We conducted a global analysis of T2D attributable to behavioral risks using data from the Global Health Data Exchange query tool (GHDx, https://ghdx.healthdata.org/gbd-results-tool). Our metrics of interest included deaths, disability-adjusted life years (DALYs), the age-standardized mortality rate (ASMR), and the age-standardized disability-adjusted life years rate (ASDR), expressed per 100,000 people. The dataset covered various demographic groups across 204 countries, 21 global burden of disease regions, and 5 socio-demographic index (SDI) regions from 1990 to 2021. The SDI is a composite indicator reflecting a country’s development level. A higher SDI value, ranging from 0 to 1, indicates greater socio-economic development [16]. Countries and regions were classified into 5 SDI groups: High, High-middle, Middle, Low-middle, and Low SDI.

T2D is a chronic condition occurring primarily in adults, characterized by insulin resistance or pancreatic insulin production deficiency. The GBD defines T2D as having a fasting plasma glucose level of ≥7 mmol/L (126 mg/dL) or currently taking drugs or insulin therapy. According to the International Classification of Diseases (ICD), Ninth Revision, the code directly assigned to T2D deaths is 250, and the Tenth Revision code is E11 [17].

Data on behavioral risks were extracted from the GHDx following the definitions and classifications applied in the GBD 2021 study. The GBD study applies a hierarchical framework of risk factors, allowing both specific risks (e.g., Diet low in fruits, Diet low in vegetables) and aggregated categories (e.g., Dietary risks) to be evaluated simultaneously. The four behavioral risk categories for T2D included tobacco, high alcohol use, dietary risks, and low physical activity. In order to quantify the burden attributable to each behavioral risk, exposures were assessed relative to a theoretical minimum risk exposure level (TMREL)—the counterfactual level of exposure that would minimize the risk of diabetes, according to epidemiological evidence [4, 15]. On this basis, tobacco included both active smoking and exposure to secondhand smoke, with the TMREL defined as lifelong non-use. High alcohol use was defined as alcohol consumption in excess of the TMREL, the level of alcohol consumption at which all-cause risk is minimized [18]. However, the contribution of each risk factor to overall health loss varies by geography, age, period, and gender, indicating that the amount of alcohol that minimizes health loss also differs across these dimensions (Supplementary Appendix S3). Dietary risks were subdivided into 7 components, each with a TMREL defined according to the GBD 2021 criteria. Protective factors included intake above the TMREL, such as fruits (≥340–350 g/day), vegetables (≥306–372 g/day), whole grains (≥160–210 g/day), and dietary fibre (≥22–25 g/day). Harmful factors were defined as intake below the TMREL, including processed meat (0 g/day) and sugar-sweetened beverages (0 g/day). A J-shaped association was observed between red meat intake and T2D, with the lowest risk occurring at 0–200 g/day. Low physical activity was assessed using metabolic equivalent minutes per week (MET-min/week) across leisure, occupational, household, and transport domains. The TMREL was set at 3,600–4,400 MET-min/week, corresponding to the level associated with the lowest risk of adverse outcomes. Detailed descriptions of the risk factor estimation methodology are provided in Supplementary Appendix S4.

To appropriately measure the burden of T2D, we utilized the count of deaths, DALYs, ASMR, and ASDR attributed to behavioral risks with 95% uncertainty intervals (95% UIs). The 95% UI refers to the 2.5th and 97.5th percentiles of the 1,000 draw-level estimates for each parameter [19]. Countries and territories were divided into 21 GBD regions based on geographical similarities and were grouped into 5 SDI regions depending on their SDI levels. Age-standardized rates (ASR) were used to estimate the T2D burden attributable to behavioral risks at the global, regional, national, and territorial levels in 1990 and 2021. We calculated the Average Annual Percentage Changes (AAPCs) using the Joinpoint regression model, where the number of joinpoints was determined by the Permutation Test with a default maximum number of three [20]. The model calculations are as follows: ln y = α + β i x + ε Y = exp ∑ w i β i ∑ w i − 1 × 100

In the above equations, y represented ASMR or ASDR, x was the calendar year, β_i indicated the slope coefficients of each segment in the expected year range, Y referred to AAPCs and w_i was the length of each segment in the year range [21]. An increasing trend in ASMR or ASDR was indicated when both the AAPCs and its 95% confidence interval (95% CI) were greater than 0, and a decreasing trend when both were below zero. If the 95% CIs of AAPCs included 0, the ASMR and ASDR were considered stable.

