Lessons from a long cohort study of diabetes in South Asia

Abstract

Population-based investigations from Indian megacities reveal insights into diabetes and broader chronic disease among native South Asian individuals.

With over 1 in 10 adults of the world’s population diagnosed with diabetes, the disease is a major health, economic and developmental threat, and among the fastest-growing causes of death and disability worldwide. This burden disproportionately affects individuals of South Asian ancestry; that is, over 2 billion people residing in or with origins from the Indian subcontinent. Dating back to the 1980s, cross-sectional studies in numerous countries suggested that people from South Asia who have migrated to other countries exhibit a remarkably high prevalence of diabetes and related cardiometabolic conditions compared with the majority population in those settings, often at lower body mass index and at younger ages. By the 2000s, this distinctive presentation of diabetes and accompanying cardiovascular risk gained recognition as the putative ‘South Asian phenotype’¹.

Despite the wealth of research examining South Asian diaspora communities, considerable gaps remained. First, diabetes was being investigated according to etiological frameworks developed in well-nourished high-income nations, ignoring contextual factors and creating theoretical blind spots. Second, methodologically, the predominance of cross-sectional studies limited the temporal understanding of causal mechanisms, pathophysiology and disease progression. Third, no large representative studies contained robust phenotypic and genotypic measures (such as biomarker panels, organ imaging, multi-omics profiling, and genome-wide association studies). Fourth, the focus on middle-aged immigrant populations largely left open fundamental questions about how diabetes manifests in native populations, especially in younger people, who have remained in the subcontinent.

To address these limitations, in 2010, we established the Center for Cardiometabolic Risk Reduction in South Asia (CARRS) cohort and biorepository in India and Pakistan. The CARRS cohort was designed as a population-based study to longitudinally investigate the development of diabetes, cardiovascular diseases and accompanying morbidities in South Asian individuals living in the subcontinent, beginning early in the life course before overt disease usually manifests. We highlight key features and findings over the 15-year evolution of CARRS as an exemplar of cohort studies generating context-specific knowledge while remaining responsive to emerging scientific opportunities, ultimately contributing to local and global cardiometabolic disease prevention efforts.

CARRS cohort design and methods

Using a multi-stage complex survey design, CARRS was assembled as a representative, geocoded cohort of adults from three South Asian megacities: Chennai and Delhi in India, and Karachi in Pakistan^2,3. Recruited in two waves using identical methods in 2010–2012 and 2014–2016, the cohort achieved high participation rates (around 90%) and enrolled nearly 31,000 individuals, with a robust biorepository of stored DNA, plasma and serum samples from over 27,000 participants (Fig. 1). Enrolling adults as young as 20 years of age enabled the study of early-stage disease development. Study eligibility was intentionally agnostic to disease status, enabling investigation of a wide range of chronic conditions.

Fig. 1 |. Cohort design and measures.

CARRS was assembled using a multi-stage complex survey design, and enrolled nearly 31,000 adults from Chennai, Delhi and Karachi, recruited in two waves (2010–2012 and 2014–2015).

Since baseline, the cohort has been evaluated biennially on pathophysiological risk factors, lifestyle and health behaviors, social, economic and environmental contexts, and health outcomes. Active follow-up continues at the Chennai and Delhi sites, with about 95% of participants assessed at one or more follow-ups and over 77% overall retention with minimal bias due to losses in follow-up⁴.

Epidemiology of diabetes in South Asian individuals.

Table 1 summarizes selected diabetes-focused findings from the CARRS cohort. Data from CARRS challenged the narrative that diabetes risk in South Asian individuals arose largely from migration to obesogenic environments. Instead, we found a substantial burden of diabetes among South Asian individuals who never migrated. Based on definitions that combine HbA1c (a measure that reflects average blood sugar levels over the past 2–3 months) and fasting plasma glucose (reflecting blood sugar in a fasted state at the time of testing), diabetes prevalence ranged from 16% in Karachi, Pakistan, to 25% in Delhi, India⁵. Roughly 60% of diabetes cases were previously undiagnosed. After age adjustment, South Asian individuals residing in Chennai had a higher prevalence of diabetes than South Asian individuals living in the USA, roughly 38% versus 24%⁶. Furthermore, the lifetime risk of developing diabetes among adults in India was alarming; 65% of women and 56% of men in Indian cities were projected to develop diabetes during their lifetime⁷.

Table 1 |.

