Age Final PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Diabetologia (2021) 64:521–529

https://doi.org/10.1007/s00125-020-05330-1

ARTICLE

Lifetime risk of diabetes in metropolitan cities in India


Shammi Luhar 1,2 & Dimple Kondal 3 & Rebecca Jones 4 & Ranjit M. Anjana 5 & Shivani A. Patel 6 & Sanjay Kinra 7 &
Lynda Clarke 2 & Mohammed K. Ali 6 & Dorairaj Prabhakaran 3,8 & M. Masood Kadir 9 & Nikhil Tandon 10 &
Viswanathan Mohan 5 & K. M. Venkat Narayan 6

Received: 29 July 2020 / Accepted: 5 October 2020 / Published online: 23 November 2020
# The Author(s) 2020

Abstract
Aims/hypothesis We aimed to estimate the lifetime risk of diabetes and diabetes-free life expectancy in metropolitan cities in
India among the population aged 20 years or more, and their variation by sex, age and BMI.
Methods A Markov simulation model was adopted to estimate age-, sex- and BMI-specific lifetime risk of developing diabetes
and diabetes-free life expectancy. The main data inputs used were as follows: age-, sex- and BMI-specific incidence rates of
diabetes in urban India taken from the Centre for Cardiometabolic Risk Reduction in South Asia (2010–2018); age-, sex- and
urban-specific rates of mortality from period lifetables reported by the Government of India (2014); and prevalence of diabetes
from the Indian Council for Medical Research INdia DIABetes study (2008–2015).
Results Lifetime risk (95% CI) of diabetes in 20-year-old men and women was 55.5 (51.6, 59.7)% and 64.6 (60.0, 69.5)%,
respectively. Women generally had a higher lifetime risk across the lifespan. Remaining lifetime risk (95% CI) declined with age
to 37.7 (30.1, 46.7)% at age 60 years among women and 27.5 (23.1, 32.4)% in men. Lifetime risk (95% CI) was highest among
obese Indians: 86.0 (76.6, 91.5)% among 20-year-old women and 86.9 (75.4, 93.8)% among men. We identified considerably
higher diabetes-free life expectancy at lower levels of BMI.
Conclusions/interpretation Lifetime risk of diabetes in metropolitan cities in India is alarming across the spectrum of weight and
rises dramatically with higher BMI. Prevention of diabetes among metropolitan Indians of all ages is an urgent national priority,
particularly given the rapid increase in urban obesogenic environments across the country.

Keywords Body mass index . Diabetes . Diabetes-free life expectancy . India . Lifetime risk . Metropolitan cities . Urban

Abbreviations ICMR-INDIAB Indian Council of Medical


CARRS cArdiometabolic Risk Reduction Research–INdia DIABetes
in South Asia study LMIC Low- to middle-income country
CBG Capillary blood glucose SAGE Study on global AGEing and adult health
FPG Fasting venous plasma glucose SRS Sample Registration System
HIC High-income country

* Shammi Luhar 6
Hubert Department of Global Health, Emory University,
sl989@medschl.cam.ac.uk Atlanta, GA, USA

7
1 Department of Non-Communicable Disease Epidemiology, London
Department of Public Health and Primary Care, University of School of Hygiene and Tropical Medicine, London, UK
Cambridge, Cambridge, UK
2 8
Department of Population Health, London School of Hygiene and Public Health Foundation of India, Gurgaon, India
Tropical Medicine, London, UK
9
3
Centre for Chronic Disease Control (CCDC), New Delhi, India Department of Community Health Sciences, Aga Khan University,
4
Karachi, Pakistan
Nutrition and Health Sciences, Laney Graduate School, Emory
University, Atlanta, GA, USA 10
Department of Endocrinology and Metabolism, All India Institute of
5
Madras Diabetes Research Foundation, Chennai, India Medical Sciences, New Delhi, India
522 Diabetologia (2021) 64:521–529

Introduction among 45-year-old individuals in the USA and 38.0% among


those aged 25 years in the Netherlands. The lifetime risk
Diabetes is a major global public health problem currently among severely obese people in the USA was reported to be
affecting 463 million individuals and projected to affect 700 as high as ten times that of underweight individuals [2]. The
million by 2045 [1]. Estimates of prevalence suggest that the combination of a high BMI distribution, high incidence of
diabetes burden is increasing at a faster pace in low- to middle- diabetes and high life expectancy drive a high overall lifetime
income countries (LMICs) than in high-income countries risk in HICs. It is not clear, however, how the lifetime risk of
(HICs). In India, 77 million adults currently have diabetes developing diabetes may differ in LMICs wherein individuals
and this number is expected to almost double to 134 million may have a comparatively lower BMI and life expectancy.
by 2045 [1]. Whereas prevalence estimates are useful for In this study, we estimated the lifetime risk of developing
expressing the overall burden of diabetes at a particular point diabetes in India’s metropolitan cities where an already large
in time, they do not inform future risk of developing this number of adults have diabetes and where there are rapid
disease at the individual level. Lifetime risk provides an esti- increases in population-level weight status and life expectancy
mate of the cumulative probability of eventually developing [7]. Drawing on recent diabetes incidence data and on multi-
diabetes over the course of life. It is a valuable estimate for ple Indian databases, we constructed a Markov model to esti-
effectively communicating diabetes risk to individuals, even mate lifetime risk of diabetes in Indian metropolitan cities and
at young ages, and is also a powerful tool with which to galva- also stratified by age, sex and BMI. We define metropolitan
nise public health and policy responses. The few studies esti- cities as per the Indian Census Commission’s definition: cities
mating the lifetime risk of developing diabetes are focused on with over four million inhabitants (including Delhi, Mumbai,
HICs [2–4]. The population of India, while at the epicentre of Kolkata, Chennai, Hyderabad, Bangalore, Ahmedabad, Pune,
the global diabetes epidemic, has a lower BMI distribution Surat and Nashik [8]; making up just over 5% of the popula-
and lower overall life expectancy; Indian people also show a tion around the time of the last census [9]).
comparatively higher propensity to develop diabetes, both at
younger ages and lower BMI levels [5], suggesting a substan-
tially different epidemiology. Methods
Studies from HICs such as the USA [2, 6], Australia [3]
and the Netherlands [4] have found high lifetime risks of Lifetime risk estimation required, as data inputs, age-, sex-
developing diabetes. For example, lifetime risk was 31.3% and BMI-specific estimates of diabetes incidence and age-,
Diabetologia (2021) 64:521–529 523

