-3.
282. Dean, A. D., Dean, J. A., Burton, A. H., and Dicker, R. C. Epi Info: a word processing,
database and statistics program for epidemiology on microcomputers. (5). 1990. Stone
Mountain, Georgia, USD Incorporated. (Computer program).
283. National Childbirth Trust. Breastfeeding; A good start. 1990. London, National Childbirth
Trust.
284. Health Education Authority. Breastfeeding; Your questions answered by the Health Education
Authority. 1992. London, Health Education Authority.
285. Launer LJ, Forman MR, Hundt GL, Sarov B , Chang D, Berendes HW et al. Maternal recall
of infant feeding events is accurate. Journal of Epidemiology and Community Health
1992;46:203-6.
286. Freer CB. Health diaries: a method of collecting health information. Journal of the Royal
College of General Practitioners 1980;30:270-82.
287. SPSS Inc. SPSS for Windows: Base System User's Guide, Release 10.0. 1999. Chicago, IL,
SPSS Inc. 24-9-1999. (Computer Program)
288. Stata Corp. Stata Statistical Software. (Release 6.0). 1999. College
Station, Texas, Stata Corporation. (Computer Program)
Reference list 289

289. Sudman S, Bradburn NM. Asking Questions. San Francisco: Jossey-Bass Inc, 1982.
290. Streiner DL, Norman GR. Health measurement scales. A practical
guide to their development and use. Oxford, New York, Tokyo: Oxford University
Press, 1989.
291. Fallowfield L. The Quality of Life; Methodological Issues. In Fallowfield L, ed. The Quality
of Life; The missing measurement in health care., pp 31-74. London: Souvenir Press (E&A)
ltd, 1990.
292. Fallowfield L. Questionnaire design. Arch Dis Child 1995;72:76-9.
293. Leff EW, Gagne MP, Jefferis RN. Maternal perceptions of successful breastfeeding. Journal
of Human Lactation 1994;10:99-104.
294. Leff EW, Jefferis RN, Gagne MP. The development of the maternal breastfeeding evaluation
scale. Journal of Human Lactation 1994;10:105-11.
295. Office of Population Censuses and Surveys. Standard Occupational Classification. London:
Her Majesty's Stationery Office, 1991.
296. MIGENT Inc. Ability Plus. (1). 1987. Incline Village, Nevada, MIGENT Inc. 1987.
(Computer Program)
297. Collett D. Modelling Survival Data in Medical Research. London: Chapman and Hall, 1994.
298. Understanding controlled trials: Baseline imbalance in randomised controlled trials. BMJ
1999:319:185.
299. MRC Health Services and Public Health Research Board. A framework for development and
evaluation of RCTs for complex interventions to improve health. 2000. London, Medical
Research Council.
300. Schulz KF, Grimes DA, Altman DG, Hayes RJ. Blinding and exclusions after allocation in
randomised controlled trials: survey of published parallel group trials in obstetrics and
gynaecology. BMJ 1996;312:742-5.
301. Pringle M,.Churchill R. Randomised controlled trials in General Practice. Gold standard or
fool's gold? BMJ 1995;311:1382-3.
302. Wilson, S., Delaney, B., Roalfe, A., Roberts, L., Redman, V., Wearn, A. M., and Hobbs, F. D.
R. Randomised controlled trials in primary care: case study. BMJ 2000; 321:24-27.
303. Fairhurst K,.Dowrick C. Problems with recruitment in a randomized controlled trial of

counselling in general practice: causes and implications. J Health Serv Res Policy 1996;1:77-
80.
304. Foy R, Parry J, McAvoy B. Clinical trials in primary care. Targeted payments for trials might
help improve recruitment and quality. BMJ 1998;317 :1168-9.
Reference list 290

305. Gulick EE. Informational correlates of successful breastfeeding. Maternal and Child Nursing
Journal 1982;7:370-5.
306. Bates AS, Fitzgerald JF, Wolinsky FD. Reliability and validity of an instrument to measure
maternal health beliefs. Med Care 1994;32:832-46.
307. Kelley MA, Kviz FJ, Richman JA, Kim JH, Short C. Development of a scale to measure
gender-role attitudes toward breast-feeding among primiparas. Womens Health 1993;20:47-
68.
308. Baranowski T, Rassin DK, Richardson CJ, Bee DE, Palmer J. Expectancies of infant-feeding
methods among mothers in three ethnic groups. Psychology and Health 1990;5:59-75.
309. Cusson RM. Attitudes toward breastfeeding among female high school students. Pediatric
Nursing 1985;11:189-91.
310. Barnes J, Stein A, Smith T, Pollock JI, ALSPAC Study Team. Extreme attitudes to body
shape, social and psychological factors and a reluctance to breast feed. J R Soc Med
1997;90:551-9.
311. Dungy CI, Losch M, Russell D. Maternal attitudes as predictors of infant feeding decisions. J
Assoc Acad Minor Phys 1994;5:159-64.
312. Hughes, R. B. The development of an instrument to measure perceived emotional,

instrumental, and informational support in breastfeeding mothers. Issues in Comprehensive
Pediatric Nursing 7, 357-362. 1984.
313. Hauck Y,.Reinbold J. Criteria for successful breastfeeding: mothers' perceptions. Australian
College of Midwives Incorporated 1996;9:21-7.
314. Houston MJ, Howie PW, Cook A, McNeilly AS. Do breastfeeding mothers get the home
support they need? Health Bulletin 1981;39:166-72.
315. Reid, M. and Glazener, C. Postnatal support - does it work? Maternity Action (85), 6-7. 1999.
316. Oakley A. Experimentation and social interventions: a forgotten but important history. BMJ
1998;317:1239-42.
317. Auerbach KG. The influence of lactation consultant contact on breastfeeding duration in a
low-income population. Nebraska Medical Journal 1985;341-6.
318. Bagwell JE, Kendrick OW, Stitt KR, Leeper JD. Knowledge and attitudes toward breast-
feeding: differences among dietitians, nurses, and physicians working with WIC clients. J Am
Diet Assoc 1993;93:801-4.
319. McClurg-Hitt D,.Olsen J. Infant feeding decisions in the Missouri WIC Program. J Hum Lact
1994;10:253-6.
320. Strauss A, Corbin J. Basics of qualitative research; Techniques and procedures for
developing grounded theory. Thousand Oaks, Ca: Sage, 1998.
321. Bryman A, Burgess RG. Analysing Qualitative Data. London: Routledge, 1994.
Reference list 291

322. Denzin N. The Research Act in Sociology. London: Butterworth, 1970.
323. Bottorff JL. Persistence in breastfeeding: a phenomological investigation. Journal of

Advanced Nursing. 1990;15:201-9.
324. Raphael D. The midwife as doula: a guide to mothering the mother. Journal of Nurse-
Midwifery 1991;26:13-5.
325. Reason P, Rowan J. Human enquiry: A Sourcebook for New Paradigm Research. Chichester:
John Wiley, 1981.
326. Silverman D. Interpreting Qualitative Data; Methods for Analysing Talk, Text and
Interaction. London: Sage, 1993.
327. Abrams P. Evaluating soft findings: some problems of measuring informal care. Research
Policy and Planning 1984;2 :1-8.
328. Britton C. 'Feeling letdown': an exploration of an embodied sensation associated with

breastfeeding. In Nettleton S, Watson J, eds. The Body in Everyday Life, pp 64-81. London:
Routledge, 1998.
329. O'Connor Y. Education for Parenthood - A Time for Change. Midwives Chronicle
1993;106:198-210.
330. Nolan M. Antenatal education: failing to educate for parenthood. British Journal of Midwifery
1997;5:21-6.
331. Britton C. The influence of antenatal information on breastfeeding experiences. British

Journal of Midwifery 1998;6 :312-5.
332. Woolridge MW, Baum JD, Drewett RF. Effect of a traditional and of a new nipple shield on
sucking patterns and milk flow. Early Human Development 1980;4:357-64.
333. Auerbach KG. The effect of nipple shields on maternal milk volume. Journal of Obstetric,
Gynaecologic and Neonatal Nursing 1990;19:419-27.
334. Cronenwett LR,.Reinhardt R. Support and breastfeeding:a review. BIRTH 1987;14:199-203.
335. Tarkka MT, Paunonen M, Laippala P. Factors related to successful breast feeding by first-
time mothers when the child is 3 months old. Journal of Advanced Nursing. 1999;29:113-8.
336. Matthews K, Webber K, McKim E, Banoub-Baddour S, Laryea M. Maternal infant-feeding

decisions: reasons and influences. Canadian Journal of Nursing Research 1998;30:177-98.
337. Smale M. Working with breastfeeding mothers: the psychosocial context. In Clements S, ed.
Psychological perspectives on pregnancy and childbirth, pp 183-204. Churchill Livingstone,
1998.
338. The Expert Maternity Group. Changing Childbirth. 1993. London, HMSO.
339. Bick, D. The provision of community-based midwifery postnatal care in the UK - Making a
difference? MIDIRS Midwifery Digest 10(2), 227-231. 2000.
Reference list 292

340. Carson, C. and Thompson, R. Thinking nationally, acting locally . Fyle, J., Dodds, R., and
Allanach, V. 10-11. 2000. London, Royal College of Midwives, National Childbirth Trust,
Community Practitioners and Health Visitors Association and Royal College of Nursing. 15-
5-2000. (Conference Proceeding)
341. McInnes, R. The Glasgow Infant Feeding Action Research Project: An evaluation of a
community based intervention designed to increase the prevalence of breastfeeding in a
socially disadvantaged urban area. PEACH Paper No 7. 1998. Glasgow, Paediatric
Epidemiology and Community Health (PEACH) Unit, University of Glasgow.
342. Grant, J., Fletcher, M., and Warwick, C. South Thames Evidence Based Practice (STEP)
Project. Supporting Breastfeeding Women. 3. 2000. South Thames Evidence Based Practice
(STEP) Project.
343. Durdle T, Price S, Gabbott D. Promoting a brand image for breast milk. Health Visitor
1996;69:185-7.
344. Wilcock, V. Rebranding breastfeeding: promoting National Breastfeeding Awareness Week.