Necessary stratified analyses by age and gender were additionally employed to identify sensitive populations for T2D attributable to behavioral risks. Restricted cubic spline (RCS) regression models were further performed to explore changes of ASMR and ASDR in 21 regions with different SDI levels from 1990 to 2021. The association of AAPCs with SDI in 2021 for ASMR and ASDR at the national level was further examined using locally estimated scatterplot smoothing (LOESS) regression.

To project future trends in T2D-related deaths and DALYs up to 2050, we utilized Bayesian age-period-cohort (BAPC) analysis, a widely applied approach in epidemiology for forecasting disease burden based on historical patterns. The BAPC model integrates both prior knowledge and observed data to provide probabilistic forecasts of disease burden, with particular attention to uncertainties [22]. Using data from 1990 to 2021, we trained the model and produced forecasts for 2050, with separate trend analyses for males and females. Uncertainty intervals were calculated using Bayesian inference, providing robust estimates for future trends [23, 24].

Data management and analyses were performed using Joinpoint Regression Program (version 5.2.0) and R Statistical Software (version 4.4.1).

Globally, deaths of T2D attributable to behavioral risks grew significantly (AAPCs = 2.75, 95% CI: 2.58−2.92, P < 0.001), increasing from approximately 267,955 cases in 1990 to 626,661 cases in 2021, while the corresponding ASMR did not change much, with AAPCs of 0.06 (95% CI: −0.06−0.17, P > 0.05). Furthermore, the DALYs of T2D attributable to behavioral risks also showed a substantial rise (AAPC = 3.47, 95% CI: 3.40−3.55, P < 0.001), increasing from 10.49 million in 1990 to 30.19 million in 2021. In addition, the ASDR of T2D increased from 260.56 in 1990 to 348.52 per 100,000 people in 2021, with AAPCs of 0.96 (95% CI: 0.92−1.00, P < 0.001) (Table 1; Supplementary Table S1).

In terms of 21 GBD regions, Southern Sub-Saharan Africa recorded the highest ASMR (31.02 per 100,000 people, 95% UI: 17.42–41.44) and ASDR (865.09, 95% UI: 479.81−1,163.58 per 100,000 people). High-income North America had a lower ASMR (5.99 per 100,000 people, 95% UI: 3.21–7.88) but still bore a high ASDR burden (458.82 per 100,000 people, 95% UI: 236.46–662.81). Eastern Europe showed the greatest increases in both ASMR (AAPCs = 3.57, 95% CI: 2.77−4.37, P < 0.001) and ASDR (AAPCs = 2.37, 95% CI: 2.10−2.64, P < 0.001) (Table 1).

At the national level, Fiji had the highest ASMR (107.66 per 100,000 people) and ASDR (3,061.31 per 100,000 people) in 2021 (Figures 1A,B). Russia had the fastest ASMR increase (AAPC = 4.52), while Guatemala showed the largest increase in ASDR (AAPC = 4.02) (Figures 1C,D). India had the highest number of deaths, reaching 109,041.84, while China recorded the highest DALYs, amounting to 4,875,274.21 (Supplementary Figure S1).

Our analysis revealed dietary risks were the predominant modifiable factors associated with T2D globally, with ASMR of 4.52 (95% UI: 0.88−7.36) and ASDR of 221.34 (95% UI: 47.97–368.92) per 100,000 people (Table 2). Specifically, the highest ASMR related to dietary risks was observed in Oceania (26.27 per 100,000 people, 95% UI: 4.25–44.14). Similarly, the highest ASDR was found in Oceania (862.05 per 100,000 people, 95% UI: 154.31–1,465.63) (Figure 2; Supplementary Table S2). Dietary risks, which encompassed seven secondary behavioral risks in the GBD study, showed significant regional variations in their impact on T2D. Globally, high consumption of processed meat, along with low intake of whole grains, were the top three dietary risk factors (Supplementary Figure S2).

Tobacco emerged as the second leading behavioral risk factor for T2D, with a global ASMR of 1.91 and ASDR of 102.08 per 100,000 people (Table 2). The highest ASMR (13.52 per 100,000 people, 95% UI: 7.63–19.49) and ASDR (487.53 per 100,000 people, 95% UI: 290.44–695.67) for T2D was observed in Oceania (Figure 2; Supplementary Table S2). The risks of tobacco are made up of two main components, smoking and secondhand smoke. These factors showed similar geographic patterns, with the highest ASMR and ASDR in Oceania, Southern Sub-Saharan Africa, Caribbean and South Asia (Supplementary Figure S2).

