Severe Maternal Morbidity and Mental Health Hospitalizations or Emergency Department Visits

Key Points Question Is severe maternal morbidity (SMM) associated with long-term mental health–related hospitalizations or emergency department (ED) visits after delivery? Findings In this cohort study among 1 579 392 individuals with hospital births in Canada, 35 825 individuals had SMM within pregnancy or up to 42 days post partum, and 1 543 567 individuals did not. SMM was associated with a 1.3-fold increased rate of hospitalization or ED visit for a mental health condition up to 13 years post partum. Meaning These findings suggest that SMM was associated with adverse mental health conditions beyond the conventional postpartum period.


Introduction
2][3][4] Beyond the known increased risks of short-term mortality and prolonged hospital stay in individuals with SMM compared with unaffected individuals, 5,6 those who survive SMM are more likely to develop chronic health conditions, including cardiovascular disease, 6,7 long-term impaired functional ability, and chronic pain. 8,9The trauma of SMM and its consequences could adversely affect psychological health. 10Given that up to 36% of mortality in the first year post partum is due to suicide, 11 it is critical to assess the mental health burden after SMM.
Studies of associations between various adverse pregnancy events and postpartum mental health outcomes have principally examined specific conditions, like preeclampsia, or have relatively short follow-up periods. 12,13Furthermore, very few studies have been completed within Canada, [12][13][14][15] which has a multiethnic population with universal antenatal health coverage yet variable access to postpartum mental health services.Knowledge of the short-and long-term risks of serious mental health conditions after SMM and its subtypes could inform the need for enhanced postpartum supportive resources.We therefore performed a population-based cohort study assessing mental health hospitalizations and emergency department (ED) visits over a 13-year period, comparing individuals who experienced SMM with unaffected postpartum individuals in Canada.

Methods
This cohort study was approved by the research ethics board of the McGill University Health Centre (MUHC).Since the study used secondary aggregate data, the need for individual informed consent was waived by MUHC.We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Study Design and Data Sources
We conducted a cohort study of individuals with a first recorded hospital delivery between April 1,   2008, and March 31, 2021.Data were extracted from the Canadian Institute of Health Information (CIHI) Discharge Abstract Database (DAD), including administrative, clinical, and demographic information on all hospital deliveries within Canada, excluding Québec, as this province does not submit hospitalization data to CIHI.The DAD accounts for approximately 98% of deliveries in Canada outside Québec. 16,17While most mental health hospitalizations are also typically recorded in the DAD, in some Ontario facilities, these events are reported through the Ontario Mental Health Reporting System, a data source that was unavailable at the time of this study.The accuracy of DAD records has been validated against medical records, demonstrating high specificity and sensitivity for most maternal conditions and high specificity and low to moderate sensitivity for most mental health conditions. 18,19ing unique patient identifiers, the CIHI DAD was linked to the National Ambulatory Care Reporting System (NACRS) dataset, which includes patient data from EDs. 16,20 Because coverage of ED data is variable across provinces and fiscal year, 20,21  to characterize the sample and define study exposures and outcomes. 16,22

Study Population
Individuals aged 18 to 55 years with a first recorded liveborn or stillborn delivery with pregnancy lasting between 20 and 43 weeks' gestation were included. 23Individuals who delivered after 43 weeks' gestation and those younger than 18 or older than 55 years at delivery were excluded to optimize the accuracy of obstetric codes; individuals younger than 18 or older than 55 years were also excluded to account for unique mental health experience among pregnancies at extremes of age. 24,25As the interest was in new-onset mental health visits related to SMM, the primary analysis excluded individuals with a previous mental health hospitalization or ED visit within 2 years before the index birth, an approach that has been used previously.pregnancy and the postpartum period. 1,28Acute psychosis was excluded from the SMM definition as this was a component of the primary outcome.

Study Outcomes
The primary outcome was a composite of mental health hospitalizations or ED visits occurring 43 days or more after the index birth hospitalization, defined by a primary or secondary coded diagnosis in the DAD for mood or anxiety disorder, substance-related or addictive disorder, schizophrenia spectrum or other psychotic disorder, or suicidality or self-harm event 29 (eTable 2 in Supplement 1).
Secondary outcomes were individual components of the primary outcome.

