Since 1990, the Global Burden of Disease (GBD) database has systematically collected and integrated epidemiological data from 21 GBD regions and 204 countries and territories (11). The GBD 2021 study provides the most recent analysis of epidemiological data for 371 diseases and injuries and 88 risk factors (12). All required data were extracted from the Global Health Data Exchange (GHDx) query tool.¹ Our dataset included sex-, age-, incidence-, and disability-adjusted life year (DALY)-specific estimates related to mental disorders across 21 GBD regions and 204 countries and territories.

The SDI is a composite indicator reflecting societal development levels and is strongly correlated with health outcomes (13). It is calculated as the geometric mean of three components: total fertility rate among individuals under 25 years, mean educational attainment among those aged 15 and older, and lag-distributed income per capita, scaled between 0 and 1 (14). In the GBD 2021 study, countries and territories were categorized into five SDI quintiles: low, low-middle, middle, high-middle, and high SDI. To assess temporal trends in health inequality, we compared data from 204 countries and territories between 1990 and 2021.

To mitigate bias and heterogeneity, we employed robust regression models (RLM) instead of ordinary linear regression (LM) for health inequality analyses. Robust regression reduces sensitivity to outliers, minimizes bias from data heterogeneity or extreme values, and provides a more accurate representation of health disparities (15). Additionally, the concentration index was calculated by matching cumulative proportions of incidence and DALYs with cumulative population distributions ranked by SDI, followed by numerical integration of the area under the Lorenz curve.

To evaluate the relationship between the burden of mental disorders in older adults and socio-demographic development, we constructed a frontier model using SDI as the predictor and ASIR and ASDR as outcomes. Unlike traditional regression models that describe variable relationships or predict outcomes, frontier analysis employs advanced statistical techniques to account for nonlinear associations between SDI and disease burden, capturing multidimensional drivers of mental disorder burden in aging populations.

The frontier approach identifies the theoretical minimum ASR achievable for each country or territory given its current SDI level, serving as a benchmark for optimal performance. This method quantifies the gap between a country’s current burden and its potential minimum burden, highlighting areas for improvement. The optimal performance boundary was fitted using locally weighted regression (LOESS) (smoothing span tested at 0.3–0.5 for sensitivity). To ensure robustness, we performed 1,000 bootstrap iterations to generate 95% uncertainty intervals (UIs) and computed the efficiency gap (absolute distance between observed ASDR and the frontier) as an indicator of improvement potential.

We applied the BAPC model to forecast future disease burden trends. This model extends the traditional generalized linear model (GLM) framework within a Bayesian framework, dynamically integrating age, period, and cohort (APC) effects. These effects were modeled as time-varying parameters smoothed via second-order random walks, improving posterior probability estimation. A key strength of the BAPC model is its use of integrated nested Laplace approximation (INLA) for approximating marginal posterior distributions (16). This approach avoids convergence and mixing challenges typical of Markov chain Monte Carlo (MCMC) methods while maintaining computational efficiency (17). The model’s flexibility in handling time-series data makes it particularly suitable for long-term burden projections (18). Due to its comprehensive coverage and ability to capture temporal trends, the BAPC model has been widely validated in epidemiological studies, particularly those involving age-structured populations and complex cohort effects (19). In this study, we implemented the BAPC model using the “BAPC” R package, leveraging GBD 2021 data and Institute for Health Metrics and Evaluation (IHME) population projections. This approach enabled nuanced predictions of future mental disorder burden in older adults, accounting for the complex interplay of age, period, and cohort effects.

