2.1 Sampling
The data were obtained from the baseline survey of a large cohort study conducted between 2004 to 2008 in 10 geographically defined areas. Details of the study design and sample characteristics could be found elsewhere [23]. The regional study sites were selected carefully to retain geographic and social diversity, and to maximize difference in disease rates and risk exposure, so that they could approximate nationally representative samples. Potential participants were approached by community health workers. Over 99% consented to participate in the baseline assessment. In total, 512,891 adults, including 302,632 (59%) women aged 30–79 years, approximately 30% of the total population of the 10 regions sampled, were recruited and completed interviewer-administered electronic questionnaire and clinic visits. A structured interview was conducted by community health workers during the clinical visit, covering socioeconomic factors, mental health related questions, and sleeping patterns. Physical metrics (e.g. height, weight, waist and hip circumference) were also recorded.
2.2 Exposures
Number of children was based on the self-report by the participants. The question: “How many children do you have” was asked by the interviewers of the questionnaire and participants reported the current numbers of their children.
2.3 Outcomes
In the present study, Chinese version of computerized Composite International Diagnostic Inventory-Short Form (CIDI-SF) was employed to access major depression (MD). The evaluation was delivered face-to-face by trained community health workers at local clinics [24]. As a diagnostic instrument, the CIDI is based on criteria of the Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV) which is proven to be generally equal to clinical psychiatric interviews [24]. The estimates of MD in the population level as accessed by the Chinese version of the CIDI were similar to those accessed by the Structured Clinical Interview for DSM [25, 26]. In the diagnostic tool, the participants with the presence of dysphoria and/or anhedonia accompanied by a clustering of somatic, cognitive, and behavioral disturbances, including appetite or weight change, feelings of guilt or worthlessness, sleeping problems, fatigue, psychomotor changes, concentration problems, and thoughts of suicide that lasted two weeks or more were diagnosed as having MD [25, 26].
The present study employed a tool similar to that used in Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) or Research Diagnostic Criteria (RDC) or International Classification of Sleep Disorders (ICSD) to access insomnia. During the interview, there were three questions to access insomnia: (1) taking >30 min to fall asleep after going to bed or waking up in the middle of the night, (2) waking up early and not being able to go back to sleep and (3) needing to take medicine (including herbal or sleeping pills) at least once a week to help sleep. The interviewers asked the participants if they had any of the above three symptoms in at least three days or more in a week during the last month. If the participants answered ‘Yes’ to any of the above symptoms, then they were considered as having insomnia [27].
2.4 Other co-variates
Demographic and socio-economic characteristics collected in the baseline survey, specifically age at study date, sex, household size, highest level of education, and occupation, were included as co-variates in the analysis. Highest level of education was categorized into primary school/below and high school/above. Occupation was categorized into agriculture and related workers, factory workers, clerks and unemployed. Participants' health behaviors, including smoking status and alcohol use, were classified as “never,” “ever use”. BMI was calculated as weight in kilograms divided by height in meters squared [23]. BMI (kg/m2) was categorized as <20, 20-25, and >25 kg/m2, which was based on standard classification specific for the Chinese population [28].
Self-rated health was accessed by asking the participants: “How is your current general health status: excellent, good, fair, or poor?” in baseline interview and classified into four categories accordingly.
2.5 Data analysis
Descriptive analyses were used to illustrate the basic demographic, socioeconomic and lifestyle factors for those people with no child, one child, two children, and more than two children. Means (SD) and percentages were presented as descriptive results. Logistic regression models were fitted to explore the association between number of children and MD and insomnia prevalence. Odds Ratios (ORs) and 95% CIs were calculated to explore the association between MD and insomnia by the number of children, stratified by sexes. One child group was chosen as the reference group for the analysis. Two types of logistic regression models were fitted: 1) crude 2) fully adjusted (adjusted for rural/urban, marital status, age at study date, level of attained education, household income, smoking status, alcohol use, self-rated health, occupation, BMI). To understand how rural/urban, age, education, household income, smoking, alcohol use, and BMI potentially modify the associations between number of children and MD/insomnia, adjusted sex-specific ORs and 95% CIs for MD and insomnia per additional child stratified by the variables described above were calculated, after adjusting for study region, marital status, age at study date, highest education, household income, smoking status, alcohol use, self-rated health, occupation, BMI, where appropriate. All the data analysis was conducted by SAS version 9.4 (SAS Institute, Cary, NC, USA).