Logistic Regression for Predicting Early Postpartum Depressive Symptoms in Japanese First-Time Mothers Aged 35 and Over

Purpose: Currently, Japan is showing a trend towards later childbearing. In 2006, 5.6% of women giving birth were first-time mothers at age 35 and over (1); this increased to 8.4% in 2011 (2). Older first-time mothers have been shown to be at increased risk for pregnancy complications compared with younger women (3, 4). These women also possess several psychosocial characteristics such as increased autonomy, limited social support and limited prior experiences with children (5), which may potentially cause poorer maternal adaptation after childbirth. Older first-time mothers may therefore demand more careful consideration from health care professionals not only during pregnancy but also after childbirth.

Previous studies suggest that women with advanced maternal age, particularly primiparae, are more vulnerable to postpartum depression (PPD) (6, 7). PPD is a common complication of childbirth. Prevalence of PPD in Japan is estimated to be between 8.6% and 17.0% (8). It is a significant health problem that impacts not only women, but also their children and families. Studies suggest that mother-infant interactions are compromised by PPD, which negatively impacts infant behaviors and development. The detrimental effects of PPD in more extreme cases include greater risks for child abuse, self-harm and suicidal ideation. Thus, preventing PPD is an important health issue in Japan.

In the present study, we report some of the findings of a prospective cohort study that we conducted to examine physical and psychosocial well-being during the first 6 months postpartum in Japanese mothers, one of three studies from a research project begun in 2011 for this purpose with older Japanese first-time mothers. The project is still in progress with a goal of developing a nursing guideline to support childrearing for older Japanese first-time mothers.

The purpose of this study was to identify factors for predicting postpartum depressive symptoms during hospitalization after childbirth in Japanese first-time mothers aged 35 and over.

Methods: The sample of this study came from the prospective cohort study that we conducted to examine physical and psychosocial well-being during the first 6 months postpartum in Japanese mothers. The study was undertaken at 12 hospitals in the Kanto and Kansai areas in Japan from May 2012 through September 2013. Women were recruited by researchers or research nurses during their hospitalization after delivery. Eligible participants were Japanese women who were 16 and over, 0 to 4 days postpartum, had a singleton live birth vaginally or operatively, and had a fixed address during the 6 months after the time of recruitment. Exclusion criteria included those who had difficulty with communication in Japanese and those with serious health problems, either in the mother or newborn. We obtained informed consent forms from all participants. Institutional ethics approvals were obtained if the institution had any review board before initiation of the study.

Data were collected using self-report questionnaires at 5 points in time: the day before hospital discharge, 1 month, 2 months, 4 months, and 6 months postpartum. The present study specifically reports the findings during the hospital stay. The self-report questionnaire during hospitalization consisted of the following contents: demographics and background information; depressive symptoms, as measured by the Japanese version of the Edinburgh Postnatal Depression Scale (the Japanese-EPDS) (9); fatigue, as measured by the Postnatal Accumulated Fatigue Scale (PAFS), which we developed; maternal confidence and maternal satisfaction, as measured by the Postpartum Maternal Confidence Scale and the Postpartum Maternal Satisfaction Scale, respectively (10); childcare values, as measured by the Childcare Value Scale (CVS) (11); checklists for physical symptoms; perceptions on childcare and daily life; and perceptions on the nursing care they received during the hospital stay. In addition to questionnaire responses by participants, researchers or research nurses obtained additional data from vital records data at the hospitals.

Data were analyzed using stepwise logistic regression through SPSS ver. 21. The two-category outcome was whether a woman scored >9 in the Japanese-EPDS. To generate candidate variables for entry into a stepwise logistic regression analysis, we initially performed univariate analysis via χ² tests for categorical variables and Mann-Whitney U-tests or Kruskal-Wallis tests for those with nonparametric distributions. We introduced significant (p<.05) or nearly significant (p<.10) independent variables from the univariate analysis into a stepwise logistic regression analysis. When independent variables that were known to be predictors of postpartum depression from previous studies failed to show significant correlations with the outcome (the Japanese-EPDS), we decided to enter those variables into a stepwise logistic regression analysis. Independent variables were entered in two sets: type of delivery (vaginal vs. elective cesarean section (CS) vs. emergency CS) was entered first, and then all remaining variables were entered based on the maximum likelihood-ratio statistic.

