Comparison of 16 metabolic-adiposity composite indices and cardiovascular risk in adults with cardiovascular-kidney-metabolic stages 0-3: cohort evidence from CHARLS and NHANES

Background Stage 0–3 of cardiovascular-kidney-metabolic (CKM) syndrome identifies a population characterized by the interaction of obesity, abnormal glucose metabolism, dyslipidemia and chronic inflammation. Composite indices built from routine anthropometric and laboratory measurements are convenient tools for characterizing this shared burden, but metabolic indices and adiposity indices are usually evaluated separately and are rarely compared head-to-head across multiple populations. Methods This study constructed all 16 multiplicative combinations of 4 metabolic indices (triglyceride-glucose index, TyG; C-reactive protein-triglyceride glucose index, CTI; cholesterol-high-density lipoprotein-glucose index, CHG; atherogenic index of plasma, AIP) and 4 adiposity indices (waist-to-height ratio, WHtR; body roundness index, BRI; Chinese visceral adiposity index, CVAI; a body shape index, ABSI). The China Health and Retirement Longitudinal Study (CHARLS) provided 5,304 adults aged 45 years or older with CKM stages 0–3 and 1,366 incident self-reported cardiovascular disease (CVD) events, plus a separate 2,811-participant wave 3 landmark cohort for cumulative-exposure and two-point exposure pattern analyses. The US National Health and Nutrition Examination Survey (NHANES) provided 5,487 participants and 431 cardiovascular deaths across six cycles from 1999–2010, with mortality linkage. Analyses used Cox models, restricted cubic splines, in-sample incremental performance metrics and a direction-constrained weighted quantile sum (WQS) extension, all fitted within source, retaining their respective weights and sampling designs. Results Per 1 source-sample standard deviation (SD) increase, CTI-CVAI was associated with incident CVD in CHARLS (hazard ratio, HR 1.213, 95% confidence interval, CI 1.147–1.282) and with cardiovascular death in NHANES (HR 1.270, 1.053–1.533), and CHG-CVAI with 1.221 (1.154–1.292) and 1.226 (1.009–1.490), respectively. The two cohorts were consistent for the metabolic component, with CTI and CHG both higher than TyG and AIP, but the best adiposity index differed: CVAI combinations were highest in CHARLS, whereas CTI-WHtR (1.331, 1.127–1.572), CTI-BRI (1.323, 1.128–1.552) and CTI-ABSI (1.318, 1.139–1.526) were highest in NHANES, and AIP combinations were unassociated in NHANES. In CHARLS, cumulative CTI-CVAI and CHG-CVAI were associated with incident CVD (HR 1.079, 1.034–1.127; 1.081, 1.034–1.130), and in the two-point pattern the highest mean-level WHtR group had HRs of 1.517–1.822. In neither cohort did the discrimination improvement of any index pass the multiple-correction threshold, although some indices improved reclassification metrics. The six-component positive WQS associations in the internal validation samples were positive in both cohorts (CHARLS 1.263, 1.139–1.401; NHANES 1.544, 1.125–2.121), with weights concentrated mainly on waist circumference, CRP, fasting glucose and total cholesterol, but component weights were unstable across repeated splits. Conclusions In adults with CKM stages 0–3, combinations of CTI or CHG with adiposity indices showed the strongest and most consistent associations with two different cardiovascular endpoints, and cumulative exposure provided additional information in CHARLS; the optimal adiposity index varied by cohort, and AIP combinations were weakest in both cohorts. Because no discrimination improvement was observed, these indices are better suited as convenient summaries of existing risk and should not be used directly for screening until absolute risk has been externally validated.