Japan has been affected by the coronavirus disease 2019 (COVID-19) pandemic as elsewhere, with the first case reported on December 31, 2019. Although COVID-19–related deaths have been relatively on a smaller scale, health and financial uncertainty coupled with the resulting social isolation have been severe, and the pandemic has turned into a crisis. Studies conducted to explore the psychological impact in a state of emergency (mild lockdown) on the general population reported a significantly higher proportion of people displaying psychological distress than that reported in previous national survey data due to deterioration of the household economic situation, difficulties in work and study, loneliness, and lack of social interations. ^{Reference Fukase, Ichikura and Murase1–Reference Odani, Shinozaki and Shibuya3} Mothers with infants are one of the most vulnerable groups. In fact, depression during the first year after childbirth, frequently called as postpartum depression (PPD) is not uncommon and estimated to be around 13% worldwide, ^{Reference O’Hara, Rehm and Campbell4,Reference O’Hara and Swain5} and according to a recent review from Japan based on studies published from January 1994 to December 2017, the prevalence of depression is 15.1% within the first month of delivery, 11.6% in 1-3 mo, 11.5% in 3-6 mo, and 11.5% in 6-12 mo after birth. ^{Reference Tokumitsu, Sugawara and Maruo6 [} Footnote ^{a ]} To understand the mechanisms of PPD, there are 2 main classical models, and a newly proposed one based on these classical models. First, classical biological models indicate the dramatic rise (before delivery) and drop (after delivery) of reproductive and stress hormones, which is hypothesised to trigger system dysregulation and depressive symptoms in a subset of vulnerable women. ^{Reference Douma, Husband and O’Donnell9–Reference Moses-Kolko, Berga and Kalro11} Second, psychological models focus more on socio-economic factors that raise the stress level of mothers. The deleterious effect of stressors, including financial strain and father abandonment of social support have been reported earlier. ^{Reference Pearlin, Lieberman and Menaghan12,Reference O’Hara, Rehm and Campbell13} Furthermore, underlying cognitive vulnerabilities, negative attributional style for instance, and individual psychological resources, such as self-esteem, would be harmful if it is lower or depreciated. ^{Reference Pearlin, Lieberman and Menaghan12,Reference O’Hara, Rehm and Campbell13} More recently, the bio-psycho-socio-cultural model was proposed ^{Reference Halbreich7} that emphasizes the integrated role of biological and psychological factors and that stress can trigger depression in postpartum women with genetic, hormonal, and cognitive vulnerabilities, and that it may interact reciprocally with the social and cultural environment.

Based on the bio-psycho-socio-cultural model, which suggests that the mechanism of PPD is dynamic and multidimensional, the COVID-19 crisis is speculated to have increased the risk for depression in mothers with infants due to unexpected labor environment, including prohibited entry of family members or a partner into the delivery room, and change in hospitals for delivery.^[ Footnote ^{b ]} Moreover, they are likely to be anxious regarding the risk this novel virus posed to them and their babies. Travel restrictions and self-quarantine results in lowered social support, both informal and formal, and are an extra mental and physical burden especially for mothers with infants who require constant attention.

Recent reports across the world revealed that mothers during the postpartum periods were found to show a higher percentage of depressive symptoms than before the crisis, indicated by the Edinburgh Postnatal Depression Scale (EPDS) as follows: 28.6% (EPDS >12) in Italy, ^{Reference Zanardo, Manghina and Giliberti15} 23.6% (EPDS >12) in Belgium, ^{Reference Ceulemans, Hompes and Foulon16} and 30% (EPDS ≥ 10) in China. ^{Reference Liang, Wang and Shi17} Because these reports used a nonuniform definition of the point prevalence of depressive symptoms in terms of EPDS cutoff and the timing of the survey, comparison between the countries are difficult. However, it is important to note that the prevalence was consistently more than double the values in the pre-crisis period for each country. Notably, these countries differ in the severity of the infection spread. The study in Italy used the data from northeastern Italy, where the infection was a “hot spot”, while China and Belgium used the data from a less severely infected area. One common thing in all the 3 countries was that they all experienced lockdown, and their results suggested the negative effects regardless of the infection rate. Japan experienced a low infection rate with mildly imposed lockdown, and studies in Japan also found negative impacts of COVID-19 on postpartum women. ^{Reference Suzuki18,Reference Tsuno, Okawa and Matsushima19} The shortcomings of the previous studies were, however, they have not explored details of adverse events related to pregnancy/delivery/childcare, economic crisis and uncertainty during the crisis, and fear of the infection of COVID-19, which are important for policy implications.

