The COVID-19 pandemic and the experience of postpartum depression

Abstract

Background: Early studies assessing peripartum mood disorders during the COVID-19 pandemic have conflicting results.

Objective: The primary aim was to examine if postpartum depression (PPD) was more common or more severe during the COVID-19 pandemic compared to earlier periods, and to assess what individual factors may worsen the impact of the pandemic.

Methods: Electronic health records at an academic pediatric practice in Michigan identified 242 biological mother-child dyads with delivery between 1/1/2017 and 12/31/2021. PPD was evaluated using the Edinburgh Postnatal Depression Scale (EPDS) during well-child visits. Participants were divided into three groups for analysis: Pre-Pandemic (n = 100), Early Pandemic (n = 93), and Later Pandemic (n = 49). Logistic regression analysis was used to predict PPD development, controlling for background factors.

Results: After controlling for confounders, the three groups did not differ significantly. Preexisting mental health conditions was a significant (p<.001) moderator; PPD rates peaked early in the pandemic (60%), compared to late pandemic (42%) and pre-pandemic (36%). Women without a mental health diagnosis pre-pregnancy experienced the lowest levels of PPD during the pandemic.

Conclusions: Rates of PPD were not significantly affected by the COVID-19 pandemic for most women. However, biological mothers with preexisting mental health conditions had significantly higher rates of PPD.

Keywords:

• Postpartum depression (PPD)
• COVID-19
• EPDS: Edinburgh Postnatal Depression Scale
• peripartum depression

Introduction

Postpartum depression (PPD), typically diagnosed if a new biological mother meets standard DSM-5 criteria for Major Depressive Episode for one to twelve months in the postpartum period [1], is a significant individual and public health concern. In a 2020 Morbidity and Mortality Weekly Report (MMWR) from the Centers for Disease Control and Prevention (CDC), 13.2% of women in the United States (U.S.) who were surveyed within a period of six months after a live birth reported experiencing depressive symptoms [2]. The CDC notes how postpartum depression rates have been known to vary by geographic area and demographic, and the state of Michigan has a higher incidence than the national average at 16.1% [3]. It’s important to note how many studies have shown an association between PPD and effects on both biological mothers and their children. Biological mothers experience impacts on their physical health, psychological health, social relationship, and quality of life. Children experience long-term effects on emotional and cognitive development [4]. Therefore, understanding the factors influencing PPD is not only beneficial to patients, their children, and their families but also for the healthcare system itself. This was particularly crucial during the time period of the COVID-19 pandemic in which many individuals experienced new or heightened socioeconomic stressors as a result of illness, job loss, and societal lockdowns.

On March 11, 2020, the World Health Organization (WHO) declared COVID-19 a pandemic, with the novel virus affecting billions across the globe. Overall, the WHO reported the global prevalence of anxiety and depression increased by 25% in the first year of the pandemic. It was suspected that this may be the case for biological mothers giving birth during this time as well. For the vulnerable population of biological mothers expecting to deliver during the pandemic, several unique challenges and anxieties arose. The United States Department of Health and Human Services reported that the pandemic specifically worsened existing disparities in access to care and health outcomes, particularly for rural and low-income populations [5]. New biological mothers during the pandemic reported elevated emotional stress, adverse breastfeeding experiences, changes to birth plans, and even some surprising decreases in the numbers of adverse birth events secondary to the pandemic. A large multisystem review of the effects of the COVID-19 pandemic indicated increases in several perinatal maternal psychosocial measures including anxiety, postnatal depression, trauma, dissociation, and affect [6]. Additional impacts noted in the review include significantly increased rates of cesarean sections, stillbirths, ectopic pregnancies, hypertension, birth weight, and gestational diabetes. The decreases in adverse birth events noted in the review include significantly reduced preterm births (<37 weeks) in certain subgroups and a reduction in very low (<1500 g) birthweight infants [6].