Results: Of the original 3769 women who agreed to participate, 3633 women responded to the questionnaire during the hospital stay (96.4% response rate). After deletion of 292 cases (47 women with diagnosed or previous history of prenatal psychological disorders including depression, 22 women who did not respond within the specified time limit, 152 women who did not practice rooming-in during the hospital stay, 70 women with missing values for main outcome measures, and 1 woman without vital records data), data from 3341 women were available for analysis. From this sample, we created a group of 461 first-time mothers aged 35 and over, which we used for the analysis of the present study.

              Nearly 21% of women scored >9 on the Japanese-EPDS during the hospital stay. Univariate analysis resulted in 14 independent variables that showed significant correlations with the Japanese-EPDS. Depressive symptoms were more strongly associated with: emergency cesarean section, gestational diabetes or diabetes in pregnancy, husband/partner with more advanced age, higher physical burden during hospital stay, lower satisfaction with sleep, not exclusively breastfeeding, taking more minutes for baby feeding, lower satisfaction with birth experience, more concerns about newborn caretaking after discharge, more concerns about one’s own life after discharge, more concerns about social support after discharge, newborn lower birth weight, newborn long-term complications, and the pregnancy desired most not by the mother or couple. A stepwise logistic regression analysis showed that only six predictors provided an adequate model among Japanese first-time mothers aged 35 and over. Emergency cesarean section (odds ratio (OR) of 2.88, 95% confidence interval (CI) of 1.47, 5.63), lower satisfaction with birth experience (OR 2.07, 95% CI 1.24, 3.45), higher physical burden in daily life (OR 3.04, 95% CI 1.11, 8.29), newborn long-term complications (OR 4.63, 95% CI 1.79, 11.96), more concerns about newborn caretaking after discharge (OR 1.21, 95% CI 1.05, 1.39), and more concerns about one’s own life after discharge (OR 1.24, 95% CI 1.08, 1.41) were associated with depressive symptoms. For the logistic regression model, the Hosmer and Lemeshow test was not significant (χ²=10.72, df 7, p=0.218), indicating that the data fit the model well. Of women who scored <9 in the Japanese-EPDS, 95.0% were correctly classified by the model; of those who scored >9, 29.6% were correctly classified. The overall success rate of prediction was 81.1%.

Conclusion: The findings of logistic regression analysis indicated that women with lower birth experience satisfaction showed more depressive symptoms, as did women who experienced emergency CS compared with those with vaginal delivery. The birth experience is potentially traumatizing for some women. Mothers may experience distress if they require medical interventions such as vacuum extraction and emergency CS, especially when they expected to have a natural delivery. Because older first-time mothers often require more medical interventions during delivery, nurses should be sensitive to women’s perceptions of their birth experience, especially when women experience emergency CS unexpectedly. Higher physical burden was found to be another significant predictor for depressive symptoms. This indicates the importance of assessing women’s perceptions of their physical burden during the hospitalization. Postpartum hospital stays in Japan are usually 4 to 5 days following a vaginal birth, a few days longer in cases of CS. Asking postpartum women about their perceived physical burden and providing appropriate care during hospitalization will help to prevent depressive symptoms. Women who had more concerns about their own life and newborn caretaking after discharge also showed more depressive symptoms. An additional predictor of depressive symptoms was newborn long-term complications that included congenital diseases. Nursing assessment of women should include their concerns about life including newborn caretaking after discharge and perceptions of newborn long-term complications. These assessments will lead to appropriate nursing support which may help to prevent PPD for older first-time mothers.

In summary, logistic regression analysis identified the following six predictors for PPD in Japanese first-time mothers aged 35 and over: emergency cesarean section, lower satisfaction with birth experience, higher physical burden during hospital stay, newborn long-term complications, and more concerns about one’s own life including newborn caretaking after discharge. Recognition of women with these factors will help nurses to identify those at risk for developing PPD and to provide appropriate care during hospitalization after childbirth.