Prevalence of Depression or Anxiety During Antepartum Hospitalizations for Obstetric Complications: A Systematic Review and Meta-analysis

Abstract

Objective:

To systematically review the prevalence of antenatal depression and anxiety in women hospitalized on an antepartum unit for obstetric complications.

Data Sources:

We searched Pubmed, Embase, Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycINFO, and ClinicalTrials.gov for English-language articles published from database inception through March 2020.

Methods of Study Selection:

We included cross-sectional, cohort, case-control, quasi-experimental, and randomized-controlled studies from any country that reported the proportion of pregnant women with an elevated depression or anxiety screening scale or diagnostic interview during antepartum hospitalization of any duration and at any gestational age.

Tabulation, Integration, and Results:

We identified 8,799 articles and reviewed 79, thirty-nine of which were included in a systematic review and 18 in meta-analysis of the primary outcome. Two raters independently assessed quality of individual studies using a 14-question tool. A random effects meta-analysis model was used to estimate prevalence and 95% confidence interval of depression or anxiety. Heterogeneity was examined with the I² test and Funnel plots were used to assess publication bias. After meta-analysis, the estimated prevalence of depression was 34% (95% CI 27–41%) and of anxiety 29% (95% CI 16–43%). There was expected substantial clinical and methodologic heterogeneity between studies which persisted even after planned a priori subgroup analyses and meta-regression. Even so, the direction of effect was consistent across studies. No publication bias was found.

Conclusion:

The current meta-analysis suggests that one in three women hospitalized during pregnancy for obstetric complications report clinical levels of depression or anxiety symptoms, twice the reported prevalence of antenatal depression or anxiety in the general obstetric population.

Précis:

There is a high prevalence of antenatal depression and anxiety in women hospitalized on an antepartum unit for obstetric complications.

Introduction:

Perinatal mood and anxiety disorders are conditions that impact a woman’s health during pregnancy or up to one year postpartum. Maternal mental health conditions are one of the leading causes of pregnancy-related deaths in the United States¹. The incidence of suicidal ideation and intentional self-harm in pregnant or postpartum women has been increasing over the past 15 years² and it has been estimated that that suicide accounts for 5–6.5% of all maternal deaths.^1,3 Maternal mental health conditions contribute to 12.9% of all maternal deaths¹, with the majority occurring in the postpartum period⁴.

Women who experience pregnancy complications are at particular risk for onset of perinatal mood disorders or relapse of underlying mood disorders^5–8. In the general obstetric population, between 12% and 18% of women are reported to experience antenatal depression and 13 to 21% experience antenatal anxiety, though estimates vary widely and are higher in certain groups of women with additional risk factors, such as women with obstetric complications^9–12. If compounded by antepartum hospitalization, women may also experience social isolation, separation from social supports, poor sleep, and high levels of stress¹³. These women are also more likely to have a neonate requiring neonatal intensive care unit (NICU) admission, prolonged hospitalization and complicated post-natal course, which are also risk factors for symptoms of postpartum mood disorders^14–16.

The aim of this systematic review and meta-analysis is to comprehensively estimate the prevalence of antenatal depression and anxiety in women hospitalized on an antepartum unit for obstetric complications. This is a necessary first step in delineating disease burden to help guide and prioritize future research addressing screening, preventive interventions, and treatment strategies for these at-risk women.

Sources:

We followed PRISMA (Preferred Reporting Items for Systematic review and Meta-Analysis Protocols) 2009 methodology¹⁷ for this review and registered the protocol in the PROSPERO International Prospective Register of Systematic Reviews (CRD42020172111).

A medical librarian (DC) conducted a comprehensive database search in March 2020 to identify studies that discussed depression, anxiety, or stress in hospitalized pregnant women. The librarian developed search strategies for PubMed, Embase, Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycINFO, and ClinicalTrials.gov. The search strategies included a combination of keywords and controlled vocabulary terms (Appendix 1, available online at http://links.lww.com/AOG/C263). The search strategies did not include any limits on dates or language. EndNote Version X9.2 (Clarivate analytics, Philadelphia, USA) was used to remove duplicate articles and managed citations.

Study selection:

A priori inclusion and exclusion criteria were established. Non-English language manuscripts were excluded, however those with corresponding English language abstracts were included if the abstract contained sufficient detail to allow inclusion criteria to be assessed. Cross-sectional, case-control, cohort, quasi-experimental and randomized controlled trials were included. Studies that were conducted for purposes other than determining the prevalence of depression or anxiety in our population of interest (such as randomized clinical trials of an intervention) were included when they reported on relevant outcomes. Mixed-methods and qualitative studies were excluded. There were no time restrictions on publishing year or place of publication. Conference abstracts were excluded if it was not possible to determine from the abstract whether the study population was drawn from hospitalized antepartum patients or antepartum outpatients. In cases of conference abstracts with subsequent manuscript publications, the original conference abstract was excluded and the manuscript only was included so as not to repeat sources of data. Doctoral dissertations, guidelines, expert opinions, editorials, letters to the editor, and comments were excluded.