sex- and BMI-specific rates of mortality by diabetes status. independent of age used to account for external causes of
Differential risks of dying depending on diabetes and weight death, which may be especially high in early adulthood [14].
status, prevalence of diabetes and the prevalence of different The mortality rates derived from the SRS lifetables are includ-
BMI classes were used to modify population-level rates, ed in ESM Table 6 and the fit of the population mortality rates
making them specific to particular subgroups (i.e. BMI groups is shown in the ESM Figs 1–2. From this, we extracted
and individuals with and without diabetes). predicted mortality rates by single years [15].
This study was approved by the LSHTM Observational/
Interventions Research Ethics Committee (ref. 17568). Differential mortality rates by diabetes and BMI status A
meta-analysis of 35 published articles between 1990 and
Incidence of diabetes Incidence of diabetes was calculated 2010 (220,689 individuals in HICs, mean follow-up
using data from the India cArdiometabolic Risk Reduction 10.7 years) informed differential mortality rates in individuals
in South Asia study (CARRS) cohort [10, 11], a metropolitan with vs without diabetes. The study found 85% and 100%
based cohort of 9812 participants with high retention and higher risk of all-cause mortality among men and women,
multiple points of follow-up, representative of Delhi and respectively, for those with vs without type 2 diabetes [16].
Chennai. CARRS was established in 2010–2011 and is under We accounted for differential mortality among populations
ongoing follow-up. Fasting venous plasma glucose (FPG) and with different BMI status by applying RRs of dying by BMI
HbA1c data were collected at baseline in 2010–2011 as well as group from a study that examined the association of BMI with
at the second and fourth follow-up assessments in 2013 and mortality risk in urban Mumbai between 1991 and 2003 [17].
2016, respectively. Age-specific population estimates of inci- The study, comprising 148,173 individuals, reported 11%
dence were restricted to participants without diabetes at base- lower risk and 22% higher risk, of dying among overweight
line and the annualised incidence rate was calculated at the and obese men aged 35–59 years, respectively, compared with
second (2013) and fourth (2016) follow-up assessments. normal-weight men. Furthermore, they identified a 67% and
Incident diabetes was defined as occurring in an individual 102% higher risk of death in underweight women and men,
reporting no diabetes at baseline and baseline biochemical respectively, relative to normal-weight counterparts. Slightly
variables in the non-diabetes range (FPG <7 mmol/l and attenuated RRs were found in individuals aged 60 years or
HbA1c< 48 mmol/mol [6.5%]) who subsequently recorded more. For details of how differential mortality rates were
an FPG ≥7 mmol/l or HbA1c ≥48 mmol/mol (6.5%) or report- calculated for the analysis, see ESM Methods.
ed diabetes treatment during follow-up. Incidence for the
following age groups was available: 20–24 years; 25–34 years; Age-specific prevalence of diabetes Diabetes prevalence data
35–44 years; 45–54 years; 55–64 years; and ≥65 years. Using were used to compartmentalise population rates of mortality
the RR of diabetes among different BMI groups, we modified into a set of rates specific to the population with and without
the population rates to estimate incidence separately for the diabetes. Age-specific diabetes prevalence was obtained from
underweight/normal weight (BMI <25.0 kg/m2), overweight the Indian Council of Medical Research–INdia DIABetes
(BMI 25.0–29.9 kg/m2) and obese population (BMI ≥30.0 kg/ (ICMR-INDIAB) study [18]. The ICMR-INDIAB study is
m2) [12]. We report the incidence rates, in addition to the the largest representative study with fasting glucose measures
number of incident cases, from the CARRS study in the elec- in India and includes data collected between 2008 and 2015
tronic supplementary material (ESM) Tables 1–5. Incidence from 57,117 Indians from 14 states and one Union Territory.
rates by single years were subsequently extracted after fitting a The ICMR-INDIAB study used 8 h fasting capillary blood
cubic smoothing spline. glucose (CBG) measured by trained field workers to identify
adults with previously undiagnosed diabetes. CBG is seen as a
Population mortality rates Mortality rates by 5 year intervals reasonable alternative to venous blood glucose in developing
were extracted from the latest Sample Registration System countries and has reasonable sensitivity (83.3–90.9%,
(SRS), which contains abridged lifetables for 2013–2017 by depending on the diagnostic criteria) in India [19]. Using an
sex and urban residence [13]. The SRS provides conditional 82.5 g oral glucose load (equivalent to 75 g of anhydrous
probabilities of death (lifetable notation = nqx), which we used glucose), 2 h post-load CBG was measured in individuals
to obtain age-specific mortality rates using an inversion of who did not self-report diabetes. Among those with self-
Chiang’s formula [14]. We obtained mortality rates by single reported diabetes, only fasting glucose was measured.
years by fitting an age at death distribution to the 5 year Individuals with an 8 h CBG ≥7 mmol/l, 2 h CBG
mortality rates. Specifically, we used the Gompertz– ≥12.2 mmol/l, or both, were categorised as having diabetes.
Makeham law of mortality, which is made up of two separate Age was categorised into groups of 5 years.
components: an age-dependent part that states that the rate of
mortality increases exponentially with age, representing the Age-specific prevalence of BMI categories We used the
underlying mortality age pattern; and a component proportion of the urban population in broad BMI categories
524 Diabetologia (2021) 64:521–529