Fyle, J., Dodds, R., and Allanach, V. 13. 2000. Royal College of Midwives, National
Childbirth Trust, Community Practitioners and Health Visitors Association and Royal College
of Nursing. 15-5-2000. (Conference Proceeding)
345. Crombie IK,.Coid DR. Voluntary organisations: from Cinderella to white knight? BMJ
2000;320 :392-3.
Reference list 293

APPENDIX A
Study One: Antenatal Questionnaire
INFANT FEEDING OUESTIONNAIRE
Date: ………………….…….
Husband present? YES/ NO
Part 1: To be completed at the antenatal home visit.
EXPLANATION FOR PATIENTS: “This questionnaire is part of a project to find out why women decide on
breast or bottle feeding and how they get on with it. The answer will be treated in strict confidence, but if you do
not wish to participate, you do not have to.”
(A) PREVIOUS CHILDREN: “First of all I would like to ask you whether you breast or bottle-fed your
previous children, if you have any.
la) Name ………………………………….. lb) Date of birth …………………………………..
lc) How did you plan to feed? BREAST/BOTTLE
1 d) How did you actually feed and how long for? ………………………………………………
1 e) How did you feel about it? POSITIVE/ NEGATIVE
2a) Name: ……………………………….. 2b) Date of birth: ………………………………….
2c) How did you plan to feed? BREAST/ BOTTLE
2d) How did you actually feed and how long for? …………………………………………………
2e) How did you feel about it? POSITIVE/ NEGATIVE
3a) Name: …………………………………. 3b) Date of birth: …………………………………..
3c) How did you plan to feed? BREAST/BOTTLE
3d) How did you actually feed and how long for? …………………………………………………
3e) How did you feel about it? POSITIVE/NEGATIVE
(B) THE CURRENT PREGNANCY AND FEEDING PLANS:
1) How many weeks pregnant are you? …………………………………
2) How do you intend to feed this child? BREAST/ BOTTLE/ UNCERTAIN
3) When did you decide this? BEFORE/ DURING PREGNANCY
4) Why did you decide on this form of feeding? ………………………………………………..………

……………………………………………………………………………………………………………
…………………………………………………………………………………………………………...
5) Have you ever thought of breast/ bottle feeding? YES/ NO
6) Which form of feeding would you enjoy most? BREAST/ BOTTLE/ UNCERTAIN
Appendices 294
APPENDIX A
7) Which form of feeding would baby enjoy most? BREAST/ BOTTLE/ UNCERTAIN
8) Which is most convenient? BREAST/ BOTTLE/ UNCERTAIN
9) Which is best for baby? BREAST/ BOTTLE/ UNCERTAIN
10) Would the problem of finding quiet places to feed influence you? YES/NO
11) Would you be embarrassed about breastfeeding in front of:

(a) Male family member YES/NO
(b) Female family member YES/NO
(c) Males outside the family YES/NO
(d) Females outside family YES/NO
(Intended breastfeeders only)

12) How confident do you feel about your ability to breastfeed successfully?
(a) VERY CONFIDENT
(b) FAIRLY CONFIDENT
(c) RATHER ANXIOUS
13) How long do you intend to breastfeed for? (a) UP TO SIX WEEKS
(b) SIX WEEKS TO 4 MONTHS
(c) MORE THAN 4 MONTHS
(C) FAMILY AND SOCIAL FACTORS:
1) How were you fed as an infant? BREAST/ BOTTLE/ UNCERTAIN
2) When you were younger, which type of feeding do you remember used most in your family?
BREAST/ BOTTLE/ UNCERTAIN
3) (Examples of questions asked for husband’s views)
(a) Have you discussed infant feeding with your husband?
(b) Which would he favour?
(c) Do you think he influenced you?
DISCUSSED PREFERENCE INFLUENCE

(a) (b) (c)
(i) Husband/ Boyfriend: YES/NO BREAST/BOTTLE/UNCERTAIN YES/NO
(ii) Mother: YES/NO BREAST/BOTTLE/UNCERTAIN YES/NO
(iii) Sister: YES/NO BREAST/ BOTTLE/ UNCERTAIN YES/NO
(iv) Sister-in -law: YES/NO BREAST/ BOTTLE/ UNCERTAIN YES/NO
(v) Friends: YES/NO BREAST/BOTTLE/UNCERTAIN YES/NO
(vi) G.P. YES/NO BREAST/BOTTLE/UNCERTAIN YES/NO
(vii) Hospital doctor: YES/ NO BREAST/ BOTTLE/ UNCERTAIN YES/ NO
(viii) Midwife (before now): YES/NO BREAST/BOTTLE/UNCERTAIN YES/NO
(ix) Health visitor: YES/NO BREAST/ BOTTLE/ UNCERTAIN YES/ NO
Appendices 295
APPENDIX A
(D) THE DECISION

“Now I would like to ask you a few detailed questions about your decision on how to feed. I am going
to read out a list of possible factors in your decision. For each, could you say whether it influenced you
a lot, a little or not at all.” (Read list through quickly then individually to record response.)
(i) Personal preference: A LOT / A LITTLE I NOT AT ALL
(ii) Pleasure: A LOT I A LITTLE I NOT AT ALL
(iii) cost A LOT / A LITTLE / NOT AT ALL
(iv) Family attitude: A LOT / A LITTLE / NOT AT ALL
(v) Convenience: A LOT I A LITTLE / NOT AT ALL
(vi) Your Health: A LOT / A LITTLE / NOT AT ALL
(vii) Baby’s Health: A LOT I A LITTLE I NOT AT ALL
(viii) Plans to return to work: A LOT / A LITTLE I NOT AT ALL
(ix) Getting family help with feeding: A LOT I A LITTLE I NOT AT ALL
(B) PERSONAL INFORMATION: “Finally, I would like to ask a few questions about yourself.”
1) Name:……………………………………………………………..
2) Age: ………………………
3) Are you married? YES/ NO
3a) How long have you been married? …………………………
3b) (If unmarried) How long together? ………………………..
4) Have you worked? YES/ NO
4a) Last job? ……………………………………………………
4b) Do you plan to return to work? If so, when? (a) No plans to return
(b) Within 6 months
(c) After 6 months
5) What is your husband/ boyfriend’s job? …………………………………………………………….
6) Can you tell me about your father’s job? ……………………………………………………………
7) How old were you when you left school? (a) 15

(Include college if full time) (b) 16
(c) 17
(d) 18 or over
_________________________________________________________________________________________.
Thank you for your help.
__________________________________________________________________________________________
Appendices 296
APPENDIX A
Study One: Six-week Postnatal Questionnaire
INFANT FEEDING QUESTIONNAIRE

Code No…………………..
Feeding plan:…………….
Husband present: Yes/ No
Part 2: To be completed by health visitor at six week post-natal visit.
EXPLANATION FOR PATIENTS: ‘This questionnaire is the second part of a project to find out how women
get on with feeding their babies. You may remember that your midwife asked you various questions when she
visited you at home early in the pregnancy. This follows from that and, as before, the answers will be treated in
strict confidence. If you would rather not participate, you do not have to.”
(A) 1) Name of child: ………………………………

2) Date of birth: ………………………………..
3) Age of child (weeks): ………………………
4) Home/ GP unit/ Consultant delivery?
(B) THE DELIVERY AND HOSPITAL EXPERIENCE:
1) When you went into hospital, how did you intend to feed? BREAST/ BOTTLE/ UNCERTAIN.
I a) (If different from when interviewed by midwife)
Why did you change your mind? …………………………………………………………
…………………………………………………………………………………………………………
2) What type of delivery did you have? (a) NORMAL DELIVERY
(b) FORCEPS
(c) CAESARIAN GA. -
(d) CAESARIAN EPIDURAL

-
2a) How did you feel about the delivery? ………………………………………………………………
……………..………………………………………………….……… POSITIVE/ NEGATIVE
3) Was the baby born prematurely, or unwell in any way in the first few days? YES/ NO
3a) If so, what was the matter? ………………………………………………………………
3b) Was the baby in a special care baby unit, and if so, for how long? …………………….
4) Did you put the baby to breastfeed at any stage? YES/ NO

(Include all attempts, even if unsuccessful).
________________________________________________________________________________________
Question 4 defines a ‘breastfeeder”. If the answer is no, omit the rest of section B, sections C and D, moving
directly to section E.
_________________________________________________________________________
Appendices 297
APPENDIX A
5) How long after the delivery was this? (a) IMMEDIATELY

(b) 0-6 HOURS
(c)6- I2HOURS
(d) 12 -24 HOURS
(e) MORE THAN 24 HOURS
6) What was your first reaction to feeding?

……………………………………………………………….
……………………………………………………………………………………………………………
…..
7) Did you have any difficulty getting the baby to take to the breast? YES/ NO
8) How much of the time was the baby with you? (a) ALL OF THE TIME
(b) ALL DAY BUT NOT AT NIGHT
(c) PART OF THE DAY ONLY
(d) NONE OF THE TIME
9) Did your baby receive any bottle milk in hospital? YES/ NO
9a) (Jf yes) How often? …………………………………………………………………………………
9b) Why was this done? …………………………………………………………………………………
9c) How did you feel about it? ………………………………………………………………………….
10) How long did you stay in hospital? (a) LESS THAN 24 HOURS
(b) 24-48 HOURS
(c) 48 HOURS 5 DAYS
-
(d) MORE THAN 5 DAYS
11) How were you feeding when you came out of hospital? (a) ONLY BREAST FEEDING
(b) BOTH BREAST AND BOTTLE
(c) ONLY BOTTLE FEEDING
11a) (If both breast and bottle feeding)

When were you giving each type of feed? ……………………………………………………...
………………………………………..……………………………………………………………………
(“Bottle feeding” refers to the use of artificial milk but does not include orange juice or water given in a bottle).
(C) PROGRESS WITH FEEDING:
1) How long did you breastfeed for? …………………………………………………………………….

(Include those giving any breast milk).
(If still feeding, omit C2, 3, 4,& 5).

