Association between TyG-FI and new-onset stroke in a Chinese elderly cohort

Abstract Objective To investigate the association between the triglyceride-glucose–frailty index (TyG-FI) and stroke risk, and to compare its discriminatory ability with other TyG-derived composite indices. Method This study included 5,010 participants from the China Health and Retirement Longitudinal Study (CHARLS). TyG-FI and other TyG-derived indices (TyG-BMI, TyG-WC, TyG-WHtR, TyG-ABSI, and TyG-BRI) were calculated using standardized formulas. Logistic regression models, restricted cubic spline (RCS) analysis, receiver operating characteristic (ROC) curve analysis with DeLong tests, and subgroup analyses were performed to evaluate the associations and discriminative ability of these indices for stroke, adjusting for demographic, lifestyle, and metabolic covariates. Results Elevated TyG-FI quartiles were associated with a graded increase in stroke risk, with significant elevations observed in Q3 (OR = 1.901, 95% CI: 1.378–2.644) and Q4 (OR = 2.090, 95% CI: 1.515–2.910) in the fully adjusted model. RCS analysis revealed a non-linear threshold effect (P-overall < 0.0001; P-non-linear = 0.0190). TyG-FI demonstrated modest discriminatory ability for stroke (AUC = 0.653, 95% CI: 0.624–0.681), which showed a statistically significant but marginal improvement over FI alone (AUC = 0.635, P = 0.007). It significantly outperformed the TyG index (AUC = 0.630, P = 0.009) and all other TyG-derived composite indices (TyG-BMI, TyG-WC, TyG-WHtR, TyG-ABSI, and TyG-BRI; all P < 0.01). However, when compared with FI alone, TyG-FI showed only a slight, yet statistically significant, improvement. Associations of most conventional TyG-obesity composite indices with stroke risk were substantially attenuated after full adjustment and lost statistical significance. Subgroup analyses suggested more pronounced associations in younger participants, females, and individuals with lower HDL-C levels. Conclusion TyG-FI is independently associated with incident stroke in older Chinese adults, with modest discriminative ability comparable to FI alone. While it may serve as a practical risk stratification tool, its clinical utility should be interpreted with appropriate caution.