to estimate weighted averages of lifetime risk for the whole of our model to the initial prevalence of diabetes (to modify
population. The age- and sex-specific prevalences of ‘under- diabetes and non-diabetes-specific mortality) using age-specific
weight/normal weight’ (BMI <25.0 kg/m2), ‘overweight’ (BMI diabetes prevalence reported for the two Indian cities in the
25.0–29.9 kg/m2) and ‘obesity’ (BMI ≥30.0 kg/m2) groups CARRS study, which used FPG rather than CBG used in the
across the whole age range in urban Indians were obtained from ICMR-INDIAB study, to determine diabetes status [25]. We
a study that forecasted weight category prevalence in India using also estimated the lifetime risk of developing diabetes using
survey data [20]. The study obtained BMI group prevalence incidence rates that omitted HbA1c in diagnosing diabetes. As
from the 2015–2016 National Family and Health Survey, which individuals with incident diabetes can have normal FPG while
collected nationally representative health and demographic data having HbA1c ≥48 mmol/mol (6.5%), this may have led us to
on approximately 625,000 adult women aged 15–49 years and overestimate the main lifetime risk estimates as they may repre-
93,000 adult men aged 15–54 years [21]. The Study on global sent a group with higher glucose tolerance than those satisfying
AGEing and adult health (SAGE) wave 0 (2002–2004) and both FPG and HbA1c-related diagnostic criteria [26]. In addi-
wave 1 (2007–2010) were used to obtain estimates of age- tion to BMI, we also examined the lifetime risk of diabetes by
specific proportions of the urban population aged 50 years or categories of waist circumference. Finally, we examined the
more in broad BMI categories. SAGE is a nationally represen- effect on the total population lifetime risk of a change in the
tative health and demographic study containing information on proportion of urban Indians classified as overweight or obese.
2559 adults in wave 0 and approximately 3000 men and 3000 Age-specific forecasts of the prevalence of excessive weight
women in wave 1 [22]. and obesity in urban India were extracted from a recent study
[20]. All analyses were performed using the R (version 3.6.2)
The model We adopted a multistate Markov model to estimate statistical software package (available for download at https://
the lifetime risk of diabetes [23]. A detailed and formal cran.r-project.org/bin/windows/base/old/3.6.2/).
description of the methodology we adopted is described in a
similar study [2]. The multistate model compartmentalises a
population into three separate age-, sex- and BMI-specific Results
groups: no diabetes; diabetes; and dead. The single-year inci-
dence and mortality rates were inputted into a matrix of rates Main findings Overall, the lifetime risk (95% CI) of develop-
of which the exponential was subsequently taken to obtain a ing diabetes among Indian metropolitans at age 20 years was
matrix of transition probabilities between the states based on 64.6 (95% CI 60, 69.5)% among women and 55.5 (51.6,
starting age and state. This was performed separately by BMI 59.7)% among men. At ages 40 and 60 years, the remaining
category. These single-year transition probabilities, derived lifetime risk (95% CI) among women was 59.2 (52.4, 64.9)%
from the variables, simulated movements between the three and 37.7 (30.1, 46.7)%, respectively, and among men aged 40
mutually exclusive states (ESM Fig. 3) and the resultant tran- and 60 years was 47.3 (42.4, 52.3)% and 27.5 (23.1, 32.4)%,
sition matrix was used to generate measures of both lifetime respectively (Table 1).
risk of developing diabetes and diabetes-free life expectancy As expected, we found a higher age-specific lifetime risk of
by single years. We were able to calculate total lifetime risk of developing diabetes in higher BMI groups compared with
developing diabetes as a weighted average of separate lifetime lower ones. At age 20 years, the lifetime risk (95% CI) of
risk estimates for different BMI groups. developing diabetes among underweight/normal-weight men
We performed multiple sensitivity analyses. First, we was 41.2 (36.7, 45.7)% compared with 71.3 (64.6, 77.6)%
performed multiple simulations whereby we simultaneously among overweight men and 86.9 (75.4, 93.8)% among obese
selected random variables estimates from the distributions that men. Similarly, at age 20 years, the lifetime risk (95% CI) of
informed their uncertainty. In total, 5000 simulations were developing diabetes among underweight/normal-weight
carried out and the median lifetime risk estimate was reported women was 51.6 (46.0, 58.4)% compared with 71.0 (61.0,
as the final point estimate; the range of values at any age 80.6)% and 86.0 (76.6, 91.5)% among overweight and obese
informed the uncertainty interval. Second, we supplemented women, respectively.
our main lifetime risk estimates by BMI using South Asian- Across BMI groups, the largest difference in lifetime risk
specific BMI cut-offs (BMI <23.0 kg/m2 for the underweight/ of developing diabetes was between underweight/normal
normal-weight group, BMI ≥23.0 kg/m2 to <27.5 kg/m2 for the weight and overweight individuals. To illustrate, among urban
overweight group and BMI ≥27.5 kg/m2 for the obese group men aged 20 years, the lifetime risk was higher by approxi-
[24]). The higher positive relationship between body fat and mately 30 percentage points in the overweight group (71.3%)
BMI observed in South Asians when compared with white compared with the underweight/normal weight group
Europeans may make use of conventional BMI cut-offs inap- (41.2%), whereas the lifetime risk was around 15 percentage
propriate and lead to a relatively high disease risk at compara- points higher in obese men aged 20 years (86.9%) when
tively lower body weight [24]. Third, we tested the sensitivity compared with overweight counterparts.
Diabetologia (2021) 64:521–529 525