Selected findings from the CARRS cohort

	Finding and representative references	Contribution
Diabetes epidemiology and pathophysiology
Prevalence comparison	Diabetes prevalence higher in India (Chennai) than in South Asian populations in the USA (Chicago and San Francisco); age-adjusted diabetes was 38% in CARRS and 24% in the MASALA study	Challenges conventional wisdom about migration and diabetes risk, suggesting local environmental factors in South Asia may contribute more significantly than previously thought
Lifetime risk	65% of women and 56% of men in Indian cities will develop diabetes over their lifetime	Quantifies the alarming burden of diabetes risk in urban South Asian populations, highlighting the urgent need for prevention strategies
BMI and diabetes risk	Higher rates of diabetes in South Asian individuals at all BMI levels compared to white individuals. In men aged ≥40 years: diabetes prevalence in normal weight South Asian adults was 24% versus 6% in white individuals; in overweight: 28% versus 8%; in obese: 40% versus 23%	Demonstrates that traditional BMI cutoffs are inadequate for assessing metabolic risk in South Asian individuals and supports the need for ethnicity-specific thresholds.
Glycemic physiology	South Asian individuals have higher mean FPG (versus 2-h plasma glucose in US Black adults and Pima Indians) and lower insulin secretion rather than higher insulin resistance; age and BMI-adjusted HOMA-B scores substantially lower in South Asian adults compared with US Black adults and Pima Indians.	Identifies distinct pathophysiological pathways across different high-risk populations, revealing β-cell dysfunction may be more important than insulin resistance in South Asian individuals, contrary to other high-risk groups
Rapid progression	Individuals spend 35–40 years in normoglycemia followed by only 6–10 years in prediabetes before developing diabetes.	Identifies a compressed timeline for intervention in the prediabetic phase, emphasizing the need for earlier screening and more aggressive management of prediabetes in South Asian individuals
Cardiometabolic comorbidities and biomarkers
Hypertension	30.1% of men and 26.8% of women had hypertension at baseline; only 7–55% of those with hypertension were treated; incidence: 82.6 cases per 1,000 person-years	Highlights notable treatment gaps in hypertension management and exceptionally high incidence rates compared to global standards, suggesting inadequate detection and treatment systems
Dyslipidemia	Over 60% of the cohort had some form of dyslipidemia, with low HDL-C and high triglycerides as the most common abnormalities; although 29% of the cohort had increased LDL cholesterol, only 2% were on lipid-lowering drugs.	Exposes a critical treatment gap in dyslipidemia management, pointing to needed improvements in awareness, access to care, and treatment adherence
Apolipoprotein B (ApoB)	ApoB was superior to LDL cholesterol and non-HDL cholesterol as a cardiovascular risk marker; ApoB levels were higher in Indian individuals compared to American individuals across percentiles.	Provides evidence for potentially improving clinical care by replacing conventional lipid panel with ApoB measurements for routine follow-ups in this population
Kidney and liver disease	7.5% prevalence of chronic kidney disease; 86% of participants with CKD had either abnormal A1c or hypertension; 67% prevalence of NAFLD based on ultrasound (53.7% based on liver elastography).	Shows high clustering of kidney, liver and cardiometabolic conditions
Screening and treatment gaps	Only 2% of adults with dyslipidemia on lipid-lowering drugs; between 45% and 93% of people with hypertension untreated; 50–60% of people with diabetes unaware of their condition	Identifies systemic healthcare delivery challenges requiring both clinical and public health interventions to improve disease awareness, diagnosis, and management
Behavioral and social determinants
Socioeconomic distribution	Cardiometabolic diseases not concentrated only among the poor; clear socioeconomic gradients only in low fruit/vegetable intake and smokeless tobacco use.	Challenges traditional socioeconomic models of disease distribution, suggesting that prevention strategies need to target all socioeconomic groups in South Asian urban settings
Physical activity and sleep	76% of the cohort classified as sedentary at baseline; insomnia (14%) and snoring (29%) highly prevalent	Identifies sedentary behavior as a key intervention target for reducing cardio-metabolic risk in urban South Asian populations; highlights sleep disorders as a potential target for cardio-metabolic intervention
Diet and tobacco	High dietary diversity appeared protective against cardio-metabolic risk factors in cross-sectional analyses; >40% of men report ever using tobacco; strong gendered pattern	Suggests dietary diversity as a potential intervention target, though further longitudinal analysis is needed to confirm causal relationships
Social relationships	60% of couples were concordant in chronic disease status; couples where both members had a chronic disease were more likely to be from wealthier households	Identifies household-level interventions as potentially effective targets for disease prevention, especially in higher socioeconomic strata
Environmental determinants
Built and physical environment	Indian food outlets, rather than Western ones, were positively correlated with overweight and obesity; inverse association between area of nearest park and major depression among people with existing cardiometabolic disease	Challenges the narrative that Western food is primarily responsible for obesity in urbanizing settings; identifies urban planning and green space access as potentially important for health
Air pollution	A 10 μg m−3 increase in monthly average PM2.5 exposure associated with 0.40 mg dl−1 increase in FPG, 0.021 U increase in HbA1c, and 1.22 times increased risk of incident type 2 diabetes; mean annual PM2.5 exposure: Chennai 40 μg m−3, Delhi 102 μg m−3; per interquartile range of monthly exposure: 1.77 mm Hg higher SBP	Establishes air pollution as a significant contributor to cardio-metabolic risk beyond respiratory effects, suggesting air quality improvement as an important public health strategy for diabetes and hypertension prevention

