Abstract
Objective:
We examined associations between age at type 2 diabetes (T2D) diagnosis and long-term mortality and lifetime health loss.
Research design and methods:
We analyzed data from the population-based CARRS cohort. T2D was defined by self-report, glucose-lowering medication use, or glycemic thresholds and categorized by age at diagnosis (20–29, 30–39, 40–59, or ≥60 years). Hazard ratios (HRs) were estimated using time-dependent Cox models with participants without T2D as reference group. Model-based projections estimated years of life lost (YLL), years lived with disability (YLD), disability-adjusted life years (DALYs), and excess life-years lost (LYL).
Results:
Among 21,574 participants (mean age 43.3 years), 6,251 had diabetes. Over a median follow-up of 8.7 years, 2,163 deaths occurred. Younger age at T2D diagnosis was associated with higher risks of mortality and cardiovascular events. Adjusted HRs for all-cause mortality were 2.98 (95% CI 1.60–5.54) for T2D diagnosis at 20–29 years, 2.28 (1.74–2.98) at 30–39 years, 1.73 (1.47–2.04) at 40–59 years, and 1.61 (1.30–1.99) at ≥60 years. Projected lifetime DALYs were greatest with younger diagnosis, ranging from 24.5 years for diagnosis at 20–29 years to 5.5 years at > 60 years, with similar gradients for excess LYL. This pattern was also observed for CVD events across age groups.
Conclusions:
T2D diagnosed at a younger age was associated with higher mortality and greater LYL in South Asians.
1. Introduction
The global burden of diabetes is rising, with type 2 diabetes (T2D) increasingly diagnosed at younger ages [1]. This trend is particularly pronounced in South Asia, where the age of diagnosis is 5–10 years earlier than in other populations [2] and diabetes is a major contributor to morbidity and mortality. In 2021, the region accounted for over 12 million disability-adjusted life years (DALYs) associated with T2D, representing nearly one-quarter of the global burden [3].
South Asian populations are affected by a clinically aggressive diabetes phenotype characterized by low lean mass [4], greater hepatic fat deposition [5], and intrinsic β-cell dysfunction [6]. These features contribute to earlier disease onset at lower body mass index thresholds [6] than in other populations. In India, the lifetime risk of T2D is estimated at 55% in men and 65% in women [7], with more than half of diagnoses occurring before age 50 [8]. Early-onset T2D portends more severe clinical trajectories, including elevated glycated hemoglobin at diagnosis [9], suboptimal glycemic control [10], earlier onset of vascular complications [11,12], and greater risks of disability and premature mortality [13–15].
Although several studies have examined associations between age at diabetes diagnosis and long-term outcomes such as mortality and cardiovascular disease, evidence from South Asian populations remains limited, particularly for analyses that jointly evaluate fatal and nonfatal cardiovascular outcomes alongside projected measures of cumulative lifetime health loss [13,15–17]. Moreover, most prior estimates of diabetes-related health loss rely on population-level models, such as those from the Global Burden of Disease study, which do not leverage individual-level longitudinal data or permit examination of gradients by age at diagnosis [3].
Given differences in clinical context and underlying cardiometabolic risk, the lack of South Asian–specific evidence is particularly important. Estimates derived from other populations may not fully capture the magnitude of risk associated with younger age at diagnosis in South Asians. No prior study has jointly examined fatal and nonfatal cardiovascular outcomes alongside individual-level estimates of cumulative lifetime health loss across a gradient of age at diagnosis in this population. Generating such evidence is therefore essentialfor, accurately characterizing the burden associated with early-onset diabetes in South Asians.
To address these gaps, we analyzed data from the Center for CArdiometabolic Risk Reduction in South Asia (CARRS) study [18–20], a population-based prospective cohort of 21,574 adults aged 20 years and older from Delhi and Chennai, with 185,784 person-years of follow-up. The cohort includes detailed longitudinal glycemic measures, adjudicated cause-specific mortality and cardiovascular events, and high participant retention [21]. We examined associations between age at diabetes diagnosis and all-cause mortality, cardiovascular mortality, and incidence of nonfatal cardiovascular events. In addition, we estimated projected individual-level cumulative health loss, including years of life lost (YLL), years lived with disability (YLD), disability-adjusted life years (DALYs), and excess life years lost (LYL), using participant-specific data and national life tables to quantify lifetime health loss relative to the general population of India.
2. Research design and methods
2.1. Study design and participants
We analyzed data from the prospective, population-based CARRS cohort, which was designed to assess the burden and determinants of cardiometabolic diseases in South Asians [18–20]. Participants aged 20 years or older were recruited from urban households in Delhi and Chennai using multistage cluster random sampling to obtain a sociodemographically representative sample of each city. Detailed methods have been published previously [16].
Enrollment occurred in two waves using identical protocols. CARRS-1 enrolled 12,271 adults in 2010–2011, with follow-up through 2024. CARRS-2 enrolled an independent sample of 9,591 adults between 2014 and 2016, also followed through 2024. The most recent follow-up visit (2023–2024) represented the latest follow-up for both cohorts. Retention was high, with over 95% of CARRS-1 and 70% of CARRS-2 participants completing at least one follow-up visit [21]. Detailed methods, instrumentation, and quality control procedures are provided in the eMethods in the Supplement.
For the present analysis, we excluded individuals younger than 20 years at baseline (n = 34), those with implausible or missing information on age at diabetes diagnosis (n = 25), those lost to follow-up after baseline (n = 228), and one transgender participant (n = 1) due to inability to assign sex-specific life expectancy required for life-table–based analyses. The final analytic sample included 21,574 participants. Details of the analytic sample are shown in Fig. 1. The study adhered to the STROBE reporting guidelines [22].
Fig. 1.
CARRS study flow diagram.
2.2. Age at diabetes diagnosis
Individuals with T2D included those with a prior diagnosis at baseline, those newly diagnosed at baseline, and those who developed T2D during follow-up. Prior diabetes was defined by self-reported physician diagnosis or current use of glucose-lowering medication [23,24]. For these participants, age at diagnosis was derived from self-reported diabetes duration.
Newly diagnosed T2D was defined by the absence of a prior diagnosis and meeting at least one glycemic criterion: fasting plasma glucose ≥126 mg/dL, 2-hour post–oral glucose challenge glucose ≥200 mg/dL, or HbA1c ≥ 6.5%.[24] For these participants, age at diagnosis was assigned as the age at the baseline visit. Among participants who developed T2D during follow-up, age at diagnosis was approximated as the midpoint between the last diabetes-free visit and the first visit at which diabetes was identified, consistent with standard epidemiologic practice [15].
Participants were categorized according to age at T2D diagnosis (20–29, 30–39, 40–59, or ≥60 years). Individuals without diabetes served as the reference group.
2.3. Outcomes
The primary outcome was all-cause mortality. Secondary outcomes included cardiovascular mortality, non-cardiovascular mortality, nonfatal myocardial infarction, and nonfatal stroke. Outcomes were assessed over follow-up visits through December 2024. Participants contributed person-time from enrollment until the first occurrence of a nonfatal cardiovascular event, death, or last follow-up.