Table 1 Lifetime risk of developing diabetes by global BMI cut points expect to live 39.9 years of their remaining 52.6 years free of
in Indian metropolitan cities
diabetes (75.9%). The remainder of life spent with diabetes
BMI group Lifetime risk varied considerably between BMI groups. Among
underweight/normal weight women aged 20 years, on aver-
Men Women age, 43.4 years of an expected 54.5 remaining years (79.6%)
Underweight/normal weight (BMI <25 kg/m2)
were estimated to be diabetes-free, compared with 27.1 years
of the remaining 52.4 years of life of obese women aged
Age 20 years 41.2 (36.7, 45.7) 51.6 (46, 58.4)
20 years (51.8%).
Age 40 years 34.5 (30.7, 38.8) 47.1 (40.5, 54.6)
Age 60 years 21.5 (16.7, 27.5) 30.7 (23.2, 40.1)
Sensitivity analysis findings As expected, the overall life-
Overweight (BMI ≥25 kg/m2 to <30 kg/m2)
time risk of developing diabetes was attenuated when
Age 20 years 71.3 (64.6, 77.6) 71.0 (61.0, 80.6)
using South Asian-specific BMI cut-offs (Table 3). To
Age 40 years 63.8 (54.0, 72.1) 64.9 (54.1, 77.8)
illustrate, at age 20 years, the lifetime risk (95% CI) of
Age 60 years 44.9 (32.2, 56.1) 45.9 (30.8, 65.6)
developing diabetes was 37.4 (33.1, 44.3)% among
Obese (BMI ≥30 kg/m2)
underweight/normal-weight men, 63.6 (55.6, 71.9)%
Age 20 years 86.9 (75.4, 93.8) 86.0 (76.6, 91.5)
among overweight men and 83.3 (73, 91.1)% among
Age 40 years 77.2 (56.7, 87.8) 79.1 (65.6, 88.8)
obese men. For women aged 20 years, the lifetime risk
Age 60 years 55.9 (34.1, 74.6) 58.5 (32.9, 78.6)
(95% CI) was 46.8 (41.4, 53.0)% among underweight/
Total population
normal-weight women, 67.6 (56.9, 75.2)% among over-
Age 20 years 55.5 (51.6, 59.7) 64.6 (60.0, 69.5)
weight women and 80.3 (69.5, 88.3)% among obese
Age 40 years 47.3 (42.4, 52.3) 59.2 (52.4, 64.9) women. We also assessed the effect of using age-specific
Age 60 years 27.5 (23.1, 32.4) 37.7 (30.1, 46.7) prevalence of diabetes from CARRS, rather than the
Lifetime risk is presented as % (95% CI) ICMR-INDIAB prevalence. The effect of this on the life-
time risk estimates was negligible (a change of 0–1
percentage points compared with initial estimates).
Women at age 20 years can expect to live 38.9 of their When using incidence rates that excluded HbA1c in the
remaining 54.4 years spent free of diabetes (71.6% of their determination of incident diabetes, we identified slightly
remaining life) (Table 2). Similarly, men aged 20 years can different, although not notable, lifetime risk estimates (ESM

Table 2 Percentage of remaining life spent with diabetes in Indian metropolitan cities by global BMI cut points at ages 20, 40 and 60

BMI group Men Women

Life expectancy Diabetes-free life % of remaining Life Diabetes-free % of remaining


(years) expectancy (years) life without diabetes expectancy (years) life expectancy life without
(years) diabetes

Underweight/normal weight (BMI <25 kg/m2)


Age 20 years 52.8 44.7 84.7 54.5 43.4 79.6
Age 40 years 34.4 28.8 83.5 35.7 27.6 77.2
Age 60 years 17.6 15.6 88.8 18.1 15.4 84.9
Overweight (BMI ≥25 kg/m2 to <30 kg/m2)
Age 20 years 51.5 34.8 67.4 53.9 36.2 67.1
Age 40 years 33.6 22.2 66.2 35.2 23.0 65.3
Age 60 years 17.3 13.1 76.0 17.9 13.6 76.1
Obese (BMI ≥30 kg/m2)
Age 20 years 49.3 22.6 45.7 52.4 27.1 51.8
Age 40 years 32.3 15.4 47.7 34.2 17.8 51.9
Age 60 years 17.3 11.1 64.1 17.6 11.6 65.9
Total population
Age 20 years 52.6 39.9 75.9 54.4 38.9 71.6
Age 40 years 34.3 25.5 74.2 35.6 24.4 68.4
Age 60 years 17.7 15.1 85.3 18.2 14.7 80.7
526 Diabetologia (2021) 64:521–529

