Association between the cardiometabolic index and cardiometabolic multimorbidity in older adults: uric acid as a potential mediator and combined predictive value

[Objective] To examine the association between cardiometabolic index (CMI) and cardiometabolic multimorbidity (CMM) in older adults undergoing health examinations, and to explore the mediating role of serum uric acid (UA), both overall and stratified by gender, and to evaluate the discriminative performance of a combined model including CMI and UA for identifying CMM. [Methods] In this retrospective cross-sectional study, 11 336 adults aged ≥60 years who underwent health examinations from January to December 2024 were included. Multivariable logistic regression was used to assess the association between CMI and CMM. Restricted cubic spline (RCS) analysis was performed to examine the dose-response relationship. Mediation analysis was used to estimate the indirect effect of UA on the association between CMI and CMM, with sex-stratified analyses. Receiver operating characteristic (ROC) curves were used to assess the discriminative performance of CMI, UA, and the combined model. [Results] The prevalence of CMM was 5.5%. After adjustment for potential confounders, CMI was independently associated with CMM (OR=1.28, 95%CI: 1.11-1.44). RCS analysis indicated a nonlinear association between CMI and the risk of CMM (P for nonlinearity=0.03), with a steeper increase in risk at lower CMI levels (<0.522). Mediation analysis showed an inconsistent mediation/suppression effect of UA: the direct effect of CMI on CMM was positive (β=0.324, 95%CI: 0.216-0.432), whereas the indirect effect through UA was negative (β=-0.082, bootstrap 95%CI: -0.118－-0.050), and this pattern was more pronounced in men. ROC curve analysis showed limited discriminative ability for CMI and UA alone [the area under the curve (AUC) was 0.551 and 0.531, respectively)], while the combined model including CMI, UA, and basic covariates achieved an AUC of 0.709(95%CI: 0.689－0.729), which was significantly higher than that of either single-indicator model (both P<0.001). [Conclusions] In older adults undergoing health examinations, CMI is independently and nonlinearly associated with CMM. UA shows an inconsistent mediation/suppression effect in the association between CMI and CMM, which should be interpreted primarily as a statistical association rather than causality. The combined model including CMI, UA, and basic covariates showed moderate discriminative value for identifying CMM.