While the overall burden of T2D attributable to high alcohol use was comparatively lower, there existed a notable increase in ASDR (AAPCs = 1.83, 95% CI: 1.68−1.97, P < 0.001) (Table 2). Southern Sub-Saharan Africa (1.73 per 100,000 people, 95% UI: 0.62–3.29) exhibited the highest ASMR for T2D linked to excessive alcohol intake. When considering ASDR, Southern Sub-Saharan Africa (47.59 per 100,000 people, 95% UI: 14.51–94.26), Central Latin America (42.99 per 100,000 people, 95% UI: 8.48–84.04), and High-Income North America (37.04 per 100,000 people, 95% UI: 7.55–74.61) were detected as the regions most affected by alcohol-related T2D (Figure 2; Supplementary Table S2).

Low physical activity was another key risk factor. Southern Sub-Saharan Africa had the highest ASMR (9.89 per 100,000 people, 95% UI: 4.18–15.08) and ASDR (239.25 per 100,000 people, 95% UI:102.53–361.94), followed closely by Oceania (ASMR: 8.68 per 100,000 people, 95% UI: 3.61−13.94; ASDR: 268.48 per 100,000 people, 95% UI:110.53–431.69) (Figure 2; Supplementary Table S2).

In 2021, T2D deaths attributable to behavioral risks varied by age, peaking in males aged 70–74 and females aged 80–84 (Figure 3A; Supplementary Table S3). The age-specific mortality rate increased exponentially with age, with higher rates observed in males than in females from age 30 onward (Figure 3B; Supplementary Table S4). The age groups most affected by DALYs due to T2D were males at ages 60–64 and females at ages 65–69, with total DALYs of 16.12 million for males and 14.07 million for females (Figure 3C; Supplementary Table S3). Consistent with the age-specific mortality rate, males showed higher age-specific DALYs rate than females, with rates increased substantially with age, except for males in the 90–94 age group (Figure 3D).

Gender differences in major behavioral risks were evident: males had higher exposure to tobacco and high alcohol use, while females showed lower levels of physical activity (Figure 3; Supplementary Table S5). In contrast, dietary risks were identified as the leading behavioral risk factor for T2D in both genders.

From 1990 to 2021, deaths and DALYs of T2D attributable to behavioral risks increased globally and across five SDI regions, highest in middle SDI regions (Supplementary Figures S3A, B). Low-middle and low SDI regions had higher ASMRs and ASDRs than the global average (Supplementary Figures S4A, B).

In the 21 GBD regions, T2D deaths and DALYs from behavioral risks followed an inverted U-shaped trend with SDI, peaking at 0.59 before declining (Supplementary Figures S5A, B). It is worth noting that Southern Sub-Saharan Africa and Oceania had very high ASMR, with a significant increase between 1990 and 2021 (Supplementary Figures S6A). Oceania had a particularly high ASDR of 1,418.20 (95% UI: 783.5–1946.81) per 100,000 in 2021 and demonstrated a rapid upward trend (AAPC = 0.52, 95% CI: 0.48–0.57), surpassing regions with similar SDI (Supplementary Figure S6B; Table 1).

At the national level, a negative correlation between AAPCs and SDI was observed (ASMR: r = −0.503, P < 0.001; ASDR: r = −0.274, P = 0.002) when SDI was ≥0.53. However, the association weakened when the SDI fell below 0.53 (ASMR: r = 0.102, P = 0.470; ASDR: r = 0.105, P = 0.458) (Supplementary Figure S7).

Over the next 30 years, the projections indicate that the ASMR for both males and females will remain relatively stable (Figures 4A,B). In contrast, the projected ASDR has a substantial increase for both genders, with a more pronounced rise in males. Specifically, the ASDR of T2D attributable to behavioral risks will increase from 392.30 (95% CI: 392.11–392.49) in 2021 to 684.87 (95% CI: 207.38–1,162.37) per 100,000 people by 2050 in males, and from 309.80 (95% CI: 309.64–309.96) to 493.27 (95% CI: 191.90–794.63) per 100,000 people in females (Figures 4C,D). We also projected the global deaths and DALYs from 2022 to 2050. By 2050, the number of DALYs is expected to increase to 32.71 million in males and 23.56 million in females, highlighting a substantial disease burden (Supplementary Figure S8).

The present study reveals a marked escalation in the global burden of T2D attributable to behavioral risks, with a particularly pronounced increase in ASDR. Males, older adults, and populations in low and low-middle SDI regions are disproportionately affected. Among the risk factors identified, dietary risks contribute the most, while high alcohol use exhibits the steepest increase. By identifying the most influential behavioral risks, our study provides evidence to guide future research and policy formulation, underscoring the need for targeted interventions as well as comprehensive T2D prevention and control strategies.