Fatal cardiovascular events were ascertained using multiple sources, including death certificates from civil registration systems, structured verbal autopsies with next-of-kin, and hospital records for in-hospital deaths. Nonfatal myocardial infarction and stroke were identified through self-report at follow-up visits (“Has a doctor told you that you have had a heart attack or stroke since your last visit?”), supplemented by available hospital records and supporting clinical documentation.
All outcomes were adjudicated independently by two study physicians using standardized criteria based on clinical records, interviews, and death documentation, with discrepancies resolved by consensus with a third investigator.
Myocardial infarction was defined as new-onset chest pain with elevated cardiac biomarkers or electrocardiographic changes consistent with infarction. Stroke was defined as a focal neurologic deficit lasting at least 24 h, with neuroimaging confirmation when available. Cardiovascular mortality was defined as death within 28 days of myocardial infarction or stroke; all other deaths were classified as non-cardiovascular.
2.4. Statistical analysis
Baseline characteristics were summarized across diabetes age-at-diagnosis groups; missing covariate data were addressed with multiple imputations (10 imputed datasets; details provided in the eMethods in the Supplement) [25].
Associations between diabetes status, age at diagnosis, and outcomes were estimated using time-dependent Cox proportional hazards models. Diabetes status was modeled as a time-varying binary exposure. Participants without diabetes at baseline contributed person-time to the non-diabetes category until meeting diagnostic criteria during follow-up, after which exposure status was updated and remained fixed. This approach ensures that pre-diagnosis person-time is correctly attributed to the non-diabetes state, avoiding misclassification that would arise from assigning baseline diabetes status to the entire follow-up period.
Age at diagnosis was treated as time-fixed covariate, assigned at diabetes onset and held constant thereafter. Participants were classified into predefined age-at-diagnosis groups for all subsequent person-time. Follow-up time from baseline to event, death, or censoring was used as the underlying time scale, and individuals without diabetes status served as the reference group.
Three sequential models were specified: Model 1 adjusted for age, sex, site, and study wave; Model 2 additionally adjusted for education, smoking, body mass index, and prevalent cardiovascular disease. To examine potential mediating pathways, Model 2 was separately extended with additional adjustment for five domains related to diabetes complications: glycemia (fasting glucose, HbA1c, glucose-lowering medication use), blood pressure (systolic blood pressure, antihypertensive use), lipids (LDL-C, HDL-C, log triglycerides, lipid-lowering therapy), kidney function (estimated glomerular filtration rate), and systemic inflammation (log high-sensitivity C-reactive protein). Model 3 further adjusted for diabetes duration and was considered a secondary analysis to explore the extent to which associations were explained by cumulative exposure. We modeled glycemic measures and other available repeated measures as time-varying covariates. Time-weighted glycemic measures were not incorporated for two reasons: first, since not all participants attended all follow-up rounds and laboratory values were missing at varying rounds due to non-attendance, deriving reliable time-weighted averages would require assumptions that introduce additional imputation uncertainty; second, serial post-baseline glycemic values may lie on the causal pathway between age at diagnosis and outcomes, such that conditioning on them risks over-adjustment bias, attenuating the estimated associations toward the null and obscuring the contribution of glycemic deterioration to the excess risk observed at younger ages of diagnosis. Models incorporated survey weights to account for the sampling design (eMethods in the Supplement).
Effect modification was assessed by sex, site, study wave, smoking status, and baseline cardiovascular disease. Nonlinearity was assessed with fractional polynomials, and proportional hazards assumptions were assessed with Schoenfeld residuals. Competing risks were evaluated with Fine–Gray subdistribution hazards [26] for cardiovascular outcomes, treating non-cardiovascular death as the competing event. Sensitivity analyses excluded events within two years of baseline and modeled diabetes duration in 5-year increments.
2.5. Estimation of projected lifetime health loss
Model-based projected cumulative health loss among individuals with diabetes was quantified as YLL, YLD, and DALYs using Global Burden of Disease methodology applied at the individual level with participant-specific data and national life tables [27–29]. Excess life-years lost were estimated using restricted mean lifetime relative to age- and sex-matched general population life expectancy [30]. These estimates represent model-based projections of lifetime health loss, rather than quantities directly observed during follow-up. Full equations and additional methodological details are provided in the Supplement.
3. Results
3.1. Baseline characteristics.
The study included 21,574 participants (mean age, 43.3 ± 13.2 years; 47.2% men). Among those with prevalent or incident diabetes (n = 6251), 242 (3.9%) were diagnosed at 20–29 years, 1220 (19.5%) at 30–39 years, 3,751 (60.0%) at 40–59 years, and 1038 (16.6%) at 60 years or older. The mean age at diagnosis was 48.1 years (Supplemental Fig. S1).
Baseline sociodemographic characteristics were largely similar across age-at-diagnosis groups with respect to sex, study site, education, and smoking status (Table 1). Men had higher blood pressure and fasting glucose than women while women had higher body mass index (eTable S1). Participants diagnosed at 20–29 years had the highest mean fasting glucose (160.6 ± 75.1 mg/dL), yet the lowest use of medications. Among participants with diabetes, mean glycated hemoglobin was highest in those diagnosed at 20–29 years (8.2%), reflecting the longest diabetes duration prior to enrollment, and decreased progressively with older age at diagnosis, reaching 7.2% in those diagnosed at ≥60 years. The prevalence of pre-existing cardiovascular disease increased with older age at diagnosis, from 4.1% before age 40–15.8% at 60 years or older.
Table 1.
Baseline characteristics.
