The Global Burden of Disease (GBD) 2021 database, maintained by the Institute for Health Metrics and Evaluation (IHME), provides comprehensive data on health loss for 369 diseases, injuries, and impairments, as well as 87 risk factors across 204 countries and territories (4). The dataset spans the years 1990 to 2021 and integrates information from diverse sources, including vital registration systems, population surveys, health records and epidemiological studies. In this study, the estimates and their 95% uncertainty interval (UI) for disability-adjusted life-years (DALYs), deaths, prevalence, and incidence of drug use disorders (DUDs) were downloaded from GBD 2021. Social Demographic Index (SDI), constructed from three key indicators, including income per capita, educational attainment and total fertility rate, was used to examine health inequalities, optimize resource allocation, and ultimately improve population health across diverse global contexts (11). The SDI ranges from 0 to 1, with higher value meaning a higher level of socioeconomic development.

Joinpoint regression analysis was applied to identify significant temporal trends in the Age Standardized Rates (ASRs) of DUDs from 1990 to 2021. This technique fits a series of joined straight lines, or segments, to the data, allowing for changes in the slope at specific points, called “joinpoints.” A log-linear model was selected, as the outcome variables (DALYs, deaths, prevalence, and incidence) are typically right-skewed and better conform to a Poisson-like distribution. The grid search method was used to identify the optimal number and location of joinpoints. The maximum number of joinpoints was set to 5, consistent with the software’s default recommendation. Model selection was performed using the Monte Carlo permutation test with 4,499 permutations and overall significance level of α = 0.05. To account for multiple testing across DUDs subtypes, the False Discovery Rate (FDR) correction was applied. The heteroscedastic error option was set to “Standard Error (Calculated)” to accommodate potential variability in the ASRs across years. For each segment between joinpoints, the Annual Percent Change (APC) and its 95% confidence interval (CI) were calculated. To summarize the overall trend across the entire study period, the Average Annual Percent Change (AAPC) was computed as a weighted average of the APCs (12). The values in both APC and AAPC greater than 0 indicate an upward, meaning the measured indicator is increasing over time on a yearly basis, while the values less than 0 indicate a downward trend, meaning the decreasing over time. The analysis was conducted using the Joinpoint Regression Program 4.9.1.0, developed by the National Cancer Institute (NCI).

The Nordpred statistical tool was employed to predict the rates of burden of diseases and the number of new cases. Data for four measures were categorized into 5-year age groups and 5-year calendar periods from 1990 to 2021, enabling a detailed analysis of trends across different age cohorts over time. ASRs were calculated using the world standard population to ensure comparability across populations and over time. Nordpred uses a log-linear age-period-cohort (APC) model, which considers the effects of age, period (calendar time), and cohort (birth cohort) on burden of disease, to extrapolate recent trends using a power-5 link function to moderate growth rates and mitigate the risk of overestimating future burden (13). In our study, we analyzed the global burden of DUDs for 5 years (1992–1996, 1997–2001, …, and 2017–2021) and the prediction was conducted in the 5 years (2022–2026, 2027–2031, …, 2042–2046).

This analysis, aligned with the World Health Organization’s framework for health inequality monitoring, is used to measure, analyze, and address health inequalities, with a special focus on low- and middle-income countries (14). The Slope Index of Inequality (SII) is a measure of absolute inequality that captures the gradient in health outcomes across populations ranked from the most disadvantaged to the most advantaged (15). To calculate the SII, country-level data were ranked based on their SDI value from the GBD 2021 database. A weighted least squares regression was then performed, using the health outcome of interest as the dependent variable and the fractional rank as the independent variable (16). Population weights were applied to account for the size of country’s population. A higher value of SII indicates greater inequality, with a steeper gradient between the least and most advantaged countries. Concentration Index (CI) is a measure used to assess relative health inequality across different SDI countries, which are assigned a cumulative proportion of the global population, ranging from the poorest to the richest (17). The value of CI is computed as twice the area between the concentration curve and the line of equality (the 45 degrees line from the bottom left to the top right of the graph), ranging from −1 to 1, with a negative value indicates that countries with low SDI level are disproportionately affected by disease burden, while a positive value indicates that countries with high SDI level are disproportionately affected.