When specifically assessing the association between the COVID-19 pandemic and PPD, early studies have incongruent results. A recent study sought to describe postpartum depression and associated risk factors among postpartum patients in the United States between February and July 2020, and this study found one in three participants who delivered during the pandemic suffered from increased postpartum depressive symptoms if they experienced certain risk factors such as NICU admission or formula feeding [7]. Another study from the Pacific Northwest found a stable incidence of PPD despite an increasing prevalence of mood disorders [8]. International studies also had conflicting results on the association between PPD rates and the pandemic, with one study in Iran showing large increases in PPD and another in Greece showing no significant increase [9,10]. Given the impact of geographical location and conflicting results from newly released studies, the goal of this study was to determine whether PPD in the U.S., both rates and severity, were influenced by the COVID-19 pandemic by examining differences in rates prior to and after March 2020. Additionally, this study also sought to assess individual factors that put new biological mothers at risk for PPD during the pandemic.

Materials and methods

This IRB-approved retrospective chart review study collected data on biological mothers who gave birth at a single academically affiliated medical center in the Midwestern United States. In order to be included, the biological mother must have attended at least two follow-up visits within 24 months at the affiliated pediatric practice.

Selection and description of participants

Participants were initially identified for inclusion through pediatric outpatient medical charts with subsequent access to the biological mother’s (all participants identified as female) medical chart. Inclusion criteria were: (1) delivery at the affiliated hospital system between 01/01/2017 and 12/31/2021, (2) child was seen for a newborn well-child visit and at least one additional well-child visit, and (3) at least two available Edinburgh Postnatal Depression Scale (EPDS) screenings conducted with the biological mother at well-child visits. Participants were excluded if the biological mother or baby passed away within the baby’s first year.

Data source and variable collection

A data collection sheet was generated after identifying variables that may be involved in the association between PPD and the COVID-19 pandemic. Participant data for the relevant study variables were manually collected through a review of maternal and pediatric electronic medical charts in Epic. Maternal charts were reviewed for pertinent past medical history, past and/or current diagnoses of PPD, substance use, obstetrical history, and other important socioeconomic demographics. Pediatric charts were reviewed for birth history and standard well-child visit information at newborn, one month, two months, four months, six months, and twelve months. The general practice at the clinical site was to screen biological mothers for depression symptoms at each well-child visit up through the 9-month well-child visit, utilizing the EPDS and inquiry about whether they had been diagnosed with and/or treated for PPD since the birth of the child.

Participants were divided into three study groups based on when they gave birth. The Pre-Pandemic group included births between 01/01/2017 and 04/01/2019 (i.e. the youngest child was ten months of age at the start of the pandemic). The Early Pandemic group included births between 3/10/2020 and 12/31/2020. The Late Pandemic group included births between 01/01/2021 and 12/31/2021, which represented deliveries after lockdowns were lifted and vaccines were available. For purposes of the current study, the Pandemic Transitional group, births between 10/01/2019 and 02/28/2020, were excluded due to exposure to both pre-pandemic and pandemic conditions.

The primary outcome was postpartum depression. The EPDS was used to quantify symptoms of PPD with the screening conducted at each attended well-child visit via paper questionnaire or verbal interview and reviewed by a social worker. For the purpose of this study, a score of 10 or greater on one or more EPDS, with actual diagnosis confirmed by a follow-up interview by a social worker, was diagnostic of PPD. Patients were also considered to have PPD if they responded affirmatively to having been diagnosed or treated by another provider for PPD. For analysis, EPDS scores were analyzed both as a continuous score (highest score for each biological mother), indicating level/severity of PPD, and as a dichotomous presence or absence of PPD based on cutoff scores and clinical diagnosis, for which we examined rates for each study group. As noted, all biological mothers had at least two EPDS screenings, with the average number across the sample greater than three screenings.

Data analysis was performed using SPSS version 26. Examination of differences between the three pandemic groups on background factors utilized ANOVA F tests for continuous factors and chi-square analysis for categorical factors. These analyses were also used to examine group differences in rates and level of PPD. Logistic regression analysis was performed to look at the link between timing of delivery and PPD status controlling for background factors (from Table 1) significant at p<.10, and multiple regression was used to look at the link between timing of delivery and PPD level controlling for background differences. Finally, to examine possible risk factors for the development of PPD related to the pandemic, subgroup analyses were performed using chi-square analysis, with interaction terms tested in parallel logistic regression analysis. Variables considered here were non-parental childcare (biological mothers with help caring for children other than father or current partner), health of the newborn (admitted to NICU at birth stay), history of maternal substance use, and history of pre-pregnancy mental health issues (Figures 1–5). For all analyses except the identification of potential confounders, a p value less than .05 was considered significant.