| Characteristic | No Diabetes | Age at diabetes diagnosis | Total | |||
|---|---|---|---|---|---|---|
| 20–29 yrs | 30–39 yrs | 40–59 yrs | ≥60 yrs | |||
| Number of participants | N = 15,323 | N = 242 | N = 1,220 | N = 3,751 | N = 1,038 | N = 21,574 |
| Prior Diabetes Diagnosis | n/a | 143 (58.6%) | 771 (63.1%) | 2,581 (68.8%) | 714 (68.7%) | 4,209 (19.5%) |
| New Diabetes Diagnosis at Baseline | n/a | 94 (38.5%) | 421 (34.5%) | 1,074 (28.6%) | 297 (28.6%) | 1,886 (8.7%) |
| Prevalent Diabetes | n/a | 237 (97.1%) | 1,192 (97.5%) | 3,655 (97.4%) | 1,011 (97.2%) | 6,095 (28.3%) |
| Incident Diabetes | 5 (2.0%) | 28 (2.3%) | 96 (2.6%) | 27 (2.6%) | 156 (0.8%) | |
| Years since diabetes diagnosis, mean (SD) | n/a | 7.1 (7.9) | 5.7 (7.3) | 4.5 (5.9) | 2.6 (4.4) | 4.5 (6.1) |
| Age in years, mean (SD) | 40.1 (12.4) | 31.4 (8.3) | 39.2 (7.2) | 51.6 (7.5) | 67.7 (6.2) | 43.3 (13.2) |
| Age at diabetes diagnosis, mean (SD) | n/a | 26.4 (2.4) | 35.4 (2.7) | 48.5 (5.4) | 66.2 (5.7) | 48.1 (11.4) |
| Men, N (%) | 7,275 (47.5%) | 96 (39.3%) | 568 (46.5%) | 1,695 (45.2%) | 544 (52.3%) | 10,178 (47.2%) |
| Site | ||||||
| Chennai | 8,405 (54.8%) | 155 (63.5%) | 702 (57.4%) | 1,954 (52.1%) | 447 (43.0%) | 11,663 (54.1%) |
| Delhi | 6,918 (45.1%) | 87 (35.7%) | 518 (42.4%) | 1,797 (47.9%) | 591 (56.8%) | 9,911 (45.9%) |
| CARRS wave | ||||||
| CARRS-1 | 8,193 (53.5%) | 159 (65.2%) | 766 (62.7%) | 2,347 (62.5%) | 668 (64.2%) | 12,133 (56.2%) |
| CARRS-2 | 7,130 (46.5%) | 83 (34.0%) | 454 (37.2%) | 1,404 (37.4%) | 370 (35.6%) | 9,441 (43.8%) |
| Educational status | ||||||
| Up to high school | 7,988 (52.1%) | 99 (40.6%) | 621 (50.8%) | 1,963 (52.3%) | 627 (60.3%) | 11,298 (52.4%) |
| College | 4,406 (28.8%) | 91 (37.3%) | 352 (28.8%) | 1,086 (28.9%) | 247 (23.8%) | 6,182 (28.7%) |
| Graduate | 2,929 (19.1%) | 52 (21.3%) | 247 (20.2%) | 702 (18.7%) | 164 (15.8%) | 4,094 (19.0%) |
| Tobacco smoking | ||||||
| Never | 12,939 (84.4%) | 218 (89.3%) | 1,064 (87.1%) | 3,200 (85.3%) | 873 (83.9%) | 18,294 (84.8%) |
| Previous | 1,483 (9.7%) | 16 (6.6%) | 105 (8.6%) | 400 (10.7%) | 139 (13.4%) | 2,143 (9.9%) |
| Current | 901 (5.9%) | 8 (3.3%) | 51 (4.2%) | 151 (4.0%) | 26 (2.5%) | 1,137 (5.3%) |
| Prevalent CVD, N (%) | 154 (1.0%) | 10 (4.1%) | 50 (4.1%) | 255 (6.8%) | 164 (15.8%) | 633 (2.9%) |
| Blood pressure medications, N (%) | 599 (3.9%) | 28 (11.5%) | 269 (22.0%) | 1,314 (35.0%) | 523 (50.3%) | 2,733 (12.7%) |
| Lipid lowering medications, N (%) | 483 (3.2%) | 23 (9.4%) | 151 (12.4%) | 710 (18.9%) | 189 (18.2%) | 1,556 (7.2%) |
| Glucose lowering medications, N (%) | 0 (0.0%) | 94 (38.5%) | 494 (40.4%) | 1,606 (42.8%) | 326 (31.3%) | 2,520 (11.7%) |
| Systolic Blood Pressure (mmHg), mean (SD) | 121.4 (18.9) | 121.5 (15.4) | 127.6 (18.2) | 134.6 (21.0) | 141.2 (24.0) | 125.0 (20.5) |
| Diastolic Blood Pressure (mmHg), mean (SD) | 79.7 (11.9) | 81.6 (10.5) | 84.9 (11.8) | 85.8 (12.1) | 84.1 (13.4) | 81.3 (12.2) |
| Body Mass Index (kg/m2), mean (SD) | 24.8 (4.8) | 26.9 (4.9) | 27.9 (4.9) | 27.8 (5.0) | 26.4 (5.0) | 25.6 (5.0) |
| Total Cholesterol (mg/dl), mean (SD) | 177.8 (35.5) | 186.0 (41.6) | 187.3 (39.8) | 191.9 (40.0) | 192.4 (41.3) | 181.6 (37.4) |
| LDL-Cholesterol (mg/dl), mean (SD) | 107.5 (30.2) | 105.7 (37.7) | 108.1 (33.6) | 113.7 (34.5) | 116.9 (35.4) | 109.1 (31.7) |
| HDL-Cholesterol (mg/dl), mean (SD) | 43.6 (11.1) | 40.5 (11.4) | 40.1 (9.6) | 42.5 (10.5) | 45.3 (12.4) | 43.2 (11.0) |
| Triglycerides (mg/dl), median (IQR) | 116.3 (85.7, 158.9) | 150.8 (106.3, 228.0) | 159.0 (114.0, 235.2 | 155.4 (113.0, 212.0) | 136.0 (102.0, 182.0) | 125.0 (92.0, 174.0) |
| Fasting Plasma Glucose (mg/dl), mean (SD) | 95.0 (9.5) | 160.6 (75.1) | 158.2 (70.6) | 148.8 (64.4) | 128.8 (49.9) | 110.3 (43.0) |
| Glycated Hemoglobin (%), mean (SD) | 5.6 (0.4) | 8.2 (2.3) | 7.9 (2.1) | 7.7 (1.9) | 7.2 (1.6) | 6.2 (1.5) |
| eGFR (ml/min/1.73 m2), mean (SD) | 111.6 (15.0) | 119.6 (17.5) | 112.6 (14.4) | 100.9 (15.8) | 85.2 (17.5) | 108.6 (16.7) |
| hsCRP (mg/L), median (IQR) | 2.9 (1.4, 4.8) | 4.0 (1.7, 7.8) | 4.2 (2.2, 7.3) | 4.4 (2.2, 7.4) | 3.8 (2.0, 6.7) | 3.2 (1.6, 5.6) |
| Median follow-up time, years (IQR) | 8.7 (7.5, 11.5) | 9.7 (7.7, 11.7) | 9.3 (7.7, 11.7) | 8.9 (7.4, 11.6) | 8.1 (5.0, 10.9) | 8.7 (7.4, 11.5) |
Data are presented as N (%) or mean (SD) or median (IQR); CVD: cardiovascular disease;
3.2. Risk of mortality and cardiovascular events according to age at diagnosis
Over a median follow-up of 8.7 years (interquartile range, 7.4–11.5), encompassing 185,785 person-years at risk, there were 2163 deaths, including 1114 from cardiovascular causes and 1049 from non-cardiovascular causes. A total of 527 nonfatal cardiovascular events occurred, including 360 coronary heart disease events, 329 nonfatal myocardial infarctions, and 178 nonfatal strokes.
In fully adjusted models (Table 2; Fig. 2), diabetes diagnosed at younger ages was associated with progressively higher relative hazards of death and cardiovascular events compared with individuals without diabetes. As compared with individuals without diabetes, the HR for all-cause mortality was 2.98 (95% CI, 1.60–5.54) for participants diagnosed with diabetes between 20–29 years, 2.28 (95% CI, 1.74–2.98) for those diagnosed between 30–39 years, 1.73 (95% CI, 1.47 to 2.04)for those diagnosed between 40–59 years, and 1.61 (95% CI, 1.30 to 1.99) for those diagnosed at ≥60 years. HRs for cardiovascular mortality showed a similar inverse gradient: 3.89 (95% CI, 1.44–10.50), 2.84 (95% CI, 1.95–4.14), 2.41 (95% CI, 1.89–3.06), and 1.65 (95% CI, 1.23–2.20), respectively.