PPD has a serious repercussion as a recent review, which included 122 studies, revealed that mothers with PPD may suffer prolonged depression/ depressive symptoms, have difficulties in their relationship with their partners, and have suicidal ideation. Infants are also affected, including poor health, and slow motor, cognitive, and language development. Mother-to-infant bonding is often negatively affected, which has long-term consequences. ^{Reference Slomian, Honvo and Emonts20} Thus, the significance of this study is to explores the depressive symptoms in postpartum women during the coronavirus disease (COVID-19) crisis in Japan with focuses on crisis induced experiences they have undergone.

Methods

We used a cross-sectional study design, in which an online survey was conducted from May 31 to June 6, 2020, through 2 companies (Karada note Inc. and baby calendar Inc.) that provided services to pregnant and postpartum women and mothers with children. We sent emails to all the users soliciting voluntary participation in our survey. Because users of the services of these companies included nonactive and nontarget groups, we could not calculate the exact percentage of respondents among the women to whom emails were sent. We sent 607,458 emails, among which approximately 1.3% of the participants accessed our survey Web page. Among those who accessed our survey Web page, approximately 74% replied (5650 replies). The respondents were from all prefectures, except Wakayama, corresponding to the population size of each prefecture.

Among the 5650 replies, 3406 were from women with babies under 12 mo of age. Among them, we used 3073 responses of women who completed the entire survey for the present study.

Data Collection

We collected the data from the self-administered questionnaires. The Japanese version of the EPDS was used to measure depressive symptoms. ^{Reference Okano21} In this study, good internal consistency of the EPDS (10 items) was shown, with a Cronbach’s alpha of 0.87. The questionnaires also included perceived risk with respect to COVID-19 infection and its negative consequences, unexpected experiences under the influence of this crisis, and socio-demographic and economic information.

Analyses

We divided our sample into 2 groups: primiparas and multiparas.^[ Footnote ^{c ]} Four procedures were conducted for each group. First, we calculated the total score and the percentage of women with an EPDS score of ≥ 9 each month after delivery. The validity of the cutoff score being 9 has been validated by Okano, ^{Reference Okano21} and this is the commonly used one in Japan. Second, factor scores of anhedonia (EPDS items 1 and 2), anxiety (EPDS items 3, 4, and 5), and depression (EPDS items 7, 8, and 9), which was validated by Kubota et al., similar to those in Western countries, ^{Reference Kubota, Inada and Nakamura23} were calculated. Third, logistic regression analysis was performed by setting the dependent variable as 1: EPDS score of ≥ 9; 0: EPDS score of < 9. Fourth, ordinary least squares (OLS) regression analysis was conducted for each factor score after adjusting for the respondents’ socio-demographic and economic backgrounds (Table 1, the list of covariates and their definitions). The latter analyses are important to explore risk factors in detail. Although the cutoff of EPDS score is useful to capture the general prevalence of depressive symptoms, it may not fully cover the risks associated with elevated unwellness of mental health among mothers because just below the cutoff does not mean the mother is free from depression risks. Thus, variables contributing to an increase in the score of each factor should not be overlooked.

Table 1.

Summary statistics

As our focused variables, we included 3 COVID-19 crisis-related variables as follows: experiences during the COVID-19 crisis, perceived risk, and place of residence. Places of residence were grouped into 5 based on the day the state of emergency was lifted in their prefecture. The longest duration was in Tokyo/Kanagawa/Saitama/Chiba followed by Hokkaido, Osaka/Kyoto/Hyogo, and Aichi/Fukuoka. All statistical analyses were performed using STATA/MP 15.1.

