Divorce or Separation Following Postpartum Psychiatric Episodes: A Population-Based Cohort Study
Objective: Psychiatric disorders are an established risk factor for divorce or separation. Despite the fact that 10%–15% of new mothers experience postpartum psychiatric episodes (PPEs), no previous studies have investigated the effects of PPEs on the probability of divorce in these new families. Therefore, this study aimed to investigate and quantify the probability of subsequent divorce/separation among women with either mild/moderate or severe PPE compared to mothers without PPE.
Methods: This cohort study based on the national Danish registers included all cohabitating, primiparous women without previous psychiatric history who gave birth from 1996 through 2014. At 6 months postpartum, each woman’s PPE status was evaluated and categorized as follows: (1) mild/moderate PPE (prescription of psychotropic medication—Anatomical Therapeutic Chemical Classification codes N03–N07), (2) severe PPE (psychiatric inpatient or outpatient treatment—International Classification of Disease, 10th Edition codes F00–F99, excluding codes for organic mental disorders, substance abuse, and mental retardation), and (3) no PPE (reference group). Subsequently, the status of cohabitation was assessed a maximum of 5 times (every January 1).
Results: A total of 266,771 new mothers were included; 4,442 had a first mild/moderate PPE and 1,141 had a first severe PPE within 6 months postpartum. Compared to mothers without PPE, women with mild/moderate PPE had a significantly higher probability of later divorce (adjusted hazard ratio [HR] = 1.23; 95% CI, 1.15–1.31); for women with severe PPE, the probability was even greater (adjusted HR = 1.64; 95% CI, 1.45–1.85).
Conclusions: Women experiencing their first-ever PPE following childbirth have a higher probability of divorce in the years following their diagnosis than mothers without PPE. Further, this study showed a dose-response relationship between the severity of PPE and the probability of divorce.
J Clin Psychiatry 2021;82(3):20m13555
To cite: Johannsen BMW, Mægbæk ML, Bech BH, et al. Divorce or separation following postpartum psychiatric episodes: a population-based cohort study. J Clin Psychiatry. 2021;82(3):20m13555.
To share: https://doi.org/10.4088/JCP.20m13555
© Copyright 2021 Physicians Postgraduate Press, Inc.
aNational Centre for Register-based Research, Aarhus University, Aarhus, Denmark
bDepartment of Public Health, Research Unit of Epidemiology, Aarhus University, Aarhus, Denmark
cCIRRAU, Centre for Integrated Register-based Research, Aarhus University, Aarhus, Denmark
*Corresponding author: Benedicte M. W. Johannsen, MD, National Centre for Register-based Research, Aarhus University, Fuglesangs Allé 26, 8210 Aarhus; Denmark (email@example.com).
Postpartum Psychiatric Episodes
For most families, having a baby is a joyful event. However, the postpartum months represent a high-risk period for psychiatric disorders in mothers.1,2 Around 10%–15% of women who give birth will experience varying degrees of depressive symptoms in the postpartum period, while approximately 0.1% of new mothers are diagnosed with postpartum psychosis, the most severe form of postpartum psychiatric episode (PPE), necessitating specialized psychiatric treatment.3–5
Postpartum Psychiatric Episodes and the Probability of Divorce or Separation
Divorce can have adverse consequences for the financial situation as well as for the physical and mental health of the adults and children involved.6–10 Various factors can influence a couple’s probability of divorce or separation, including young age at marriage, the total number of years of marriage completed, and socioeconomic status.11,12 Another risk factor for marital dissatisfaction and divorce is one partner’s having a psychiatric disorder, especially within the spectrum of anxiety and mood disorders.13
Earlier studies on PPEs have described increased marital dissatisfaction among mothers following varying degrees of postpartum depression14–16 and postpartum psychosis.17 However, to our knowledge, no previous study has systematically investigated the probability of divorce or separation among couples affected by the broad spectrum of PPEs. Therefore, this study aimed to investigate and quantify the probability of subsequent divorce or separation among women suffering from either mild/moderate or severe PPE compared with new mothers without PPE.
Study Design, Setting, and Participants
We conducted an epidemiologic cohort study using Danish national register data. Our population included all women born in Denmark who gave birth for the first time between January 1, 1996, and December 31, 2014. We restricted our cohort to mothers who were at least 15 years of age at the time of giving birth and were alive and living in Denmark at the start of follow-up.