Table 2.
Associations between age at diagnosis of diabetes and the clinical outcomes.
| Outcome\Age-at-diabetes diagnosis | N events/ n at risk | IR per 1000 PY | Adjusted HRs (95%CI) | ||
|---|---|---|---|---|---|
| Model1* | Model 2* | Model 3** | |||
| All-cause mortality | |||||
| No diabetes | 973/15,323 | 7.5 | 1 (ref) | 1 (ref) | 1 (ref) |
| 20–29 yrs | 17/242 | 7.8 | 2.59 (1.32–5.07) | 2.98 (1.60–5.54) | 1.59 (0.74–3.43) |
| 30–39yrs | 118/1220 | 10.9 | 2.09 (1.62–2.70) | 2.28 (1.74–2.98) | 1.44 (0.97–2.13) |
| 40–59yrs | 625/3751 | 19.6 | 1.56 (1.33–1.82) | 1.73 (1.47–2.04) | 1.35 (1.05–1.74) |
| ≥ 60yrs | 430/1038 | 57.7 | 1.57 (1.28–1.92) | 1.61 (1.30–1.99) | 1.74 (1.30–2.34) |
| Per decade earlier | 2163/21,574 | 11.9 | 1.18 (1.08–1.30) | 1.17 (1.06–1.30) | 0.90 (0.75–1.09) |
| Diabetes vs No diabetes | 1190/21,574 | 22.7 | 1.64 (1.45–1.87) | 1.79 (1.56–2.05) | 1.49 (1.18–1.87) |
| CVD mortality | |||||
| No diabetes | 385/15,323 | 3.0 | 1 (ref) | 1 (ref) | 1 (ref) |
| 20–29 yrs | 5/242 | 2.3 | 3.34 (1.15–9.71) | 3.89 (1.44–10.50) | 1.67 (0.57–4.93) |
| 30–39yrs | 56/1220 | 5.2 | 2.73 (1.92–3.86) | 2.84 (1.95–4.14) | 1.42 (0.87–2.31) |
| 40–59yrs | 401/3751 | 12.6 | 2.24 (1.80–2.78) | 2.41 (1.89–3.06) | 1.55 (1.15–2.07) |
| ≥ 60yrs | 267/1038 | 35.9 | 1.66 (1.28–2.15) | 1.65 (1.23–2.20) | 1.45 (0.99–2.15) |
| Per decade earlier | 1114/21,574 | 6.1 | 1.20 (1.07–1.35) | 1.19 (1.05–1.35) | 0.86 (0.68–1.08) |
| Diabetes vs No diabetes | 729/21,574 | 13.9 | 2.14 (1.79–2.56) | 2.25 (1.84–2.77) | 1.51 (1.15–1.98) |
| Non-CVD mortality | |||||
| No diabetes | 588/15,323 | 4.5 | 1 (ref) | 1 (ref) | 1 (ref) |
| 20–29 yrs | 12/242 | 5.5 | 2.04 (1.00–4.14) | 2.35 (1.16–4.74) | 1.44 (0.52–3.95) |
| 30–39yrs | 62/1220 | 5.7 | 1.65 (1.13–2.41) | 1.87 (1.28–2.74) | 1.39 (0.73–2.66) |
| 40–59yrs | 224/3751 | 7.0 | 1.04 (0.84–1.30) | 1.19 (0.96–1.48) | 1.08 (0.69–1.70) |
| ≥ 60yrs | 163/1038 | 21.9 | 1.50 (1.11–2.03) | 1.59 (1.18–2.15) | 2.21 (1.42–3.42) |
| Per decade earlier | 1049/21,574 | 5.7 | 1.16 (1.01 −1.33) | 1.14 (0.99–1.31) | 0.99 (0.78–1.25) |
| Diabetes vs No diabetes | 461/21,574 | 8.8 | 1.25 (1.05–1.49) | 1.40 (1.17–1.67) | 1.41 (0.94–2.11) |
| Non-fatal CVD | |||||
| No diabetes | 218/15,323 | 1.7 | 1 (ref) | 1 (ref) | 1 (ref) |
| 20–29 yrs | 9/242 | 4.1 | 3.76 (1.74–8.14) | 4.01 (1.88–8.55) | 1.47 (0.58–3.73) |
| 30–39yrs | 50/1220 | 4.6 | 3.75 (2.46–5.72) | 3.60 (2.35–5.52) | 1.55 (0.88–2.73) |
| 40–59yrs | 183/3751 | 5.7 | 2.35 (1.64–3.36) | 2.27 (1.58–3.25) | 1.19 (0.76–1.88) |
| ≥ 60yrs | 67/1038 | 9.0 | 1.39 (0.90–2.14) | 1.38 (0.89–2.13) | 0.93 (0.46–1.88) |
| Per decade earlier | 527/21,574 | 2.9 | 1.18 (0.96–1.45) | 1.18 (0.96–1.46) | 0.75 (0.53–1.04) |
| Diabetes vs No diabetes | 309/21,574 | 5.9 | 2.40 (1.80–3.19) | 2.33 (1.74–3.13) | 1.20 (0.79–1.81) |
| Non-fatal MI | |||||
| No diabetes | 126/15,323 | 1.0 | 1 (ref) | 1 (ref) | 1 (ref) |
| 20–29 yrs | 5/242 | 2.3 | 2.71 (1.05–7.01) | 2.86 (1.10–7.48) | 1.29 (0.35–4.74) |
| 30–39yrs | 36/1220 | 3.3 | 4.25 (2.63–6.87) | 4.37 (2.64–7.23) | 2.71 (1.36–5.37) |
| 40–59yrs | 122/3751 | 3.8 | 2.21 (1.61–3.04) | 2.28 (1.60–3.24) | 1.78 (0.99–3.20) |
| ≥ 60yrs | 40/1038 | 5.4 | 1.28 (0.79–2.07) | 1.60 (0.94–2.74) | 2.03 (0.90–4.58) |
| Per decade earlier | 329/21,574 | 1.8 | 1.03 (0.79–1.34) | 1.02 (0.79–1.33) | 0.55 (0.40–0.76) |
| Diabetes vs No diabetes | 203/21,574 | 3.9 | 2.67 (1.95–3.65) | 2.51 (1.83–3.46) | 1.96 (1.15–3.34) |
| Non-fatal Stroke | |||||
| No diabetes | 87/15,323 | 0.7 | 1 (ref) | 1 (ref) | 1 (ref) |
| 20–29 yrs | 2/242 | 0.9 | 2.16 (0.45–10.26) | 2.51 (0.53–1.86) | 0.50 (0.09–2.77) |
| 30–39yrs | 13/1220 | 1.2 | 2.09 (0.93–4.68) | 2.17 (0.94–5.00) | 0.48 (0.18–1.28) |
| 40–59yrs | 54/3551 | 1.7 | 2.31 (1.09–4.89) | 2.34 (1.11–4.95) | 0.62 (0.29–1.31) |
| ≥ 60yrs | 22/1038 | 3.0 | 1.07 (0.52–2.19) | 1.11 (0.53–2.34) | 0.33 (0.10–1.03) |
| Per decade earlier | 178/21,574 | 1.0 | 1. 38 (1.03–1.83) | 1.40 (1.06–1.85) | 1.18 (0.73–1.91) |
| Diabetes vs No diabetes | 91/21,574 | 1.7 | 1.99 (1.11–3.58) | 2.04 (1.09–3.82) | 0.52 (0.27–1.04) |
Model 1: age (years), sex (men/women), study site (Chennai/Delhi) and study wave (CARRS-1/CARRS-2) adjusted;
Model 2 (Primary model): Model 1 + education (up to high school/college/graduate), smoking (no/previous/current), body mass index, and prevalent cardiovascular disease.