This study was approved by the Research Ethics Committee of the University of Tsukuba (No. 2020-1), and no potential conflict of interest exists relevant to this research. To protect respondents from mental disturbance possibly caused by answering our questionnaires, the letter of request included a summary of the contents of our survey. Moreover, participants could withdraw from the survey anytime, and all questions were set as non-mandatory.

Results

The point prevalence of postpartum women with an EPDS score of ≥ 9 was 28.66% for primiparas and 25.83% for multiparas. Exploring the scores and the point prevalence for each month after delivery by groups, primiparas and multiparas, the depressive symptoms were prevalent throughout. For primiparas, the highest prevalence was at 1 mo (37.2%) and the lowest was at 3 mo (20.7%), and a continuous and steady increase was observed after 6 mo up to 10 mo. For multiparas, there was an increasing trend after 3 mo of delivery, and it reached the highest at 11 mo (36.8%). When comparing the average point prevalence of depressive symptoms, primiparas was higher than multiparas, showing 29.9% and 22.1%, respectively. Nevertheless, after 3 mo of delivery, the average percentage became higher for multiparas (29.2%) than for primiparas (27.7%). These results are shown in Figure 1.

Figure 1.

Point prevalence of EPDS ≥ 9 for each month.

The trends for 3 factors are shown in Figure 2. For primipara, the lowest total EPDS scores at 3 mo were 2.728 (anxiety), 0.315 (anhedonia), and 1.000 (depression), and the highest total EPDS scores at 1 mo were 3.769 (anxiety), 0.500 (anhedonia), and 1.769 (depression). For multiparas, the lowest total EPDS scores at 2 mo were 2.309 (anxiety), 0.573 (anhedonia), and 1.136 (depression), and the total EPDS scores at 11 mo were 3.500 (anxiety), 0.789 (anhedonia), and 1.579 (depression). Among the 3 factors, the anxiety score was the highest, even when it was divided by the number of factors,^[ Footnote ^{d ]} and it showed a general increasing trend for mothers having older infants. At 11 mo after delivery, the anxiety and anhedonia scores of multiparas were higher than for primiparas.

Figure 2.

Factor scores for each month.

Seven types of experiences during the COVID-19 crisis were listed. The top 3 most frequently experienced unintended events during COVID-19 were as follows: for primiparas, lifestyle changes with a partner working from home (35.96%), the prohibition of visitors at hospitals (34.12%), and cancellation of planned formal support experienced (32.97%); and for multiparas, the prohibition of visitors at hospitals (46.02%), cancellation of planned formal support experienced (35.96%), and cancellation of parenting classes (34.01%). Although less than 3%, some mothers needed to change the place of delivery, and more that 15% of people were affected by the cancellation of informal support.

Another COVID-19–related variable was the perceived risk. The highest perceived risk pertained to not being able to receive formal childcare support, followed by COVID-19 infection, financial difficulties, and not being able to receive informal childcare support.

The respondents’ socio-demographic/economic variables showed that nearly 40% were in the age range of 30-34 y with multipara being older in general, and nearly 98% were married. Approximately 80% of the respondents had more than 16 y of education with a higher percentage among the primiparas. More than 40% had an annual income (including taxes) of above 5 million yen with a higher percentage among the multiparas. Regarding working status, approximately 55% and 45% of the primiparas and multiparas, respectively, were working full-time.

Analysis of Primiparas

Table 2 shows the results of the analysis on primiparas. Logistic regression analysis for primiparas revealed the significant effects of COVID-19–related issues. Concerning experiences of unexpected changes, participants who had experienced cancellation of planned informal support and formal support had a higher risk of having depressive symptoms (odds ratio [OR], 1.358; 95% confidence interval [CI], 1.024-1.802 and OR, 1.292; CI, 1.013-1.647, respectively). For perceived risk variables, mothers perceiving higher risks of financial difficulties and those who were not able to receive informal childcare support were independently associated with an EPDS score of ≥ 9 (OR, 1.099; 95% CI, 1.030-1.173 and OR, 1.227; CI, 1.142-1.319, respectively). Postpartum mothers who lived in areas with a longer duration of the state of emergency had a higher probability of EPDS ≥ 9 than the regions that had the shortest duration of the state of emergency (OR, 1.405; 95% CI, 1.066-1.851 for Tokyo/Kanagawa/Saitama/Chiba). Except for COVID-19–related variables, only those with lower incomes were associated with a higher probability of having depressive symptoms.

