Estimated glucose disposal rate shows the broadest associations with cardiometabolic multimorbidity progression in early cardiovascular–kidney–metabolic stages: a longitudinal cohort study

Abstract

Early risk stratification may help prevent cardiometabolic multimorbidity (CMM) among adults at cardiovascular–kidney–metabolic (CKM) stages 0–3. We compared the estimated glucose disposal rate (eGDR), triglyceride–glucose (TyG) index, and atherogenic index of plasma (AIP) for incident CMM, disease transitions, and incremental prediction. This longitudinal cohort analysis included 6,407 China Health and Retirement Longitudinal Study participants aged 45–79 years at CKM stages 0–3 and free of CMM in 2011, with follow-up through 2020. CMM was defined as the coexistence of at least two of diabetes, heart disease, and stroke. Cox models estimated hazard ratios (HRs) per standard deviation in the adverse direction (lower eGDR or higher TyG or AIP). Restricted cubic-spline and multistate models assessed dose–response patterns and transitions; five repetitions of stratified fivefold out-of-fold validation evaluated incremental 9-year prediction. During a median follow-up of 9.0 years, 593 participants developed CMM. Primary-model HRs per adverse standard deviation were 1.92 (95% confidence interval [CI], 1.77–2.09) for eGDR, 1.69 (95% CI, 1.57–1.82) for TyG, and 1.51 (95% CI, 1.39–1.64) for AIP. After CKM-stage adjustment, eGDR remained associated with all three disease-accumulation transitions (false-discovery-rate-adjusted P ≤ 0.009); TyG remained associated only with the earliest transition, and no AIP transition survived correction. Among single-marker additions, eGDR yielded the largest point gain in discrimination (ΔC-index, 0.041; 95% CI, 0.027–0.055). Adding TyG to eGDR further increased the C-index by 0.012 (95% CI, 0.005–0.020), whereas adding AIP to eGDR plus TyG changed the C-index by 0.001 (95% CI, −0.003 to 0.005). Calibration slopes were 0.910–0.945. Interval-aware and competing-risk analyses were consistent. All three indices were associated with incident CMM. eGDR showed the most consistent transition associations and the numerically largest single-marker gain in internally validated prediction; TyG was complementary, whereas AIP added little beyond eGDR and TyG. Independent external validation and clinical-utility assessment are needed before use in early risk stratification.