Potential Predictors

All information of potential predictors was collected at baseline. Five sociodemographic factors were collected and categorized, including age (18–39, 40–49, 50–59, 60 years, and over), gender (female or male), residential area (urban or rural), educational level (literate or below primary school, primary school, junior high school, senior high school, and college or university and above), and marital status (married, never married, separated, or divorced). Information of chronic physical diseases was collected based on self-reports, including heart disease, high blood pressure, asthma, chronic lung disease, tuberculosis, diabetes, stroke, stomach ulcers or intestinal ulcers, rheumatic fever or arthritis, and chronic headaches. Three categories, including no physical disease, one or two physical diseases, and three physical diseases and over were defined by using this information. Comorbidity of other mental disorders, including depressive disorders, bipolar disorders, drug use disorders, alcohol use disorders, was made by the CIDI-3.0 and divided into no comorbidity with other mental disorders, and comorbid at least one mental disorder. Data of lifetime and 12-month treatments using mental health services among ADs patients was also included as potential predictors.