Table 3 Lifetime risk of developing diabetes by South Asian BMI cut risk was consistently higher among women compared with
points in Indian metropolitan cities
men and declined with age. Two in five underweight/
BMI group Lifetime risk normal-weight 20-year-old men and over half of the
underweight/normal-weight 20-year-old women were
Men Women projected to develop diabetes, and the remaining risk
Underweight/normal weight (BMI <23 kg/m2)
increased to over eight in ten among obese individuals aged
20 years. We also found that obese metropolitans can expect
Age 20 years 37.4 (33.1, 44.3) 46.8 (41.4, 53.0)
to spend the greatest percentage of their remaining life with
Age 40 years 32.0 (27.5, 39.2) 42.2 (36.0, 48.8)
diabetes (around 50% from age 20 years).
Age 60 years 20.3 (15.2, 28.3) 27.0 (20.8, 34.6)
These estimations relied on a number of assumptions. First,
Overweight (BMI ≥23 kg/m2 to <27.5 kg/m2)
the model assumed that BMI-specific age-specific diabetes
Age 20 years 63.6 (55.6, 71.9) 67.6 (56.9, 75.2)
incidence will be constant throughout the lifetime of the
Age 40 years 55.7 (46.9, 64.8) 63.1 (51.3, 71.9)
cohort. This is a reasonable assumption given the lack of
Age 60 years 37.8 (25.5, 49.4) 44.2 (31.1, 56.8)
evidence as to how future BMI-specific diabetes incidence
Obese (BMI ≥27.5 kg/m2)
will develop. Second, we assumed that the transition to diabe-
Age 20 years 83.3 (73.0, 91.1) 80.3 (69.5, 88.3)
tes is permanent and that an individual cannot transition back
Age 40 years 73.8 (60.8, 83.4) 72.1 (59.6, 83.7)
to not having diabetes, an assumption we decided was reason-
Age 60 years 53.5 (31.5, 71.8) 49.5 (29.3, 72.6)
able given the relative rarity of spontaneous remission from
Lifetime risk is presented as % (95% CI) diabetes. Third, we assumed that the RR of dying among those
with diabetes relative to those without will remain constant
into the future. The extent to which results changed through
Table 7). Overall lifetime risk (95% CI) at age 20 years among relaxation of this assumptions was negligible.
men and women, respectively, was 55.0 (51.1, 58.7)% and Our study has a number of strengths. Blood tests were used
59.2 (54.9, 63.7)% and decreased to 29.0 (25.0, 33.5)% and to identify diabetes and BMI was also objectively measured
32.8 (27.5, 37.7)% at age 60 years. Again, this masked consid- rather than using self-reports. Studies assessing the validity of
erable variation by BMI, whereby the lifetime risk (95% CI) at self-reported diabetes have found considerable misclassifica-
age 20 years in men and women, respectively, was as follows: tion. For instance, one study found a sensitivity of self-
40.0 (36.4, 43.3)% and 40.6 (35.9, 45.2)% among reported diabetes in China was around 58.2% in urban areas
underweight/normal-weight individuals; 69.9 (63.7, 75.1)% and 35.0% in rural areas [27]. Low sensitivity of self-reported
and 69.3 (62.8, 75.2)% among overweight individuals; and diabetes has the potential to underestimate the lifetime risk
88.9 (82.2, 93.3)% and 85.3 (77.3, 92.4)% among obese estimates of previous studies [2], potentially explaining differ-
individuals. ences between ours and previous findings from different
We also estimated lifetime risks of developing diabetes settings. We also used data sources that are designed to be
according to waist circumference. At age 20 years, the lifetime geographically representative to measure mortality rates,
risk (95% CI) was 71.0 (64.4, 80.9)% for men with a waist BMI and diabetes. As data on national-level incidence of
circumference ≥90 cm and 45.2 (40.9, 48.8)% for men with a diabetes are not available, we obtained incidence rates from
waist circumference <90 cm (ESM Table 8). Among women a large well-characterised, well-retained and followed up,
aged 20 years, the lifetime risk (95% CI) was 75.5 (65.2, representative cohort in urban areas of North and South
84.2)% for those with waist circumference ≥80 cm and 44.5 India, making our results both geographically representative
(39.6, 49.4)% for those with waist circumference <80 cm. and generalisable to the wider metropolitan population.
Using the 2040 forecasted BMI group distribution in urban Finally, we performed extensive sensitivity analyses,
India, we estimated that the overall lifetime risk (95% CI) of reporting a range of uncertainty around our estimates, provid-
developing diabetes among individuals aged 20 years would ing supplementary lifetime risk estimates using South Asian-
increase to 64.0 (60.0, 66.8)% and 77.1 (71.0, 81.7)% for men specific BMI cut-offs, accommodating for potential overesti-
and women, respectively (ESM Table 9). mation of diabetes incidence when using HbA1c by using
solely FPG and diabetes treatment to classify diabetes cases,
and examining the sensitivity of the estimates to initial diabe-
tes prevalence by using diabetes prevalence estimated in the
Discussion CARRS cohort.
Our study also has a number of limitations. First, assuming
We found that in metropolitan India, the lifetime risk of devel- that age-specific rates will prevail over the projection period
oping diabetes among women and men without diabetes at may have led us to both underestimate lifetime risk of diabetes
age 20 years was 64.6% and 55.5%, respectively. The lifetime within a BMI group (as the individual BMI distribution within
Diabetologia (2021) 64:521–529 527

