Global Burden, Trends, and High‐ BMI ‐Attributable Risk of Hip Osteoarthritis: An Analysis of GBD 2021 Data

BACKGROUND: The purpose of this study is to evaluate the global burden of hip osteoarthritis (OA) and investigate its risk factors using Global Burden of Disease (GBD) 2021 data and Mendelian randomization (MR). METHODS: Hip OA burden was assessed by region, age, sex, and Socio-Demographic Index (SDI). Temporal trends were evaluated using Joinpoint regression-derived average annual percentage change (AAPC). Two-sample MR and mediation analysis explored causal risk factors. Future burden from 2022 to 2050 was primarily projected using a Bayesian age-period-cohort (BAPC) model, with autoregressive integrated moving average (ARIMA) analysis used as a sensitivity comparison. RESULTS: The global age-standardized hip OA prevalence was 416.01/100000 in 2021. The incidence and prevalence varied widely across countries. Males who were above 70 years old had a greater incidence of hip OA. The incidence and prevalence positively correlated with SDI, with the largest rise in incident cases occurring in middle SDI regions. Joinpoint analysis showed that the global burden of hip OA increased significantly from 1990 to 2021, with AAPCs of 0.2206 for incidence, 0.1879 for prevalence, and 0.1846 for DALYs. The most pronounced increases were observed in low-middle and middle SDI regions. High body mass index (BMI) contributed to a 21.2% increase in disability-adjusted life years (DALYs) from 1990 to 2021. MR mediation supported a positive association between higher BMI and hip OA and suggested that palmitoyl sphingomyelin may represent a potential mediator. The primary BAPC model suggested that the burden of hip OA would remain substantial through 2050, whereas comparison with ARIMA indicated model-dependent uncertainty in the long-term magnitude of the forecasts. CONCLUSIONS: Our study provides valuable insights into the global epidemiological trends and risk factors of hip OA. These findings can help formulate policies and allocate resources, although long-term projections should be interpreted cautiously because they are sensitive to model choice.