Because family status is updated in the registers every January 1, we identified the women’s cohabitation status in the year of childbirth. We restricted our study population to women cohabiting with a Danish-born partner, excluding women living with their parents or missing information on family status or partner identification number. Further, as we aimed to define first-time (incident) psychiatric episodes as our exposure (see further description in the Exposure of Interest section), we excluded women with a history of psychiatric disorders before childbirth, both mild/moderate disorders, defined as redeemed prescription within the Anatomical Therapeutic Chemical Classification (ATC) codes N03–N07, as and severe disorders, defined as inpatient or outpatient contacts to psychiatric facilities (International Classification of Diseases, Eighth Revision [ICD-8] 290xx–315xx; Tenth Revision [ICD-10] F00–F99) (see Supplementary Appendix 1).
Additionally, to evaluate the probability of divorce or separation after a PPE following the second childbirth, we created a secondary cohort of women, defined by the date of birth of their second child. We used the aforementioned cohort definition and exclusion criteria.
Exposure of Interest
To evaluate the probability of divorce or separation in the broad spectrum of postpartum psychiatric episodes, we defined two non-overlapping exposure groups within 0–6 months postpartum: those with mild/moderate PPE and those with severe PPE. In the Danish treatment system, initial assessment and treatment of psychiatric disorders are managed by general practitioners. For more severe psychiatric episodes or insufficient treatment response, general practitioners will refer the patient to a specialized psychiatric treatment facility for further evaluation and treatment as either an outpatient or an inpatient. Consequently, mild/moderate PPE was ascertained on the first-ever redemption of psychotropic medication (ATC codes N03–N07) within 6 months after childbirth, provided the woman had no contact with a specialized psychiatric facility. Severe PPE was ascertained on the first-ever inpatient or outpatient contact with a psychiatric facility within 6 months after giving birth to a child (ICD-10 codes F00–F99, excluding organic mental disorders ICD-10 codes F00–F09, substance abuse ICD-10 codes F10–F19, and mental retardation ICD-10 codes F70–F79). It should be noted that a hierarchy was introduced in situations in which a woman experienced both a mild/moderate episode and a severe episode during our exposure period; in such cases, the severe episode trumped the mild/moderate episode. The woman would consequently be categorized as having a severe case of PPE, regardless of which of the two episodes occurred first.
Outcome of Interest
The outcome of interest was divorce or separation among women living with a partner (including same-sex partnerships) after the birth of their first child. This outcome, labeled as divorce (cohabiting/non-cohabitating) in the remainder of the article, was assessed during 5 consecutive measurements of family status after the postpartum period based on available information on the cohabiting status.
To evaluate any possible familial instability following a PPE, we considered the relationship to the registered cohabitating partner in the year of childbirth as the relationship that could be affected by the diagnosis of PPE. Even though the cohabitating partner, in some cases, might not be the biological parent of the child, we assumed the cohabitating partner to be coparenting if they were living together in the year of childbirth.
The data in this study came from Danish national registers. The National Danish Data Agency approved the study, and legal permission from the participants is not necessary when conducting analyses on anonymized register data. Every individual born and/or residing in Denmark is assigned a unique identification number (ie, the Civil Registration System [CRS] number), which is used throughout every national register and therefore enables accurate linkage of information within and between registers. For the present study, we used information from the following national registers:
1. The Danish CRS18 contains information on CRS number, gender, date of birth, vital status, and all potential children (since 1968).
2. The Danish National Prescription Registry (DNPR)19 includes data on all redeemed prescription drugs sold in all Danish pharmacies since 1994.
3. The Danish Psychiatric Central Register (PCR)20 contains information on all admissions to psychiatric hospitals in Denmark since 1969, with outpatient information included from 1995 onward.
4. The Population Statistics Register21 includes data on, eg, marriages, divorces, living spouses, relocations, and migration from 1971 onward.
5. The Danish Student Register and Qualification Register,22 in combination, provide data on the highest completed educational level at any given time for individuals residing in Denmark.