This study analyzed the burden of mental disorders among people over 60 years across age groups, sexes, years (1990–2021), and geographic locations using Global Burden of Disease (GBD) 2021 data. ASIR and ASDR per 100,000 population were calculated with 95% uncertainty intervals (UIs) using GBD’s standardized demographic methods. Temporal trends were assessed via EAPC, derived from a log-linear regression model: ln(ASR) = α + β × calendar year + ε, where EAPC = 100 × (exp(β) − 1). Trends were classified as increasing (EAPC and 95% confidence interval [CI] lower bound >0), decreasing (EAPC and 95% CI upper bound <0), or stable (95% CI spanning 0). Socioeconomic patterns were evaluated by stratifying countries into quintiles based on the SDI (SDI; range: 0–1). Correlation analyses (Pearson’s r) examined associations between ASR, SDI, and EAPC. Frontier analysis identified optimal achievable ASR-SDI relationships. All statistical analyses and data visualizations were performed using R (version 4.4.2) and JD_GBDR (V2.37, Jingding Medical Technology Co., Ltd.). In this study, the R software package (version 4.2.3) and JD_GBDR (V2.22, Jingding Medical Technology Co., Ltd.) was used for the drawing of the figures. Uncertainty was propagated throughout all calculations using 1,000 posterior draws from GBD’s cause-specific mortality and disability models.

In 2021, there were 74.9 million (95% UI: 59.8–94.2 million) new cases of mental disorders among people over 60 years globally, with an ASIR of 6,867.63 per 100,000 population (95% UI: 5,470.33–8,645.08). Although the absolute number of cases increased significantly compared to 1990, the rise in ASIR was minimal (Table 1; Figure 1A). From 1990 to 2021, the EAPC in ASIR was 0.01 (95% CI: −0.05 to 0.07), indicating a stable trend.

Regionally, Central Sub-Saharan Africa had the highest ASIR (13,024.14 per 100,000; 95% UI: 9,381.3–17,860.35), nearly three times that of the lowest-ranking region, Oceania (4,643.89; 95% UI: 3,304.71–6,415.88) (Table 1; Figure 2A). The most pronounced decline in ASIR was observed in Eastern Europe (EAPC: −0.58; 95% CI: −0.7 to −0.45), whereas East Asia exhibited the fastest increase (EAPC: 0.55; 95% CI: 0.43–0.67) (Table 1; Figure 3A). When stratified by SDI, low-SDI regions had the highest ASIR (10,748.45; 95% UI: 8,232.15–13,994.39) but also the largest decline (EAPC: −0.22; 95% CI: −0.31 to −0.12). In contrast, middle-high SDI regions had the lowest ASIR (6,805.33; 95% UI: 5,391.03–8,525.27) and showed a moderate decline (EAPC: −0.18; 95% CI: −0.23 to −0.13) (Table 1; Figure 3A). These findings highlight persistent disparities, with low-resource regions bearing the highest burden, while the rising incidence in East Asia may be linked to demographic shifts and disparities in healthcare access.

In 2021, the global age-standardized DALY rate (ASDR) due to mental disorders among individuals aged 60 + years was 2,095.43 per 100,000 population (95% UI: 1,537.49–2,718.99), reflecting a slight increase from 1990 (2,003.53; 95% UI: 1,464.06–2,598.96). However, the EAPC was 0 (95% CI: −0.04 to 0.05), suggesting an overall stable burden (Table 2; Figures 1B, 2B, 3B). Among the 21 GBD regions, Central Sub-Saharan Africa had the highest ASDR (2,681.48; 95% UI: 1,861.61–3,601.57), while the high-income Asia Pacific region had the lowest (1,487.41; 95% UI: 1,113.79–1,869.56). The largest increase in EAPC was observed in the high-income Asia Pacific region (0.20; 95% CI: 0.16–0.23), whereas Eastern Europe experienced the most significant decline (−0.25; 95% CI: −0.31 to −0.18) (Table 2; Figures 2B, 3B). When stratified by SDI, low-SDI regions had the highest ASDR (2,423.86; 95% UI: 1,726.38–3,173.06), while high-SDI regions had the lowest (1,934.16; 95% UI: 1,429.33–2,502.99). The EAPC analysis revealed a notable increase in middle-SDI regions (0.09; 95% CI: 0.03–0.15) and a marked decline in high-middle SDI regions (−0.07; 95% CI: −0.11 to −0.04) (Table 2; Figure 3B). These results underscore significant geographic disparities, with low-SDI regions and aging populations remaining key intervention priorities, while high-income regions may have achieved effective disease burden control through optimized healthcare systems.