Model 3 (Secondary model): Model 2 + diabetes duration. This model is presented to explore the extent to which associations with age at diagnosis are explained by cumulative exposure; attenuation of estimates reflects partial overlap between age at diagnosis and diabetes duration.
Model 3 includes additional adjustment for diabetes duration; attenuation of estimates reflects partial overlap between age at diagnosis and duration.
BMI: body mass index, CVD: cardiovascular disease, IR: incidence rate; MI: Myocardial infarction, PY: person-years;
Fig. 2.
Adjusted HRs for mortality and cardiovascular events according to groups defined by age at diabetes diagnosis (No Diabetes, 20–29 years, 30–39 years, 40–59 years, and ≥60 years). Model is adjusted for age (years), sex (men/women), study site (Chennai/Delhi) and study wave (CARRS-1/CARRS-2), education (up to high school/college/graduate), smoking (no/previous/current), prevalent CVD (No/Yes) and BMI. Error bars denote group-specific 95%CIs. Box sizes are inversely proportional to the variance in risk estimates.
For non-cardiovascular mortality, the pattern was non-monotonic, with HRs of 2.35 (95% CI, 1.16–4.74) for diagnosis at 20–29 years, 1.87 (95% CI, 1.28–2.74) at 30–39 years, 1.19 (95% CI, 0.96 to 1.48) at 40–59 years, and 1.59 (95% CI, 1.18–2.15) at ≥ 60 years. For nonfatal cardiovascular events, HRs were 4.01 (95% CI, 1.88–8.55), 3.60 (95% CI, 2.35–5.52), 2.27 (95% CI, 1.58–3.25), and 1.38 (95% CI, 0.89–2.13), respectively. Analysis restricted to newly diagnosed participants in whom timing of onset is well-defined yielded results consistent with the main analysis (eTable S2).
When modeled continuously, each decade earlier age at diagnosis was associated with 17% higher relative hazards of all-cause mortality (HR, 1.17; 95% CI, 1.06–1.30) and 19% higher relative hazards of cardiovascular mortality (HR, 1.19; 95% CI, 1.05–1.35). Associations with non-cardiovascular mortality (HR, 1.14; 95% CI, 0.99–1.31) and nonfatal cardiovascular events (HR, 1.18; 95% CI, 0.96–1.46) did not reach statistical significance. Longer diabetes duration was associated with higher mortality risk (eTable S8). After mutual adjustment for age at diagnosis, these associations were attenuated, mirroring the attenuation observed when adjusting for diabetes duration (Table 2 (Model 3); eTable S8), and indicating that age at diagnosis and duration are closely related and partially overlapping constructs.
3.3. Assessing the role of traditional metabolic markers on associations between age at diabetes diagnosis and clinical outcomes
Adjustment for glycemic markers substantially attenuated the excess risks associated with diabetes (eTable S3). Among participants diagnosed at 20–29 years of age, the HR for all-cause mortality decreased from 2.98 to 1.91 after adjustment for fasting glucose, and to 1.71 after adjustment for HbA1c. Corresponding estimates for those diagnosed at ≥ 60 years declined from 1.61 to 1.46 (fasting glucose) and 1.34 (HbA1c). A similar attenuation was observed for cardiovascular mortality and nonfatal cardiovascular events, particularly among younger age-at-diagnosis groups, with many associations becoming non-significant.
Further adjustment for systolic blood pressure, lipid concentrations, kidney function, and systemic inflammation produced minimal changes in HRs (eTable S3).
3.4. Effect modification and sensitivity analyses
No substantial effect modification was observed by sex, smoking status, study site, enrollment wave, or history of cardiovascular disease. (eTable S4). HRs declined progressively with increasing age at diabetes diagnosis, with the highest relative risks observed among individuals diagnosed before age 40 when we used fractional polynomials (eFig. S2). Continuous dose–response relationships between age-at-diagnosis of diabetes and study outcomes using restricted cubic splines were consistent with the primary analysis (eFig. S3). Sensitivity analyses excluding events occurring within the first two years of follow-up yielded similar estimates (eTable S5). When we censored all follow-up at December 31, 2019, prior to the onset of the COVID-19 pandemic, and repeated the primary Cox models, hazard ratios were directionally consistent and of broadly similar magnitude to the primary analyses, suggesting that pandemic-era excess mortality did not materially alter the overall pattern of findings (eTable S6). Fine–Gray competing risk models produced results consistent with the main findings, although the pattern of cardiovascular mortality across age-at-diagnosis groups was less distinct (eTable S7). Analyses using categories of diabetes duration demonstrated a log-linear relationship between longer duration and higher risk of mortality and cardiovascular outcomes (eTable S8 and eFig. S4).
3.5. Projected years of life lost, disability, and excess life-years lost
Diabetes diagnosed at younger ages was associated with greater projected YLL, YLD, and DALYs. Among individuals diagnosed at 20–29 years, the mean projected YLL was 22.99 years (95% CI, 18.44–28.36), compared with 5.19 years (95% CI, 3.75–6.99) among those diagnosed at ≥ 60 years (Table 3; eFig. S5). Corresponding YLD estimates were 1.48 years (95% CI, 0.94–2.03) and 0.34 years (95% CI, 0.12–0.55), respectively. Furthermore, the mean projected DALYs among individuals with diabetes ranged from 24.47 years (95% CI, 19.64–30.15) among those diagnosed at 20–29 years to 5.52 years (95% CI, 3.98–7.45) among those diagnosed at ≥60 years. Across all ages of diagnosis, the overall projected burden of diabetes was 12.10 (95%CI, 9.27–15.53) YLLs, 0.83 (95%CI, 0.45–1.22) YLDs, and 12.92 (95% CI, 9.89–16.58) DALYs.
Table 3.