Approval of research projects and access to the population-based data in the Danish national registers are provided by the Danish authorities, including the Danish Data Proctection Agency, the Danish Health Data Authority, the Ethical Committee, and Statistics Denmark. As this study was approved by the Danish authorities and based on register information, further personal consent from the individuals in the study cohort was not required. However, the data included are anonymized and presented in a way that ensures no single individual can be recognized or identified.
As our outcome of interest (divorce/separation) is updated yearly each January, follow-up started on the first January 1 after the women’s 6-months-postpartum exposure period, at least 180 days (6 months) after the date of the birth of their first live-born child, with the earliest start date for follow-up being January 1, 1997. From this date, every woman’s status of cohabitation was assessed each January a maximum of 5 times or up until the year the woman got divorced/separated or had a missing record of cohabitation status; December 31, 2016; or the January after a woman emigrated from Denmark or died, whichever came first (see Supplementary Appendix 2 for further illustration of this procedure).
Within our main cohort, we considered two exposure groups: those with mild/moderate PPE episodes and those with severe PPE episodes, both evaluated within 6 months postpartum. Both groups were compared to the same reference group: first-time mothers with no PPE registered before or within the period 6 months postpartum, categorized as “disease-free.” We analyzed our data using a parametric regression model with interval-censoring to take into account that our outcome of interest (divorce) is recorded only once every year. We used PROC LIFEREG in SAS 9.4 (2013; SAS, Inc) with an accelerated failure time model and based on graphs, and, by using likelihood ratio tests, we found that an exponential distribution of event times (a constant hazard function) fitted the data sufficiently.
Analyses were adjusted for calendar time, age at birth of the child, and the completed educational level registered on the last October 1 preceding the date of birth. Educational status was categorized into 5 groups: (1) mandatory school, with a maximum of 9–10 years of completed education; (2) short education, with a maximum of 12–13 years of education; (3) medium education, with 16–17 years of education; (4) higher education, with a long-term education of up to approximately 20 years; and (5) missing (encompassing women for whom information on the educational level was missing). Further, we considered the partner’s previous psychiatric history by adjusting for any inpatient or outpatient contact with a psychiatric facility prior to the cohabitating woman’s giving birth. For the analyses of second-time childbirths, the same setup was applied.
To address potential misclassification of women getting divorced from their partner shortly after the end of the 6-month exposure period, we conducted a sensitivity analysis to compare estimates from women giving birth in the first 6 months of the calendar year versus those giving birth in the last 6 months of the calendar year (see Supplementary Appendix 2).
Among the 266,771 new mothers, a total of 4,442 were defined as having mild/moderate PPE cases (women who redeemed their first-ever prescription for psychotropic medication within the 6-months-postpartum period). Further, 1,141 women were defined as having severe cases of PPE (first-ever psychiatric disorder, which was recorded as inpatient or outpatient contact with a psychiatric treatment facility).
The most common age group for first childbirth was 26–30 years (49.0%), including mothers with mild/moderate PPE (44.4%) and mothers with severe PPE (45.9%). Fewer women with mild/moderate PPE had completed higher education at the time of their first childbirth (19.8%) compared to mothers in the reference group (31.3%) and women with severe PPE (28.7%) (Table 1).
Probability of Divorce in Women With Postpartum Psychiatric Episodes Following First Childbirth
Of the 4,442 women with onset of mild/moderate PPE, 857 women divorced or ceased to cohabitate with their partner during the follow-up period (Table 1), corresponding to an adjusted hazard ratio (HR) for divorce of 1.23 (95% CI, 1.15–1.31) compared to mothers without any record of PPE (Table 2). In comparison, among the 1,141 mothers with severe PPE, a total of 254 women were divorced during the follow-up period (Table 1), resulting in an adjusted HR of 1.64 (95% CI, 1.45–1.85) for women with severe PPE compared to mothers without PPE (Table 2). Thus, both women with mild/moderate PPE and those with severe PPE had a significantly higher probability of divorce than mothers without PPE. However, when our two defined PPE groups are compared, mothers with severe PPE had the higher probability of divorce.
Cumulative incidence curves of divorce for all mothers—those with mild/moderate PPE, those with severe PPE, and those without PPE—are shown in Figure 1. At the time of first family status assessment, 5% of women with mild/moderate PPE had divorced, and 7% of women with severe cases of PPE had divorced. In comparison, mothers without PPE had a cumulative incidence of divorce of 2.5%. In the second year, 10% and 12% of women with mild/moderate and severe PPE, respectively, had divorced, whereas 5.5% of women without PPE had divorced. Ultimately, by year 5, 19% of women with mild/moderate PPE, 22% of women with severe PPE, and 13% of women without PPE were divorced.