Across all age groups, the burden of mental disorders was consistently higher in females than males. The ASIR was 7,483.7 per 100,000 (95% UI: 6,004.6–9,367.12) in females compared to 5,181.55 (95% UI: 4,155.14–6,507.66) in males (Table 1). Similarly, the ASDR was 2,198.26 (95%UI: 1,593.36–2,871.1) in females versus 1,760.22 (95% UI: 1,305.97–2,255.15) in males (Table 2). While incidence trends remained stable for both sexes (female EAPC: 0.01; 95% CI: −0.05 to 0.08; male EAPC: 0.06; 95% CI: 0–0.12) (Table 1), DALY rates exhibited divergent patterns: females showed a slight decline (−0.02; 95%CI: −0.07 to 0.04), whereas males experienced a significant increase (0.07; 95%CI: 0.04–0.09) (Table 2). These findings highlight the urgent need for sex-specific interventions, particularly to address the growing disease burden among aging males, while continuing to manage the persistently higher absolute burden in females.

Figure 4 and Supplementary Figure S1 illustrate the relationship between the ASIR and ASDR of mental disorders among the people over 60 years, and the SDI globally, regionally, and nationally in 2021. Figure 4A shows a significant negative correlation between ASIR and SDI (r = −0.561, p < 0.001), indicating that regions with lower SDI have higher ASIR. Figure 4B shows a significant negative correlation between ASDR and SDI (r = −0.477, p < 0.001), similarly indicating that regions with lower SDI have higher ASDR. Supplementary Figure S1 further breaks down the analysis to the national level. Supplementary Figure S1A shows a significant negative correlation between ASIR and SDI (r = −0.522, p < 0.001), and Supplementary Figure S1B shows a significant negative correlation between ASDR and SDI (r = −0.335, p < 0.001). These results suggest that the burden of mental disorders is more severe in regions with lower SDI, highlighting the need to develop more targeted interventions for these areas.

Supplementary Figure S2 presents the relationship between the ASIR and ASDR of mental disorders among the people over 60 years in 2021, and the SDI as well as the EAPC across 204 countries globally. Supplementary Figure S2A shows a weak relationship between incidence and EAPC, with a correlation coefficient of 0.11 and a p-value of 0.11. Supplementary Figure S2B reveals a significant positive correlation between SDI and EAPC, with a correlation coefficient of 0.21 and a p-value of 0.0027, indicating that regions with higher SDI have higher EAPC. Supplementary Figure S2C shows a similarly weak relationship between DALYs and EAPC, with a correlation coefficient of 0.11 and a p-value of 0.11. Supplementary Figure S2D shows a significant positive correlation between SDI and the EAPC of DALYs, with a correlation coefficient of 0.24 and a p-value of 0.00064. These results suggest that regions with higher SDI exhibit more significant changes in the incidence of mental disorders and DALYs, indicating that the level of socio-economic development may have an important impact on the epidemiological trends of mental disorders.