Diabetes-related mean years of life lost (YLL), years lived with disability (YLD) and disability adjusted life years (DALYs) by age at diabetes diagnosis.
| Cumulative health loss | Participants with diabetes; Cumulative health loss | ||||
|---|---|---|---|---|---|
| 20–29 years n = 242 |
30–39 years n = 1,220 |
40–59 years n = 3,751 |
≥60 years n = 1,038 |
All Diabetes | |
| YLL | |||||
| – Both sexes | 22.99 (18.44–28.36) | 18.07 (14.17–22.74) | 11.36 (8.61–14.72) | 5.19 (3.75–6.99) | 12.10 (9.27–15.53) |
| – Men | 22.55 (18.13–27.79) | 17.74 (13.93–22.30) | 11.13 (8.46–14.40) | 5.00 (3.62–6.74) | 11.66 (8.94–14.95) |
| – Women | 23.27 (18.65–28.73) | 18.36 (14.37–23.12) | 11.55 (8.74–14.97) | 5.39 (3.90–7.27) | 12.48 (9.55–16.02) |
| YLD | |||||
| – Both sexes | 1.48 (0.94–2.03) | 1.21 (0.72–1.71) | 0.80 (0.41–1.19) | 0.34 (0.12–0.55) | 0.83 (0.45–1.22) |
| – Men | 1.42 (0.90–1.94) | 1.19 (0.71–1.67) | 0.77 (0.40–1.14) | 0.31 (0.12–0.51) | 0.79 (0.43–1.15) |
| – Women | 1.52 (0.967–2.09) | 1.23 (0.73–1.74) | 0.82 (0.42–1.23) | 0.37 (0.14–0.60) | 0.86 (0.46–1.27) |
| DALY | |||||
| – Both sexes | 24.47 (19.64–30.15) | 19.28 (15.11–24.23) | 12.16 (9.20–15.73) | 5.52 (3.98–7.45) | 12.92 (9.89–16.58) |
| – Men | 23.97 (19.28–29.50) | 18.93 (14.86–23.76) | 11.90 (9.03–15.38) | 5.31 (3.83–7.16) | 12.44 (9.53–15.95) |
| – Women | 24.80 (19.88–30.58) | 19.59 (15.33–24.64) | 12.36 (9.34–16.02) | 5.76 (4.14–7.76) | 13.34 (10.20–17.12) |
YLL: years of life lost; YLD: years lived with disability; DALY: disability adjusted life years.
Age-standardized projected DALY rates per 100,000 individuals were highest among those diagnosed at 20–29 years, with estimates of 2.5 million (95% CI, 2.0–3.1 million), and lowest among those diagnosed at ≥60 years (512,356; 95% CI, 367,388–692,578). No sex differences were observed in the burden of YLLs, YLDs, or DALYs (eFig. S6–A).
Compared with the general population of India, younger age at diagnosis was associated with greater excess LYL (eTable S9; eFig. S6–B; eFig. S7). The mean excess LYL was 9.56 years (95% CI, 8.65–10.52) for those diagnosed at 20–29 years, and 5.28 years (95% CI, 4.55–5.87) for those diagnosed at ≥60 years. Across all age groups, excess LYL was greater among men than women. Among participants diagnosed at 20–29 years, excess LYL was 12.40 years (95% CI, 10.54–14.46) in men and 6.90 years (95% CI, 5.69–7.89) in women. Corresponding estimates for diagnosis at ≥60 years were 5.77 years (95% CI, 4.86–6.56) and 4.52 years (95% CI, 3.36–5.36), respectively. Cardiovascular mortality accounted for the majority of excess LYL in all groups, mirroring patterns observed for all-cause mortality.
4. Discussion
In this large, population-based cohort of South Asian adults, diabetes diagnosed at younger ages was associated with higher relative risks of all-cause mortality, cardiovascular mortality, and nonfatal cardiovascular events compared with individuals without diabetes. Individuals diagnosed before age 40 years had up to three times the risk of death and four times the risk of cardiovascular disease compared with those without diabetes, even after accounting for diabetes duration. These associations were accompanied by substantial projected cumulative health loss. Among those diagnosed between ages 20–29 years, the mean projected YLL exceeded 22 years, compared with approximately 5 years among those diagnosed at ≥60 years. Projected YLD and total DALY followed a similar gradient. Men diagnosed before age 40 years experienced particularly pronounced reductions in life expectancy, with excess LYL exceeding 12 years. Across all age-at-diagnosis groups, cardiovascular causes accounted for the majority of excess LYL. These findings extend prior literature by integrating fatal and nonfatal outcomes with individual-level estimates of lifetime health loss and suggest that the prognostic burden of younger age at diabetes diagnosis may be particularly severe in South Asian populations.
Adjustment for glycemic markers attenuated the magnitude of associations between age at diabetes diagnosis and cardiovascular outcomes. Residual associations with all-cause mortality persisted after adjustment, suggesting that factors beyond contemporaneous glycemic measures, including cumulative exposure to metabolic risk, delayed diagnosis, or long-term patterns of care, may contribute to excess mortality risk. Although the prospective design limits protopathic bias, early deaths following diagnosis may still reflect reverse causality. Sensitivity analyses excluding events occurring within the first two years of follow-up yielded consistent results, supporting the robustness of the findings.
Although our primary analyses focused on age at diagnosis as a proxy for the accrued lifetime burden of diabetes, supplementary analyses showed that longer diabetes duration was associated with higher risks of all-cause and cardiovascular mortality, consistent with cumulative exposure. When modeled jointly, associations with age at diagnosis were attenuated but remained directionally consistent, indicating that age at diagnosis and duration are closely related and partially overlapping dimensions of disease exposure. Together, these findings suggest that both earlier age at diagnosis and longer duration contribute to risk, but their effects are interrelated and not fully separable.
Despite the magnitude of this burden, individuals diagnosed at younger ages are rarely included in clinical trials. Most randomized studies enroll middle-aged and older adults in order to maximize event rates over shorter follow-up. In a review of 90 major diabetes trials, fewer than 5% of participants had been diagnosed between ages 20 and 40 years [31]. Although trials such as TODAY [32,33], and RISE [34], have focused on adolescents, little evidence exists for adults diagnosed between 20–40 years of age. To address this gap, eligibility criteria in clinical trials may need to better reflect the global epidemiology of diabetes, including regions where diagnosis often occurs decades earlier. Failure to include these younger populations risks perpetuating evidence gaps and undermining care for those with the highest projected lifetime burden.
Prior studies estimating the impact of diabetes on life expectancy have largely relied on model-based projections or population-level aggregates primarily derived from high-income countries, where the mean age at diagnosis in available cohorts tended to be considerably older than contemporary populations and those in low- and middle-income regions such as South Asia. For example, the Emerging Risk Factors Collaboration pooled data from 97 prospective cohorts of high-income regions and estimated that each decade of earlier diabetes onset reduced life expectancy by three to four years, based on hazard ratios applied to summary mortality rates [15]. These projections, while informative, depend on strong assumptions and lack granularity on cause-specific or nonfatal outcomes. Similarly, a Swedish registry study among individuals with type 2 diabetes, in which the mean age at diagnosis was 62 years and the median follow-up was just over five years, found increased mortality with earlier onset but did not assess nonfatal cardiovascular events or account for competing risks [16]. In a secondary analysis of the ADVANCE trial involving older adults (mean age 66 years), both age at diagnosis and diabetes duration were associated with macrovascular events and all-cause mortality; however, only duration was linked to microvascular complications, with the strongest effects seen in younger patients [35]. A secondary analysis of NHANES data [36] has described trends in glycemic control or cardiovascular risk but has not evaluated the cumulative impact of age at diagnosis on long-term health loss. Our study extends this literature by leveraging individual-level, longitudinal data from a rigorously phenotyped South Asian cohort, where the average age at diabetes diagnosis was more than a decade younger than in these prior studies [15,16,36].