Figures 1-5. Logistic regression analyses controlled for significant background factors and entered the interaction term on the final step.

Table 1. Participant background characteristics by timing of delivery.

	Pre-Pandemic	Early Pandemic	Later Pandemic
	(n = 100)	(n = 93)	(n = 49)	F/χ2	p
Maternal age (years)	25.6 (3.2)	26.3 (3.3)	27.9 (4.1)	2.91	.056
Marital status (% married/cohabitating)	76.2%	65.3%	73.5%	3.01	.220
Race (% white non-Hispanic)	42.6%	35.8%	51.0%	3.15	.207
Medical insurance (% private)	10.9%	18.9%	24.5%	4.92	.085
Childcare (% with some amount of non-parental care)	36.6%	21.1%	40.8%	8.02	.018
Existing maternal mental health issues (%)	44.6%	26.3%	49.0%	9.81	.007
Current maternal substance use (% any)	50.5%	41.1%	55.1%	3.08	.215
Current maternal substance use (% excluding tobacco use)	44.6%	31.6%	51.0%	6.05	.049
Newborn admitted to NICU (%)	27.7%	17.9%	26.5%	2.89	.235
Low birth weight (% <2.5 kg)	13.0%	12.4%	16.3%	.46	.796
Preterm delivery (% <37 weeks)	19.2%	11.4%	12.2%	2.59	.274

Results

The final study sample included 100 women in the Pre-Pandemic group, 93 women in the Early Pandemic group, and 49 women in the Later Pandemic group. First examined were group differences on background characteristics. The Later Pandemic group was more likely to be older, have private insurance, utilize some form of non-parental childcare, have existing maternal mental health diagnoses, and have current maternal substance use excluding tobacco compared to the other two groups. Marital status, race, medical insurance status, number of children in the home, current maternal substance use, newborn admittance to NICU, low birth weight, and preterm delivery were significantly different between the cohorts (Table 1). Overall, the sample was demographically representative of the larger clinic population from which it was drawn, and it did not differ significantly on any of these factors (analyses not shown).

After controlling for potentially confounding differences, study group membership did not significantly predict postpartum depression rates or levels. A comparison of the three groups found no significantly different levels of PPD nor rates of PPD. Further, no control variables included in the models significantly predicted PPD rates or levels (Table 2).

Table 2. Regression results predicting PPD status and level.

Logistic Regression Predicting PPD Status	Odds Ratio (95% confidence interval)
Background factors
Maternal age	.98 (.93 to 1.03)
Income dependent insurance (ref: private)	.75 (.34 to 1.66)
Some level of childcare (ref: complete parental)	1.05 (.57 to 1.93)
Maternal drug use (ref: no use except tobacco)	1.28 (.71 to 2.30)
Timing of birth
Early pandemic (ref: pre-pandemic & later pandemic)	.81 (.42 to 1.54)
Later pandemic (ref: pre-pandemic & early pandemic)	.89 (.41 to 1.93)
Linear Regression Predicting PPD Level	Adjusted Mean Difference (95% confidence interval)
Background factors
Maternal age	−0.07 (-0.22 to .07)
Income dependent insurance (ref: private)	−0.53 (-2.73 to 1.67)
Some level of childcare (ref: complete parental)	.39 (-1.29 to 2.07)
Maternal drug us (ref: no use except tobacco)	.73 (-0.90 to 2.34)
Timing of birth
Early pandemic (ref: pre-pandemic & later pandemic)	−0.55 (-2.17 to 1.07)
Later pandemic (ref: pre-pandemic & early pandemic)	−0.06 (-2.19 to 2.08)

Note: Associated odds ratios and adjusted mean differences are from the final step of the regression with all potential confounders (variables related to timing of birth at p < 0.01 from Table 1) entered into the equation.

