Setting and sample
In the current study, we used the data from the baseline of the Chinese Longitudinal Ageing Social Survey (CLASS) that was collected by the National Survey Research Center, Renmin University of China. This survey was stratified, multi-stage, probabilistic sampling survey, which coved 28 provincial areas in China. Details of the design and conduct of the study have been described elsewhere [24]. A total of 11,511elderly aged 60 or above was interviewed by face to face. In the present study, the sample is comprised of 7505 subjects aged 60 or above who had finished the Center for Epidemiological Studies Depression Scale (CES-D) and other independent variables of interest.
Depressive symptoms Assessment
Depressive symptoms was assessed using an abbreviated nine-item Center for Epidemiological Studies Depression Scale (CES-D), which is reliable and valid for detecting depressive symptoms among Chinese older adults [25]. In the CLASS study, the internal Cronbach's alpha for CES-D used was 0.75. Using Kohout’s formula [26], standardized cut scores are determined by dividing the total possible score on a short CES-D scale by 60 (the total possible score on the full 20-item CES-D) and multiplying that number by 16 (the established cut score on the full 20-item CES-D. For current study, on a 9-item scale, the total possible score is 18 (9 items multiplied by 2, the highest response). That total score is divided by 60, which equals 0.3. Then, the 0.3 is multiplied by 16, resulting in a standardized cut score of 4.8 for the 9-item form of the CES-D.
Independent Variable of Interest
We included the following socioeconomic characteristics in our study: age (60-64, 65-69, 70-74, 75-79, 80+), gender (male, female), residence (rural, urban), marital status (married, widowed/divorced/unmarried), education level (junior high school and above, primary school, never attended school), ethnicity (Han, Others), and living arrangements (lives alone, lives with others). Income is categorized into five levels using the quintiles of household income (Yuan) (Q1: ≤ 3000, Q2: >3000 and ≤ 10000, Q3: >10000 and ≤ 24000, Q4: >24000 and ≤ 36000, Q5: >36000). We dichotomized the physical disability status, which was assessed by using ten-item version of the activities of daily living (ADL) scale, into two groups (“no functional problems”= 0, “has at least one limitation”=1) [27]. The participants were also asked whether they had one of the following health problems (covered 23 chronic disease), including hypertension (Yes, No), diabetes mellitus (Yes, No), arthritis (Yes, No), cerebrovascular disease (Yes, No), liver disease, and so on. Number of comorbid chronic disease were further categorized into “0”, “1” and “≥2”.
Statistical analysis
The difference between subjects with/without depressive symptoms was tested by c2 test with proportions. The logistic regression model was used to calculate the adjusted odds ratios (OR) and 95% confidence interval (CI) of depressive symptoms (yes/no) with the covariates: age group, gender, residence, marital status, education, ethnicity, physical disability, living arrangement, and wealth quantiles. Separate multiple logistic regression models were used to assess the association of depressive symptoms with each chronic disease. In each model, the exposure of interest was one chronic disease and other covariates (mentioned above) were included. All two-way interactions between each chronic disease and covariates were assessed in multivariable adjusted models. The trend of the association was assessed with ordinal scores assigned to number of comorbid chronic disease. Statistical significance was declared with a two-sided p-value < 0.05. Statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC, USA).