CKD, chronic kidney disease; FPG, fasting plasma glucose; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MASALA, Mediators of Atherosclerosis in South Asians Living in America; NAFLD, non-alcoholic fatty liver disease; SBP, systolic blood pressure. CARRS Cohort study publications that support these findings are listed at https://www.carrsprogram.org/publications-1.

Reconsidering diabetes pathophysiology.

CARRS baseline data showed that the prevalence of diabetes in Asian Indian adults was higher than in white adults, even at body mass indices within the underweight and normal ranges⁸. These observations sparked curiosity about diabetes pathophysiology in non-obese individuals and led to the hypothesis that insulin deficiency rather than insulin resistance — the classical hallmark of type 2 diabetes — contributes to the disease in South Asian individuals⁹. Further data about the incidence of diabetes were needed to rule out reverse-causation.

Comparative analysis of diabetes incidence between the CARRS cohort and Black adults in the USA proved revealing¹⁰. The age-standardized rate of diabetes in South Asian and Black adults was comparable (28.7 and 27.6 cases per 1,000 person-years, respectively), yet the groups exhibited highly different profiles with respect to classical measures of diabetes risk. Compared with Black adults in the USA, South Asian individuals presented with lower body mass index (−3.3 BMI units), lower insulin secretion (−23 pmol l⁻¹ fasting insulin), less insulin resistance (−0.8 μIU ml⁻¹ × mmol l⁻¹ HOMA-IR; fasting insulin × fasting glucose), and lower beta cell function (−31.1 μIU ml⁻¹ mmol l⁻¹ HOMA-B; fasting insulin/fasting glucose) — biomarkers that indicate substantially higher insulin deficiency in the absence of insulin resistance. South Asian individuals also had a higher incidence of diabetes among adults classified as having normal weight. These contrasts further cemented the notion of insulin deficiency as a key driver of diabetes in South Asian populations.

Analyses considering newer subtyping methods later showed that roughly 77% of diabetes¹¹ and 67% of prediabetes¹² in CARRS was classified as related to severe or mild insulin deficiency. Through further longitudinal investigation, we found that this insulin-deficient phenotype seems to accelerate progression from prediabetes to diabetes, creating a compressed intervention window of only 6–10 years and demonstrating serious disease progression with heightened mortality risk¹³.

Other cardiometabolic diseases.

Table 1 describes a large burdens of hypertension (30% of men and 27% of women at baseline) and dyslipidemia (approximately 23% had high levels of low-density lipoprotein cholesterol although only 2% of these individuals were on treatment; over half the cohort had low levels of high-density lipoprotein cholesterol) in the cohort. In addition, 7.5% of participants had evidence of chronic kidney disease, and among those with kidney disease, 86% presented with either high HbA1c or hypertension. Based on vibration-controlled transient elastography, over half of the CARRS participants had indications of non-alcohol fatty liver disease, which was itself, unsurprisingly, associated with diabetes, central obesity and insulin resistance. Collectively, these data point to the relatively high prevalence of cardiovascular-kidney-metabolic syndrome and metabolic-associated liver disease — conditions that are rising globally.

Health behaviors, social factors and environment.

In addition to the unique pathophysiology of diabetes in South Asian populations, established lifestyle and social factors have a crucial role in cardiometabolic health, placing the contemporary South Asian population at double jeopardy of developing diabetes (Table 1). The prevalence of sedentary behavior (76%), insomnia (14%) and snoring (29%), lifetime tobacco use (approximately 45% in men), and poor nutrition (over 50% reporting less than two servings of fruits and vegetables per day) is on par with reports in high-income countries.

The social patterning of major health behaviors shows both similarities and differences with populations in high-income countries (Table 1), and we find that features of the built and physical environment are relevant to the distribution of cardiometabolic health in this population. Importantly, the CARRS portfolio of environment-related research initiatives has shown the role of ambient air pollution with diabetes¹⁴ and health outcomes and contributes to the global understanding of the relationship between environment and health.