Covariates
Covariates selected a priori as potential confounders in multivariable models were captured within 2 years prior to index birth hospitalization, and guided by a directed acyclic graph (eFigure 2 in Supplement 1).These included maternal age at delivery, income quintile, delivery year, province or territory of delivery, maternal comorbid conditions (preexisting hypertension; diabetes; chronic kidney disease; chronic liver disease; cardiovascular condition; sickle cell disease; HIV; autoimmune syndrome, such as systemic lupus erythematosus; asthma; obesity; or smoking), urban or rural residential status, and hospital type (teaching tertiary care hospital vs community hospital). 30

JAMA Network Open | Obstetrics and Gynecology
Severe Maternal Morbidity and Maternal Mental Health Hospitalizations or ED Visits

Statistical Analysis
Baseline characteristics for the cohort were described, stratified according to the presence or absence of SMM in first recorded birth, using means and SDs or medians and IQRs for continuous data and frequencies and percentages for categorical data.A graph was generated displaying the temporal trend in frequency and rates per 1000 deliveries of SMM in all births by fiscal year.
Incidence rates were calculated per 10 000 person-years with 95% CIs for the composite outcome and individual components in individuals with SMM and unaffected individuals.Univariable and multivariable Cox proportional hazards models were used to calculate crude hazard ratios (HRs) and adjusted HRs (aHRs) and 95% CIs, estimating the association between SMM and a mental health hospitalization or ED visit.
In secondary analyses, the association between SMM and each component of the primary outcome was assessed.Also, the risks of outcomes according to common individual SMM diagnoses were examined.Finally, associations of SMM with hospitalizations and with ED visits were assessed separately, as individuals requiring hospitalization may differ in important ways from individuals with an ED visit with subsequent discharge to the community. 31In all models, follow-up was censored on death or end of study period (March 2021).Follow-up was also censored on subsequent pregnancy in the primary analysis to address misclassification bias arising from individuals who experienced SMM in subsequent pregnancies but not the first.
A complete-case analysis was conducted after we determined that 2.4% of the records in the dataset had missing values and deleting them would be unlikely to meaningfully impact estimates. 32seline characteristics were compared among those with and without missing data to assess the validity of our findings.Log(−log[survival]) by log(time) plots were generated to test the proportional hazards assumption.
We performed 7 sensitivity analyses.First, we assessed hospitalizations with an ICD-10-CA code for a mental health condition in any of the 25 DAD fields to capture visits with as opposed to for a mental health condition. 33Second, we reincluded individuals with preexisting mental health conditions identified prior to the index pregnancy in the cohort and stratified models according to the presence or absence of a previous mental health hospitalization or ED visit to assess for exacerbated disease.Third, we excluded Ontario births from the cohort, because mental health hospitalizations in this province are reported largely through the Ontario Mental Health Reporting System. 21,33Fourth, we ran a model in which there was no censoring on a subsequent pregnancy, as some individuals at high risk for the outcome of interest may be less likely to conceive or have longer interpregnancy intervals.Fifth, we excluded stillbirths and preterm births from the cohort, as these events have been associated with postpartum mental illness.Sixth, we ran a model in which SMM was limited to events occurring at or before delivery and follow-up started the day after hospital discharge.Finally, we ran models for 3 separate follow-up periods: up to 1 year, 1 to 5 years, and more than 5 years, to assess how proximity to delivery might modify the association between SMM and mental health hospitalization or ED visit.
P values were 2-sided, and statistical significance was set at P = .01.Data were analyzed using SAS software version 9.4 (SAS Institute).Data were analyzed from January to June 2023.

Baseline Characteristics of Study Sample
We  2).
While ED visits were more frequent than hospitalizations for all mental health conditions, the relative risks for ED visit and hospitalization after SMM were similarly elevated (Table 3).Similar increased relative risks for mental health hospitalizations and mental health ED visits were seen across all provinces examined (eTable 3 in Supplement 1).

