Longitudinal frailty index–single point insulin sensitivity estimator and incident cardiometabolic multimorbidity: a nationwide prospective cohort study in China

Cardiometabolic multimorbidity (CMM) is an increasing public health concern in middle-aged and older adults. The frailty index (FI) and metabolic indices may help predict CMM. This study examined longitudinal patterns of FI combined with atherogenic index of plasma (AIP), lipid accumulation product (LAP), Chinese visceral adiposity index (CVAI), and single-point insulin sensitivity estimator (SPISE) and compared their predictive performance. Repeated measurements from 2012 to 2015 in the China Health and Retirement Longitudinal Study were analyzed. K-means clustering identified longitudinal patterns of FI and composite FI-AIP, FI-LAP, FI-CVAI, and FI-SPISE indices. Cox regression, Kaplan–Meier analysis, restricted cubic splines, mediation and subgroup analyses assessed associations with incident CMM. Time-dependent ROC analyses compared predictive performance, and the incremental value of FI-SPISE was evaluated beyond PREVENT, SCORE2, and China-PAR. Among 3,586 participants, 270 incident CMM events occurred during follow-up. Two-waves exposure and longitudinal patterns of FI-related indices showed greater discrimination for CMM than baseline measurements. Participants with higher-risk pattern (Cluster 3) or the highest two-waves exposure levels(Q4) of FI-SPISE had a significantly higher risk of CMM; specifically, the hazard ratios were 5.05 (3.25–7.85) for Cluster 3, and 4.95 (3.03–8.09) for Q4. Adding FI-SPISE to PREVENT-, SCORE2-, and China-PAR-based reference models was associated with higher C-index, NRI, and IDI values for incident CVD. Exploratory mediation analyses indicated that glycemic, inflammatory, hemodynamic, and renal biomarkers partially mediated the association. Longitudinal assessment of FI-related indices, particularly FI-SPISE, provided additional information beyond baseline measurements and provided incremental value beyond established cardiovascular risk models.