The era of precision measurement in CARRS

Current research initiatives are integrating cutting-edge approaches around a central theme of precision prevention and early detection across multiple organ systems — with emphasis on cardiovascular conditions and the brain — to improve individual-level outcomes in the context of these population-level features.

Towards precision prevention, CARRS is conducting genome-wide association studies of the diabetes and prediabetes phenotypes, combined with whole-genome sequencing to identify both rare and common genetic variants for improved risk prediction. Using multi-omics approaches, CARRS is exploring the relationships between genetic factors, epigenetic factors, metabolites and cardiometabolic disease phenotypes to distinguish the role of innate genetic endowments from that of lifestyle and environmental factors.

For the early detection of diseases, the cohort is pursuing state-of-the-art imaging of the arteries, heart, liver, brain and eye as well as biomarkers to capture kidney disease and subclinical cardiovascular conditions. These multimodal assessments enable identification of disease processes before clinical manifestation.

Together, these precision medicine approaches offer new opportunities for targeted intervention aimed at arresting disease progression long before clinical symptoms manifest.

Implications for screening, prevention and management

The distinctive cardiometabolic profile of South Asian individuals has implications for screening, prevention and management strategies for diabetes and associated diseases in this population.

With respect to screening, the substantially higher diabetes prevalence at all BMI levels and rapid progression from prediabetes to diabetes supports earlier initiation of routine screening for all weight classes, including normal weight and underweight South Asian individuals. In addition, the insulin-deficient phenotype characteristic may go undetected using standard single measures such as HbA1c and fasting plasma glucose alone. Multi-step testing protocols that include post-challenge glucose measurements, such as the 2-h oral glucose tolerance test and possibly 30- or 60-min glucose levels, may be needed to identify high-risk individuals who would otherwise be missed during the critical prediabetes stage.

Considering prevention strategies, the distinctive insulin-deficient phenotype warrants attention to beta-cell function preservation alongside traditional lifestyle modification interventions targeting insulin resistance. This also motivates additional research on pharmacological interventions to preserve beta-cell function.

For clinical management, similar tailoring to South Asian populations is warranted. For example, there is already evidence that DPP4 and SGLT2 inhibitors seem to work better in individuals of East Asian or South Asian ethnicity than in white European individuals¹⁵. In addition, comorbidity clustering — particularly among diabetes, hypertension, and kidney disease — calls for multi-system approaches rather than treatment of each condition in isolation. Furthermore, CARRS highlights diagnosis and treatment gaps (Table 1) that call for integrated screening and management approaches that capture and aggressively treat multiple comorbidities simultaneously.

Traditional risk stratification algorithms for cardiovascular disease — a major complication of diabetes and leading cause of death — should be revisited to evaluate whether non-traditional lipid measures such as apolipoprotein B and lipoprotein(a) more accurately predict adverse events in this group.

The cohort also shows that South Asian individuals experience health behaviors and environmental exposures that are particularly resistant to change across all populations. These shared challenges underscore the importance of cross-learning on social, behavioral and environmental levers to address cardiometabolic disease across diverse settings.

Conclusions

Given the substantial burden of diabetes in South Asia, CARRS has provided much needed longitudinal data from within the region itself. Through laboratory-confirmed measures of incident diabetes in a representative sample, the cohort has shown that over half of individual living in India’s megacities are expected to develop diabetes in their lifetime, and that 80% of people with diabetes exhibit an insulin-deficient phenotype with rapid disease progression. This risk profile stands apart from the majority populations in Europe and North America, where the bulk of diabetes research has been conducted. In addition, CARRS challenged the prevailing belief that high diabetes rates existed largely among South Asian individuals who had migrated. We have also learned that people with diabetes in South Asia commonly accumulate comorbidities and engage in suboptimal health behaviors, mirroring global patterns and suggesting some universal aspects of disease development.

Building on bio-banked longitudinal infrastructure, CARRS is now well-positioned to generate data that can inform risk stratification and management strategies tailored for this distinct diabetes phenotype. These approaches must account for the combination of insulin deficiency, rapid disease progression, and multi-organ involvement that characterizes the condition in South Asian populations. These approaches also position the cohort for investigating the key issues of our time — such as aging, cognitive decline and dementia.

The success of CARRS also underscores the critical crucial of global collaboration. International partnership around the CARRS cohort has brought together a study population, scientific expertise, financial resources and research technologies that would otherwise be inaccessible to any single researcher or institution. CARRS has contributed data to global surveillance infrastructure and comparative epidemiological studies, creating a collaborative framework for addressing these global health determinants. The CARRS experience demonstrates how collaborative approaches can advance understanding of population-specific health challenges with global implications.