ObjectiveThis study aimed to develop and validate a multimorbidity index using self-reported chronic conditions for predicting 5-year mortality risk.MethodsWe analyzed data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS) and included 11,853 community-dwelling older adults aged 65–84 years. Restrictive association rule mining (ARM) was used to identify disease combinations associated with mortality based on 13 chronic conditions. Data were randomly split into the training (N = 8,298) and validation (N = 3,555) sets. Two multimorbidity indices with individual diseases only (MI) and disease combinations (MIDC) were developed using hazard ratios (HRs) for 5-year morality in the training set. We compared the predictive performance in the validation set between the models using condition count, MI, and MIDC by the concordance (C) statistic, the Integrated Discrimination Improvement (IDI), and the Net Reclassification Index (NRI).ResultsA total of 13 disease combinations were identified. Compared with condition count (C-statistic: 0.710), MIDC (C-statistic: 0.713) showed significantly better discriminative ability (C-statistic: p = 0.016; IDI: 0.005, p < 0.001; NRI: 0.038, p = 0.478). Compared with MI (C-statistic: 0.711), the C-statistic of the model using MIDC was significantly higher (p = 0.031), while the IDI was more than 0 but not statistically significant (IDI: 0.003, p = 0.090).ConclusionAlthough current multimorbidity status is commonly defined by individual chronic conditions, this study found that the multimorbidity index incorporating disease combinations showed supreme performance in predicting mortality among community-dwelling older adults. These findings suggest a need to consider significant disease combinations when measuring multimorbidity in medical research and clinical practice.

Read Full Article (External Site)