Sensitivity Analyses
The relative risks of hospitalization using the broader definition for a mental health condition (ie, a mental health diagnostic code in any of the 25 diagnostic fields) were similar to that in the primary analysis (eTable 4 in Supplement 1).Absolute rates of mental health visits were substantially higher in individuals with vs without previous mental illness within the last 2 years, but relative risks were not appreciably different from the primary analysis (eTable 5 in Supplement 1).Similarly, the removal of births in Ontario did not significantly impact our findings (eFigure 4 in Supplement 1) nor did ignoring subsequent pregnancy (eTable 6 in Supplement 1), excluding preterm births and stillbirths from the cohort (eTable 7 in Supplement 1), or starting follow-up the day after hospital discharge (eTable 8 in Supplement 1).Individuals with SMM had the highest relative risk of hospitalization or ED visit for a mental health condition in the first year post partum (aHR, 1.38 [95% CI, 1.24-1.53]).Individuals with more than 1 year and less than 5 years of follow-up had increased risk of hospitalization or ED visit for a mental health condition (aHR, 1.23 [95% CI 1.14-1.34]),as did those with more than 5 years of follow-up (aHR, 1.21 [95% CI, 1.07-1.37])(eTable 9 in Supplement 1).5][36][37] Maternal ICU admission often co-occurs with SMM, resulting in separation of parent and neonate and subsequent maternal distress.This study has several strengths including the large sample size, which facilitated the analysis of a rare exposure and a rare outcome.The population-based nature of the study increases the generalizability of our findings, while the relatively long follow-up period and the use of robust, validated definitions enhanced their internal validity.

Limitations
This study has some limitations, primarily related to the observational design using administrative health data.Due to lack of data on ambulatory clinic visits and prescriptions, we could not capture individuals with less severe health conditions, such as some mood and anxiety disorders and other comorbid conditions that are commonly managed in outpatient settings.This could have led to both misclassification of outcomes and unmeasured confounding by comorbidities and previous mental illness.In an effort to capture the whole of pregnancy morbidity, our cohort included both stillbirths and live births, as well as term and preterm births.We acknowledge that mothers who experienced stillbirth or extreme prematurity would be more at risk for mental health issues.In sensitivity analyses restricted to term liveborn deliveries, the relative risks of SMM on mental health visits were unchanged.Therefore, our findings suggest that SMM, with or without stillbirth and preterm birth, was associated with future mental health visits in the mother.We acknowledge that our principal results apply mostly to individuals with intermediate or high risk; however, findings of a sensitivity analysis demonstrating similar relative risks of the outcome due to SMM among individuals with preexisting mental illness suggest that SMM was a risk factor associated with poor mental health regardless of baseline predisposition.
Results may not be generalizable to all births.We acknowledge that individuals with the most severe SMM subtypes may be less likely to conceive again and more likely to have psychological complications; however, our results were consistent in models in which we did not censor on subsequent pregnancy.Operational definitions used to define SMM in CIHI DAD have been validated 8][49] Any recent changes to diagnostic definitions may not be captured, but there have been no major changes to SMM diagnoses since 2016.Despite few missing data, we acknowledge the possibility that we may have misestimated the true associations due to exclusion of individuals with incomplete data.Additionally, this study does not capture data on completed suicide and thus may underestimate the severity of mental health issues after SMM.

Conclusions
In this cohort study of postpartum individuals with and without SMM in pregnancy, SMM was associated with hospitalization or ED visit for mental health conditions several years after obstetric delivery.These findings suggest that individuals who experience severe pregnancy complications may benefit from additional mental health screening.

