One-year change in the triglyceride-glucose index and subsequent incident hyperuricemia: a retrospective landmark cohort study

Abstract Background The triglyceride-glucose (TyG) index is associated with hyperuricemia, but the longitudinal interpretation of within-person change is complicated when repeated exposure measurements overlap with outcome ascertainment. We examined whether one-year change in the TyG index was associated with subsequent incident hyperuricemia among adults undergoing repeated annual health examinations. Methods We conducted a retrospective landmark cohort study using annual health-examination data from 2023 to 2025 at a tertiary hospital in China. Adults had to remain normouricemic through the 2024 landmark. The prespecified primary analysis evaluated 2023–2024 TyG change in a joint model with baseline TyG, age, sex, and the most recent pre-outcome serum urate and estimated glomerular filtration rate (eGFR) values from 2024. A complementary baseline-covariate model used 2023 serum urate and eGFR. Modified Poisson regression with robust variance estimated risk ratios (RRs) and 95% confidence intervals (CIs). Results Among 21,188 participants, 1,171 (5.5%) developed hyperuricemia in 2025. In the prespecified primary model, a 1-standard-deviation (SD) greater one-year TyG change was associated with incident hyperuricemia after adjustment for baseline TyG, age, sex, 2024 eGFR and 2024 serum urate (RR 1.120, 95% CI 1.061–1.181). The complementary baseline-covariate model using 2023 eGFR and serum urate gave a similar estimate (RR 1.139, 95% CI 1.080–1.201). Adding one-year TyG change increased the apparent C-statistic only slightly, from 0.801 to 0.803. Conclusions In this repeated health-examination cohort, greater one-year TyG change was associated with subsequent incident hyperuricemia after accounting for baseline TyG and the most recent pre-outcome urate and renal-function values. The incremental gain in discrimination was very small, and the findings should be interpreted as observational associations rather than evidence of clinical predictive utility or causality.