Development and validation of CMI5: a novel composite metabolic index for predicting cardiometabolic multimorbidity using routine health check-up data

<title>Abstract</title> <p>Background Routine health check-ups generate extensive clinical data, yet the predictive utility of many commonly measured parameters for cardiometabolic multimorbidity (CMM) remains insufficiently characterized. CMM, defined as the coexistence of two or more cardiometabolic diseases , including diabetes, heart disease, and stroke, substantially reduces life expectancy and quality of life. Although several composite indices have been developed for predicting individual cardiometabolic diseases, few have been specifically designed or validated for CMM as a composite endpoint. We therefore aimed to identify the most predictive routine indicators from a broad panel of 48 clinical parameters and to construct a novel composite index for CMM risk stratification. Methods We extracted cross-sectional data from the National Health and Nutrition Examination Survey (NHANES) 2013–2023 as the discovery set (n = 32,843) and used the China Health and Retirement Longitudinal Study (CHARLS) 2015–2016 as the external validation set (n = 13,725). Forty-eight routine clinical indicators were evaluated for their individual predictive performance for CMM using univariate receiver operating characteristic (ROC) analysis. The five indicators with the highest area under the curve (AUC) were selected to construct a composite metabolic index (CMI5). A regularized random forest model was trained on the NHANES data and tested in both internal and external validation cohorts. Results Hemoglobin A1c (AUC = 0.831), age (AUC = 0.777), fasting blood glucose (AUC = 0.762), blood urea nitrogen (AUC = 0.658), and body mass index (AUC = 0.621) ranked as the top five predictors and were integrated into CMI5. In the NHANES internal test set, the CMI5-based model achieved an AUC of 0.904 (95% CI: 0.896–0.912). Multivariable logistic regression showed that each one-unit increase in CMI5 corresponded to a 6.5-fold higher odds of CMM (OR = 7.53, 95% CI: 7.30–7.88, p &lt; 0.001). Restricted cubic spline analysis revealed a nonlinear dose–response relationship, with a sharp risk escalation when CMI5 exceeded 9.5. External validation in CHARLS yielded an AUC of 0.742, indicating attenuated but still discriminative performance across populations. Conclusions CMI5, derived from five readily accessible routine indicators, shows good discriminative ability for CMM in the US population. Its cross-population generalizability is limited by ethnic heterogeneity and low-dimensional feature space, requiring calibration before application in non-US settings. The index may serve as a low-cost screening tool for primary CMM risk stratification. Trial registration Not applicable.</p>