JAMA Network Open | Obstetrics and Gynecology Severe
Maternal Morbidity and Maternal Mental Health Hospitalizations or ED Visits ED outcomes were assessed among individuals with available NACRS data (eFigure 1 in Supplement 1).Up to 25 diagnostic and 20 procedural codes per visit were captured using International Statistical Classification of Diseases and JAMA Network Open.2024;7(4):e247983.doi:10.1001/jamanetworkopen.2024.7983(Reprinted) April 23, 2024 2/15 Downloaded from jamanetwork.comby guest on 04/25/2024 Related Health Problems, Tenth Revision, Canadian version (ICD-10-CA) and the Canadian Classification of Health Interventions (CCI) respectively.Diagnostic and procedural codes were used 12,26,27Duplicate delivery records, delivery records with missing identifiers, and individuals with ectopic pregnancy or miscarriage were also excluded.Therapeutic abortions and out-of-hospital deliveries were not available in the data source, but out-of-hospital deliveries account for less than 2% of deliveries in Canada.17 1xposure to SMM was captured between 20 weeks' gestation and 42 days after delivery hospital discharge in the first recorded hospital birth; individuals without SMM were considered unexposed.SMM was defined using the validated Canadian Perinatal Surveillance System definition,1which included 1 or more of the following diagnoses: severe preeclampsia or eclampsia; severe hemorrhage; cardiac complications (cardiomyopathy, cardiac arrest, resuscitation, myocardial infarction, pulmonary edema, and heart failure); complications of anesthesia; surgical complications; cerebrovascular accidents; acute kidney failure; embolism, shock, or disseminated intravascular coagulation; severe sepsis; uterine rupture; acute fatty liver or liver failure; cerebral edema; and coma.SMM also included critical illness interventions: urgent hysterectomy, dialysis, assisted ventilation, and intensive care unit (ICU) admission (eTable 1 in Supplement 1); SMM indicators are

Table 2 ). Association Between SMM and Mental Health-Related Hospitalization or ED Visit
Overall, 1287 (96.1 per 10 000) individuals with SMM and 41 779 (72.7 per 10 000) unaffected individuals had a mental health hospitalization or ED visit (HR, 1.31 [95% CI, 1.24-1.39];aHR,1.26[95%CI,1.19-1.34])(Table3).The median (IQR) time to event was 2.8 (1.3-6.5) years for individuals with SMM-affected deliveries and 2.6 (1.3-6.4) years for individuals with deliveries without SMM.The relative increased risk after SMM was seen for all components of the mental health outcome except for schizophrenia spectrum and other psychotic disorder.The greatest risk of hospitalization or ED visit was observed for suicidality and self-harm (aHR, 1.54 [95% CI, 1.26-1.88])(Table3).The risk of a mental health hospitalization or ED visit was highest if the SMM subtype was embolism, shock, and a This cohort was used to calculate outcomes of mental health hospitalization alone.bThis cohort was used to calculate outcomes that included emergency department (ED) visits, ie, mental health hospitalization and/or mental health ED visit.

Table 1 .
Baseline Characteristics of Study Cohort at First Recorded Hospital Delivery, Stratified According to the Absence or Presence of Severe Maternal Morbidity disseminated intravascular coagulation (aHR.1.71 [95% CI, 1.38-2.12]).Procedural indicators of SMM, such as use of assisted ventilation and maternal ICU admission, were similarly associated with mental health outcomes (Figure

Table 1 .
Baseline Characteristics of Study Cohort at First Recorded Hospital Delivery, Stratified According to the Absence or Presence of Severe Maternal Morbidity (continued) a Comorbidity includes preexisting hypertension, diabetes, chronic kidney disease, chronic liver disease, cardiovascular condition, sickle cell disease, HIV, autoimmune syndrome (eg, systemic lupus erythematosus), asthma, obesity, or smoking.bPreterm birth is defined as gestational age at delivery less than 37 weeks.c Stillbirth is defined as pregnancy loss at gestational age more than 20 weeks.

Table 2 .
Frequency and Incidence Rates of Mental Health Hospitalizations or ED Visits According to Characteristics of First Recorded Hospital Delivery b Preterm birth is defined as gestational age at delivery less than 37 weeks.c Stillbirth is defined as pregnancy loss at gestational age less than 20 weeks.

