Routine metabolic and nutrition-related composite indices for identifying adverse cardiovascular-kidney-metabolic phenotypes: a cross-sectional study
Cardiovascular-kidney-metabolic (CKM) syndrome provides an integrated framework for capturing the overlap of adiposity, metabolic dysfunction, kidney impairment, and cardiovascular risk. Whether low-cost composite biomarkers derived entirely from routine health examinations and brief questionnaires can identify adverse CKM phenotypes in preventive-care populations remains insufficiently characterized. We conducted a single-center, hospital-based cross-sectional study of adults who underwent routine health examinations at the Health Examination Center of the Second Affiliated Hospital of Shandong First Medical University between 2024 and 2025. Participants were included in the complete-case analysis based on complete data availability. Demographic, lifestyle, anthropometric, blood pressure, complete blood count, and fasting biochemical variables were used to construct anthropometric, insulin-resistance, lipid-metabolic, inflammatory, and nutrition-related indices. The primary outcome was prevalent adverse CKM phenotype, defined as CKM stage > = 2. Multivariable logistic regression, receiver operating characteristic analysis, restricted cubic spline modeling, and internally validated machine-learning models were used to evaluate associations, discrimination, non-linearity, and model interpretability. Among 1,226 participants (mean [SD] age, 47.42 [13.14] years; 631 men [51.47%]), 718 (58.56%) had adverse CKM phenotype. CKM stage 2 accounted for most adverse CKM cases (693/718, 96.52%), whereas stages 3 and 4 were uncommon (9 [0.73%] and 16 [1.31%] of the total sample, respectively). After adjustment for age, sex, smoking status, alcohol status, exercise frequency, and sleep duration, the strongest associations were observed for the atherogenic index of plasma (AIP; odds ratio [OR] per 1-SD increase, 6.70; 95% CI, 5.24–8.56; p < 0.01), triglyceride-glucose (TyG) index (OR, 6.38; 95% CI, 5.00–8.13; p < 0.01), non-HDL-C to HDL-C ratio (OR, 4.97; 95% CI, 3.86–6.41; p < 0.01), TyG-BMI (OR, 4.53; 95% CI, 3.67–5.58; p < 0.01), and TyG-WHtR (OR, 4.08; 95% CI, 3.33–5.01; p < 0.01). AIP and TyG showed similarly high single-index discrimination (AUC, 0.87 for both; p < 0.01). In the held-out testing set, the random forest model achieved the best internal performance (AUC, 0.89; 95% CI, 0.85–0.92; p < 0.01; PR-AUC, 0.94; accuracy, 0.82). In this single-center cross-sectional routine health examination population, adverse CKM phenotype was common but predominantly represented stage 2 disease. AIP and TyG were similarly high-performing low-cost summary markers of prevalent CKM-related metabolic risk. These findings suggest that routine composite indices may help summarize and preliminarily stratify prevalent CKM risk in health examination settings, pending external and longitudinal validation; they should not be interpreted as evidence of readiness for clinical deployment or prediction of future events.