According to the International Diabetes Federation Saudi Arabia is among the top five countries with diabetes mellitus (DM) with age-standardised prevalence of 23.1% 1. The national pooled prevalence of prediabetes was estimated at 24.1% (95% confidence interval (CI) 19.5%–29.4%) 2 while the pooled prevalence of type 2 diabetes mellitus (T2DM) was 16.4% (95% CI: 11.6–17.5) 3. This study aimed to assess the diabetes epidemic in Saudi Arabia through quantifying age-standardised rates and temporal trends with sex stratification, examine shifting balance between mortality and disability, and generate validated forecasts to 2030. Age-standardised rates with 95% uncertainty intervals for prevalence, incidence, mortality, disability-adjusted life years (DALYs), years of life lost (YLLs) and years lived with disability (YLDs) were extracted from the Global Burden of Disease Study 1990–2023 while United Nation World Population Prospects 2024 were used for population data. Significant trend breakpoints, with annual percentage change (APC) were calculated using Join-point regression. Decomposition analysis partitioned total burden change into three components: population growth effect, age structure effect, and rate effect. Effects of age, period and birth cohort on diabetes trends were analysed using Age-Period-Cohort analysis. We calculated The YLLs: YLDs ratio over 5-year periods to assess whether burden was driven by premature mortality or long-term disability. Autoregressive integrated moving average (ARIMA), exponential smothing state space (ETS) and Poisson forecast models were developed using expanding window cross-validation: an initial 15-year training window (1990–2004) generated one-step ahead forecasts sequentially through 2023. Forecast accuracy was assessed by root mean square error (RMSE) and mean absolute percentage error (MAPE). Using Diebold-Mariano tests, the best-performing model was selected. Sensitivity analyses examined training window size (10, 15 and 20 years), forecast horizon (1, 3 and 5-step), error metrics (RMSE, MAE, MAPE, MASE) and temporal stability (early vs. late periods). R 4.3.1 was used for doing analysis with statistical significance set at p-value < 0.05. From 1990 to 2023, Saudi Arabia experienced a 117.3% increase in age-standardised diabetes prevalence, with a consistent 2.41% average annual increase. This was accompanied by a near-doubling of incidence (+97.2%) and a 112.9% surge in DALYs. Most alarmingly, diabetes-related mortality increased by 132.5%, from 25.26 to 58.74 per 100 000. T2DM constitutes most of the burden, with prevalence increasing 118.4%. Table 1 Temporal trends demonstrated steadily increasing trajectories across all metrics, particularly from 2000 onwards. Males consistently show higher rates than females across all metrics, with the gender gap widening over time (Table S1 and Figure 1). The rate effect demonstrates that a substantial proportion of the diabetes burden in Saudi Arabia is driven by true increases in disease incidence and health loss per person, independent of population growth and ageing. For DM, changes in rates explained 38.2% of the increase in prevalence, 30.3% of the rise in DALYs, 32.2% of the increase in incidence, 8.0% of the increase in deaths and 10.1% of the growth in YLLs. For T2DM, changes in rates accounted for 38.7% of the rise in prevalence, 32.5% of incidence, 31.2% of DALYs, 38.1% of YLDs, 8.9% of deaths and 11.6% of YLLs. The rate effect for type 1 diabetes mellitus (T1DM) was −23.3% for prevalence, −133.3% for incidence, −43.2% for deaths, −31.5% for DALYs, −27.9% for YLDs and −33.9% for YLLs. (Tables S6–S8). Age–period–cohort analysis shows substantial temporal increases in diabetes burden, varying by type. DM, age-related effects ranged from 0.22 to 1.84 for prevalence and from 0.06 to 20.78 for deaths. Period rate ratios increased from 0.64 to 1.48 for prevalence and from 0.84 to 1.47 for deaths, with average APCs of +1.72% for prevalence, +1.16% for deaths and +1.65% for DALYs. Cohort effects were modest, except for incidence (0.75–2.22). T2DM showed strong age effects (0.21–1.85 prevalence; 0.04–22.50 deaths), period RRs 0.64–1.48 (prevalence) and 0.79–1.50 (deaths) and mean APCs +1.74% (prevalence) and +1.33% (deaths). Cohort effects for prevalence (0.91–1.25) suggest higher risk in later generations; deaths were stable (0.99–1.01). T1DM had narrow age effects (0.74–1.13 prevalence), modest prevalence increase (0.86–1.17, APC +0.64%), and largely stable incidence (−0.19%), deaths, and DALYs (95% CI includes 1) (Table S8). The YLLs-to-YLDs ratio for total DM decreased from 0.75 (1990–1994) to 0.55 (2010–2014), then increased to 0.72 (2020–2023) (Table S5). DM prevalence is projected to reach 27 305 per 100 000 by 2030 (95% PI: 25 220–29 391), representing a further 18.0% increase from 2023 levels. T2DM prevalence is expected to reach 27 022 per 100 000 by 2030, whereas T1DM prevalence shows relative stability at 284 per 100 000. DM incidence is forecast to reach 1014 per 100 000 (95% PI: 962–1066), a 12.8% increase from 2023. Diabetes-related mortality is forecasted to decline to 44.3 per 100 000 (95% PI: 18.6–70.1), a 24.6% decrease from 2023, whereas DALYs are projected to rise to 2668 per 100 000 (95% PI: 2159–3177), a modest 0.7% increase from 2023. (Table S3) ARIMA and ETS models both demonstrated superior forecast accuracy compared to Poisson models across all metrics (Diebold-Mariano test p < 0.05) (Table S4). The concurrent projection of falling mortality, together with increasing prevalence and DALYs, points to a major epidemiological transition from a predominantly fatal condition to one characterised mainly by chronic disability due to likely reflecting improved survival and expanded treatment coverage 4. The increase in the YLLs/YLDs ratio from 0.55 to 0.72 after 2010 reflects a worrisome rollback in mortality gains, pointing to new shortcomings in complication control. This will shift resource demand towards dialysis, retinopathy services, diabetic foot care, cardiovascular prevention and long-term