Table 3 .
Severe Maternal Morbidity and Associated Rate and Risk of Mental Health-Related Hospitalization or ED Visits, 2008 to 2021, Overall and for Specific Mental Health Diagnoses (N = 1 579 392) Alberta contributed ED data for follow-up from 2010 to 2021, British Columbia contributed ED data for follow-up from 2008 to 2010 and 2011 to 2021, Manitoba contributed ED data for follow-up from 2009 to 2021, New Brunswick contributed no ED data for follow-up, Newfoundland and Labrador contributed no ED data for follow-up, Nova Scotia contributed ED data for follow-up from 2008 to 2021, Ontario contributed ED data for follow-up from 2008 to 2021, Saskatchewan contributed ED data for follow-up from 2010 to 2021, and the Northern Territories contributed ED data for follow-up from 2008 to 2021.Data were available for 1 579 392 women overall, including 35 825 with SMM and 1 543 567 without SMM.HRs adjusted for maternal age at delivery, income quintile, comorbidity, delivery year and urban or rural residential status.Overall, 2.43% of records in the dataset had missing data for at least 1 study variable.Baseline characteristics comparing 1 541 105 individuals with complete data with 38 287 individuals with incomplete data were overall similar, except that individuals with missing data were younger at the time of the index birth, were more often from the lowest income quintile and rural or remote areas, and delivered in the years 2019, 2020, or 2021 (eTable 10 in Supplement 1).In this population-based Canadian cohort study, individuals who experienced SMM had increased risk of mental health hospitalization or ED visit up to 13 years after delivery compared with those who did not experience SMM.This increase in risk was consistent across provinces and for both hospitalizations and ED visits.The risk of a mental health hospitalization or ED visit was highest during the first postpartum year and among individuals who were treated in a maternal ICU during pregnancy and those with an embolism, shock, or disseminated intravascular coagulation.Individuals with SMM had the highest risk of hospitalization or ED visit for suicidality and self-harm.These results corroborate and expand on findings from previous work conducted in the US and Sweden. 12,13A 2019 study by Lewkowitz et al 12 analyzed 1 229 835 pregnant individuals in Florida and found that SMM was associated with a 74% higher odds of hospital admission for depression, anxiety, and psychosis in the first year post partum.Unlike our study, Lewkowitz et al 12 did not find an association between SMM and suicidality, likely due to differences in sample size.The Swedish study by Wall-Wieler et al 13 included 25 674 deliveries and found that SMM was associated with higher odds of inpatient psychiatric treatment for mood disorders, neuroses, and behavioral disorders also in the first postpartum year.Estimates from the analysis by Wall-Wieler et al 13 (adjusted odds ratio, 1.22 [95% CI, 1.03-1.45)are similar to those in this study, although Wall-Wieler et al did not adjust for maternal comorbidity.

eTable 2 .
Outcome Definition With Accompanying ICD-10 and CCI Codes eTable 3. Association Between Severe Maternal Morbidity and the Risk of Mental Health Hospitalization, or Mental Health ED Visit, Separately, by Province or Territory eTable 4. Association Between Severe Maternal Morbidity and the Risk of Hospitalization and/or ED Visit Capturing a Mental Health Condition Recorded in Any of the 25 Diagnostic Fields During Hospitalization eTable 5. Association Between Severe Maternal Morbidity and the Risk of Mental Health Hospitalization or ED Visit Among Individuals With Previous Mental Illness eTable 6. Association Between Severe Maternal Morbidity and the Risk of Mental Health Hospitalization or ED Visit Without Censoring on Subsequent Pregnancy eTable 7. Association Between Severe Maternal Morbidity and the Risk of Mental Health Hospitalization or ED Visit in a Cohort Where Stillbirth and Preterm Birth Are Excluded eTable 8. Association Between Severe Maternal Morbidity at or Before Delivery and the Risk of Mental Health Hospitalization or ED Visit eTable 9. Association Between Severe Maternal Morbidity and the Risk of Mental Health-Related Hospitalization or ED Visit Within the First Year Post Partum, From 1 Year to 5 Years Post Partum, and Beyond 5 Years Post Partum eTable 10.Baseline Characteristics of Study Cohort at First Recorded Hospital Delivery, Stratified According to the Absence or Presence of Missing Data