a broad BMI group shifts to the right) and underestimate the is driven by a relatively high predisposition to developing
overall lifetime risk for the total population (as the prevalence diabetes among Indians at both lower ages (up to a
of excessive weight and obesity has increased in recent decade earlier) and lower BMIs when compared with
decades [21], and will likely increase into the future). One white European populations [5].
study in the USA identified increases in overall population On average, 70.1% of the remaining lifespan of the average
lifetime risk of 20 and 13 percentage points in men and woman and 74.0% of the remaining life of the average man
women, respectively between 1985 and 2011 [6]. The latter without diabetes at the age of 20 years can be expected to be
concern was addressed in sensitivity analysis using forecasted spent diabetes free; this proportion decreases considerably
prevalence of excessive weight and obesity in urban India. with increasing BMI. Attenuated results have been found in
Second, life expectancy at all ages in India is expected to HICs; one study reported that Australians aged 25 years will
continue to increase over the coming years [28]. Longer life spend 86% of their remaining years diabetes free [3], whereas
expectancies may increase the lifetime risk of diabetes, a USA-based study found that women aged 18 years who
making our results possible underestimates. Third, the socio- were obese and severely obese could expect to spend around
economic distribution of excessive weight and obesity is like- 80% and 65%, respectively, of their remaining life diabetes
ly to continue to change. A recent study has documented an free [2].
increasing proportion of the overweight and obese population The remarkably high lifetime risk of developing diabetes
coming from lower socioeconomic backgrounds [29] among and the low diabetes-free life expectancy in urban India, espe-
whom the RR of dying compared with people from higher cially for individuals with high BMI, implies that interven-
socioeconomic backgrounds is likely to be higher. This may tions targeting the incidence of diabetes may be of paramount
have led us to potentially overestimate the lifetime risk for the importance moving forward. The importance is reinforced by
obese people. Our results should be interpreted with some the early onset of diabetes among South Asians, implying a
caution, especially given India’s considerable heterogeneity, greater proportion of life spent with diabetes and greater expo-
somewhat limiting the generalisability of our findings to all of sure to diabetes-related complications. Potentially effective
urban India. The results from this study relate to an average interventions include high and sustained sugar-sweetened
urban-dwelling individual in two highly urbanised areas; beverage taxation [31] or diabetes prevention programmes
however, the extent of variation in lifetime risk and years of involving a combination of culturally tailored lifestyle inter-
life lost to diabetes will depend on how much a particular ventions with a course of medication [32]. The success of any
individual differs from this average in regards their risk of intervention will also require improved detection of individ-
diabetes [2, 30]. uals with diabetes and either impaired glucose tolerance or
To our knowledge, this is the only study estimating the impaired fasting glucose, given that an estimated 57% of
lifetime risk of diabetes in urban India. A comparison of diabetes cases are undiagnosed in India [1]. Although in
our findings from those reported in HICs indicates an India the advancement from either impaired glucose tolerance
alarmingly high lifetime risk at every age in urban India. or impaired fasting glucose to diabetes has been found to be
Recent data from the Netherlands indicates a 31.3% life- comparatively faster than in other populations [7, 33], any
time risk of diabetes among 45-year-olds; this contrasts level of insulin resistance adds to a very sobering picture of
with the reported 53.8% in urban Indian men and 55.4% dysglycaemia and associated complications in a country
in urban Indian women of the same age [4]. A study from where urban obesogenic environments are on the increase
the USA, using data from 2000–2011, reported a lifetime and public health infrastructure is strained. Additional strain
risk of diabetes of 40.2% among men and 39.6% among is also likely to be exerted on individual families as personal
women aged 20 years. Our results are much closer to expenditure on diabetes care can be as high as 34% of house-
estimates of lifetime risk of diabetes among the black hold income for some low-income urban families [34].
and Hispanic populations in the USA, groups considered Metropolitan Indians at every age and BMI have an
at a higher risk of developing diabetes than the general alarmingly high probability of developing diabetes
population. A recent study found lifetime risks of devel- compared with results from HICs, necessitating a strong
oping diabetes in excess of 55.2% among black women demand for proactive efforts to prevent diabetes in
and 51.5% in Hispanic women aged 20 years in the USA metropolitan cities, given the rapid increase in urban
[6]. The higher lifetime risk among black and Hispanic obesogenic environments across the country [35].
women in the USA is driven by higher incidence when Prevention of diabetes, especially in younger metropoli-
compared with white women. A study of diabetes inci- tan Indians, should be a high priority for India. Future
dence comparing urban Indian adults with black and white research into rural lifetime risks of diabetes, using
US residents reported consistently higher incidence in diabetes incidence rates from such settings, is encour-
urban South Asians at all ages and BMI levels [11]. aged to provide a more complete snapshot of the overall
Overall, we posit that the considerably higher lifetime risk lifetime risk of diabetes at the national level.
528 Diabetologia (2021) 64:521–529