The population studied was pregnant women at any age and any gestational age at the time of the study. The exposure of interest was antepartum hospitalization. Studies were included if participants were hospitalized in an antepartum unit for any length of stay for medical or obstetric complications. Subjects with any co‐morbid psychiatric or other medical conditions were included. Outcome studied was presence of depression or anxiety identified either via validated, self‐report scale, or a diagnostic interview. The threshold scores corresponding to a positive screen for the respective scales were defined by the individual study investigators. Self-report scales were included because of the very low number of studies which reported on diagnosis of depression or anxiety by diagnostic interview in our study population. The primary analysis was the meta-analysis of studies estimating the point prevalence of depression via an elevated depression screen or diagnostic interview (further referred to as depression outcome) and the point prevalence of anxiety via elevated anxiety screen or diagnostic interview (further referred to as anxiety outcome).

The search strategy was developed and executed by an experienced librarian (DC) with input from co-investigators with clinical training. The screening title search, subsequent abstract screening, and application of eligibility and inclusion criteria with full manuscript review were performed by independently by two reviewers (MT & RR). Any disagreements between the two reviewers was settled via adjudication by the same two reviewers. Because both authors reviewed all article titles, abstracts, and manuscripts and discussed all discrepant opinions until consensus was obtained, inter-rater reliability was not calculated. Data were extracted from studies by a single author (MT) and inputted into electronic database using previously prepared data extraction forms. No attempts were made to obtain further data from authors of studies included in the study. References were managed with EndNote Version X9.2 (Clarivate analytics, Philadelphia, USA).

Study quality was assessed using the National Institutes of Health (NIH) quality assessment tool for observational cohort and cross-sectional studies^18,19. The quality of each study was evaluated by two independent investigators (MT and RR) and disagreements were resolved by adjudication by the same two reviewers. Quality assessment was used for descriptive purposes only, and no studies were withheld from analysis on the basis of quality appraisal. Quality scores were assigned to each included study as a percentage of the number of questions marked “yes” divided by the total number of possible points. Total number of possible point was calculated as the number of questions with either a “yes” or “no” answer excluding questions marked “not applicable (NA)” or “not reported (NR)”.

A priori hypotheses for study heterogeneity were developed. Authors anticipated heterogeneity would arise from the use of different screening tools or validated surveys for depression and anxiety employed by different investigators. For this reason, a subgroup analysis of studies using the most common single screening tool for depression and for anxiety were performed (Edinburgh Postnatal Depression Scale (EPDS) and State-Trait Anxiety Inventory (STAI)). Authors also anticipated heterogeneity would arise from differing study designs, study populations, type of depression or anxiety scale used, and quality assessment score assigned. We therefore performed a sensitivity analysis for these additional contributors to heterogeneity.

Population variance of the prevalence was estimated based on the sampling distribution of the included studies. Variance of the prevalence was estimated based on the sampling distribution of a sample proportion. Random effects models were used to obtain summarized prevalence and their 95% confidence intervals in all eligible studies and subgroup analysis. Cohen’s Q statistics were used to evaluate the heterogeneity among studies. The inconsistency index (I²) was further used to quantify the level of heterogeneity with 25%, 50%, and 75% representing the evidence of low, moderate, and high heterogeneity respectively. Based on the heterogeneity test results (p < 0.001), random effects models were chosen for the estimations of weighted prevalence. Funnel plots were used to assess the publication bias and the asymmetry of the funnel plots were tested using the egger regression asymmetry test. The metafor package in the statistical analysis software R was used for the meta-analysis.

Results:

The combined database search yielded 8,799 studies (Appendix 1, available online at http://links.lww.com/AOG/C263). Studies not related to mood disorders in pregnancy were then excluded (n = 7166). After title screening, 446 studies remained for further abstract screening, of which 79 underwent manuscript review for inclusion. Forty were excluded (Figure 1).

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram of study selection.

A total of 39 studies met inclusion criteria for systematic review, of which 18 were included in the meta-analysis of the primary outcome. Twenty-one studies^20–40 reported raw depression or anxiety scale scores and were included in systematic review but not included in the meta-analysis because the proportion of subjects with an elevated score was not reported.