Following the primary analyses, further exploration of possible moderating factors in the relationship between the study group and PPD was undertaken. Preexisting mental health diagnosis was found to be a significant moderator. For those women without a preexisting mental health diagnosis, PPD rates were lower during the pandemic compared to prior to the pandemic. However, for those women with a preexisting mental health diagnosis, PPD rates during the early pandemic were nearly double the pre-pandemic rates, with rates nearly returning to pre-pandemic levels during the later pandemic. Other potentially moderating factors (e.g. marital status, substance use, NICU admission, childcare) were not found to be statistically significant.

Discussion

Current reports from both the WHO and the CDC have indicated that rates of depression in the general population increased during the pandemic. The same was expected to have occurred for PPD rates, but this study did not find an increase for most participants. The results of this study did not directly align with the anticipated relationship regarding a possible correlation between the timing/period of the pandemic and the incidence/severity of postpartum depression, but as seen in Table 2, the data did reflect an important and distinctive association between the pandemic and rates of PPD for women who had a prior history of mental health diagnoses. The connection between a prior history of mental health diagnoses and PPD is in concordance with existing evidence on risk factors that are associated with the development of PPD. A large umbrella review conducted just prior to the pandemic in January of 2020 sought to identify the most common risk factors for PPD, and these factors were high life stress, lack of social support, current or past abuse, prenatal depression, and marital or partner dissatisfaction, with the strongest risk factors being prenatal depression and current abuse [11].

There are several theories to support these findings despite how they differ from previous findings by other studies. Explanatory factors are likely multifaceted with key differences in the individual experience and circumstances of each biological mother. Certain societal changes influenced by the pandemic may have uniquely benefited postpartum women in that the risk factors listed above could have decreased. The pandemic may have allowed for a longer duration of maternity leave that created more time with newborns, which is supported by a study finding an association between an increase in leave duration and a decrease in depressive symptoms until six months postpartum [12]. Some biological mothers may have had more support from partners and family who were at home during the pandemic, which is supported by a study that found 55.7% of new biological mothers with postpartum depressive symptoms could be predicted by low partner support [13]. New biological mothers may have had fewer economic stressors during the pandemic due to economic stimuli and less need for paid daycare, which is supported by a study identifying how a poor economic foundation was a moderate risk factor for PPD [14].

While we did not find an overall association between post-pandemic delivery and PPD, we did find that prior mental health diagnosis did significantly moderate this relationship. Specifically, biological mothers without preexisting mental health issues had lower PPD rates in the Early Pandemic period, and even the Later Pandemic period, than the rates that were observed for the Pre-pandemic cohort. However, for biological mothers with preexisting mental health issues, rates of PPD during the Early Pandemic period were 65% higher than they were Pre-pandemic period, and 44% higher than they were in the Late Pandemic period. Thus, having no prior mental health diagnoses may have represented a protective factor during the pandemic time period, and this aligns with current evidence mentioned above that prenatal depression history is one of the strongest risk factors for PPD [11].

New biological mothers without a history of mental health diagnoses may have had coping mechanisms that allowed them to handle pandemic-induced stressors, which have been shown in other studies to be a predictive factor in postpartum depression [14]. It is possible that financial, familial, occupational, and physical stressors could have exacerbated mental health diagnoses in biological mothers with an existing history. Maternal perspectives regarding the pandemic may have also contributed to these findings, as the ability to be home with their newborns and have more help may have enabled new biological mothers to feel more optimistic, which is a perspective that was shown to be associated with a lower risk of developing clinically significant depressive symptoms at six and twelve months postpartum compared to those who were pessimistic [15]. This is supported by a preliminary study out of New York that noted how pregnant women from lower socioeconomic status reported improved mood during the social restrictions secondary to the pandemic [16]. There were also several studies on the ability to access healthcare during the COVID-19 pandemic that indicated there is an association between decreased access to healthcare and increased symptoms of depression and anxiety. A longitudinal study out of Canada specifically evaluated disruptions to prenatal and health services and found substantial changes to prenatal care due to the pandemic, with 23.0% of their participants having prenatal appointments canceled and 60.8% reporting birth plan changes [17]. This difficulty in getting access healthcare could have resulted in biological mothers with existing history of mental health disorders being unable to obtain appointments with mental health professionals or maintain medication prescriptions during this time.

