Multimorbidity Patterns and Depression: Bridging Epidemiological Associations with Predictive Analytics for Risk Stratification.

Background

Late-life depression is a critical public health concern, particularly among older adults with chronic multimorbidity. Existing studies often focus on single-disease associations, neglecting the complex interplay of coexisting conditions. Understanding how multimorbidity patterns contribute to depression risk and identifying high-risk subgroups through integrated statistical and machine learning approaches remain underexplored, limiting targeted prevention strategies.

Methods

Using data from the China Health and Retirement Longitudinal Study (CHARLS), latent class analysis (LCA) was employed to cluster multimorbidity patterns. Associations between these patterns and depression were analyzed using multivariable logistic regression, while predictive performance and interaction effects were evaluated via an XGBoost machine learning model.

Results

Four distinct multimorbidity patterns were identified: cardio-metabolic, digestive-joint, respiratory, and cardiovascular-digestive pattern. All clusters showed significant independent associations with depression, with the cardiovascular-digestive pattern exhibiting the strongest association (OR = 4.56). However, the digestive-joint pattern demonstrated the highest predictive effects for depression. Sociodemographic factors-low income, limited education, female gender, and rural residence-emerged as robust predictors, amplifying depression risk in older adults with multimorbidity.

Conclusions

This study bridges epidemiological insights with predictive analytics to inform depression risk stratification. We recommend routine depression screening for all individuals with cardiovascular-digestive diseases and prioritize screening for women with digestive-joint diseases. Additionally, low-income and rural-dwelling older adults with chronic conditions warrant heightened clinical vigilance. These findings provide a framework for integrating multimorbidity profiling into depression prevention protocols, addressing both biological and socioeconomic determinants.