Supplementary Information The online version contains peer-reviewed 2. Narayan KMV, Boyle JP, Thompson TJ, Gregg EW, Williamson
but unedited supplementary material available at https://doi.org/10.1007/ DFW (2007) Effect of BMI on lifetime risk for diabetes. Diabetes
s00125-020-05330-1. Care 30:1562–1566. https://doi.org/10.2337/dc06-2544
3. Magliano DJ, Shaw JE, Shortreed SM, Nusselder WJ, Liew D, Barr
Data availability The publicly available SRS data have been deposited in ELM et al (2008) Lifetime risk and projected population prevalence
the Office of the Registrar General & Census Commissioner repository of diabetes. Diabetologia 51:2179–2186. https://doi.org/10.1007/
(available at: http://censusindia.gov.in/vital_statistics/SRS_Based/SRS_ s00125-008-1150-5
Based.html). Data from the CARRS and the ICMR-INDIAB study can 4. Ligthart S, van Herpt TTW, Leening MJG, Kavousi M, Hofman A,
be accessed through liaison with the study’s principal investigators. Stricker BHC et al (2016) Lifetime risk of developing impaired
glucose metabolism and eventual progression from prediabetes to
Funding This specific research was funded by the Economic and Social type 2 diabetes: a prospective cohort study. Lancet Diabetes
Research Council (https://esrc.ukri.org; grant no. ES/J500021/1 to SL). Endocrinol 4:44–51. https://doi.org/10.1016/S2213-8587(15)
This funder had no role in study design, data collection and analysis, 00362-9
decision to publish or preparation of the manuscript. CARRS was 5. Narayan KMV, Kanaya AM (2020) Why are South Asians prone to
funded in part by the National Heart, Lung, and Blood Institute type 2 diabetes? A hypothesis based on underexplored pathways.
(NHLBI), National Institutes of Health (NIH), Department of Health Diabetologia 63(6):1103–1109. https://doi.org/10.1007/s00125-
and Human Services, under Contract No. HHSN268200900026C, and 020-05132-5
the United Health Group, Minneapolis, MN, USA. KMVN, MKA and 6. Gregg EW, Zhuo X, Cheng YJ, Albright AL, Narayan KMV,
SAP were funded in part by the National Institute of Diabetes and Thompson TJ (2014) Trends in lifetime risk and years of life lost
Digestive and Kidney Diseases of the NIH under award no. due to diabetes in the USA, 1985-2011: a modelling study. Lancet
P30DK111024. KMVN was funded in part for ‘Worksite Lifestyle Diabetes Endocrinol 2:867–874. https://doi.org/10.1016/S2213-
Program for Reducing Diabetes and Cardiovascular Risk in India’ 8587(14)70161-5
project funded by NHLBI, NIH, Department of Health and Human 7. Tandon N, Anjana RM, Mohan V, Kaur T, Afshin A, Ong K et al
Services under award no. R01HL125442. SAP, KMVN, MKA, NT and (2018) The increasing burden of diabetes and variations among the
DP were supported in part by the NHLBI of the NIH, award no. states of India: the Global Burden of Disease Study 1990–2016.
5U01HL138635 under the Hypertension Outcomes for T4 Research Lancet Glob Health 6:e1352–e1362
within Lower Middle-Income Countries (Hy-TREC) program. DK has 8. Sikarwar A, Chattopadhyay A (2020) Analyzing population and
been supported by Fogarty International Center for PH Leader Course, land use change: selected case studies of Indian metropolitan cities.
NIH under grant no. D43TW009135. Springer Nature, Singapore
9. Office of the Registrar General & Census Commissioner, India
Authors’ relationships and activities The authors declare that there are no (2011) List of towns and their population. Available from https://
relationships or activities that might bias, or be perceived to bias, their censusindia.gov.in/towns/town.aspx. Accessed 11 Sept 2020
work. 10. Narayan VKM, Kondal D, Daya NR et al (2019) Incidence of
diabetes in young adult South Asians compared with Pima
Contribution statement SL, RJ, SK, LC and KMVN all helped develop Indians. Diabetes 68(Suppl 1):1597. https://doi.org/10.2337/db19-
the concept and design the study. The acquisition of data was done by 1597-P
DK, RMA, MKA, SAP, DP, MK, VM, NT and KMVN. Statistical anal- 11. Narayan VKM, Kondal D, Daya NR et al (2019) 1598-P: Incidence
ysis of the data was performed by SL. All authors made substantial of diabetes in South Asian adults in urban India/Pakistan compared
contributions to the interpretation of results. SL and KMVN drafted the with blacks and whites in U.S. Diabetes 68(Suppl 1):1598. https://
manuscript and all authors contributed to the critical revision of the manu- doi.org/10.2337/db19-1598-P
script for important intellectual content. SL had full access to all of the 12. WHO (2016) Obesity and overweight. Available from http://www.
data in the study and takes responsibility for the integrity of the data and who.int/mediacentre/factsheets/fs311/en/. Accessed 13 July 2017
the accuracy of the data analysis. All authors approved the final manu- 13. Office of the Registrar General & Census Commissioner, India
script for publication. (2019) SRS based abridged life tables 2013-17. Available from
https://censusindia.gov.in/vital_statistics/Appendix_SRS_Based_
Open Access This article is licensed under a Creative Commons Life_Table.html. Accessed 13 June 2020
Attribution 4.0 International License, which permits use, sharing, adap- 14. Preston SH, Heuveline P, Guillot M (2000) Demography: measur-
tation, distribution and reproduction in any medium or format, as long as ing and modeling population processes. Wiley-Blackwell,
you give appropriate credit to the original author(s) and the source, Hoboken, p 6
provide a link to the Creative Commons licence, and indicate if changes 15. Pascariu M (2018) MortalityLaws: parametric mortality models,
were made. The images or other third party material in this article are life tables and HMD. R package version, 1(0). Available from
included in the article's Creative Commons licence, unless indicated https://cran.r-project.org/web/packages/MortalityLaws/index.html
otherwise in a credit line to the material. If material is not included in 16. Nwaneri C, Cooper H, Bowen-Jones D (2013) Mortality in type 2
the article's Creative Commons licence and your intended use is not diabetes mellitus: magnitude of the evidence from a systematic
permitted by statutory regulation or exceeds the permitted use, you will review and meta-analysis. Br J Diabetes Vasc Dis 13:192–207.
need to obtain permission directly from the copyright holder. To view a https://doi.org/10.1177/1474651413495703
copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. 17. Pednekar MS, Hakama M, Hebert JR, Gupta PC (2008)
Association of body mass index with all-cause and cause-specific
mortality: findings from a prospective cohort study in Mumbai
(Bombay), India. Int J Epidemiol 37:524–535. https://doi.org/10.
1093/ije/dyn001
18. Anjana RM, Deepa M, Pradeepa R, Mahanta J, Narain K, Das HK
References et al (2017) Prevalence of diabetes and prediabetes in 15 states of
India: results from the ICMR – INDIAB population-based cross-
1. International Diabetes Federation (ed) (2019) IDF Diabetes Atlas, sectional study. Lancet Diabetes Endocrinol 5:585–596. https://doi.
9th edn. International Diabetes Federation, Brussels, Belgium org/10.1016/S2213-8587(17)30174-2
Diabetologia (2021) 64:521–529 529

