Measures

The World Mental Health Composite International Diagnostic Interview (CIDI) version 3.0 was used to establish life-time Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) diagnosis of mental disorders [27]. Given the respondent burden, only select mental disorders were included in the survey. These were major depressive disorder (MDD), dysthymia, bipolar disorder (BD), generalized anxiety disorder (GAD), obsessive-compulsive disorder (OCD) and alcohol use (alcohol abuse and dependence) disorders (AUD). Diagnosis was obtained using established algorithms with rules of hierarchy [27]. Lifetime prevalence of ‘any mental disorder’ was derived when the individuals had experienced at least one of the above conditions in their lifetime. Due to the small sample having dysthymia, it was excluded from this analysis.

The Multidimensional Scale of Perceived Social Support (MSPSS) was used to estimate the levels of perceived social support in three domains - support from Significant Other, Family and Friends [13]. The scale comprises 12 items, with 4 items in each subscale. Respondents were asked to indicate their level of agreement to each item by using a seven-point Likert scale ranging from 1 “very strongly disagree” to 7 “very strongly agree”. Global MSPSS and domains (Significant Other, Family and Friends) scores were derived by summing the responses from the respective items, with higher scores indicating better social support. This scale has been extensively applied and validated for assessment of perceived social support, including a study among Singaporean Chinese, Malay and Indian ethnic groups [28–30]. The original English version of the scale was translated into two predominant local languages - Chinese and Malay using two independent forward translations, followed by cognitive testing. All language versions were tested in a subgroup of the local population (n = 15) using cognitive interviews and pre-testing to assess their acceptability. No cross-language issues were identified, and modifications were not required as the scales were well-understood and received. In this sample the internal consistency reliability for the Global MSPSS and Significant Other, Family and Friends subscales had Cronbach’s alphas of 0.91, 0.90, 0.90 and 0.93, respectively.

Respondents were asked to self-report history of chronic physical illnesses using a modified version of the CIDI chronic conditions checklist which included the following categories: (1) asthma, (2) diabetes, (3) hypertension and high blood pressure, (4) chronic pain, (5) cancer, (6) cardiovascular disorders, (7) ulcer and chronic inflamed bowel, (8) thyroid disease, (9) neurological condition, (10) chronic lung diseases, and, (11) hyperlipidemia [31]. In the current analysis, presence of any of these conditions was classified into a dichotomous variable (Yes or No).

Detailed sociodemographic information was obtained from the respondents during the survey. This included age, gender (male or female), ethnicity (Chinese, Malay, Indian, or Others), marital status (never married, married, divorced/ separated or widowed), educational level (primary and below, secondary, vocational, pre-university/ junior college, diploma or university), employment status (employed, unemployed or economically inactive i.e., students, homemakers and retirees) and average monthly household income in thousand Singapore dollars (less than 2, 2–3.9, 4–5.9, 6–9.9 or 10 and over). All variables were captured for the ‘current’ state i.e. status at the time of the survey. The marital status variable was classified in three ways for the present analysis: (1) using 4 original groups i.e. never married, married, divorced/ separated or widowed, in multivariable analysis and to assess estimates in reference to married sample, (2) each group dichotomized into dummy coded variables eg, never married = 1 versus rest (all others) = 0, used to study association of specific marital status and perceived social support, and (3) marital group dummy coded as married or single (combined from never married, divorced/ separated and widowed) for assessing mediation.