The rate of DALYs attributed to DUDs increased notably between 1990 and 2021. For males, DALYs rose from 198.52 (UI: 161.30–234.34) to 235.88 (UI: 199.50–270.25), while for females, it increased from 133.73 (UI: 102.80–161.92) to 145.25 (UI: 113.42–175.63) (Figure 1A). A similar upward trend was observed in deaths related to DUDs, with male deaths rising from 1.74 (UI: 1.58–1.92) to 2.37 (UI: 2.24–2.51), and female deaths increasing from 0.79 (UI: 0.70–0.89) to 0.93 (UI: 0.85–1.01) (Figure 1B). Despite these trends in DALYs and deaths, no significant increase was observed in the rates of prevalence and incidence, although males continue to shoulder a disproportionately higher burden of DUDs compared to female (Figures 1C,D). When analyzing age-specific distributions, individuals aged 25–29 years consistently bore the highest DALYs in both 1990 and 2021 (1990: 424.88, UI: 324.73–521.56; 2021: 460.35, UI: 363.92–561.78) (Figure 1E). This was followed by those aged 20–24 and 30–34 years, indicating that the burden of DUDs is most prevalent among younger adults. As for deaths, the highest rates were observed in individuals aged 95 and older (1990: 4.16, UI: 3.22–4.73; 2021: 6.43, UI: 4.82–7.53), although younger age groups also exhibited a significant burden (Figure 1F). Overall, from 1990 to 2021, the number of DALYs, deaths, prevalence, and incidence attributed to DUDs has risen (Figure 1G). However, the male population consistently showed higher rates compared to females across all four measures.

In 2021, the United States (1944.08, UI: 1632.99, 2249.41), Canada (877.19, UI: 759.19, 998.21), and Estonia (733.92, UI: 586.79, 889.92) had the highest rates of DALYs attributed to DUDs (Figure 2A). Conversely, the countries with the lowest DALYs rates were Mali (36.87, UI: 25.47, 48.51), Guinea (35.54, UI: 25.19, 46.40) and Nigeria (33.93, UI: 24.41, 43.68) (Figure 2A). For deaths rates, Palestine (0.03, UI: 0.02, 0.03), Ghana (0.02, UI: 0.01, 0.03) and Palau (0.01, UI: 0.01, 0.01) recorded the lowest figures, while the United States (19.52, UI: 17.73, 21.61), Canada (8.85, UI: 8.08, 9.60), and Iceland (5.26, UI: 4.59, 5.93) were the top (Figure 2B). Additionally, the United States, Canada and United Kingdom were the top three countries with the highest prevalence rates of DUDs, and the United States, Australia, and Canada recorded the highest incidence rates of DUDs (Figures 2C,D). At the opposite end, Burkina Faso (228.58, UI: 193.64, 278.67), Nigeria (224.12, UI: 190.37, 260.09) and Togo (209.86, UI: 180.47, 248.47) had the lowest prevalence rates, while Togo (91.32, UI: 72.08, 107.44), Niger (90.93, UI: 75.51, 108.49) and Kenya (86.25, UI: 71.30, 101.66) showed the lowest incidence rates. In all analyses of DUDs burden, regions with higher SDI levels had the highest rates, while regions with lower SDI levels consistently had the lowest rates (Table 1).

The study evaluated the change in rates of global burdens of DUDs from 1990 to 2021, and further included subgroup analysis of amphetamine, cannabis, cocaine, opioid, and other drug use disorders to estimate their contributions to overall DUDs burden. The analysis revealed an increasing trend in DALYs for cocaine use disorders (AAPC, 0.8), DUDs (AAPC, 0.4), and opioid use disorders (AAPC, 0.9) (Figure 3A). In contrast, amphetamine use disorders (AAPC, −1.1) and other drug use disorders (AAPC, −0.8) showed a declining trend. Similar patterns were observed in the analysis of deaths rates (Figure 3B). In terms of prevalence, amphetamine use disorders (AAPC, −1.5), cannabis use disorders (AAPC, −0.2), cocaine use disorders (AAPC, −0.2), DUDs (AAPC, −0.2), and other drug use disorders (AAPC, −0.1) exhibited a decreasing trend, while opioid use disorders (AAPC, 0.8) showed the faster growth (Figure 3C). Furthermore, the incidence of opioid use disorders (AAPC, 0.1) also increased, while the incidence of amphetamine use disorders (AAPC, −1.6), cannabis use disorders (AAPC, −0.2), cocaine use disorders (AAPC, −0.2), DUDs (AAPC, −0.3) and other drug use disorders (AAPC, −0.2) demonstrated a declining trend (Figure 3D). The detail of optimal joinpoints and corresponding APCs were shown in Table 2.