19. Priya M, Mohan Anjana R, Pradeepa R, Jayashri R, Deepa M, the China Health and Retirement Longitudinal Study. BMJ Open
Bhansali A et al (2011) Comparison of capillary whole blood 5:1–7
versus venous plasma glucose estimations in screening for diabetes 28. Yadav A, Yadav S, Kesarwani R (2012) Decelerating mortality
mellitus in epidemiological studies in developing countries. rates in older ages and its prospects through Lee-Carter approach.
Diabetes Technol Ther 13:586–591. https://doi.org/10.1089/dia. PLoS One 7:e50941. https://doi.org/10.1371/journal.pone.0050941
2010.0218 29. Luhar S, Mallinson PAC, Clarke L, Kinra S (2018) Trends in the
20. Luhar S, Timæus IM, Jones R, Cunningham S, Patel SA, Kinra S socioeconomic patterning of overweight/obesity in India: a repeated
et al (2020) Forecasting the prevalence of overweight and obesity in cross-sectional study using nationally representative data. BMJ
India to 2040. PLoS One 15:e0229438. https://doi.org/10.1371/ Open 8:e023935. https://doi.org/10.1136/bmjopen-2018-023935
journal.pone.0229438 30. Narayan KMV, Boyle JP, Thompson TJ, Sorensen SW,
21. International Institute for Population Sciences (2017) National Williamson DF (2003) Lifetime risk for diabetes mellitus in the
Family Health Survey (NFHS-4) 2015-16 India. International United States. J Am Med Assoc 290:1884–1890. https://doi.org/
Institute for Population Sciences (IIPS) and ICF Available from 10.1001/jama.290.14.1884
http://rchiips.org/nfhs/NFHS-4Reports/India.pdf. Accessed 1 31. Basu S, Vellakkal S, Agrawal S, Stuckler D, Popkin B, Ebrahim S
Feb 2018 (2014) Averting obesity and type 2 diabetes in India through sugar-
22. Arokiasamy P, Parasuraman S, Sekher TV, Lhungdim H (2013) sweetened beverage taxation: an economic-epidemiologic model-
Study on global AGEing and adult health (SAGE), Wave 1. ing study. PLoS Med 11:e1001582. https://doi.org/10.1371/journal.
World Health Organization. Available from https://apps.who.int/ pmed.1001582
healthinfo/systems/surveydata/index.php/catalog/sage. Accessed
32. Weber MB, Ranjani H, Staimez LR, Anjana RM, Ali MK, Narayan
Jan 2018
KMV et al (2016) The stepwise approach to diabetes prevention:
23. Roth G, Caswell H (2018) Occupancy time in sets of states for
results from the D-CLIP randomized controlled trial. Diabetes Care
demographic models. Theor Popul Biol 120:62–77. https://doi.
39:1760–1767. https://doi.org/10.2337/dc16-1241
org/10.1016/j.tpb.2017.12.007
24. Nishida C, Barba C, Cavalli-Sforza T, Cutter J, Deurenberg P, 33. Anjana RM, Rani CSS, Deepa M, Pradeepa R, Sudha V, Nair HD
Darnton-Hill I et al (2004) Appropriate body-mass index for et al (2015) Incidence of diabetes and prediabetes and predictors of
Asian populations and its implications for policy and intervention progression among Asian Indians: 10-year follow-up of the
strategies. Lancet 363:157 Chennai urban rural epidemiology study (CURES). Diabetes Care
25. Deepa M, Grace M, Binukumar B, Pradeepa R, Roopa S, Khan HM 38:1441–1448. https://doi.org/10.2337/dc14-2814
et al (2015) High burden of prediabetes and diabetes in three large 34. Ramachandran A, Ramachandran S, Snehalatha C, Augustine C,
cities in South Asia: the Center for cArdio-metabolic Risk Murugesan N, Viswanathan V et al (2007) Increasing expenditure
Reduction in South Asia (CARRS) Study. Diabetes Res Clin on health care incurred by diabetic subjects in a developing country:
Pract 110:172–182. https://doi.org/10.1016/j.diabres.2015.09.005 a study from India. Diabetes Care 30:252–256. https://doi.org/10.
26. Gujral UP, Prabhakaran D, Pradeepa R, Kandula NR, Kondal D, 2337/dc06-0144
Deepa M et al (2019) Isolated HbA1c identifies a different 35. UN-DESA (2018) World urbanization prospects: the 2018 revision.
subgroup of individuals with type 2 diabetes compared to fasting Department of Economic and Social Affairs. Available from https://
or post-challenge glucose in Asian Indians: the CARRS and population.un.org/wup/. Accessed 5 Sept 2019
MASALA studies. Diabetes Res Clin Pract 153:93–102. https://
doi.org/10.1016/j.diabres.2019.05.026 Publisher’s note Springer Nature remains neutral with regard to jurisdic-
27. Yuan X, Liu T, Wu L, Zou ZY, Li C (2015) Validity of self- tional claims in published maps and institutional affiliations.
reported diabetes among middle-aged and older Chinese adults:

You might also like