2) Why did you stop breastfeeding? ……………………………………………………………………
3) What did you do first when you had this problem? (i.e. the reason in question 2) …………………
….……………………………………………………………………………………………………….
Appendices 298
APPENDIX A
4) What advice, if any, did you get from the following?
4a)Family and friends: …………………………………………………………………………
4b) Midwife: ……………………………………………………………………………………
4c) Health Visitor: ……………………………………………………………………………...
4d) Doctor: ……………………………………………………………………………………...
5) How did you feel about changing over? ………………………………………………………………
…………………………………………………………………………………………………..
…………………………………………………………………………………………………..
(D) FEEDING DIFFICULTIES:

(Need not duplicate information recorded above)
1) Have you had any painful swelling or engorgement of the breasts? YES/ NO
1a) (If yes) What did you do about it? ………………………………………………………….
Ib) Did you get advice from anyone? Who from? ……………………………………...
What advice? …………………………………...
2) Have you ever felt that you weren’t making enough milk for the baby? YES/ NO
2a) (If yes) When? (a) DURING FIRST WEEK

(b) DURING SECOND WEEK
(c) SINCE SECOND WEEK
2b) What made you concerned? ………………………………………………………………..
………………………………………………………………………………………………
2c) Why do you think it happened? ……………………………………………………………..
………………………………………………………………………………………………
2d) What did you do about it? …………………………………………………………………
………………………………………………………………………………………………
2e) Did you get advice from anyone? Who from? …………………………………….
What advice? ……………………………………
………………………………………………… ………………………………………………
Appendices 299
APPENDIX A
3) Have you had any difficulty finding quiet places to feed? YES/ NO
4) Have you felt, or would you feel embarrassed about breastfeeding in front of the following?
(a) Male family member: YES/NO

(b) Female family member: YES/ NO
(c) Males outside the family: YES/ NO
(d) Females outside the family: YES/ NO
5) Have you, or baby had any other difficulties with breastfeeding? YES/ NO
5a) If so, what? ……………………………………………………………………………….
5b) What did you do about it yourself’? ……………………………………………………..
5c) Did you get advice from anyone? Who from? ……………………………………
What advice? …………………………………
6) Have you found breastfeeding easier, about the same, (a) EASIER
or more difficult than you expected? (b) ABOUT THE SAME
(c) MORE DIFFICULT
(E) CURRENT FEEDING AT SIX WEEKS.

(Ask all mothers. If this interview is delayed after the sixth week, questions El - 5 should still relate to
-
feeding during the baby’s sixth week.)
1) How have you fed the baby during the last week? (a) ONLY BREASTFEED1NG
(b) BOTH BREAST AND BOTTLE
(c) ONLY BOTTLE FEEDING
I a) (If both breast and bottle feeding)

When were you giving each type of feed? ……………………………………………………...
…………………………………………………………………………………………………..
2) How many times a day do you feed? ………………………………………………………………….
3) Do you feed at night? NO / ONLY ONCE/ MORE THAN ONCE
4) Have you been unwell, depressed, or had any illnesses since the baby was born? YES/ NO
4a) (If yes) What was the matter? ………………………………………………………………………..
4b) Did you get advice from anyone? Who from? ……………………………………………….
What advice? ……………………………………………..
4c) Do you think it affected your feeding? YES/ NO
Appendices 300
APPENDIX A
5) Has baby been unwell or had any illness at all? YES/ NO
5a) (If yes) What was the matter? …………………………………………………………..
5b) Did you get advice from anyone? Who from? ………………………………….
5c) Do you think it affected baby’s feedings? YES/ NO
6) Have you had any difficulties with bottle feedings? YES/ NO
6a) (If yes) What? …………………………………………………………………………..
6b) What did you do about it? ………………………………………………………………
6c) Did you get any advice from anyone? Who from? ………………………………….
What advice? ………………………………..
7) Have you found bottle feeding easier, about the same, (a) EASIER
or more difficult than you expected? (b) ABOUT THE SAME
(c) MORE DIFFICULT
__________________________________________________________________________________________
“Thank you for your help.”
__________________________________________________________________________________________
Name of interviewer: ……………………………………
Date of interview: ………………………………………..
Appendices 301
APPENDIX A
Study One: Six-month Postnatal Questionnaire
INFANT FEEDING OUESTIONNAIRE

Code No …….………..
Feeding: ……………..
Date: …………………
Part 3: To be completed six months post-natally.
EXPLANATION FOR PATIENTS: “This questionnaire is the final part of a project to find out how women get
on with feeding their babies. As before the answers will be treated in strict confidence. However, if you would
rather not participate, you do not have to.”
1) Name of child: …………………………………………..
2) Date of birth: ……………………………………………
3) Age of child (months) …………………………………..
4a) Have you breastfed your baby on any occasion during the last week? YES/ NO
4b) Have you given your baby any bottle feeds during the last week? YES/ NO
4c) (If both breast and bottle feeding)

When were you giving each type of feed? ……………………………………………………………….
……………………………………………………………………………………………………………
….
(“Bottle feeding” refers to the use of cow’s milk or artificial milk but not water or orange given in a bottle.)
5) How old was baby when you started solid food? …………………………………………………………
6) How many times a day do you feed baby? …………………………………………………………….
7) Do you feed at night? NO/ ONCE/ MORE THAN ONCE
8) (If breastfeeding stopped since last questionnaire)

8a) How long did you breastfeed for? ……………………………………………………………………
8b) Did you plan to stop then or did you have some problem with feeding?
(a) PLANNED TO STOP
(b) PROBLEM WITH FEEDING
(Ask all women, regardless of answer to question 8b)

8c) What was your reason for stopping breastfeeding? ……………………………………………….....
…………………………………………………………………………………………………………….
(If breastfeeding stopped because of a problem)

8d) What did you do first when you had this problem? ……………………………………………….....
……………………………………………………………………………….……………………………
Appendices 302
APPENDIX A
Study One: Six-month Postnatal Questionnaire
8e) What advice, if any did you get from the following:
i) Family and friends? …………………………………………………………..
ii) Midwife? ……………………………………………………………………...
iii) Health visitor? ………………………………………………………………..
iv) Doctor? ……………………………………………………………………….
8f) How did you feel about changing over? …………………………………………………...
…………………………………………………………………………………………………
__________________________________________________________________________________
_
“Thank you for your help”
__________________________________________________________________________________
Name of interviewer: ………………………………
Appendices 303
Appendices 304
Appendices 305
Appendices 306
Appendices 307
Appendices 308
Appendices 309
Appendices 310
Appendices 311
Appendices 312
Appendices 313
Appendices 314
Appendices 315
APPENDIX C
Study two: Six week postnatal questionnaire
INFANT FEEDING PROJECT [ ]
Directions: This questionnaire is part of a project to look at the support women get with breastfeeding. Please read it carefully
and tick the boxes which apply. If you have brought your feeding diary with you, please return it with this questionnaire.
As before, your answers will be treated in strict confidence. Thank you for your help.
A. The birth:
1) What type of delivery did you have? Please tick one box
Normal delivery
Forceps or Ventouse delivery
Caesarian - general anaesthetic
Caesarian - epidural
2) Was your baby premature, or unwell in the first few days?

No
Yes If yes, What was the matter?............................................................................
2a) Was your baby in the Special Care Baby Unit?

No
Less than 24 hrs
More than 24 hrs
3) Did you put your baby to the breast to feed at any stage?
(Please include all attempts, even if unsuccessful)
Yes
No If no, Why did you decide not to?......................................................................
______________________________________________________________________________________________________
If your answer to question 3 is "No" and you never breastfed this child, you do not need to complete the rest of the questionnaire.
______________________________________________________________________________________________________
Appendices 316
APPENDIX C
B. Current feeding: You may use your feeding diary as a reminder if you wish. "Bottle feeding" refers to the use of formula
milk but does not include juice or water given in a bottle. "Breast feeding" includes the use of your own expressed breast milk.
4) At present is your baby.... .
Breast fed → Go to question 4a
Bottle fed → Go to question 5
Both breast and bottle → Go to question 6
4a) Do you give your baby any milk in a bottle at present, (apart from expressed breast milk)?
Yes (even if occasionally) → Go to question 6
No → Go to question 7
5) When did you last breast feed your baby? ......................................................................... Please write in the age or date.
5a) What were your reasons for stopping? Please give all reasons, but indicate which you feel is most important.
.....................................................................................................................................................................................................
.....................................................................................................................................................................................................
5b) Would you like to have continued breastfeeding for longer, or had you fed for as long as you intended?
I would have liked to have breast fed for longer
I breast fed for as long as I intended
6) When did your baby first have any bottle milk? .................................................... Please write in the age or date.
6a) What was the reason for giving formula feeds? .............................................................................................................
6b) How much bottle milk are you now giving daily? (In 24 hrs) Either ........oz or .........ml
7) Do you give your baby any juice or water? Yes If Yes, How much a day? Either ........oz or ..........ml
No
Appendices 317
APPENDIX C
C. The next questions ask about your experience with breastfeeding. You may add any comments you wish.
8) How have you found breastfeeding?

Much easier than I expected
A little easier than I expected
As I expected
A little harder than I expected
Much harder than I expected
9) Have you felt confident or unsure about your ability to breastfeed?

I have felt very confident
I have felt fairly confident
I have felt fairly unsure
I have felt very unsure
10) Have you found breastfeeding stressful?

Most of the time
Some of the time
A little of the time
Very little of the time
11) Have you enjoyed breastfeeding?

A lot
A fair amount
A little
Not at all
12) Have you worried that your baby may not be gaining enough weight?
A lot
A fair amount
A little
Not at all
Appendices 318
APPENDIX C
13) Have you felt you would be embarassed about breastfeeding in front of people you don't know?
A lot
A fair amount
A little
Not at all
14) Have you tried to contact a breastfeeding counsellor since your baby was born?
(Please do not include contacts with your health visitor or midwife).
No
Yes If "Yes" Please give details: ............................................................................
..............................................................................................................................................................................................................
14a) How did you hear of the counsellor? ...................................................................................................................................
14b) Did you have any difficulty contacting the counsellor?

No
Yes If "Yes" Please give details: ...........................................................................
14c) Did you find the counsellor helpful?

