Investigating intrinsic and situational predictors of depression among older adults: An analysis of the CHARLS database

Abstract

Background

This study aimed to investigate the intrinsic and situational predictors of depression under the health ecological model.

Methods

Two waves (2011 and 2013) of survey data were collected from the CHARLS. A total of 5845 older adults (≧60) were included, and depression was defined as CESD-10 score ≧10. Random forest combined with interpretable methods were utilized to select important predictors of depression. Multilevel logit model was used to examine the associations of intrinsic and situational predictors with depression.

Results

After a 2-year follow up, 1822 individuals (31.17 %) developed depression. Interpretable analyses showed that both intrinsic and situational variables were predictive for depression. Multilevel logit model showed that age, gender, number of chronic diseases, number of pain areas, life satisfaction, and toilet distance were significantly associated with depression.

Conclusion

Both intrinsic and situational factors were found to be associated with depression among community older population, highlighting their significance for early prevention from the perspective of public health.