Comparing a New Multimorbidity Index with Other Multimorbidity Measures for Predicting Disability Trajectories.

Hui-Wen Xu,Hui Liu,Yan Luo,Kaipeng Wang,My Ngoc To,Yu-Ming Chen,He-Xuan Su,Zhou Yang,Yong-Hua Hu,Beibei Xu

DOI: https://doi.org/10.1016/j.jad.2023.11.014

IF: 6.533

2024-01-01

Journal of Affective Disorders

Abstract:Background: The optimal multimorbidity measures for predicting disability trajectories are not universally agreed upon. We developed a multimorbidity index among middle-aged and older community-dwelling Chinese adults and compare its predictive ability of disability trajectories with other multimorbidity measures.Methods: This study included 17,649 participants aged >= 50 years from the China Health and Retirement Longitudinal Survey 2011-2018. Two disability trajectory groups were estimated using the total disability score differences calculated between each follow-up visit and baseline. A weighted index was constructed using logistic regression models for disability trajectories based on the training set (70 %). The index and the condition count were used, along with the pattern identified by the latent class analysis to measure multimorbidity at baseline. Logistic regression models were used in the training set to examine associations between each multimorbidity measure and disability trajectories. C-statistics, integrated discrimination improvements, and net reclassification indices were applied to compare the performance of different multimorbidity measures in predicting disability trajectories in the testing set (30 %).Results: In the newly developed multimorbidity index, the weights of the chronic conditions varied from 1.04 to 2.55. The multimorbidity index had a higher predictive performance than the condition count. The condition count performed better than the multimorbidity pattern in predicting disability trajectories. Limitation: Self-reported chronic conditions.Conclusions: The multimorbidity index may be considered an ideal measurement in predicting disability trajectories among middle-aged and older community-dwelling Chinese adults. The condition count is also suggested due to its simplicity and superior predictive performance.