Very helpful Please explain: ................................................................................
Fairly helpful
A little helpful ...........................................................................................................
Not helpful
Appendices 319
APPENDIX C
D Feeding difficulties: The questions on the next three pages ask about common feeding problems and the help you have had.
Please tick one box; however, if you have not had the problem referred to, you can leave the rest of the page.
15) Have you had difficulty getting the baby to take the breast?
Most of the time → Go to question 15a
Some of the time → Go to question 15a
A little of the time → Go to question 15a
Not at all → Turn to next page, question 16
15a) What did you do yourself when you first had this problem?.............................................................................................
........................................................................................................................................................................................................
15b) What advice or suggestions, if any, did the following people give you for this problem?
- Husband or partner: .............................................................................................................................................................
- Other family and friends: .....................................................................................................................................................
- Midwife: ..............................................................................................................................................................................
- Health Visitor: .....................................................................................................................................................................
- Breastfeeding counsellor: ....................................................................................................................................................
- General practitioner: ............................................................................................................................................................
Appendices 320
APPENDIX C
16) Have you felt you weren't making enough milk for baby?
Most of the time → Go to question 16a
Some of the time → Go to question 16a
A little of the time → Go to question 16a
Not at all → Turn to next page, question 17
16a) What did you do yourself when you first had this problem?.............................................................................................
................................................................................................................................................................................................
- Husband or partner: .............................................................................................................................................................
- Other family and friends: .....................................................................................................................................................
- Midwife: ..............................................................................................................................................................................
- Health Visitor: .....................................................................................................................................................................
- Breastfeeding counsellor: ....................................................................................................................................................
- General practitioner: ............................................................................................................................................................
Appendices 321
APPENDIX C
17) Have sore nipples been a problem for you?
Yes, a severe problem → Go to question 17a

Yes, a moderate problem → Go to question 17a
Yes, a minor problem → Go to question 17a
Not a problem for me → Turn to next page, question 18
17a) What did you do yourself when you first had this problem?.................................................................................................
......................................................................................................................................................................................................
..
- Husband or partner: .................................................................................................................................................................
- Other family and friends: ..........................................................................................................................................................
- Midwife: ...................................................................................................................................................................................
- Health Visitor: ..........................................................................................................................................................................
- Breastfeeding counsellor: .........................................................................................................................................................
- General practitioner: .................................................................................................................................................................
Appendices 322
APPENDIX C
18) Have you been unwell, depressed or had any other problems since your baby was born?
No
.....................................................................................................................................................................................................
19) Has your baby been unwell or had any illness at all?
No
.....................................................................................................................................................................................................
20) Of all the advice you have received about breastfeeding, which was most helpful? .................................................................
................................................................................................................................................................................................................
20a) Who gave this advice? ............................................................................................................................................................
21) Of all the advice you have received, which was least helpful? ....................................................................................................
................................................................................................................................................................................................................
21a) Who gave this advice? .............................................................................................................................................................
Appendices 323
APPENDIX C
Please add anything else that you feel is important:-
Please give your baby’s full name:

…………………………………………………………………………………………………..
(This will help us contact you when your baby is 3 months old)
Date completed: ...................................................

__________________________________________________________________________________________
Thank you very much for completing this. Please return it to the receptionist or nurse who gave it to you. If
you were unable to complete it at the surgery, you can do so at home and return it by post. You do not need to
use a stamp.
Send it to: Dr Jonathan Graffy,
Appendices 324
APPENDIX C
Study two: Three month postnatal questionnaire
INFANT FEEDING PROJECT

[ ]
Your feeding at three months:
Please complete this while you wait to see the doctor or nurse. If you have brought your feeding
diary with you, please return it with the questionnaire. In this questionnaire, "Bottle feeding" refers
to the use of formula milk, but does not include juice or water given in a bottle. "Breast feeding"
includes the use of your own expressed breast milk.

Breast fed ⁭ → Go to question 1a
Bottle fed ⁭ → Go to question 2
Both breast and bottle ⁭ → Go to question 3
Yes (even if occasionally) ⁭ → Go to question 3
No ⁭ → You may leave the rest of the questionnaire. Please add
the date at the bottom and return it to the receptionist.
2) When did you last breast feed? .........................................................

Please write in the age or date.
2a) What were your reasons for stopping? ........................................................................................
........................................................................................................................................................
........................................................................................................................................................
2b) Would you like to have continued breast feeding for longer,
or had you fed for as long as you intended?
Please tick:
I would have liked to have breast fed for longer ⁭
I breast fed for as long as I intended ⁭
3) When did your baby first have any bottle milk? .......................................................................
3a) What was the reason for giving formula feeds? .......................................................................
........................................................................................................................................................
3b) How much bottle milk do you now give daily? (In 24 hrs)
Either ........oz or .........ml
Date Completed: .............................................

_______________________________________________________________________________________
Thank you very much for completing this. Please return it to the receptionist or nurse who gave it to
you. If you were unable to complete it at the surgery, you can do so at home and return it by post.
You do not need to use a stamp.
If you are still breastfeeding, please take the attached Feeding Diary 3 home and return it when you
bring your baby for the immunisation or check-up at four months. If you have stopped breastfeeding,
you do not need to complete the Feeding Diary.
Appendices 325
APPENDIX C
Study two: Four month postnatal questionnaire
INFANT FEEDING PROJECT

[ ]
Your feeding at four months:
Please complete this while you wait to see the doctor or nurse. If you have brought your feeding
diary with you, please return it with the questionnaire. In this questionnaire, "Bottle feeding" refers
to the use of formula milk, but does not include juice or water given in a bottle. "Breast feeding"
includes the use of your own expressed breast milk.

Breast fed ⁭ → Go to question 1a
Bottle fed ⁭ → Go to question 2
Both breast and bottle ⁭ → Go to question 3
Yes (even if occasionally) ⁭ → Go to question 3
No ⁭ → You may leave the rest of the questionnaire. Please add
the date at the bottom and return it to the receptionist.
3) When did you last breast feed? .........................................................

2a) What were your reasons for stopping? ........................................................................................
........................................................................................................................................................
........................................................................................................................................................
2b) Would you like to have continued breast feeding for longer,
or had you fed for as long as you intended?
Please tick:
I would have liked to have breast fed for longer ⁭
I breast fed for as long as I intended ⁭
3) When did your baby first have any bottle milk? .......................................................................
3a) What was the reason for giving formula feeds? .......................................................................
........................................................................................................................................................
3b) How much bottle milk do you now give daily? (In 24 hrs)
Either ........oz or .........ml
Date Completed: ...................................

_______________________________________________________________________________________
Thank you very much for completing this. Please return it to the receptionist or nurse who gave it to
you. If you were unable to complete it at the surgery, you can do so at home and return it by post.
You do not need to use a stamp.
Appendices 326
Appendices 327
Appendices 328
Appendices 329
Appendices 330
Appendices 331
Appendices 332
Appendices 333
Appendices 334
APPENDIX D
Creating a measure of duration of feeding for survival analyses
Note on generating outcome variables for duration of breastfeeding
Data from the six weeks, three months and four months questionnaires was used to generate single
measures of the duration of breastfeeding. The three postnatal questionnaires allowed women to
enter either the date, or number of days they breast fed, or until they first introduced bottle feeds.
- When both were entered, if there was a discrepancy between the reported duration and the date of
the change, the date was used to calculate the number of days from the baby's date of birth.
- Where there was a discrepancy between the feeding diary, completed at the time, and the
questionnaire, completed retrospectively at six weeks, three or four months, the data from the
feeding diary was used in preference.
- In a number of cases, information was available from more than one questionnaire. When there
were inconsistencies between these, the original forms were checked and the data from the
earliest questionnaire used whenever possible, on the basis that the recall period was less and the
data was therefore more likely to be accurate.
- The survival analyses were based on variables calculated for the number of days that participants
were known to have breast fed, (DAYSBR), or the number of days that they were known not to
have bottle fed, (DAYSBO). Additionally, two further variables, (STATUSBR and STATUSBO),
were calculated to record whether or not the actual time to the change in feeding behaviour was
known. STATUSBR and STATUSBO were given the value of "0" when participants were lost to
follow-up or had not changed their feeding behaviour by four months when data collection
ceased, (censored observations) or "1" when they had either stopped breastfeeding, or introduced
bottle feeds.
- When only the six-week questionnaire was available, DAYSBR was given the value of 42 for
women still giving any breastfeeds at six weeks. When the three month questionnaire was the last
one received for women giving any breastfeeds at three months, DAYSBR was given the value of
91 and for all women still giving any breastfeeds at four months, DAYSBR was given a value of
120. Similarly, DAYSBO was given the values of 42, 91 and 120 for women who had not
introduced bottle feeds by those ages, where no subsequent data was available. In all these cases,
STATUSBR or STATUSBO was recorded as "0".
Appendices 335
APPENDIX E
Additional data for section 3.3.4
Appendix E: Table 3.3.4

Recruitment by practice27
Potential Recruited % Excluded % Questionnaire not %
Practice recruits returned
1 427 105 24.6% 251 58.8% 71 16.6%
2 163 21 12.9% 72 44.2% 70 42.9%
3 211 31 14.7% 72 34.1% 108 51.2%
4 375 71 18.9% 99 26.4% 205 54.7%
5 188 21 11.2% 61 32.4% 106 56.4%
6 300 65 21.7% 121 40.3% 114 38.0%
7 211 22 10.4% 66 31.3% 123 58.3%
8 161 33 20.5% 38 23.6% 90 55.9%
9 96 26 27.1% 39 40.6% 31 32.3%
10 140 31 22.1% 48 34.3% 61 43.6%
11 194 38 19.6% 82 42.3% 74 38.1%
12 68 16 23.5% 20 29.4% 32 47.1%
13 51 0 0% 20 39.2% 31 60.8%
14 106 18 17.0% 33 31.1% 55 51.9%
15 122 28 23.0% 52 42.6% 42 34.4%
16 92 16 17.4% 44 47.8% 32 34.8%
17 75 18 24.0% 23 30.7% 34 45.3%
18 167 30 18.0% 60 35.9% 77 46.1%
19 94 27 27.7% 21 22.3% 46 48.9%
20 96 24 25.0% 37 38.5% 35 36.5%
21 110 30 27.3% 34 30.9% 46 41.8%
22 63 10 15.9% 28 44.4% 25 39.7%
23 79 7 8.9% 15 19.0% 57 72.2%
24 62 12 19.4% 17 27.4% 33 53.2%
25 137 43 31.4% 49 35.8% 45 32.8%
26 139 37 26.6% 55 39.6% 47 33.8%
27 54 16 29.6% 15 27.8% 23 42.6%
28 153 25 16.3% 33 21.6% 95 62.1%
29 22 2 9.1% 7 31.8% 13 59.1%
30 75 8 10.7% 15 20.0% 52 69.3%
31 61 17 27.9% 20 32.8% 24 39.3%
32 72 20 27.8% 24 33.3% 28 38.9%
Total: 4364 868 19.9% 1571 36% 1925 44.1%
27
In table 3.3.4, women who were eligible, but were either not asked, or declined to participate in the
randomised controlled trial or observation study are included in the excluded group for simplicity. The
"recruited" column includes those recruited to either the RCT or observation study.
Appendices 336
APPENDIX E
Appendices 337
APPENDIX E
Appendix E: Table 3.3.6

Language spoken by non-English speakers:
Language: Number of women: %

Turkish 94 42.5
Bengali 41 18.6
French 9 4.1
Somali 6 2.7
Spanish 5 2.3
Portuguese 3 1.4
Vietnamese 2 .9
Russian 1 .5
Arabic 1 .5
Ethiopian 1 .5
Not recorded 58 26.2
Total: 221
Appendices 338
APPENDIX E
Additional data on the effectiveness of randomisation
Appendix E: table 3.3.7a

