Differential associations of baseline, dynamic change, and cumulative average TyG-BMI with incident cardiometabolic multimorbidity: a longitudinal landmark analysis of CHARLS

Abstract Cardiometabolic multimorbidity (CMM) is common, yet the prognostic relevance of temporal triglyceride-glucose body mass index (TyG-BMI) measures remains unclear. A 2015 landmark analysis of the China Health and Retirement Longitudinal Study included 4597 participants. Baseline TyG-BMI was measured in 2011; dynamic change and cumulative average were derived from 2011 and 2015 measurements. Incident CMM was ascertained during 2015–2020. Discrete-time complementary log–log models estimated associations, with supplementary analyses of the 2015 landmark value, disease transitions, component outcomes, and bootstrap validation. Overall, 528 participants developed CMM. Baseline and cumulative average TyG-BMI were associated with CMM (HR per 1-SD increase 1.28, 95% CI 1.17–1.40; and 1.25, 95% CI 1.14–1.37, respectively), with linear dose–response patterns. Dynamic change was not independently associated after baseline adjustment (HR 0.96, 95% CI 0.89–1.04). The 2015 landmark value was also associated with CMM (HR 1.19, 95% CI 1.09–1.30), and residualized cumulative average remained associated after accounting for this value (HR 1.17, 95% CI 1.08–1.27). Among cases with complete landmark disease status, 80.7% had one existing condition; baseline and cumulative average associations extended across all three components and CMM transitions with or without newly acquired diabetes. Bootstrap-corrected improvement beyond the 2015 value was small (ΔAUC 0.004, 95% CI − 0.0002 to 0.0076). Baseline, current, and cumulative average TyG-BMI identified higher CMM risk, whereas short-term change added little information. TyG-BMI may complement clinical risk assessment, but its added predictive value was modest.