Overall, this study sought to raise awareness of the effects of large-scale events on maternal mental health, and the hope was that evidence gleaned from this study could be used to increase surveillance, create actionable plans, and provide support for the maternal populations who were identified as at-risk. While the findings did not match our initial hypotheses, they did identify a specific at-risk subgroup.

Future directions of study could include a deeper investigation into how the pandemic affected the risk factors and triggers that increased PPD rates in biological mothers with a history of mental health diagnoses, as well as a deeper investigation into health disparities affecting rates of PPD during the pandemic. Additionally, it would be beneficial for future studies to expand this population to other regions of Michigan, as well as the greater regions of the U.S., to look for confounding effects on variables that may be specific to location and hospital system.

Limitations

Limitations identified during the study were largely logistical. Some patients missed well-child appointments due to various reasons and those biological mothers could not be assessed for PPD. The EPDS assessment could also not be assessed if the child was brought to their well-child appointment by a family member other than the biological mother. However, as noted, the study sample did not differ significantly on background factors from the larger clinic population. Other remaining limitations include a smaller sample size and being restricted to information available in the medical record.

Conclusion

This study sought to further extend understanding of early maternal responses to COVID-19’s impact on maternal mental health in the postpartum period. The findings showed reduced rates of PPD in biological mothers without preexisting mental health conditions and increased rates of PPD in biological mothers with a history of mental health conditions early in the pandemic. The COVID-19 pandemic was not expected to have such dichotomic effects on subgroups of biological mothers who give birth, and this demonstrates how a devastating international event can have such an unexpected and varied impact on vulnerable populations that do not align with previous evidence. With these findings in mind, a key target for intervention would be ensuring clinical providers engage in heightened surveillance for pregnant patients who have reported a medical history with a mental health diagnosis, as more preventative efforts could potentially improve outcomes for those at-risk patients and their newborns well beyond the pandemic.

Author contribution

N.B., H.K, and M.S. devised the project, the main conceptual ideas, and proof outline. All authors contributed to IRB submission/exemption, data collection sheet creation, review of patient charts, and data collection. B.B. and N.B. performed the data analysis, figure/table creation, and results interpretation. K.D. table creation and results interpretation/table creation and results interpretation. K.D., N.B., H.K., and C.D. wrote the manuscript. M.H. updated citations and reference formatting/style.

Disclosures

The authors have no business and/or financial interest to disclose.

Ethics statement: human participants

The Covenant Medical Center Institutional Review Board has determined this project is exempt from IRB review according to federal regulations. This IRB is associated with Covenant Healthcare, and it was assigned the following reference number: C-22-04 PPD-COVID-19. This retrospective study had no direct contact with participants and was thus determined to be exempt from IRB review. Participant data was deidentified and stored securely.

Abbreviations
BB	=	Beth Bailey
CD	=	Connor Dyer
CDC	=	Centers for Disease Control and Prevention
DSM-5	=	The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition
EPDS	=	Edinburgh Postnatal Depression Scale
HK	=	Haley Kopkau
KD	=	Katlyn Droke
MH	=	Madeleine Hoke
MS	=	Molly Stanford
NB	=	Nathanial Bartosek
PNV	=	prenatal vitamins
PPD	=	postpartum depression
WHO	=	World Health Organization [Database]

Acknowledgments

We would also like to thank the CMU Pediatrics Department for their assistance with data collection and Victoria Zablocki for her work with data collection and data analysis.

Disclosure statement

No conflicts of interest to report. Researchers never directly interacted with patients.

Data availability statement

The datasets generated and analyzed for this study are the property of CMU Medical Education Partners and Covenant Health System, not the study authors. As such, data are not publicly available, and any request for access would need to be made to and approved by both CMU Medical Education Partners and Covenant Health System.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.