Parity of intervention and control groups:

Yes Count 94 87 181
% within Randomisation 25.90% 24.40% 25.10%
No Count 269 270 539
Total 363 357 720
Chi-Square = 0.22 (1df) p = 0.637.
Appendix E: table 3.3.7b

Age group of intervention and control groups:

Under 20 Count 20 24 44
20 - 24 Count 63 54 117
25 - 29 Count 119 111 230
30 - 34 Count 106 119 225
35 and over Count 53 45 98
Total 361 353 714
Chi-Square = 2.65 (4df) p = 0.618
Appendix E: table 3.3.7c

Ethnicity of intervention and control groups:

White (UK) Count 212 205 417
White (Other) Count 37 37 74
African / Carribean Count 61 48 109
Indian Subcontinent Count 24 31 55
Other Count 23 26 49
Total 357 347 704
Chi-Square = 2.60 (4df) p = 0.627
339
APPENDIX E
Appendix E: table 3.3.7d

Social class of intervention and control groups: 28

I Count 38 31 69
II Count 81 98 179
IIINM Count 68 56 124
IIIM Count 90 88 178
IV Count 40 36 76
V Count 7 15 22
Other Count 22 13 35
Total 346 337 683
Chi-Square = 8.82 (6df) p = 0.184
Appendix E: table 3.3.7e

Terminal educational age of intervention and control groups:

Under 16 Count 25 26 51
16 Count 86 88 174
17 Count 51 52 103
18 Count 50 59 109
19 & over Count 142 124 266
Total 354 349 703
Chi-Square = 1.98 (4df) p = 0.740
28
Social class was coded by husband or partner's occupation when the woman had a partner and by her
own if she did not. The reasons for this approach are given in section 3.3.5 note iv.
340
APPENDIX E
Appendix E: table 3.3.7f
Intended duration of intervention and control groups:

Under 6 weeks Count 22 28 50
6 weeks - 3 months Count 75 77 152
3 - 6 months Count 150 152 302
6 - 9 months Count 51 36 87
9 - 12 months Count 25 30 55
Over 1 year Count 8 15 23
Total 331 338 669
Chi-Square = 5.86 (5df) p = 0.320
Appendix E: table 3.3.7g

Intention to return to work of intervention and control groups:

No plan to return Count 85 91 176
Within 6 months Count 117 118 235
After 6 months Count 122 104 226
Total 324 313 637
Chi-Square = 1.45 (2df) p = 0.484
341
APPENDIX E
Additional data for section 3.5.8a
Appendix data for table 3.5.8a
Only those who had decided to breastfeed
Crosstab
Randomisation
Breastfeeding Any breast Count 216 210 426
at six weeks % within Randomisation 67.5% 63.6% 65.5%
Only bottle Count 104 120 224
Total Count 320 330 650
Crosstab
Randomisation
Breastfeeding Any breast Count 216 210 426
Crosstab
Randomisation
Breastfeeding at Any breast Count 142 131 273
four months % within Randomisation 48.3% 43.0% 45.6%
342
APPENDIX E
Additional data for section 3.5.8a
Crosstab
Randomisation
Any bottle feeds Any bottle feeds Count 102 125 227
by 7 days % within Randomisation 31.9% 38.0% 35.0%
No bottle feeds Count 218 204 422
Crosstab
Randomisation
Bottle feeding Any bottle feeds Count 188 211 399
Crosstab
Randomisation
Bottle feeding at Any bottle feeds Count 213 241 454
four months % within Randomisation 72.4% 79.0% 75.8%
343
APPENDIX E
Additional for section 3.4.3
Social factors and the uptake of counselling
Appendix E: Table 3.4.3a

Type of postnatal contact and parity
Type of postnatal contact l
Previous Children Face to face Telephone No contact Total
Yes 8 41 38 87
9.2% 47.1% 43.7%
No 59 102 88 249
23.7% 41.0% 35.3%
Total 67 143 126 336

19.9% 42.6% 37.5%
2
Chi = 8.57, (2df), p = 0.014
Appendix E: Table 3.4.3b

Type of postnatal contact and terminal educational age
Type of postnatal contact l
Terminal educational Face to face Telephone No contact Total

age:
Under 16 1 7 15 23
4.3% 30.4% 65.2%
16 9 43 25 77
11.7% 55.8% 32.5%
17 9 16 21 46
19.6% 34.8% 45.7%
18 12 15 17 44
27.3% 34.1% 38.6%
19 & over 35 58 44 137

25.5% 42.3% 32.1%
Total 66 139 122 327

20.2% 42.5% 37.3%
2
Chi = 21.21, (8df), p = 0.007
344
345
APPENDIX G
WHO/UNICEF Ten Steps to Successful Breastfeeding
TEN STEPS TO SUCCESSFUL BREASTFEEDING
Every facility providing maternity services and care for newborn infants
should:
1. Have a written breastfeeding policy that is routinely communicated to all

healthcare staff.
2. Train all healthcare staff in the skills necessary to implement this policy.
3. Inform all pregnant women about the benefits and management of

breastfeeding.
4. Help mothers initiate breastfeeding within a half-hour of birth.
NB: The UNICEF UK Baby Friendly Initiative have revised the wording of
this step to: Help mothers initiate breastfeeding soon after birth.
5. Show mothers how to breastfeed and to maintain lactation even if they are
separated from their babies.
6. Give newborn infants no food or drink other than breastmilk, unless medically
indicated.
7. Practice rooming-in -- allow mothers and infants to remain together -- 24 hours

a day.
8. Encourage breastfeeding on demand.
9. Give no artificial teats or pacifiers (also called dummies or soothers) to

breastfed infants.
10. Foster the establishment of breastfeeding support groups and refer mothers to
them on discharge from the hospital or clinic.
WHO/UNICEF Baby Friendly Hospital Initiative (1992). The Global Criteria for the
WHO UNICEF Baby-friendly Hospital Initiative. Geneva: WHO and UNICEF
Appendices 346
APPENDIX H
Summary tables for intervention studies
Appendix H: table 1.5.7.a Sjolin et al, 1979

Sjolin S, Hofvander Y, Hillervik C. A prospective study of individual courses of breastfeeding. Acta Paediatrica Scandinavia
1979;68:521-9.
Design
Method Participants & Normal Intervention Focus of Venue of Outcome
setting Care Intervention intervention
Single site study. Urban Sweden - "Interview" with Maintenance of Hospital, Home Visits Partial and exclusive
Duration 12 months. maternity ward of paediatrician in breastfeeding and Telephone. breastfeeding at 2, 4,
Quasi-randomised - University Hospital. hospital on days 1 and 8, 12, 16, 20 and 24
(births before and 4 and at home at 2 Postnatal only weeks.
after midnight). Inclusion criteria: and 6 weeks and at 3
- resident in Uppsala months.
No info on drop-outs. - normal birth
Outcome assessment - healthy baby Telephone contact
not independent - birth weight >3kg weekly while
breastfeeding with
Quality score 2/5 Ethnicity not stated. visit if problem noted.
28% college or
university education.
n = 146
Results (number stopping/total in group, %) Interpretation & comment
Control Intervention Odds ratio [95% CI]
2 months: This early quasi-randomised study had significant methodological problems in that women in the
18/71, 25% 13/75, 17% 0.62 [0.28 , 1.37] intervention group were contacted weekly, while the control group were interviewed
3 months: retrospectively. It seems hard to justify including it in a meta-analysis, but it does provide useful
24/71, 34% 16/75, 21% 0.54 [0.26 , 1.11] information about events during the period before women discontinue breastfeeding.
4 months:
34/71, 48% 24/75, 32% 0.52 [ 0.27 , 1.00] Summary:
6 months: Small study, methodology weak. Impact on breastfeeding not significant.
44/71, 62% 40/75, 53% 0.70 [0.37 , 1.35]
Appendices 347
APPENDIX H
Appendix H: table 1.5.7.b Jones and West, 1986

Jones DA,.West RR. Effect of a lactation nurse on the success of breastfeeding: a randomised controlled trial. Journal of
Epidemiology and Community Health 1986;40:45-9.
Design
Single site study. Duration UK Hospital. Normal UK Individual support and Maintenance of Hospital and home Breastfeeding rates at
18 months. Quasi- maternity care problem solving by breastfeeding. 4 weeks, 3, 6 and 12
randomised, using Inclusion criteria: lactation nurse. months.
alternating two-week - Women attempting at Postnatal only
periods. Follow up 96%. least one feed. Satisfaction with care
Drop-outs recorded. - admission not and intention to
Independent outcome overlapping control and breastfeed next
assessment intervention periods. pregnancy.
Potential confounder: Late 55% primips

exclusion of 66 women Social class: 22% I & II,
due to overlap of 46% III, 13% IV & V
recruitment periods.
Quality score 4/5 n = 678
3 months: This study of the effectiveness of a single lactation nurse was quasi-randomised in that women
180/355, 51% 90/228, 39% 0.64 [0.46 , 0.89] were recruited to the control and intervention groups in alternating two-week blocks, although the
6 months: investigators did exclude women whose deliveries overlapped both control and intervention weeks.
257/355, 72% 142/228, 62% 0.63 [0.44 , 0.90] There could also be concerns about recall bias because duration of feeding was assessed
retrospectively by interview at twelve months. However, the finding of a significant increase in the
number of women breastfeeding up to four weeks was also reflected in more women reporting that
they had received enough help, fewer women reporting breastfeeding problems and more mothers
in the intervention group reporting that they found breastfeeding satisfying. Women in social
classes IV and V appeared to benefit most.
Summary:
Adequate size, well-designed non-random study. Shows benefit of support, particularly for
low-income mothers.
Appendices 348
APPENDIX H
Appendix H: table 1.5.7.c Moore et al, 1985

