Identifying Important Factors for Depressive Symptom Dynamics in Chinese Middle-Aged and Older Adults Using a Multi-State Transition Model with Feature Selection

TL;DR

This study identifies key factors influencing transitions between depressive symptom states in Chinese middle-aged and older adults using a multi-state model.

Contribution

A novel approach combining multi-state transition modeling with feature selection to identify risk factors for depressive symptom dynamics.

Findings

Nine significant risk factors were identified for new depressive symptom episodes, including urban–rural residence and income.

The effects of risk factors on new symptom episodes weakened as symptoms became persistent or remitted.

Sex by age was a significant factor for depressive symptom relapse.

Abstract

Depressive symptoms are increasingly common in middle-aged and older adults and have become a major public health problem. People may experience transitions across different underlying states due to symptom variability over a course of many years. And risk factors may have different impact on different symptom states. However, existing research rarely considers the identification of important factors related to symptom conversion. The purpose of this study was to examine the risk associated with transitioning between various stages of depressive symptoms and their influencing factors, utilizing a multi-state model with a simultaneous feature selection method. We used the four waves of data from the China Health and Retirement Longitudinal Study (CHARLS) and 3916 participants were selected after screening. Five states of depressive symptoms were defined including no symptom, new symptom episode, symptom persistence, remission and relapse. We included 13 variables on demographic background, health status and functioning, and family and social connectivity, along with their interactions. Multi-state models were used to evaluate the risks of state transitions. The regularized (adaptive Lasso) partial likelihood approach was employed to simultaneously identify the important risk factors, estimate their impact on the state transition rates and determine their statistical significance. There were 1392 new depressive episodes events, 402 symptom persistence events, 639 remission events and 118 relapse events. We identified nine significant risk factors for the new onset of depressive symptoms: urban–rural residence, sex, retirement status, income, body pain, difficulty with basic daily activities, social engagement, education by income interaction and number of conditions by income interaction. The effects of the identified risk factors on new symptom episode weakened as those symptoms became persistent or went into remission. In terms of symptom relapse, sex by age was identified as a significant influencing factor. This study identified key factors and explored their effects on the various depressive symptom states among older Chinese adults. The findings could serve as a foundation for the development and implementation of targeted policies aimed at enhancing the mental well-being of China’s elderly population.

Figures