Modified Cardiometabolic Index and Incident Surrogate-Defined Cardiovascular–Liver–Metabolic Multimorbidity Among Middle-Aged and Older Adults in China: A Prospective Cohort Study from the China Health and Retirement Longitudinal Study (CHARLS)

Background: Defined by the coexistence of at least two cardiovascular diseases, type 2 diabetes and suspected metabolic dysfunction-associated steatotic liver disease, cardiovascular–liver–metabolic multimorbidity (CLMM) is an emerging phenotype that is related to the cardiovascular–kidney–metabolic continuum yet distinct from it. By combining triglycerides, fasting glucose, HDL cholesterol and central adiposity, the modified cardiometabolic index (MCMI) summarizes these features in a single fasting-panel measure. Evidence to date has been confined to cross-sectional fatty-liver and self-reported outcomes; established lipid–glucose indices have not been compared head to head with MCMI for CLMM. Methods: We included 6124 CHARLS participants aged 45 years or older, free of CLMM in 2011 and with a 2015 health questionnaire. For the primary analysis, we assigned indeterminate follow-up outcomes (n = 1919) to non-CLMM; results from determinable-outcome and multiple-imputation analyses are presented beside the primary estimates, along with additional alternative assumptions. We used multivariable logistic regression, restricted cubic splines and paired ROC analyses. Exploratory analyses compared MCMI with 12 other indices. We also examined joint MCMI–CRP associations and interaction, subgroup, sensitivity and component-specific analyses, together with incremental models and bootstrap internal validation. Results: Over approximately four years, 619 participants (10.11%) fulfilled the surrogate-defined CLMM definition, with suspected MASLD documented in 93.4% of cases. A nonlinear association was observed (p < 0.001). The fully adjusted MCMI Q4-versus-Q1 odds ratio was 6.13 (95% CI 4.34–8.66) in the primary analysis, 6.67 (4.70–9.49) among participants with determinable outcomes and 5.69 (4.24–7.65) after multiple imputation of missing components, but 1.74 (1.49–2.02) if all indeterminate outcomes were assumed to be cases. Adjusting for baseline component count reduced the odds ratio from 6.38 to 3.22 (2.17–4.78) across the same 6064 participants. Per-SD odds ratios were 3.63 (3.01–4.37) for incident suspected MASLD versus 1.10 (1.00–1.21) for incident CVD. MCMI had an apparent AUC of 0.684 (0.664–0.705); differences from BMI and waist-to-height ratio were not statistically significant. After bootstrap optimism correction, the covariate model had AUCs of 0.697 with MCMI and 0.698 with BMI, and adding MCMI to separately modelled glucose, triglycerides, HDL cholesterol and waist circumference did not significantly improve model fit (p = 0.263). The high-MCMI/high-CRP group had an odds ratio of 3.29 (2.55–4.25) versus the double-low group, but interaction tests were not significant. Conclusions: MCMI was associated with incident surrogate-defined CLMM, yet much of the observed association appeared attributable to pre-existing component burden and to the LAP-based suspected MASLD component sharing inputs with MCMI, while its magnitude varied with the treatment of indeterminate outcomes. Its exploratory apparent discrimination was not significantly different from that of BMI or waist-to-height ratio. Interpretation remains limited by the lack of independent validation; the joint MCMI–CRP findings did not establish interaction or biological synergy.