Abstract: Objective  To explore the association between the novel composite metabolic index triglyceride-high density lipoprotein-cholesterol-glucose body index(TyHGB) and the risk of stroke incidence in the middle-aged and elderly population in China. Methods  Based on the China Health and Retirement Longitudinal Study (CHARLS), 7 064 middle-aged and elder adults aged ≥ 45 years with no prior history of stroke were enrolled and followed up for 9 years，with additional validation performed using a hospital-based clinical cohort (n=204). The Cox proportional hazards regression model was employed to evaluate the association between TyHGB and the incidence of stroke. Subgroup analyses and restricted cubic spline (RCS) models were utilized to investigate potential interactions and nonlinear dose-response relationships. Additionally, receiver operating characteristic (ROC) curves were used to assess the predictive value of TyHGB in relation to stroke risk. Results  Following multivariable adjustment, each 1-unit increase in TyHGB was associated with an 18% elevated risk of stroke in the CHARLS cohort (HR=1.18, 95% CI: 1.11－1.25, P<0.001) and a 38% increased risk in the hospital validation cohort (HR=1.38, 95% CI: 1.09－1.75, P =0.008). RCS analysis revealed population heterogeneity in the association pattern, showing a nonlinear relationship in the CHARLS cohort but an approximately  linear trend in the hospital cohort. Nevertheless, both cohorts consistently supported a positive association between TyHGB and stroke risk. ROC curves suggested that TyHGB had moderate capability in predicting stroke risk among middle-aged and elder Chinese adults, with area under the curve (AUC) of 0.578 in the CHARLS cohort and 0.753 in the hospital cohort. TyHGB was significantly superior to the conventional TyG (triglyceride-glucose) index in both cohorts (both P<0.05). The positive association between TyHGB and incident stroke risk remained robust across various demographic and clinical subgroups. Conclusion  TyHGB is significantly positively associated with stroke risk in middle-aged and elder Chinese adults, and it exhibits superior predictive efficacy compared to the conventional TyG index in both the nationally representative CHARLS cohort and the hospital-based clinical cohort. The nonlinear association observed in the community population, combined with its robustness in the clinical cohort, collectively suggests that TyHGB serves as a reliable and convenient preliminary screening tool for stroke risk assessment. It is applicable for population risk evaluation and primary prevention strategy development across different scenarios.

Key words: triglyceride, high density lipoprotein-cholesterol, body mass index, CHARLS, middle-aged and elderly,  stroke