Moore WJ, Midwinter RE, Morris AF, Colley JRT, Soothill JF. Infant feeding and subsequent risk of atopic eczema. Archives of
Disease in Childhood 1985;60:722-6.
Design
Single site study. Urban UK antenatal Normal UK maternity Daily Health Visitor or Antenatal and Hospital, Home, Exclusive
19 months recruitment clinic. care Doctor visits when Postnatal Telephone breastfeeding at 3
Randomisation inpatient. months
method not stated. Inclusion criteria: Initiation and
Follow-up 90%. - Personal or partner Home visit at 4 - 6 maintenance of
Drop-out reasons not history of atopy. weeks. breastfeeding
stated.
Outcome assessment Exclusions: 24 hr telephone
not independent. - Non-white support
Possible counfounder: - Unsure of EDD
inclusion criteria - Multiple pregnancy.
designed for trial of
allergy prevention. No socio-economic
info.
Control Intervention Relative Risk [95% CI]
Stopping exclusive Although based on a failed attempt to conduct a controlled trial of the effects of breastfeeding in
breastfeeding before 3 preventing eczema, this study also tested the investigators' assumption that additional support
months: would encourage more mothers to breastfeed. Mothers were recruited if they, or their partners,
had a family history of asthma or eczema. Perhaps because their primary concern was whether
183 / 248 169/227 1.01 [0.91 - 1.12] the early introduction of cow's milk might be associated with eczema, the authors reported the
proportion exclusively breastfeeding at twelve weeks, but gave little information on the duration of
breastfeeding, or the age at which women introduced formula milk.
Twenty six percent of women in both control and intervention groups breastfed exclusively to three
months, but as the intervention made no difference, the investigators were unable to compare the
impact of different feeding regimes. The study design was however robust enough to merit
inclusion in the Cochrane review.
Appendices 349
APPENDIX H
Appendix H: table 1.5.7.d Lynch et al, 1986

Lynch SA, Koch AM, Hislop TG, Coldman AJ. Evaluating the effect of a breastfeeding consultant on the duration of breastfeeding.
Canadian Journal of Public Health 1986;77:190-5.
Design
Single site study. Urban Canada - Control group Home visit by Maintenance of Home and telephone. Duration of
Duration not stated hospital. received postnatal breastfeeding breastfeeding. breastfeeding.
Randomisation method visit by public health consultant within 5
not stated. Inclusion criteria: nurse who gave days of discharge. Postnatal only.
Follow-up unclear. - Intention to breastfeed advice determined
Outcome assessment - English speaker. largely by the Telephone calls to
independent. Exclusions: questions and mother weekly for 1
- Multiple births concerns of mother. month then monthly
Possible confounders: - Birth weight < 2.5kg for 2 - 6 months.
Significant differences - Gestation <37 weeks
in parity and intention
to return to work 41% primips
between groups.
3 months: This study evaluated the effectiveness of a single lactation consultant on the duration of
48/135, 36% 51/135, 38% 1.10 [0.67 , 1.80] breastfeeding, but the report omits information about the process used to recruit and randomise
6 months: the 270 women. This is important because the intervention group included more mothers of first
79/135, 59% 81/135, 60% 1.06 [0.65 , 1.73] babies and more women who intended to return to work. As has been found in other studies,
mothers of first babies were more likely to discontinue breastfeeding than those with previous
children, which may have contributed to the study showing no benefit for additional support.
Overall, breastfeeding rates were high in both control and intervention groups, with 80% of
mothers breastfeeding at six weeks, suggesting that relatively few women needed support. Given
this, it is perhaps not surprising that the study showed no benefit from the intervention.
Summary:
Moderate sized study, which showed no benefit from additional support. Randomisation
bias may have contributed to negative result.
Appendices 350
APPENDIX H
Appendix H: table 1.5.7.e Frank et al, 1987

Frank DA, Wirtz SJ, Sorenson JR, Heeren T. Commercial discharge packs and breastfeeding counselling: effects on infant-feeding
practices in a randomised trial. Pediatrics 1987;80:845-54.
Design
Single site study Urban USA (Boston) - Postpartum nursing Study 1: Breastfeeding Maintenance of Hospital and Exclusive breastfeeding
Recruitment 12 mths. Hospital. staff contact, counselling in hospital breastfeeding. telephone at 1, 2, 3 and 4 months.
94% follow-up. infrequent (20-40 mins) Any breastfeeding at 4
Randomisation Inclusion criteria: breastfeeding classes, Telephone call at 5,7, 14, Effect of months.
appropriate. - Breastfed once leaflet on 21, 28 days and 6, 8 & commercial Median duration of
Drop-outs recorded. - Spanish or English breastfeeding and 12 weeks. discharge packs. breastfeeding.
Independent outcome - Not on NICU > 24hrs midwife-led telephone Telephone advice line. Time to introduction of
assessment. - Telephone available line. Postnatal only solids.
Study 2: Commercial vs Hospital admission of
Quality score 5/5 57% primips research discharge pack. infants.
65% Black, 19%
Hispanic, White 13%
n = 343
4 months: This study tested two interventions independently, breastfeeding counselling delivered by a trained
70/160, 44% 60/163, 37% 0.75 [0.48 , 1.17] counsellor and the difference between commercial discharge packs and specially designed packs
which promoted breastfeeding. 91% of the women allocated to counselling received six of the eight
calls and 40% initiated a contact themselves.
The part of the study which compared commercial and pro-breastfeeding research discharge
packs showed that women who received the research packs breastfed for longer, but this may tell
us more about the ability of commercial packs to undermine breastfeeding than the research packs
to promote it. Although women who received counselling introduced solids later, (Median age
105 days vs 91 days), which was significant on survival analysis, there was no significant
difference in the duration of breastfeeding, or the proportion breastfeeding at four months.
Summary:
Complex study with two interventions, but inadequate sample size. Impact on
breastfeeding not significant.
Appendices 351
APPENDIX H
Appendix H: table 1.5.7.f Grossman et al, 1990

Grossman LK, Harter C, Sachs L, Kay A. The effect of postpartum lactation counselling on the duration of breast-feeding in low-
income women. AJDC 1990;144:471-4.
Design
Single site study, over Urban USA (Ohio) - Routine teaching Meeting with lactation Maintenance of Hospital and Breastfeeding rates at
10 months. hospital. regarding infant care counsellor, (nurse) for breastfeeding. telephone 6 weeks, 3 months
Quasi-randomised and feeding given by 30 - 45 mins. and 6 months.
(coin toss with women Inclusion criteria: nursing staff on Educational booklet. Postnatal only.
sharing room allocated - WIC programme for postnatal wards. Median duration of
to same toss.) low income mothers Telephone calls days breastfeeding.
Follow-up 90%. - intention to 2, 4, 7, 10 & 21.
Dropouts not stated. breastfeed
Outcome assessment Telephone helpline.
not independent. 1/3 primips.
54% black Lactation clinic visits
Quality score 3/5 for those with
n = 97 problems.

This was a relatively small study, and the authors wondered whether there had been a Hawthorne
3 months: effect, with control women breastfeeding for longer because of the interest in the study on the
23/44, 52% 32/49, 65% 1.70 [0.75 , 3.89] postnatal wards. Randomisation by coin toss is particularly open to bias if the person tossing the
6 months: coin has an interest in giving the "best support" to a mother. The results were not significant, but
34/44, 77% 42/49, 86% 1.75 [0.61, 4.99] showed rather higher breastfeeding rates in the control group.
Summary:
Small quasi-randomised study, which showed no benefit from support.
Appendices 352
APPENDIX H
Appendix H: table 1.5.7.g Serafino-Cross, 1992

Serafino-Cross P,.Donovan PR. Effectiveness of Professional Breastfeeding Home-Support. Journal of Nutrition Education
1992;24:117-22.
Design
Single site study over Urban USA - prenatal Control group given Home support from Maintenance of Home and telephone Breastfeeding at 2
14 months. clinics. contact name and breastfeeding breastfeeding. months and 6 months.
Randomisation number for clinic counsellor (5-8 visits,
method not stated. Inclusion criteria: nutritionist. lasting 30 - 60 mins.)
Follow-up 75% - Intention to breastfeed
Drop-out due to loss - Not previously Telephone advice if
of follow-up. breastfed > 1 month needed.
Outcome assessment - English speaking
not independent.
71% primips
Possible confounder: 40% White, 27% Black,
50% of control group 33% others.
not contactable at 6 92% eligible WIC
months. programme

2 months: This small study from New England included only 52 women, but offered women in the intervention
17/26, 65% 10/26, 38% 0.35 [0.12 , 1.02] group intensive support with an average of seven visits over a two-month period.
There are significant methodological concerns, which include a lack of information about the
method of randomisation, the differential follow-up rates between the control and intervention
groups and the fact that the researcher also provided the counselling herself. Its inclusion in a
meta-analysis seems questionable.
Summary:
Small intensive study. Limited generalisability and methodological concerns. Impact on
breastfeeding not significant.
Appendices 353
APPENDIX H
Appendix H: table 1.5.7.h Barros et al, 1994

Barros FC, Halpern R, Victora CG, Teixeira ABM, Beria JU. A randomised intervention trial to increase breast-feeding prevalence in
southern Brazil. Rev.Saude Publica 1994;28:277-83.
Design
Single site study. Urban Brazil - hospital Little statutory support Three home visits at Maintenance of Home Breastfeeding at
Stated as randomised - social assistants only 5, 10 and 20 days by breastfeeding. monthly intervals to 6
but method not Inclusion criteria visited if requested to a social assistant or months and median
described. - Family income less do so by hospital nutritionist. Postnatal duration of
93% follow-up. Drop- than twice minimum team. breastfeeding.
outs recorded. wage. (Assistants had
- Hospital stay < 5 breastfed themselves Time to introduction of
Outcome assessor days. and had training in artificial feeds.
independent of -Intention to breastfeeding
intervention. breastfeed. physiology and Breastfeeding
common problems.) problems & reasons
Quality score 4/5 n = 900 for weaning.