Sensitivity analyses on the probability of divorce among those giving birth in the first 6 months of the calendar year (mild/moderate PPE: HR = 1.22; 95% CI, 1.12–1.34; severe PPE: HR = 1.65; 95% CI, 1.40–1.94) versus those giving birth in the last 6 months of the calendar year (mild/moderate PPE: HR = 1.23; 95% CI, 1.12–1.36; severe PPE: HR = 1.62; 95% CI, 1.36–1.93) did not produce different estimates.
Probability of Divorce in Women With Postpartum Psychiatric Episodes Following Second Childbirth
In the second cohort, of mothers giving birth to their second child, we identified 264,931 mothers. Among these, 3,918 experienced their first episode of mild/moderate PPE within 6 months postpartum, while 852 women had a first episode of severe PPE within the postpartum period. The characteristics of this second cohort were similar to those of the cohort of primiparous mothers, except that the women in this cohort were naturally slightly older at the second childbirth (results not shown).
A total of 743 of the 3,918 mothers who experienced a first episode of mild/moderate PPE following the second childbirth were divorced from their partner during the follow-up period (adjusted HR = 1.25; 95% CI, 1.17–1.35). Among the 852 mothers with severe PPE following their second childbirth, 192 divorced their partner throughout this study, corresponding to a HR of 1.66 (95% CI, 1.44–1.91) (Table 3).
Overall, the analyses based on the second childbirth showed that the probability of divorce following an episode of PPE did not vary by parity. Divorce estimates were similar for mild/moderate PPE (HR = 1.23; 95% CI, 1.15–1.31) and severe PPE (HR = 1.64; 95% CI, 1.45–1.85) following first childbirth versus onset of mild/moderate PPE (HR = 1.25; 95% CI, 1.17–1.35) or severe PPE (HR = 1.66; 95% CI, 1.44–1.91) following the second childbirth. Figure 2 illustrates cumulative incidence curves for the cohort of women with PPE following their second childbirth that are similar to the curves for the cohort with PPE following their first childbirth. At the end of follow-up, 13% of second-time mothers without PPE had divorced, while 19% of mothers with mild/moderate PPE cases and 23% of mothers with severe PPE had divorced from their partners.
The present study showed that women experiencing PPE following either first or second childbirth have a higher probability of divorce in the subsequent years compared with mothers without PPE. Further, this study demonstrates that a dose-response relationship exists between the severity of PPE and the probability of divorce, as mothers with the most severe cases of PPE had a significantly higher probability of divorce than mothers with mild/moderate PPE.
Circumstances Influencing the Probability of Divorce
The proportion of marriages ending in divorce has increased throughout the last decades in most Western countries.11,12 Families with young children constitute a group that is especially at risk of divorce and separation. Each year, approximately 2.2% of Danish children experience the change from cohabitating with both parents to living with only one parent, with the risk mainly increasing during the first 5 years of the child’s life.23 A Swedish study found an increased risk of divorce among parents, peaking at 3–5 years following first childbirth. However, the relative risk of divorce was lower than for childless couples.24 In general, experiencing psychiatric disorders is associated with a higher probability of divorce following the diagnosis, especially among patients with major depressive disorders.13,25
Divorce and Relationship Problems in Women With Postpartum Psychiatric Episodes
It is well-established that a lack of relationship support is a risk factor for developing PPE.1,14 However, previous studies15–17 have noted increased occurrences of marital difficulties or unhappiness in relationships following PPEs. Nevertheless, the impact of PPE on the occurrence of subsequent divorce has not been shown previously. A recent study26 on subsequent reproduction among women with PPE found that 6.8% of women with PPE had divorced within 1 year after their diagnosis. Interestingly, these findings were all reported as additional observations in studies investigating other aspects of PPE.
Consequences of Divorce and the Importance of Findings
Previous studies show that women with PPE are at increased risk of adverse outcomes following their diagnosis, eg, self-harm,27–29 relapse in later pregnancies,30 recurrent psychiatric morbidity,31 and higher mortality especially from unnatural causes such as suicide.28,32,33 Hence, this subset of patients identifies a particularly vulnerable group of new mothers. Per definition, a diagnosis of PPE affects a family, and not only the individual patient, as results from the present study confirm.
