CHARLS Data

The CHARLS analysis treated each chronic disease in isolation, leaving the combined income effects of comorbid conditions unknown despite their frequent co-occurrence. Disease status in CHARLS was measured as a binary variable from self-reported physician diagnosis, introducing recall bias and potential misclassification. Depression missingness was handled with multiple imputation because it was judged missing at random. After excluding participants with missing key variables, the baseline sample comprised 69,457 individuals. The study used CHARLS, a nationally representative longitudinal survey of Chinese adults aged 45 and older. CHARLS used multistage stratified probability-proportional-to-size random sampling across 450 communities in 150 counties within 28 mainland Chinese provinces. CHARLS is a biennial nationally representative survey of Chinese adults aged 45 and older, with four waves spanning 2011 to 2018. CHARLS collected information across health, family, living arrangements, assets, labour, retirement, medical insurance, and related domains. The study used 2014 CHARLS data for migration and life-course variables and 2020 CHARLS data for later health outcomes and covar…