2 months: This large Brazilian study is published in Portuguese. It assessed the effectiveness of three home
157/414, 38% 114/424, 27% 0.60 [0.45 , 0.81] visits by a social assistant or nutritionist and found a significant increase in the proportion of
3 months: mothers who breastfed.
195/414, 47% 187/424, 44% 0.89 [0.68 , 1.16]
4 months: Whereas many intervention studies have been conducted in settings where normal postnatal care
228/414, 55% 225/424, 53% 0.92 [0.70 , 1.21] included a degree of breastfeeding support, women in this study's control group had little support,
6 months: enabling the investigators to evaluate the effect of any breastfeeding support, rather than just
257/414, 62% 254/424, 60% 0.91 [0.69 , 1.20] offering mothers additional support.
Summary:
Large, well-designed trial. Significant increase in proportion breastfeeding to 2 months.
Appendix H: table 1.5.7.i Brent et al, 1995

Brent NB, Redd B, Dworetz A, D'Amico F, Greenberg JJ. Breast-feeding in a low-income population: program to increase incidence
and duration. Archives of Pediatrics & Adolescent Medicine. 1995;149:798-803.
Appendices 354
APPENDIX H
Design
Single site study. Duration Urban USA - Standard 2 - 4 antenatal sessions Initiation and Ambulatory care Breastfeeding at 2
not stated. Randomisation ambulatory care hospital care: with lactation consultant maintenance of centre (clinic), hospital months. Median
partly described. Follow-up centre & hospital. Optional (10 -15 mins). breastfeeding and telephone. duration of
94%, drop-out reasons breast-feeding breastfeeding.
unclear. Assessment of Inclusion criteria: classes, Telephone call 48 hrs Antenatal and
outcome not independent. - English speaking postnatal care after discharge. postnatal Kaplan Meier survival
- Primiparae from ward staff analysis.
Potential confounders: Exclusions: & clinic follow Lactation clinic visit 1
Intention-to-treat analysis - gestation < 37 wks up. week postnatally.
not done. (7 Women who did - NICU >72 hrs
not receive 2 prenatal Lactation consultant
consultations were excluded 71% white contact at each child
from intervention group & 8 90% eligible for health clinic attendance
from control group as they WIC programme for until weaning or 1 year.
contacted a consultant.) low-income families.
Quality score 3/5 n =115
Control Intervention Odds ratio [95%
CI]
Not initiating This U.S. study included all mothers expecting their first babies, whether or not they intended to
breastfeeding: breastfeed. It therefore attempted to both encourage more mothers to breastfeed, and support those
39/57, 68% 20/51, 39% who wanted to do so.
When data for the incidence of ever breastfeeding were reanalysed on an 'intention to treat' basis,
Stopping by 2 months: adding the 8 excluded from the intervention and 7 excluded from the control group, the proportions
52/57, 91% 32/51, 63% 0.20 [0.08 , 0.48] initiating breastfeeding were 33/58 (51%) and 26/65 (40%) respectively. It is also unclear how data on
the feeding behaviour of the control group were obtained. This suggests the potential confounders did
influence the results. Although the data still show a significant increase in the numbers attempting
breastfeeding, there is a strong case for excluding this study from meta-analyses of the effectiveness
of support for women who want to breastfeed.
Summary: Small study assessing impact on both initiation rates and duration. Methodological
problems make the statistically significant impact on breastfeeding rates hard to interpret.
Appendix H: table 1.5.7.j Mongeon et al, 1995

Mongeon M,.Allard R. Controlled study of a regular telephone support program given by volunteers on the establishment of
breastfeeding (French). Canadian Journal of Public Health 1995;86:124-7.
Design
Appendices 355
APPENDIX H

Single site study. Urban Canada - Home visit by public Antenatal home visit. Maintenance of Antenatal home visit Breastfeeding rates at
Duration not stated. community based. health nurse during breastfeeding. and postnatal 1, 2, 3, 4, 5 and 6
Quasi-randomised first month and other Telephone call weekly telephone contact months.
(numbered tickets). Inclusion criteria: contacts as for 6 weeks then Antenatal and
97% follow-up. - Intention to determined by mother. fortnightly. postnatal
Reasons for drop-outs breastfeed.
recorded. - Not previously
Independent outcome breastfed.
assessment.
97% primips.
Quality score 3/5 57% University or
college education.
n = 200

2 months: This study, published in French, shows no benefit from the intervention which involved an
31/99, 31% 34/95, 36% 1.22 [ 0.67 , 2.21] antenatal home visit and postnatal telephone support. Although the sample only included 200
3 months: participants, the results do not suggest a trend that might have become significant with a larger
43/99, 43% 45/95, 47% 1.17 [0.67 , 2.06] sample size.
4 months:
51/99, 52% 54/95, 57% 1.24 [0.70 , 2.17] Summary:
6 months: Although quasi-randomised, this appears to be a well-designed study. It showed no benefit
79/99, 80% 71/95, 75% 0.75 [0.38 , 1.47] from an antenatal home visit and postnatal telephone support.
Appendices 356
APPENDIX H
Appendix H: table 1.5.7.k Redman et al, 1995

Redman S, Watkins J, Evans L. Evaluation of an Australian intervention to encourage breastfeeding in primaparous women. Health
Promotion International 1995;10:101-3.
Design
Method Participants & setting Normal Intervention Focus of Venue of Outcome
Care Intervention intervention
Single site study. Urban & suburban Usual care from Antenatal teaching Maintenance of Hospital, telephone Initiation of
Recruitment over 3.5 Australia - hospital. physician, hospital session (3 hrs) + written breastfeeding. and home. breastfeeding.
months. staff and antenatal info.
Quasi-randomised, Inclusion criteria: classes. Antenatal and Breastfeeding at six
(odd/even numbered - Intention to breastfeed. Postnatal hospital visit postnatal weeks and 4 months.
consent forms.) - Primips. (average 34 mins).
Follow-up 66%. Drop- - Live within 20km of Satisfaction with
out reasons recorded. hospital. Telephone calls at 2-3 intervention
Independent outcome - No plan to move before weeks and 3 months.
assessment by baby 4 months, Home visits on request.
questionnaire. Telephone support.
University 13.5%, full high
Quality score 3/5 school 65.%, some high Discussion group +
school 21%. n = 235 written info at 6 - 8 wks.
This Australian study tested a comprehensive intervention which included antenatal and postnatal
4 months: classes, written information, a visit from the breastfeeding consultant in hospital, phone calls at two
33/75, 44% 32/77, 42% 0.91 [0.48 , 1.72] and twelve weeks and telephone support when requested. Although the method of
randomisation, which relied on odd and evenly numbered consent forms, was not ideal, the
authors considered this was unlikely to have affected the results.
A potentially serious confounding factor was that only 66% of participants provided information on
their feeding up to four months. Those who did respond had a relatively high breastfeeding rate
and were older and more likely to be married than non-responders. This raises the possibility that
the study failed to record the experiences of those who were more likely to stop; the very women
the intervention most needed to help if it were to be effective. There was however, no evidence
that the intervention was effective for those who did reply.
Summary:
Moderate-sized study with potential bias from inadequate follow-up. No benefit shown from
intensive intervention.
Appendices 357
APPENDIX H
Appendix H: table 1.5.7.l Haider et al, 1996

Haider R, Islam A, Hamadani J, Amin NJ, Kabir I, Malek MA et al. Breast-feeding counselling in a diarrhoeal disease hospital.
Bulletin of the World Health Organization 1996;74:173-9.
Design
Single site study. Duration Mothers with infants Inpatient hospital Minimum of 3 Maintenance of Hospital inpatient and Exclusive and
not stated. Randomisation admitted to care, including daily sessions, with breastfeeding. home. predominant
appropriate. Follow-up diarrhoeal disease health education lactation counsellor or breastfeeding at
83%. Drop-outs recorded. hospital in sessions, which research physician on Postnatal discharge.
Outcome assessment not Bangladesh. included advice on days 1, 2 and day of
independent. exclusive breast- discharge. Exclusive breastfeeding
Inclusion criteria: feeding to five at 2 weeks follow-up.
Potential confounders: - Infants < 12 wks months. Encouragement to
Controls visited at home old stay in hospital until
by lactation consultant - Living within 15km diarrhoea resolved.
"without intervening for of Dhaka.
breastfeeding Home visit at one
management". Intervention Exclusions: week by counsellor.
mothers encouraged to - Severe diarrhoea
stay in hospital until - Mothers unable to
diarrhoea resolved. More stay in hospital.
primips in intervention
group than control (44 vs N = 250
65, p<0.01) , but parity
was unrelated to outcome.
Quality score 5/5

Control Intervention Rel Risk [95% CI]
Exclusive A number of aspects of this study make it very different from the others included. In particular, the
breastfeeding two fact that infants recruited were ill makes this more a study of disease management than of a health
weeks after promotion intervention.
discsharge
This is a well-designed study, but It is questionable whether it should be included within meta-
8/103 78/104 9.66 [ 4.66 - 18.96] analyses of breastfeeding support for otherwise healthy mothers and infants.
Appendices 358
APPENDIX H
Appendix H: table 1.5.7.m Gagnon et al, 1997

Gagnon AJ, Edgar L, Kramer MS, Papageorgiou A, Waghorn K, Klein MC. A randomized trial of a program of early postpartum
discharge with nurse visitation. Am J Obstet Gynecol 1997;176:205-11.
Design
Single site study over 12 Urban Canada - Hospital care. Antenatal home visit. Initiation and Home and telephone Predominant
months. Randomisation hospital-based. maintainance of breastfeeding, (less
appropriate but 50% excluded Early discharge from breastfeeding than one breast milk
after randomisation because Inclusion criteria: hospital. substitute a day) at
of requirements of early - Para 0 - 4 Early vs standard one month.
discharge programme. - Normal pregnancy Telephone call at 48 discharge programme.
Follow-up 87%. Drop-outs - English, French or hrs and 10 days.
recorded. Outcome by mailed Spanish speaker. Antenatal and
questionnaire. - Live within 30 Postnatal home visits postnatal.
mins of hospital at 3 and 5 days.
Potential confounders:
Intention to treat analysis not Exclusions based
performed due to postpartum on obstetric
exclusions. Some significant complications and
differences between control infant health
and intervention groups which problems.
were adjusted for in
multivariate analyses.
Quality score 4/5
The focus of this study was on the impact of early discharge on perceived competence in
In crude, unadjusted comparisons, those randomised to early discharge mothering, infant weight gain, identification of hyperbilirubinaemia, maternal satisfaction with health
were 1.41 (1.02 to 1.94) times more likely to be predominantly care, predominant breastfeeding and infant health service utilisation.
breastfeeding at 1 month, but this effect was reduced to 1.25 (0.88 to
1.75) after adjustment for a 15.9% difference in those planning to It is unclear from the report how much support for breastfeeding was included in the contacts with
breastfeed. intervention group mothers and although breastfeeding is quoted as an outcome measure, the
study does not appear to assess the impact of breastfeeding support. It is therefore questionable
whether it should be included in a meta-analysis of studies of breastfeeding support.
Appendices 359
Appendices 360
Appendices 361
Appendices 362
Appendices 363
Appendices 364
Appendices 365
APPENDIX J
Publications and Presentations from this work
PUBLICATIONS AND PRESENTATIONS OF THIS WORK
Publications:
(Copies of both publications are available – This list updated after submission of thesis)
Graffy, J & Taylor J. What Information, Advice, and Support Do Women Want With
Breastfeeding? BIRTH, Issues in Perinatal Care 2005: 32, 179 –186.
Graffy J, Taylor J, Williams A, Eldridge S. Randomised controlled trial of support from
volunteer counsellors for mothers considering breast feeding. British Medical Journal
2004;328:26-31.
Graffy J. Breastfeeding: the GP's role. Practitioner, 1992; 236: 322-324
Graffy J. Mother's attitudes to and experience of breastfeeding: a primary care study.
BJGP 1992; 42: 61-64
Presentations:
Infant feeding in Nottingham; Intentions and Outcome.

