Association of the triglyceride glucose-body roundness index with stroke risk in middle-aged and elderly Chinese adults without diabetes: a prospective cohort study.

Background: Both the triglyceride-glucose index (TyG) and body roundness index (BRI) are individually associated with stroke risk, yet, each single indicator has limitations in predicting long-term risk. This study aims to evaluate the prospective association of TyG, BRI, and their combined index (TyG-BRI) with stroke incidence in non-diabetic populations. Methods: From the nationally representative China Health and Retirement Longitudinal Study (CHARLS), a multi-stage probability sampled cohort, we established an analytical cohort consisting of non-diabetic adults aged 45 years or older without pre-existing cardiovascular disease (CVD) at the May 2011 to March 2012 baseline. After excluding participants with missing data on blood glucose, lipids, or anthropometric measurements, a final sample of 7,396 participants was followed prospectively until September 2020. The TyG was computed using fasting serum levels of triglycerides (TG) and glucose, while the BRI was calculated from waist circumference (WC) and height. Incident stroke events were identified through interviews, linkage with health insurance claims data, and cross-checking of death registry records. Cox proportional hazards models estimated hazard ratio (HR) and 95% confidence interval (CI), while restricted cubic spline (RCS) analyses evaluated dose-response relationships with stroke incidence. Both multiplicative interaction terms and additive interaction indices [relative excess risk due to interaction (RERI), attributable proportion (AP), and synergy index (SI)] were used to assess joint effects. To compare the predictive performance of the models, we conducted a receiver operating characteristic (ROC) curve analysis and quantified the results using the area under the curve (AUC). Results: Index-specific exclusion for missing essential calculation parameters yielded distinct sample sizes for the analysis of stroke incidence associations with TyG (n=7,396), BRI (n=6,862), and TyG-BRI (n=6,315). During a median follow-up of 6.5 years, 440 incident stroke events were documented in the non-diabetic cohort of the TyG-BRI analysis. The mean age of the cohort was 58.17 years [with a standard deviation (SD) of 9.70]. Approximately 47% of the participants were male. In the fully adjusted model, elevated levels of TyG-BRI [comparing the highest quartile (Q4) to the lowest (Q1): HR =2.29, 95% CI: 1.58-3.33], BRI (HR =2.12, 95% CI: 1.49-3.02), and TyG (HR =1.82, 95% CI: 1.30-2.53) were all significantly associated with the risk of stroke. RCS analysis indicated a linear relationship between TyG-BRI and stroke risk (P non-linear =0.084). The TyG index and BRI exhibited a joint association with incident stroke; however, this relationship was characterized by an absence of both multiplicative and additive interaction effects. The TyG-BRI model demonstrated improved predictive performance for stroke risk relative to the basic model (AUC =0.677, P=0.02). Conclusions: In non-diabetic population, the composite index of TyG-BRI demonstrated a more robust and pronounced association with stroke risk, highlighting its potential as a superior predictive marker for stroke in non-diabetic individuals.