Study design and population
The present study was based on the Tohoku Medical Megabank Project Birth and Three Generation Cohort Study (TMM BirThree Cohort Study). The TMM BirThree Cohort Study details have been shown previously [17-20]. Between July 2013 and March 2017, trained genome medical research coordinators explained the information to the possible participants in each clinic, hospital, or community support center in Miyagi and Iwate prefectures and obtained signed consent from each participant. Of the 32,986 pregnant women who were contacted, 23,406 pregnant women, including multiple different pregnancies, were registered. The following exclusion criteria were applied to 23,130 mothers and children: withdrew informed consent (n = 505), multiple participation in the survey (n = 875), missing data for maternal psychological distress in early pregnancy (n = 720), missing data for maternal psychological distress at two years postpartum (n = 8,880), missing information on children's behavior problems at four years of age (n = 5,716), missing information on childcare facilities (n = 41), and taking psychotropic drugs between pre- to early-pregnancy (n = 134). A total of 6,259 mother and child pairs were included in the analysis (Figure 1).
Maternal psychological distress
The K6 was used to assess the maternal psychological distress in early pregnancy and two years postpartum. K6 is a brief scale consisting of six questions; additionally, the Japanese version of K6 was developed and validated [21,22]. We defined K6 ≥ 5 points as psychological distress in this study. This value has been shown to maximize the sum of sensitivity and specificity in screening for mood and anxiety disorders in the Japanese version of K6 and has also been used in the previous study [23,24]. Furthermore, we classified mother and child pairs based on psychological distress into four categories: no psychological distress in both prenatal and postnatal periods (none), only the prenatal period (prenatal only), only the postnatal period (postnatal only), and both the prenatal and postnatal periods (both).
Children's behavior problems
The Child Behavior Checklist for Ages 1½-5 (CBCL) was used to assess the behavioral problems in children at an age of four years [25]. The CBCL is a parent-completed screening method. It consists of 100 items and is divided into seven syndrome scales (emotionally reactive, anxious/depressed, somatic complaints, withdrawn, sleep problems, attention problems, and aggressive behavior) [25]. Furthermore, the scale of total problems is defined as the sum score of all 100 items on the CBCL, and the scale of internalizing problems is defined as the sum score of being emotionally reactive, anxious/depressed, somatic complaints, and being withdrawn [25]. The scale of externalizing problems is defined as the sum score of attention problems and aggressive behavior [25]. The T-score (mean = 50, standard deviation = 10) for each of the scales was calculated and standardized for Japanese children [26]. A standardized T-score in the range of 60–63 indicates the borderline clinical range, while a standardized T-score of 64 or higher indicates the clinical range [26]. In this study, the clinical range (standardized T-score ≥ 64) of internalizing, externalizing, and total problems scales were analyzed as behavior problems [26].
Use of childcare facilities
Information on the use of childcare facilities was collected from the questionnaire at two years postpartum. The parents responded as to whether their children were attending childcare facilities at age two years or not.
Covariates
We have selected covariates that may influence the association between maternal psychological distress and behavioral problems in children, referring to previous studies [1-11,27-31]. We collected information regarding the children's sex from birth records. Information regarding maternal age (years) and parity (never, one, or more) was collected from the medical records. The maternal age was divided into four categories (< 25, 25–29, 30–34, ≥ 35 years). The information on maternal alcohol consumption in early pregnancy (never, ever, current), maternal cigarette smoking in early pregnancy (never, stopped before pregnancy, stopped after pregnancy, current), and paternal smoking in early pregnancy (never, stopped before pregnancy, stopped after pregnancy, current) were gathered from the questionnaire during early pregnancy [27-30]. Information on household income (< 4,000,000, 4,000,000–5,999,999, ≥ 6,000,000 Japanese yen/year) was collected from the questionnaire during mid-pregnancy [31]. The maternal educational attainment data (high school graduate or less, junior college or vocational college graduate, university graduate or above, others) were gathered a year after birth [27,31].
Statistical analysis
The characteristics of the participants according to the four categories of maternal psychological distress were shown descriptively. The data were presented as frequency and percentage since all the variables were categorical variables. To confirm the association between maternal psychological distress of four categories and scales of internalizing, externalizing, and total problems, we first conducted a multivariable logistic regression analysis, using the none category as a reference (no psychological distress in both prenatal and postnatal periods). Furthermore, to examine whether the use of childcare facilities moderates the association between maternal psychological distress and children’s behavior problems, we conducted a stratified analysis based on the use of childcare facilities at two years of age. Moreover, the interaction term between maternal psychological distress and the use of childcare facilities was included as a covariate in the multivariate logistic regression analysis to confirm the p-value for interaction. Multiple imputations imputed incomplete confounders by chained equation [32]. Plausible synthetic values were generated from the given exposure, outcome, and other confounders in the data. Twenty sets of quasi-complete data were analyzed in the multivariate analyses independently and integrated the estimates. All statistical analyses were performed using R, v.4.0.2. A P-value < 0.05 was considered as statistically significant.