Graffy J. WONCA conference, London. 1986 (Poster presentation)
Who needs most help with breastfeeding? Graffy J. Three presentations to postgraduate
clinical meetings in Haringey, Hackney and Basingstoke during 1992.
What information and support do women want with breastfeeding?

Graffy J and Taylor J. Annual London meeting of the Association of University Departments
of General Practice, Cambridge. February 1999
Practical aspects of recruiting to trials in primary care.

Graffy J. NHSE London Regional Office Residential Research Training Meeting,
Sunningdale. March 1999
Interventions to support breastfeeding, a review of the evidence.

Graffy J and Taylor J. East London and the City Health Authority conference: Breastfeeding
to the Year 2000. May 1999.
What information and support do women want with breastfeeding?

Graffy J and Taylor J. Barriers to Breastfeeding. A national conference organised by the
Royal College of Midwives, National Childbirth Trust, Community Practitioners and Health
Visitors Association and Royal College of Nursing. (Plenary presentation) May 2000.
What information, advice and support do women want with breastfeeding?

Graffy J and Taylor J. North American Primary Care Research Group Conference. Halifax,
Nova Scotia. October 2001. (Awarded Pfizer Practitioner Research Scholarship to present
this paper.)
Appendices 366
APPENDIX K
Examples of women’s comments
EXAMPLES OF WOMEN’S COMMENTS ON SIX-WEEK QUESTIONNAIRES
These examples of women’s comments are provided to give an example of the type of data
which was available for qualitative analysis in the study of the information, advice and
support women want with breastfeeding. The comments selected have been chosen to
illustrate the data available, rather than because of their content. The reference numbers given
refer to the mother’s study number and data is provided for each question from three mothers.
Of all the advice you have received about breastfeeding, which was most helpful?
a. Being told that supplement was not helpful. The sucking action required was different and
the baby may not feed as well from a bottle. [Counsellor] (1589)
b. Mum told me to try and make baby last two hours between feed and then feed ten to
fifteen minutes each side. This meant nipples weren’t constantly under pressure. [Mother]
(2509)
c. Found the booklets helpful. [From counsellor] (3350)
Of all the advice you have received, which was least helpful?
a. Give water if baby is dehydrated, thirsty or has hiccups. [General Practitioner] (1589)
b. Breastfed babies don’t get wind. Let the baby feed as long as he wants. Don’t need to
swap breasts halfway through – just start with that one next time. [Hospital midwife]
(2509)
c. None (3350)
Appendices 367
APPENDIX K
Examples of women’s comments
Please add anything else that you think is important:-
a. The more I learned about how breastfeeding worked, the more determined and confident I
felt. I think people should be warned that the first few days are the hardest where some
babies may appear to be unsatisfied by the breast but you stick at it. Also information on
how supplementing can hinder progress such as reducing supply, different sucking action.
I read a book “Breast is Best” which gave me information on the nutritional benefits, also
the breastfeeding counsellor gave me a number of leaflets that helped. (1589)
b. As a first time mum I feel the hospital should have advised me better on breastfeeding. I
believe to keep new mums sane they should be made aware babies don’t just cry for food
and can use the nipple as a comforter and that they do get wind.
After being born my baby was put to my breast for as long as he wanted which was 40
minutes one side and 30 mins the other. I believe this was too long too soon and probably
lead to cracked nipples.
If it hadn’t hurt and if I had received assistance regarding expressing some milk I may
have breastfed for longer. I gave up because it really hurt and I was dreading every feed.
(2529)
c. I wish I’d had a little bit more help - and really in the hospital. If the midwife was a bit
more helpful with breastfeeding. They were rushed and they just pulled me around a bit
and just stuck the nipple in and that. (3350)
Appendices 368
APPENDIX L
Literature Search Strategy
LITERATURE SEARCH STRATEGY
In reviewing the literature, I have sought to identify relevant work by conducting electronic
searches at intervals and drawing on reference lists included in other reviews. I hold over a
thousand references on a Reference Manager database, but whenever possible, I have retained
paper copies of original papers referred to. When my approach has been based on reviews, I
have identified these in the text.
Electronic searches:
I searched the following electronic databases on a number of occasions between 1994 and
February 2001.
• The Cochrane Library
• MEDLINE,
• Cumulative Index of Nursing Research and the Allied Literature (CINAHL)
Terms used for electronic searches:

The terms used for electronic searches varied for the different topics being investigated. For
example, I combined the terms “breast feeding”, “milk, human” and “lactation” to identify
literature likely to be relevant to breastfeeding. I then combined this selection with a number
of other topic areas including “attitude to health”, “social support”, “counselling”, “health
promotion” and terms such as “infection” for the chapter on the benefits of breastfeeding.
Additionally, I browsed recent references selected using the term “breast feeding”, to identify
references of interest which I might otherwise have missed. I normally confined my searches
to references in English, relating to human health.
Reference lists were obtained from:

• Lawrence, RA. Breastfeeding, a guide for the medical profession (Fourth edition). St
Louis, Missouri: Mosby-Year Book Inc. 1994.
• Heinig MJ,.Dewey KG. Health advantages of breast feeding for infants: A critical review.
Nutrition Research Reviews 1996;9:89-110.
• Dykes F,.Griffiths H. Societal influences upon initiation and continuation of breastfeeding. British
Journal of Midwifery 1998;6:76-80.
• Tedstone A, Dunce N, Aviles M, Shetty P , Daniels L. Effectiveness of interventions to promote
healthy feeding in infants under one year of age: a review. London: Health Education Authority,
1998.
• Sikorski, J. and Renfrew, M. J. Support for breastfeeding mothers. (Cochrane Review). Issue 1.
1999. Oxford, Update Software.
• Vallenas C, Savage F. Evidence for the ten steps to successful breastfeeding. Geneva: World
Health Organization, 1999
• Renfrew MJ, Woolridge MW, Ross McGill H. Enabling women to breastfeed. A review of
practices which promote or inhibit breastfeeding - with evidence-based guidance for practice.
London: The Stationery Office, 2000.
• Hoddinott, P. Why don't some women want to breastfeed and how might we change their
attitudes? A qualitative study. 1998.University of Wales College of Medicine. MPhil Thesis
• Smale, O. M. Women's Breastfeeding. 1996. University of Bradford. D Phil thesis
Appendices 369
APPENDIX L
Literature Search Strategy
Other references were identified from a range of sources, which were not formally searched:
• Midwives Information and Resource Service database (MIDIRS)

• Maternity Alliance Newsletter
• National Childbirth Trust Current Awareness Bulletin
• Breastfeeding Network Newsletter
• National Childbirth Trust Library
• British Medical Journal
Appendices 370

Thesis On Randomized Controlled Trial

Uploaded by

Copyright:

Available Formats

Thesis On Randomized Controlled Trial

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesis On Randomized Controlled Trial

Uploaded by

Copyright:

Available Formats

EVALUATING BREASTFEEDING SUPPORT:

A RANDOMISED CONTROLLED TRIAL OF SUPPORT FROM

Jonathan Peter Graffy

A thesis submitted to the Faculty of Medicine and Dentistry

Department of Primary Care and General Practice,

This unpublished thesis/dissertation is copyright of the author and/or third

Any use made of information contained in this thesis/dissertation must be in

In agreeing to participate in the randomised controlled trial, counsellors trusted me with

Part One: Literature Review

Chapter 1.1 How great are the benefits of breastfeeding?

1.1.1 Introduction: Women's perspectives on the benefits of breastfeeding..............................2

Chapter 1.2 Epidemiology of breastfeeding:

1.2.1 Introduction: The importance of epidemiological data...................................................26

1.3.1 Introduction ...................................................................................................................46

Chapter 1.4 Theories of behaviour change

1.4.1 Introduction ...................................................................................................................53

Chapter 1.5 Interventions to promote breastfeeding

1.5.1 Introduction ...................................................................................................................73

1.6,1 Introduction ...................................................................................................................88

Chapter 1.7 Counselling and breastfeeding counsellors

1.7.1 Introduction ...................................................................................................................97

Chapter 2.1 Rationale and objectives of study one

2.1.1 Introduction .................................................................................................................110

Chapter 2.2 Study one: method

2.2.1 Recruitment .................................................................................................................112

Chapter 2.3 Study one: results

2.3.1 Study population...........................................................................................................114

Chapter 2.4 Study one: discussion

2.4.1 Strengths and limitations of the study...........................................................................124

Chapter 3.1 Rationale and objectives of study two

3.1.1 Rationale and selection of intervention for study two ..................................................133

Chapter 3.2 Study two: Methods

3.2.1 Outline of study two .....................................................................................................137

Chapter 3.3 Results of study two: recruitment and follow up

3.3.1 Introduction and overview ............................................................................................156

3.4.1 Introduction .................................................................................................................177

Chapter 3.5 Results of study two: effect of the intervention

3.5.1 Introduction .................................................................................................................194

Chapter 3.6 Results of study two: counsellors' perspectives on the study

3.6.1 Introduction .................................................................................................................211

3.7.1 Evaluating a complex intervention ...............................................................................216

4.1.1 Introduction .................................................................................................................232

Chapter 4.2 Qualitative analysis of women’s comments: participants and methods

4.2.1 Recruitment of participants...........................................................................................234

Chapter 4.3 Results of qualitative analysis

4.3.1 Women's experiences of advice and support with breastfeeding..................................241

Chapter 4.4 Discussion of qualitative findings

4.4.1 Introduction .................................................................................................................251

Part five: Conclusions and recommendations

Chapter 5 Conclusions and recommendations

Reference list .........................................................................................................................273

Appendix A Study one questionnaires

Table 1.1.1 Mother's reasons for choice of infant feeding method.........................................3

Table 3.3.2 Reasons women were no longer receiving maternity care...............................157

Table 4.2.1 Characteristics of participants .........................................................................235

Frontispiece Breastfeeding counsellor and client....................................................................vi

Fig 1.1.3a Promotion of Breastfeeding Intervention Trial - Main results.............................7

Fig 2.3.3 Duration of breastfeeding (Study one).............................................................117

Fig 5.2 Ways to improve breastfeeding support ..........................................................263