The average annual all-age number of DALYs is projected to increase by 12.9 million from the 1992–1996 period to the 2042–2046 period (Figure 4A). The gap between males and females has also widened during this period. From 1990 to 2021, the age-standardized DALYs rate (ASDR) of DUDs showed an increasing but fluctuating trend; however, this increase is expected to accelerate from 2021 to 2044 (Figure 4B). A similar pattern was observed in the analysis of deaths, with an even wider gender gap (Figures 4C,D). As for prevalence, the average annual all-age number is expected to rise by 23 million from the 1992–1996 period to the 2042–2046 (Figure 4E). However, the age-standardized prevalence rate (ASPR) of DUDs decreased slowly from 1990 to 2021, which is predicted to grow from 2021 to 2044 (Figure 4F). Additionally, the increase in the average annual all-age number of incidence from the 2042–2046 period to the 2017–2021 period is nearly 0.76 times to that seen from the 2017–2021 period to the 1992–1996 period, indicating a slowly increase trend (Figure 4G). Of note is that age-standardized incidence rate (ASIR) of DUDs is predicted to decrease from the 2022–2026 period to the 2042–2046 period (Figure 4H).

The analysis revealed notable changes in both absolute and relative health inequalities across the years, as indicated by shifts in the SII and CI values. In 1990, the distribution of DALYs across SDI rankings were more concentrated in countries with higher SDI (Figure 5A). The value of SII was 82.40, which means that there was an estimated burden of 82.40 (per 100,000 population) DALYs between country with the highest SDI and the lowest SDI in 1990, and this gap further amplified to 289.24 in 2021. Moreover, the value of CI also increased from 1990 to 2021, which means the relative health equalities are more disproportionately concentrated in countries with higher SDI (Figure 5B). However, no significant trend was observed in the analysis of deaths burden (Figures 5C,D). In the analyses of prevalence and incidence, despite the burdens were likely centered on countries with higher SDI, this increased from 459.48 in 1990 to 675.60 in 2021 for prevalence, and from 86.11 in 1990 to 110.74 in 2021 for incidence (Figures 5E,G). Meanwhile, no significant changes in relative SDI-related health inequalities were found for prevalence and incidence (Figures 5F,H).

One of the primary strengths of this study lies in its utilization of the Global Burden of Disease (GBD) database, which offers comprehensive, standardized data on the burden of DUDs across 204 countries over a nearly three-decade period. This provides a robust foundation for identifying trends at the multiple levels, including sex, age, and SDI, and allows for both global and regional comparisons. Another strength is the inclusion of cross-country inequality analyses, which reveal disparities in the distribution of DUDs-related burden and underscore the need for targeted policies in both high- and low-SDI regions. Finally, by incorporating multiple metrics-DALYs, deaths, prevalence, and incidence, the study provides a comprehensive assessment of the multifaceted impact of DUDs on global health. However, several limitations must be acknowledged. First, despite the comprehensiveness of the GBD database, data quality and availability vary significantly across countries, particularly in low-SDI countries. This may result in underestimation or inaccurate reporting of DUDs-related burdens, as many countries lack robust healthcare infrastructure and data collection systems. Second, although we employed standard measure of inequality (SII and CI), we did not perform sensitivity analyses, such as using alternative SDI vintages or conducting regional sub-group rankings. Third, the future projections from the Nordpred model, while useful for indicating trends, are presented without prediction intervals and were not internally validated through back-testing. Finally, the study focuses primarily on aggregate, population-level data, which may obscure important within-country disparities, such as those related to socioeconomic status, urban–rural divides, and specific drug types. Further research incorporating individual-level data is needed to provide a more nuanced understanding of the factors driving DUDs and their health impacts.