Figure 5 and Supplementary Table S1, provides a detailed analysis of the absolute and relative health inequalities in the incidence and DALYs of mental disorders among the people over 60 years globally in 1990 and 2021. Figures 5A,B show the relationship between the age-standardized rates of incidence and DALYs and the SDI, revealing health inequalities across different SDI levels through regression curves. Figures 5C,D further quantify health inequalities by displaying the cumulative burden distribution across different SDI levels using Lorenz curves for cumulative incidence and DALYs. The slope coefficients in Supplementary Table S1 reflect the trends in health inequalities. For incidence, the slope coefficient was −5587.34 in 1990 and −6047.42 in 2021, indicating that over time, the burden of incidence in low-SDI regions has relatively increased, with a worsening of absolute inequality. For DALYs, the slope coefficient was −408.64 in 1990 and −440.72 in 2021, showing an increase in the burden of DALYs in low-SDI regions, albeit with a relatively smaller magnitude of change. Relative inequalities were measured using Lorenz curves and concentration indices. The Lorenz curves in Figures 5C,D illustrate the distribution of cumulative incidence and DALYs across different SDI levels in 1990 and 2021. The 95% confidence interval (CI) for the cumulative incidence Lorenz curve in 1990 was −0.12 to −0.10, while in 2021, it was −0.13 to −0.11, indicating a higher burden of cumulative incidence in low-SDI regions, with this relative inequality persisting over time. Similarly, the 95% CI for the cumulative distribution of DALYs in 1990 was−0.03 to −0.02, and in 2021, it was −0.04 to −0.02, showing a higher burden of DALYs in low-SDI regions. By comparing the Lorenz curves and their confidence intervals between 1990 and 2021, it is evident that the burden in low-SDI regions has increased, further confirming the exacerbation of relative inequalities. These findings underscore the need for more targeted interventions in low-SDI regions to reduce health inequalities.

Figure 6 presents a frontier analysis of the relationship between the SDI and the age-standardized rate (ASR) of mental disorders among the people over 60 years in 204 countries and regions from 1990 to 2021. Figures 6A,C show the temporal trends of the ASIR and ASDR, respectively. The color gradient ranging from light green (1990) to dark green (2021) illustrates that as SDI improves, the disease burden tends to increase. The frontier plots (Figures 6B,D) reveal significant differences among countries in 2021. Low-SDI countries (blue), such as Timor-Leste and Papua New Guinea, have incidence rates close to 15,000 per 100,000 population, closely aligning with the epidemiological frontier. In contrast, high-SDI countries (red), such as Switzerland and Monaco, show a substantial gap between the observed incidence rates and the theoretically achievable incidence rates. Notably, 32% of countries exhibit an upward trend in ASIR (indicated by green dots), particularly in East Asia and Oceania, while a downward trend (indicated by orange dots) is concentrated in Eastern Europe (e.g., Lithuania) and Southern Africa. The DALY frontier (Figure 6D) shows a similar divergence, with ASDR in low-SDI countries being 2.3 times higher than in high-SDI regions. Despite an overall increase in global SDI, persistent inequalities remain in the access to mental health care for the older adults.

According to the 2021 Global Burden of Disease (GBD) data, the ASIR of mental disorders among people over 60 years shows significant variation across 204 countries globally. Uganda has the highest ASIR, at 20,513.24 per 100,000 population [95% uncertainty interval (UI): 14,496–28,497], while Singapore has the lowest ASIR, at 3,492.62 per 100,000 population (95% UI: 2,485.34–4,883.32) (Supplementary Table S3; Figure 2A). From 1990 to 2021, the EAPC in ASIR ranges from −1.47 in Singapore (95% confidence interval [CI]: −1.66 to −1.27), indicating the most significant decline, to 1.17 in Spain (95% CI: 0.89–1.45), reflecting the most significant increase (Supplementary Table S3; Figure 3A). Notably, high-income regions such as Singapore and Western Europe show declining trends, while low- and middle-income regions (e.g., sub-Saharan Africa and Latin America) exhibit stable or rising ASIR. This may be related to differences in healthcare access, diagnostic capacity, and the dynamics of the older adults. These findings highlight the need for targeted interventions to address the growing burden of mental disorders among the older adults, especially in resource-limited settings.