When a couple separates, the breakup can have ramifications for all involved parties, including a decline in financial status, especially for women.6 Further, divorced individuals have a higher incidence of subsequent physical and mental disorders and higher all-cause mortality than married individuals.6,7 Also, children affected by parental divorce are at higher risk of adverse outcomes, such as behavioral problems, anxiety, affective disorders, and lower academic achievements, especially if a high level of parental conflict characterizes the breakup.8–10,34 With the results of this study underlining the higher probability of divorce following an episode of PPE, it is essential to convey this message to health care professionals working with affected women and their families. Not only are the women at higher risk of adverse outcomes directly from their psychiatric diagnosis, but their families also appear to be vulnerable in the subsequent years. Within the present study, we are not able to report specific reasons for the divorces observed in the two groups of women with PPE, and the recommendation of specific interventions is therefore not supported by our results. However, the vulnerability of the relationships affected by PPE, demonstrated in this study, suggests that both treatment and follow-up of women with PPE should include a focus on the well-being of the entire family and potentially offer appropriate, unbiased information and guidance for women experiencing difficulties in their relationships following PPE.
Strengths and Limitations
The use of the Danish national registries is a strength of the present study. These registers contain information on all individuals residing in Denmark, linked through the CRS number. All Danish-born women giving birth to their first and/or second child were included in this study, and information on exposure, outcome, and possible confounders was all derived from register information, thereby minimizing the risk of selection and recall bias. We defined mild/moderate PPE through the National Prescription Registry established in 1994. By including women giving birth from 1996 onward, we established a washout period of at least 12 months to optimize the conditions for identifying incident cases of mild/moderate PPE. Severe cases of PPE were defined by all inpatient and outpatient contacts with psychiatric facilities in Denmark during the study period.
Information on the outcome of interest in this study (divorce and separation) is updated yearly, and identifying the exact timing of the separation is not possible. With the delay in registration of divorce to the following January, we risk misclassifying women as still being married/cohabitating until the next registration, especially among those giving birth in the second half of the calendar year, when the time elapsed between childbirth and start of follow-up is longest. However, we addressed this issue in a sensitivity analysis, which showed almost identical results for those giving birth in the first half of the calendar year versus those giving birth in the second half of the calendar year. We adjusted our analyses for potential confounding from age at childbirth, education, and psychiatric disorders in the cohabitating partners, but as in any observational study, residual confounding cannot be ruled out.
Mothers with postpartum psychiatric episodes have a higher probability of divorce or separation in the immediate years following their diagnosis compared to mothers without postpartum psychiatric episodes. Further, there is a dose-response relationship between the severity of the postpartum psychiatric episode and the probability of divorce, as mothers with the most severe episodes have significantly higher probabilities of divorce than those with mild/moderate episodes and mothers without postpartum psychiatric episodes.
Submitted: June 26, 2020; accepted October 21, 2020.
Published online: March 23, 2021.
Potential conflicts of interest: The authors of this study have no conflicts of interest to declare.
Funding/support: This study was supported by a research training supplement from the School of Health at Aarhus University and Fabrikant Vilhelm Pedersen og Hustrus Legat. Dr Munk-Olsen receives funding from: iPSYCH (The Lundbeck Foundation Initiative for Integrative Psychiatric Research), The Lunbeck Foundation (grant number R313-2019-567), and AUFF Nova (Aarhus University Research Foundation).
Role of the sponsor: None of the funders had any role in the planning, design, analysis, interpretation of the results, or publication of this study.
Supplementary material: Available at Psychiatrist.com.
- Postpartum psychiatric episodes (PPEs) affect entire families; however, very little is known about the potential consequences of the diagnosis for the parental relationship.
- PPE increases the probability of subsequent divorce or separation. Health care professionals should, to the extent possible, ensure the well-being of the entire family and facilitate unbiased guidance for relationships strained by PPE.
Editor’s Note: We encourage authors to submit papers for consideration as a part of our Focus on Women’s Mental Health section. Please contact Marlene P. Freeman, MD,
Buy this Article as a PDF