The higher effect estimates observed in our cohort likely reflect differences in underlying risk profiles, healthcare context, and disease characteristics rather than methodological variation. Our study extends prior work by leveraging individual-level longitudinal data from a rigorously characterized South Asian cohort, in which the average age at diagnosis is more than a decade younger than in these prior studies. Using methods aligned with the Global Burden of Disease framework, we estimated projected years of life lost, years lived with disability, disability-adjusted life years, and excess life years lost. The integration of fatal and nonfatal outcomes, time-updated cardiometabolic measures, and high participant retention enabled more nuanced, context-specific estimates of lifetime burden. Together, these findings highlight the need for population-specific benchmarks in settings where diabetes is diagnosed earlier and progresses more aggressively.
Several features of the South Asian diabetes phenotype may contribute to this elevated risk. These include reduced β-cell reserve, greater visceral and hepatic adiposity at lower body mass index, and earlier cumulative exposure to cardiometabolic risk. Together, these factors may lead to more rapid disease progression and earlier onset of vascular injury. The younger age at diagnosis in South Asia further extends lifetime exposure to adverse metabolic conditions. These findings underscore that risk estimates derived from European or East Asian populations may not fully capture the burden associated with diabetes diagnosed at younger ages in South Asians.
Several limitations should be acknowledged. Age at diagnosis may reflect age at detection rather than true disease onset, particularly among newly diagnosed participants, and may underestimate the duration of undiagnosed diabetes. Our cohort predominantly reflects T2D, consistent with the epidemiology of diabetes in South Asian adults; however, diabetes type was not formally characterized using biomarker-based classification, and atypical subtypes cannot be excluded, particularly among those diagnosed at younger ages. Any resulting misclassification is likely limited and may bias estimates in either direction.
Among participants with prior diabetes, age at diagnosis was based on self-report and may be subject to recall error. In addition, baseline HbA1c among newly diagnosed participants approximates glycemia at diagnosis, whereas among those with prior diabetes it reflects glycemic control at enrollment, potentially underestimating cumulative glycemic burden, particularly in those diagnosed at younger ages. Accordingly, attenuation of associations after adjustment for glycemic measures should be interpreted with caution.
The year of glucose-lowering medication initiation was not systematically captured; however, sensitivity analyses restricted to newly diagnosed participants, in whom timing of onset is well defined, yielded consistent results and support the robustness of the primary findings. Residual confounding from unmeasured factors, including healthcare access and treatment patterns, cannot be excluded.
Ascertainment of nonfatal cardiovascular events may be incomplete, which would bias associations toward the null. Estimates for the 20–29 years age-at-diagnosis group should be interpreted with caution given smaller sample size and wider confidence intervals, although the overall pattern of results remained consistent. A portion of follow-up overlapped with the COVID-19 pandemic; however, sensitivity analyses yielded similar results. Finally, findings from urban populations may not be fully generalizable to rural settings.
In conclusion, diabetes diagnosis at younger ages was associated with substantial projected reductions in life expectancy and healthy life-years, particularly among those diagnosed before age 40 years. Despite greater cumulative exposure and elevated long-term risk, younger individuals remain underrepresented in clinical trials and underserved in routine care. Diabetes diagnosed at a young age carries enduring risks that underscore the importance of timely prevention, early intervention, and sustained management across the life course.
Supplementary Material
Acknowledgements
We are grateful to the CARRS study participants and the dedicated research personnel and field assistants who undertook outreach and follow-up despite challenging study conditions and deeply cared for the health and wellbeing of our study participants.
Funding statement
The Center for cArdiometabolic Risk Reduction in South Asia study was supported by grants from the National Heart, Lung, and Blood Institute (HHSN2682009900026C, P01HL154996), National Institutes of Health (NIH), National Institute on Aging, NIH (R01-AG89759), and National Institute of Diabetes and Digestive and Kidney Diseases (R01-DK139632).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.diabres.2026.113259.
Footnotes
Disclosures
Nothing to disclose.
CRediT authorship contribution statement
Ram Jagannathan: Writing – original draft, Supervision, Methodology, Conceptualization. Ayodipupo S. Oguntade: Writing – original draft, Formal analysis. Mohan Deepa: Writing – review & editing, Project administration. Dimple Kondal: Writing – review & editing, Project administration, Data curation. Ranjith Mohan Anjana: Writing – review & editing. Shivani A. Patel: Writing – review & editing. Rodrigo M. Carrillo-Larco: Writing – review & editing. Sailesh Mohan: Writing – review & editing, Project administration. Howard H. Chang: Writing – review & editing, Methodology. Mohammed K. Ali: Writing – review & editing, Funding acquisition. Arshed A. Quyyumi: Writing – review & editing, Funding acquisition. Dorairaj Prabhakaran: Writing – review & editing, Funding acquisition. Viswanathan Mohan: Writing – review & editing, Funding acquisition. K.M. Venkat Narayan: Writing – review & editing, Supervision, Funding acquisition. Nikhil Tandon: Writing – review & editing, Supervision, Funding acquisition.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Data sharing statement
Data supporting this study’s findings are available to the corresponding author (JR: ram.jagannathan@emory.edu) upon reasonable request. The R codes and the R notebook for the reproducible analysis is available to the interested readers by contacting: asogunt@emory.edu.
