Abstract
BACKGROUND: Serum uric acid (SUA) may predict incident type 2 diabetes mellitus (T2DM), but longitudinal evidence from Middle Eastern populations remains limited. METHODS: We conducted a retrospective cohort study using electronic health records from Qatar's Primary Health Care Corporation over a six-year period (2018-2023). Adults aged ≥18 years with at least one valid serum uric acid (SUA) measurement and no prior diabetes at baseline were eligible. All eligible participants were retained; no propensity score matching was performed. Baseline SUA was defined at the first valid measurement, and repeated-measure exposures included current SUA, cumulative-average SUA, and landmark time-weighted average (TWA) SUA. Sex-specific SUA categories were low <208, normal 208-428, and high >428 µmol/L in males and low <149, normal 149-357, and high >357 µmol/L in females. Sex-stratified Cox models, restricted cubic spline analyses, prespecified sensitivity analyses, and complementary explainable boosting machine (EBM) models were used to evaluate associations with incident type 2 diabetes mellitus (T2DM). RESULTS: The cohort included 169,876 adults (85,361 males and 84,515 females) and 18,714 incident T2DM events. In fully adjusted baseline Cox models, high baseline SUA was associated with higher T2DM hazard in females (hazard ratio [HR]: 1.44; 95% CI: 1.36-1.53), whereas low baseline SUA was associated with higher hazard in males (HR: 1.60; 95% CI: 1.44-1.78), and high SUA was not. In women, positive SUA-T2DM associations persisted in time-varying and landmark analyses, including current high- versus- normal SUA (HR: 1.50; 95% CI: 1.41-1.58) and 2-measurement landmark TWA SUA per 1 mg/dL (HR: 1.17; 95% CI: 1.13-1.20). In men, unlagged whole-cohort analyses showed inverse continuous associations, but lagged and repeated-measure analyses shifted toward positive associations, including 365-day lagged high- versus- normal baseline SUA (HR: 1.19; 95% CI: 1.11-1.28) and 2-measurement landmark TWA SUA per 1 mg/dL (HR: 1.06; 95% CI: 1.03-1.09). Restricted cubic splines showed a steadily rising risk gradient in females above approximately 262 µmol/L and a J-shaped pattern in males, with the lowest risk near 374 µmol/L. In EBM models, TWA SUA ranked third in women and fifth in men in the 2-measurement landmark cohorts. CONCLUSIONS: In this large Qatar cohort, longitudinal SUA was associated with incident T2DM in a sex-specific manner, with consistent positive associations in females and exposure-definition-dependent patterns in males. Repeated SUA measurements may improve diabetes risk stratification, but causal and therapeutic implications require further study.