Comparative associations and incremental predictive value of cardiometabolic and inflammatory indices for short- to mid-term cardiovascular disease risk in two Chinese cohorts

Abstract

Background Composite cardiometabolic and inflammatory indices can be calculated from routine anthropometric and laboratory measures, but their value for cardiovascular disease (CVD) risk assessment in Chinese populations remain unclear. Methods We analyzed 7,170 participants free of CVD at baseline from the China Health and Retirement Longitudinal Study (CHARLS) and the Shanghai Diabetes Study (SHDS). Mean follow-up was approximately 41 months in both cohorts. The evaluated indices included body mass index (BMI), Chinese visceral adiposity index (CVAI), lipid accumulation product (LAP), triglyceride-glucose index (TyG), atherogenic index of plasma (AIP), high-sensitivity C-reactive protein (hs-CRP), hs-CRP to high-density lipoprotein cholesterol ratio (hs-CRP/HDL-C), and hs-CRP-glucose index (CTI). Cox proportional hazards models were used to examine associations with incident CVD, with cohort interactions tested to assess cross-cohort consistency. Dose-response and nonlinearity were evaluated using quartile-based analyses and restricted cubic splines. Predictive performance and the added value of each index beyond a common clinical base model were evaluated in terms of clinical utility. Results During follow-up, 343 CVD events occurred in CHARLS and 148 occurred in SHDS. CVAI was the only index consistently associated with incident CVD across both cohorts and adjustment models. In the fully adjusted model, the HR per SD increase in CVAI was 1.221 (95% CI: 1.070–1.392, P = 0.003) in CHARLS, and 1.330 (95% CI: 1.067–1.656, P = 0.011) in SHDS. Other indices showed cohort-specific patterns: CTI, hs-CRP, and hs-CRP/HDL-C remained associated with CVD in CHARLS, whereas LAP remained associated across models in SHDS. Quartile-based and spline analyses generally supported approximately linear dose-response patterns. When added to the clinical base model, individual indices produced only modest changes in discrimination, and decision curve net benefit. Conclusions In two Chinese cohorts, CVAI showed the most reproducible association with short- to mid-term CVD risk, whereas other indices provided more cohort-specific signals. These indices may help describe cardiometabolic risk burden, but their incremental predictive value beyond conventional clinical factors was limited. Further studies with longer follow-up, verified outcomes, and more diverse populations are needed.