2.1 Study Population and Design
The research sample was derived from the Born in Shenyang Cohort Study (BISCS), which follows mothers and children from pregnancy to 12 months postpartum. The aim of the BISCS is to explore and verify risk factors for maternal and infant health. We recruited participants from April 2017 to September 2017. Pregnant women were enrolled from 54 hospitals and community health care centers providing antenatal and maternity care in the urban areas of Shenyang. The eligibility criteria included: (1) singleton pregnancy; (2) second trimester (14–27 weeks) at enrollment; and (3) no plan to move from Shenyang during the subsequent 3 years. All participants provided written information consent, and the study was approved by the ethics committee of China Medical University.
Participants were interviewed in person during the second trimester (mean ± SD: 23.63 ± 3.28 weeks). We also followed up mothers and infants at the child development clinics at an infant age of 3 months. We collected sociodemographic, environmental, behavioral, and clinical information on mothers and children by using standardized questionnaires.
In total, 1338 women agreed to participate in the study and 1260 had a live singleton birth. Among the 1260 women with live singleton births, 1152 completed the sleep quality, stress, and depression status questionnaires during the second trimester and 739 underwent postnatal depression assessment 3 months after giving birth. The response rate for the 3-month visit was 65%. In total, 1152 mother-child pairs were included in the present study. We compared the characteristics of the 1152 participants included in this study with those excluded. There were no significant differences in demographic indicators between the excluded individuals and those retained.
2.2 Exposure: Sleep Quality
We assessed sleep quality using the Pittsburgh Sleep Quality Index (PSQI) [17], a validated tool for the measurement of sleep quality in Chinese pregnant women [18, 19]. The PSQI includes a 19-item self-rating questionnaire that assesses sleep quality during the past month. The total score ranges from 0 to 21, with higher scores indicating worse sleep quality. Poor sleep quality was defined as a sum score of ≥ 5 in accordance with previous studies [19, 20]. The sensitivity and specificity of the questionnaires are 89.6% and 86.5%, respectively.
2.3 Outcome: Stress and Depression Status
We assessed pregnancy stress status using the Pregnancy Pressure Scale (PPS), which is a validated tool for Chinese pregnant women [21]. The PPS includes 30 items valued from 0 to 3 to obtain a score between 0 and 90, with higher scores indicating increased stress status. The cutoff point of a PPS standardized score > 0 reflects a state of stress [21, 22]. We assessed prenatal and postpartum depression status using the Edinburgh Postnatal Depression Scale (EPDS) in accordance with previous studies [23, 24]. The EPDS is a structured 10-item self-report measurement of depression during pregnancy. Items are scored with a value from 0 to 3, which give a sum score of 0 to 30 [23]. It is also validated for screening depression during pregnancy [25]. The cutoff point of an EPDS standardized score ≥ 9 reflects depressive symptomatology in the Chinese population (sensitivity, 80.0%; specificity, 83.0%) [26].
2.4 Covariates
We collected participants’ age, height, weight, educational level, household income, smoking status, gestational age (in weeks), and social support during the second trimester. We treated age (in years) as a continuous variable in multivariate regression analysis. Educational level was classified into four categories (junior high school or lower, senior high school, university, and postgraduate). Annual household income was calculated in Chinese Yuan. Household income was divided into five categories (<¥10,000, ¥10,000–¥30,000, ¥30,000–¥50,000, ¥50,000–¥70,000, and ≥¥70,000). Pre-pregnancy body mass index (BMI) was divided into two categories (< 23 and ≥ 23) in a multilevel logistic model [27]. Smoking status was treated as a dichotomized variable (yes/no). Gestational age at recruitment was treated as a continuous variable. We assessed the social support of pregnant women using the Social Support Rating Scale (SSRS), which has been widely applied to the Chinese population [28]. The SSRS includes 10 items and has a total score of 12 to 65. Higher scores indicate higher levels of social support. The SSRS has high reliability and validity in the Chinese population [29, 30]. The cutoff point of the SSRS is such that scores ≥ 45 mean high social support; scores below the cutoff mean low social support [31, 32]. Multiple imputation was used to impute missing values for SSRS scores (n=122) and gestational age (n=115). No missing values were observed for any of the other covariates.
2.5 Statistical Analysis
We used a t-test and chi-square tests to describe the characteristics of exposures and covariates classified by PSQI scores (< 5 vs. ≥ 5).
Multiple linear regression was used to investigate the associations of PSQI scores with PPS scores and EPDS scores. The following variables were considered potential confounders: maternal age, pre-pregnancy BMI, educational level, household income, pre-pregnancy smoking, gestational age, and social support.
We used logistic regression models to assess the odds ratio (OR) and 95% confidence interval (CI) for stress during pregnancy, antenatal depression, and postnatal depression in relation to sleep quality using PSQI scores of participants < 5 as reference. We conducted crude and adjusted analyses using the following models: Model 1, the crude model; Model 2, adjusted for maternal age, pre-pregnancy BMI, educational level, household income, pre-pregnancy smoking, and gestational age; and Model 3, additionally adjusted for social support based on Model 2. In addition, we conducted stratified analysis classified by women’s age (younger than 30 vs. older than 30 years old).
For sensitivity analyses, multiple imputation was used to impute the missing values (SSRS scores and gestational age), which may be confounders in the study. In addition, we used the age categories of younger than 30, 30 to 35, and older than 35 years old in stratified analyses for comparisons with previous results.
All analyses were conducted with Stata S.E. version 15 (Stata Corp., Texas, TX, USA).