Table 2.

Results of analysis of primiparas

Note: *** P<0.001, ** P<0.01, *P<0.05.

For OLS regression, we use robust standard errors. They are not shown for brevity.

For further analysis, anxiety, anhedonia, and depression were considered. Change in the place of delivery, the prohibition of permitting partner into the delivery room, cancellation of planned informal support, and living in Osaka/Kyoto/Hyogo were the variables that did not show significant associations with EPDS ≥ 9. The coefficients suggested that women who experienced a change in the place of delivery were likely to demonstrate an increase in the anxiety and depression score by 0.854 and 1.433, respectively. These coefficients were larger than other variables of experience during the COVID-19 crisis. Other notable findings were: having a partner at home and having more people who were involved with childcare daily, which reduced the depression score. Additionally, anhedonia score decreased with an increase in the number of people engaged in childcare support, whereas being unemployed resulted in an increase in all the factor scores. Although not significantly associated with EPDS ≥ 9, the additional analysis suggested a higher risk for unemployed mothers.

Analysis of Multiparas

Table 3 shows the results of the analysis of multiparas. Logistic regression revealed that similar to primiparas, COVID-19–related variables had a significant association with elevated depression symptoms. Additionally, these associated variables could be categorized as childcare support and financial issues. Those who experienced cancellation of planned formal support and those who perceived higher risk of not receiving informal childcare support had a higher probability of exhibiting depressive symptoms (OR, 1.39; 95% CI, 1.017-1.815 and OR, 1.253; 95% CI, 1.146-1.369, respectively). Having a higher perceived risk of financial difficulties, lower income group, and being unemployed also contributed to having depressive symptoms (OR, 1.084; 95% CI, 1.002-1.173, OR, 1.377; 95% CI, 1.033-1.835, and OR, 3.142; 95% CI, 1.224-8.964, respectively). In addition to these variables, the variable, residential area, suggested that a mother living in Tokyo//Kanagawa/Saitama/Chiba had a higher probability of EPDS ≥ 9 (OR, 1.424; 95% CI, 1.030-1.969 for Tokyo). Furthermore, the dummy variable indicating the baby’s age as being 6 months or over shows a higher probability of having depressive symptoms.

Table 3.

Results of analysis of multiparas

Note: *** P<0.001, ** P<0.01, *P<0.05.

For OLS regression, we use robust standard errors. They are not shown for brevity.

Analyses of factors added more evidence that COVID-19-related variables had a significant association with elevated anxiety, anhedonia, and depression; change in the place of delivery and cancellation of parenting classes both increased anxiety scores by 0.678 and 0.426, respectively. Cancellation of planned formal support increased anhedonia and depression scores by 0.141 and 0.241, respectively. As the results of logistic regression indicated, higher perceived risk toward financial difficulties and not receiving informal childcare support also increased all the factor scores. A lower income level was associated with anhedonia and depression, although it did not show a statistically significant association with anxiety. Compared with the never-married individuals, mothers who were either divorced, widowed, or in a different marital status had a lower score for anhedonia, and married women had a lower score for depression. Consistent with the results of logistic regression, mothers with older babies recorded higher scores for all the factors.

Limitation

Our study has some limitations. First, this study involved an online survey by sending e-mails to the registered users of 2 companies’ maternal-related services. Because registered users were not always the active users, and their e-mail may not be opened, we could obtain only an approximate response rate. In relation to that, representation of the sample needs to be carefully taken into consideration due to the nature of the online survey. Second, there is no control group for this study. Considering the impossibility to overcome these limitations totally, research to accumulate the evidence would be helpful.