Association between modified triglyceride glucose indices and stroke risk in middle-aged and older Chinese adults: a prospective cohort study.

Abstract

Objective: Various modified triglyceride and glucose (TyG) indices have been proposed, but the literature on the impact and ability of the modified TyG index to predict stroke disease remains limited. We aimed to investigate the associations between the modified TyG indices and incident stroke diseases and to compare their predictive power in a nationally representative cohort.

Methods: This was a prospective cohort study with longitudinal data from the China Health and Retirement Longitudinal Study (CHARLS), which was based on data from four CHARLS surveys in 2011, 2013, 2015 and 2018, and included a total of 8677 participants. The Cox proportional risk model, Restricted Cubic Spline (RCS), and Receiver Operating Characteristic curve (ROC) were analyzed for the associations of TyG-BMI, TyG-WC and TyG-WHtR with stroke risk. In addition, the robustness of the findings was further validated using a series of sensitivity analyses and subgroup analyses.

Results: During the follow-up period from 2011 to 2018, 807 (9.3%) of the 8677 participants experienced stroke. After multifactorial adjustment (Model III), a 10-unit increase in TyG-BMI corresponded to a 5.5% increased risk of stroke (HR = 1.055, 95% CI 1.033-1.078), a 10-unit increase in TyG-WC corresponded to a 2.0% increased risk of stroke (HR = 1.020, 95% CI 1.012-1.027), and a 1-unit increase in TyG-WHtR corresponded to a 32.4% increased risk of stroke (HR = 1.324, 95% CI 1.178-1.487). Compared with individuals in the lowest quartile (Q₁), those in higher quartiles showed progressively elevated risks. For TyG-BMI, adjusted hazard ratios increased by 39.1% (Q₂), 62.1% (Q₃), and 86.1% (Q₄); for TyG-WC, by 40.6%, 63.0%, and 88.8%; and for TyG-WHtR, by 36.5%, 65.7%, and 83.7%, respectively. The RCS showed a nonlinear dose‒response relationship between TyG-BMI and stroke and a nonlinear dose‒response relationship between TyG-WC. In addition, TyG-WHtR had a linear dose‒response relationship with stroke. The area under the curve (AUC) values were 0.5930, 0.6078 and 0.6032, respectively, and all three had comparable predictive abilities for stroke risk, with TyG-WC having slightly greater predictive ability. The results of the sensitivity analysis and subgroup analysis were consistent with the main results.

Conclusions: In middle-aged and elderly populations, TyG-BMI, TyG-WC and TyG-WHtR are positively correlated with the risk of stroke. Maintaining favorable levels of these indices through weight and waist management may help reduce stroke risk.