References
- [1].Zhou B, Rayner AW, Gregg EW, et al. Worldwide trends in diabetes prevalence and treatment from 1990 to 2022: a pooled analysis of 1108 population-representative studies with 141 million participants. Lancet 2024;404(10467):2077–93. 10.1016/S0140-6736(24)02317-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Luk A, Wild SH, Jones S, et al. Early-onset type 2 diabetes: the next major diabetes transition. Lancet 2025;405(10497):2313–26. 10.1016/S0140-6736(25)00830-X. [DOI] [PubMed] [Google Scholar]
- [3].Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet. 2023;402(10397):203–234. doi: 10.1016/s0140-6736(23)01301-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Pomeroy E, Mushrif-Tripathy V, Cole TJ, Wells JCK, Stock JT. Ancient origins of low lean mass among South Asians and implications for modern type 2 diabetes susceptibility. Sci Rep 2019;9(1):10515. 10.1038/s41598-019-46960-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Iliodromiti S, McLaren J, Ghouri N, et al. Liver, visceral and subcutaneous fat in men and women of South Asian and white European descent: a systematic review and meta-analysis of new and published data. Diabetologia 2023;66(1):44–56. 10.1007/s00125-022-05803-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Narayan KMV, Kanaya AM. Why are South Asians prone to type 2 diabetes? a hypothesis based on underexplored pathways. Diabetologia 2020;63(6):1103–9. 10.1007/s00125-020-05132-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Luhar S, Kondal D, Jones R, et al. Lifetime risk of diabetes in metropolitan cities in India. Diabetologia 2021;64(3):521–9. 10.1007/s00125-020-05330-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Gujral UP, Pradeepa R, Weber MB, Narayan KM, Mohan V. Type 2 diabetes in South Asians: similarities and differences with white Caucasian and other populations. Ann N Y Acad Sci 2013;1281(1):51–63. 10.1111/j.1749-6632.2012.06838.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Mohan V, Vijayachandrika V, Gokulakrishnan K, et al. A1C cut points to define various glucose intolerance groups in Asian Indians. Diabetes Care 2010;33(3): 515–9. 10.2337/dc09-1694. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Varghese JS, Anjana RM, Geldsetzer P, et al. National estimates of the adult diabetes care continuum in India, 2019–2021. JAMA Intern Med 2023;doi: 10.1001/jamainternmed.2023.3070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Misra S, Ke C, Srinivasan S, et al. Current insights and emerging trends in early-onset type 2 diabetes. Lancet Diabetes Endocrinol 2023;11(10):768–82. 10.1016/S2213-8587(23)00225-5. [DOI] [PubMed] [Google Scholar]
- [12].Banker KK, Liew D, Ademi Z, et al. The impact of diabetes on productivity in India. Diabetes Care 2021;44(12):2714–22. 10.2337/dc21-0922. [DOI] [PubMed] [Google Scholar]
- [13].Nanayakkara N, Curtis AJ, Heritier S, et al. Impact of age at type 2 diabetes mellitus diagnosis on mortality and vascular complications: systematic review and meta-analyses. Diabetologia 2021;64(2):275–87. 10.1007/s00125-020-05319-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Gregg EW, Zhuo X, Cheng YJ, Albright AL, Narayan KM, Thompson TJ. Trends in lifetime risk and years of life lost due to diabetes in the USA, 1985–2011: a modelling study. Lancet Diabetes Endocrinol 2014;2(11):867–74. 10.1016/s2213-8587(14)70161-5. [DOI] [PubMed] [Google Scholar]
- [15].Tomic D, Morton JI, Chen L, et al. Lifetime risk, life expectancy, and years of life lost to type 2 diabetes in 23 high-income jurisdictions: a multinational, population-based study. Lancet Diabet Endocrinol 2022;10(11):795–803. 10.1016/S2213-8587(22)00252-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Sattar N, Rawshani A, Franzén S, et al. Age at diagnosis of type 2 diabetes mellitus and associations with cardiovascular and mortality risks. Circulation 2019;139 (19):2228–37. 10.1161/circulationaha.118.037885. [DOI] [PubMed] [Google Scholar]
- [17].Seo DH, Kim M, Suh YJ, et al. Association between age at diagnosis of type 2 diabetes and cardiovascular morbidity and mortality risks: a nationwide population-based study. Diabetes Res Clin Pract 2024;208:111098. 10.1016/j.diabres.2024.111098. [DOI] [PubMed] [Google Scholar]
- [18].Nair M, Ali MK, Ajay VS, et al. CARRS surveillance study: design and methods to assess burdens from multiple perspectives. BMC Public Health 2012;12:701. 10.1186/1471-2458-12-701. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Kondal D, Patel SA, Ali MK, et al. Cohort profile: the center for cardiometabolic risk reduction in South Asia (CARRS). Int J Epidemiol Feb 9 2022. 10.1093/ije/dyac014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].The CARRS Cohort (https://www.carrsprogram.org/overview). Access Date (4/28/2022). [Google Scholar]
- [21].Kondal D, Awasthi A, Patel SA, et al. Evaluating bias with loss to follow-up in a community-based cohort: empirical investigation from the CARRS Study. J Epidemiol Community Health. Jan 10 2024;doi: 10.1136/jech-2023-220963. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Ghaferi AA, Schwartz TA, Pawlik TM. STROBE reporting guidelines for observational studies. JAMA Surg 2021;156(6):577–8. 10.1001/jamasurg.2021.0528. [DOI] [PubMed] [Google Scholar]
- [23].Deepa M, Grace M, Binukumar B, et al. High burden of prediabetes and diabetes in three large cities in South Asia: the Center for cArdio-metabolic Risk Reduction in South Asia (CARRS) study. Diabetes Res Clin Pract Nov 2015;110(2):172–82. 10.1016/j.diabres.2015.09.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Association AD. 2. Classification and diagnosis of diabetes: standards of medical care in diabetes—2018. Diabetes Care 2018;41(Supplement_1):S13–27. [DOI] [PubMed] [Google Scholar]
- [25].Eddings W, Marchenko Y. Diagnostics for multiple imputation in Stata. Stata J 2012;12(3):353–67. [Google Scholar]
- [26].Austin PC, Fine JP. Practical recommendations for reporting fine-gray model analyses for competing risk data. Stat Med 2017;36(27):4391–400. 10.1002/sim.7501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Mathers CD, Salomon JA, Ezzati M, Begg S, Hoorn SV, Lopez AD. Sensitivity and uncertainty analyses for burden of disease and risk factor estimates. In: Lopez AD, Mathers CD, Ezzati M, editors. Global burden of disease and risk factors. New York: Oxford University Press; 2006. chap 5. [PubMed] [Google Scholar]
- [28].Murray CJL, Lopez AD, Jamison DT. The global burden of disease in 1990: summary results, sensitivity analysis and future directions. Bull World Health Organ 1994;72(3):495–509. [PMC free article] [PubMed] [Google Scholar]
- [29].Murray CJL, Acharya AK. Understanding DALYs. J Health Econ 1997;16:703–30. [DOI] [PubMed] [Google Scholar]
- [30].Plana-Ripoll O, Canudas-Romo V, Weye N, Laursen TM, McGrath JJ, Andersen PK. Lillies: an R package for the estimation of excess Life Years lost among patients with a given disease or condition. PLoS One 2020;15(3):e0228073. 10.1371/journal.pone.0228073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Sargeant JA, Brady EM, Zaccardi F, et al. Adults with early-onset type 2 diabetes (aged 18–39 years) are severely underrepresented in diabetes clinical research trials. Diabetologica 2020;63:1516–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Today Study Group. A clinical trial to maintain glycemic control in youth with type 2 diabetes. N Eng J Med 2012;366(24):2247–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Today Study Group. Long-term complications in youth-onset type 2 diabetes. N Engl J Med 2021;385:416–26. 10.1056/NEJMoa2100165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].The Rise Consortium. Impact of insulin and metformin versus metformin alone on b-cell function in youth with impaired glucose tolerance or recently diagnosed type 2 diabetes. Diabetes Care 2018;41(August):1717–25. 10.2337/dc18-0787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [35].Zoungas S, Woodward M, Li Q, et al. Impact of age, age at diagnosis and duration of diabetes on the risk of macrovascular and microvascular complications and death in type 2 diabetes. Diabetologia 2014;57(12):2465–74. 10.1007/s00125-014-3369-7. [DOI] [PubMed] [Google Scholar]
- [36].Zhang H-J, Feng J, Zhang X-T, Zhang H-Z. Age at type 2 diabetes diagnosis and the risk of mortality among US population. Sci Rep 2024;14(1). 10.1038/s41598-024-80790-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data supporting this study’s findings are available to the corresponding author (JR: ram.jagannathan@emory.edu) upon reasonable request. The R codes and the R notebook for the reproducible analysis is available to the interested readers by contacting: asogunt@emory.edu.