Similarly, the age-standardized disability-adjusted life year rate (ASDR) of mental disorders among people over 60 years also shows significant variation across the 204 countries. Uganda has the highest ASDR, at 3,762.62 per 100,000 population (95% UI: 2,519.99–5,271.42), while Brunei Darussalam has the lowest ASDR, at 1,303.47 per 100,000 population (95% UI: 958.88–1,673.47) (Supplementary Table S4; Figure 2B). From 1990 to 2021, the EAPC in ASDR ranges from −0.47 in Singapore (95% CI: −0.54 to −0.40), indicating the most significant decline, to 0.42 in South Korea (95% CI: 0.30–0.53), reflecting the most significant increase (Supplementary Table S3; Figure 3B). Notably, high-income regions such as Singapore and Western Europe (e.g., Slovenia, EAPC: −0.39) show a significant reduction in ASDR, while lower-middle-income countries, particularly in sub-Saharan Africa (e.g., Sierra Leone, EAPC: 0.28) and Southeast Asia, face an increasing burden. These differences highlight varying trends influenced by healthcare access, the older adults, and diagnostic practices, emphasizing the need for targeted mental health interventions in high-burden regions.

Based on data from the 2021 Global Burden of Disease (GBD) study and projections using the Bayesian Age-Period-Cohort (BAPC) model (Supplementary Table S2; Figure 7), the burden of mental health disorders among the global people over 60 years from 1990 to 2035 shows significant gender differences and concerning epidemiological trends. Figure 7 illustrates the projected trends of ASIR and age-standardized disability-adjusted life year rate (ASDR) of mental disorders among people over 60 years globally. From 1990 to 2035, both the combined ASIR and ASDR for both sexes are projected to increase continuously. Notably, females have significantly higher ASIR and ASDR than males across all time periods, with a more pronounced upward trend since 2010 (Figures 7C,F).

Supplementary Table S2 provides detailed year-by-year projections of ASIR and ASDR for mental disorders among the older adults from 2021 to 2035. The ASIR is projected to increase from 6,866.46 per 100,000 population in 2021 to 7,679.28 per 100,000 population in 2035. The ASDR is projected to rise from 2,093.83 per 100,000 population in 2021 to 2,494.53 per 100,000 population in 2035. For females, the ASIR is projected to increase from 8,000.19 per 100,000 population in 2021 to 8,856.09 per 100,000 population in 2035, while the ASDR is projected to rise from 2,305.82 per 100,000 population in 2021 to 2,494.53 per 100,000 population in 2035, both of which are higher than the corresponding values for males. These results highlight the increasing burden of mental health disorders with global population aging, particularly the more severe challenges faced by the female population. There is an urgent need for targeted public health interventions to address this challenge.

Supplementary Figure S4 projects that from 2021 to 2036, the ASIR of anxiety disorders will stabilize initially before rising, while depressive disorders show a continued upward trend. For bipolar disorder, ASIR is expected to increase moderately, whereas schizophrenia exhibits a slight decline. In terms of disability burden (ASDR), anxiety disorders and bipolar disorder demonstrate stable or marginally decreasing trends, while depressive disorders and schizophrenia show modest reductions, with uncertainty intervals narrowing in later years. These projections suggest divergent epidemiological patterns across mental disorders in older adults (≥60 years).

Supplementary Figure S3A shows that depressive disorders (blue section) have the highest proportion across all regions and SDI strata, especially in developed areas (such as high SDI and Western Europe) where the proportion exceeds 40%; anxiety disorders (orange section) are the second most common mental illness, with a particularly prominent proportion in high-income Asia-Pacific regions (about 35%). There are significant regional differences in the distribution of mental illness types: high SDI areas are mainly characterized by depression and anxiety disorders, while low SDI areas such as Sub-Saharan Africa have significantly higher proportions of schizophrenia (purple section) and bipolar disorder (green section). Supplementary Figure S3B shows that Anxiety disorders (red section) have the highest proportion of DALY rates among people over 60, especially in developed areas (such as high SDI, Western Europe, and high-income North America) where they generally exceed 30%, becoming the primary factor in the global burden of mental health among the older adults. Depressive disorders (dark blue section) follow closely behind, forming the main burden together with anxiety disorders, with the combined proportion of the two exceeding 50% in most regions, while the proportion of schizophrenia (purple section) is relatively higher in low-income areas (such as Sub-Saharan Africa) (about 10–15%).