Systematic review yielded a small number of studies reporting the longitudinal relationship between duration of hospital length of stay and mood disorder symptoms. Trends towards increased depressive symptoms with increasing duration of hospitalization^24,41, improving depressive symptoms over hospitalization period^29,33,42, and no change in depressive symptoms over hospitalization period⁴⁰ were all reported. Worsening of depressive symptoms in the early postpartum period after hospital discharge²², with eventual improvement in symptoms during the later postpartum period was also reported^22,24,39,43. This is in contrast to anxiety symptoms, which tended to improve over the course of hospitalization^35,42 and postpartum period after hospital discharge^22,43.

A small number of studies identified by systematic review also compared rates of depression and anxiety between high-risk inpatients versus high-risk outpatients, but there was significant variation in type of depression and anxiety scales used by individual studies. Several studies reported a trend toward^22,44 or statistically significant lower mean depression and anxiety scores or rates of positive screen among antepartum outpatient groups as compared to antepartum inpatient groups^23,37,45,46. However, one study reported a non-statistically significant higher mean anxiety score among antepartum outpatients and one study reported no difference in depression diagnoses among high-risk antepartum outpatients versus inpatients^35,46.

Seventeen studies of 1,998 women were included in the meta-analysis of depression outcome^42–58 (Table 1) and 7 studies of 851 women included in the anxiety outcome^{42,43,45–47,58,59} (Table 2). Quality assessment of individual studies included in the meta-analysis are available in Appendix 2, available online at http://links.lww.com/AOG/C263. Additional pertinent study characteristics are reviewed in Appendix 3, available online at http://links.lww.com/AOG/C263. Studies were conducted in high-income or upper-middle income countries (as defined by the World Bank Country and Lending Groups⁶⁰) in Europe (n=8)^{43,45,48,50,52–54,57}, North or South America (n=4)^42,49,55,58, Australasia (n=2)^44,59, and Asia (n=2)^47,51. Two included study abstracts did not specify their study population^46,56.

Table 1.

Major Characteristics of Studies Included in the Meta-Analysis of Depression

Article citation	Study Design	Study time, Setting, and Location	n	Instrument	Cut-off	Elevated score (%)
Araujo et al. (2016)49	Randomized-controlled trial	April-June 2013; University Hospital, Vitória, Espírito Santo, Brazil.	25	EPDS	≥12	52.0%
Barber et al. (2017)44	Cross-sectional	July, 2009 – January, 2010; New Zealand.	68	EPDS	≥11	41.2%
Brandon et al. (2008)55	Cross-sectional	October, 2005 – December, 2006; University Hospital, Texas, USA.	129	EPDS, SCID	≥11 (EPDS)	44.2%
Byatt et al. (2014)42	Prospective cohort	June, 2011 – October, 2012. University Hospital, Massachusetts, USA.	62	EPDS	≥10	29.0%
Dagklis et al. (2016)53	Cross-sectional	September, 2014 – November, 2015; University Hospital, Greece.	157	EPDS	≥13	28.0%
Dagklis et al. (2018)54	Cross-sectional	September, 2014 - August 2016; University Hospital, Greece.	103	EPDS	≥13	24.3%
Doty et al. (2020)46	Cross-sectional	Not reported (abstract)	60	EPDS	≥10	30.0%
Fiskin et al. (2017)52	Cross-sectional	January - May 2014; Istanbul, Turkey.	122	CES	≥16	58.2%
Gourounti et al. (2015)50	Cross-sectional	February - June 2014; Two large public hospitals, Athens, Greece.	133	EPDS	≥11	54.9%
Hanko et al. (2020)43	Case-control	August 2015–2017; University Hospital, Dresden, Germany.	75	EPDS	≥10	49.3%
Hermon et al. (2019)57	Cross-sectional	November, 2016 – July, 2017: University Hospital, Israel.	279	EPDS	≥10	28.3%
Nagle-Yang et al. (2019)58	Cross-sectional	January – November, 2016; University Hospital, Cleveland, Ohio, USA.	98	EPDS	≥10	35.7%
Nie et al. (2017)51	Cross-sectional	March-August 2016; Three tertiary hospitals, Chongqing, Southwest China,	126	EPDS	≥13	33.3%
Thiagayson et al. (2013)47	Cross-sectional	September, 2011 – March, 2012; KK Women’s and Children’s Hospital, Singapore	200	MINI	n/a	18.0%
Tsakiridis et al. (2019)48	Cross-sectional	August 2016, - February, 2018; University Hospital, Northern Greece.	73	EPDS	≥13	32.9%
Weidner et al. (2010)45	Randomized-controlled trial	September 2003–2004. University Hospital, Dresden, Germany.	238	HADS	≥11	6.7%
Young et al. (2017)56	Cross-sectional	Not reported (abstract)	50	EPDS	≥10	24.0%

EPDS – Edinburgh Postnatal Depression Scale

HADS – