Routine metabolic–adiposity indices and incident cardiometabolic multimorbidity among adults with CKM stages 0–3: a prospective cohort study
Abstract
Cardiometabolic multimorbidity (CMM) is an increasing burden in ageing populations. Cardiovascular-kidney-metabolic (CKM) stages 0–3 provide an upstream framework for identifying metabolic, renal, and cardiovascular vulnerability before overt multimorbidity. We evaluated whether routinely derived metabolic–adiposity indices add useful risk-phenotype information beyond established clinical factors. This prospective CHARLS cohort included 7,062 adults aged ≥ 45 years who were free of CMM and had CKM stages 0–3 at baseline. Twenty-nine cholesterol–high-density lipoprotein cholesterol–glucose (CHG)-derived, triglyceride–glucose (TyG)-derived, and adiposity/body-shape indices were evaluated; the main text focused on RFM, CHG-RFM, TyG-RFM, BMI, and WC. Model 3 was used to anchor the primary association interpretation, while Model 4 was retained as an extended clinical-adjustment model to assess robustness after broader adjustment. Benjamini–Hochberg correction, inter-index correlations, paired CHG–TyG models, proportional-hazards diagnostics, prediction analyses, and sensitivity analyses were performed. Among 7,062 participants, 613 developed incident CMM (8.7%). In Model 3, the per-SD HRs (95% CIs) were 1.85 (1.52–2.25) for RFM, 1.72 (1.47–2.03) for CHG-RFM, 1.67 (1.41–1.98) for TyG-RFM, 1.30 (1.20–1.41) for BMI, and 1.34 (1.23–1.46) for WC. The corresponding estimates remained broadly consistent in Model 4. After Benjamini–Hochberg correction, 28 of 29 Model 3 per-SD associations remained supported; ABSI was the exception. Correlations were substantial among structurally related indices, and paired CHG–TyG models showed marked mutual attenuation. The clinical baseline AUC was 0.787, whereas selected index-extended models achieved AUCs of approximately 0.799–0.800, indicating modest absolute improvement. Selected routine metabolic–adiposity indices were associated with incident CMM among adults with CKM stages 0–3. RFM-related indices showed stronger association signals, whereas BMI and WC remain more accessible measures. The broadly consistent Model 3 and Model 4 estimates supported the overall association pattern. However, because inter-index overlap was substantial and incremental predictive gains were modest, these indices should be viewed as candidate adjunctive risk markers rather than independent biomarkers or clinically validated decision tools.