pharmacotherapy. Decomposition analysis demonstrates that true epidemiological change, rather than demographic growth alone, is a primary driver of the expanding burden, explaining 38.7% of the increase in T2DM prevalence and 32.5% of incidence. This finding indicates the epidemic is not an inevitable consequence of population expansion or ageing. The pronounced birth cohort effect suggests that each successive generation faces a progressively higher lifetime risk, with cohort effect estimates for T2DM incidence ranging from 0.75 to 2.24. This suggests that individuals born more recently have a markedly greater risk of developing the disease than those from earlier cohorts, largely because they were exposed to obesogenic environments driven by rapid urbanisation, shifts in diet and declines in physical activity 5. This study revealed a speedily intensifying T2DM epidemic marked by pronounced male predominance. The 118.4% T2DM increase with higher male mortality rates (APC: +2.96%) parallels the increasing prevalence of overweight 32.8% (41% in men and 28.9% in women), obesity (23% of the population are obese, 23.1% in men and 24.2% in women) 6, reduced physical activity (64.1% of the males reported sitting time ≥ 7 h/day compared to females 52.3%), and high prevalence of smoking (16.9% smoker, 28.7% among men and 4.5% among women) 7. Additionally, fat distribution, health care utilisation and treatment adherence may contribute to this gender difference 8, 9. A possible metabolic effect of benign prostatic hyperplasia (BPH) medications represents a new biologically plausible pathway that merits further study 10. United Kingdom and Taiwan cohorts compared BPH patients receiving dutasteride or finasteride with those on tamsulosin over a 5.2-year follow-up. DM incidence was higher with 5-alpha reductase inhibitors (5ARIs) at 76/10 000 person-years than with tamsulosin (60/10 000 person-years). Risk increased with dutasteride (HR 1.32) and finasteride (HR 1.26–1.49), with no significant difference between these two drugs 11 5ARI increases local glucocorticoid exposure and impairs insulin sensitivity; consequently, it raises insulin resistance and risk of T2DM 12. Interestingly, the post-2010 reversal of the YLLs/YLDs ratio temporally corresponds with wider 5ARI use, making this a hypothesis-generating observation. In contrast, T1DM mortality decreased, mainly in females, whereas male mortality showed a slight increase (+2.9%), highlighting a sex-specific disparity. A similar trend of reduced average APC was reported in China 13 and the United States 14. This may be due to better management of the disease 15, 16. The implications of this study include targeted screening of men through mandatory annual HbA1c testing in primary care; school-based prevention programs to counteract birth-cohort effects, such as sugary drink taxes and compulsory physical education; and the establishment of specialised complication clinics, as the anticipated increase in prevalence will necessitate nurse-led services to address the expanding burden of disability. This study possesses several methodological strengths, including comprehensive analysis of GBD 1990–2023 data, rigorous statistical validation, sex-specific and type-specific stratification and sensitivity analyses. However, ecological design prevents causal inference at the individual level; forecasts assume historical trends continue and cannot incorporate future interventions; and GBD estimates mask subnational variation. The T2DM epidemic in Saudi Arabia is driven mainly by real increases in disease rates, not just demographic shifts, with marked male disadvantage. The rising YLLs/YLDs ratio after 2010 suggests mortality gains have reversed. These findings support urgent targeted screening for middle-aged men, early-life prevention that accounts for cohort effects, and health system reforms to manage the projected 2030 burden. Ramy Mohamed Ghazy: conceptualisation, methodology, software, formal analysis, investigation, data curation, writing – original draft, writing – review and editing, visualisation. Safar Abadi Alsaleem: methodology, investigation, data curation, writing – review and editing. Ayoub Ali Alshaikh: investigation, data curation, writing – review and editing. Faisal Saeed Al-Qahtani: investigation, resources, writing – review and editing. Shehata Farag Shehata: investigation, data curation, writing – review and editing. Hayfa A. AlHefdhi: supervision, resources, writing – review and editing. Asma Saad Habbash: investigation, data curation, writing – review and editing. Razan Suliman Alhumayed: investigation, data curation, writing – review and editing. Awad Alsamghan: investigation, data curation, writing – review and editing. Authors want to thank IHME for approving our request to use Saudi data. The authors extend their appreciation to the Deanship of Research and Graduate Studies at King Khalid University for funding this work through a Small Research Project under grant number RGP2/348/46, academic year 1446. The authors extend their appreciation to the Deanship of Research and Graduate Studies at King Khalid University for funding this work through a Small Research Project under grant number RGP2/348/46, academic year 1445. This study utilised aggregated, anonymised and publicly available data from the Global Burden of Disease (GBD) Study 1990–2023. No individual-level data were accessed or used. Therefore, ethical approval was not required for this secondary data analysis. The authors have nothing to report. The authors declare no conflicts of interest. The datasets generated and/or analysed during the current study are publicly available from the Global Burden of Disease (GBD) Results Tool (https://vizhub.healthdata.org/gbd-results/) and the Institute for Health Metrics and Evaluation (IHME) data repository. Population data were obtained from the UN World Population Prospects 2024. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.70650. Data S1: dom70650-sup-0001-Supinfo.zip. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Ghazy et al. (Wed,) studied this question.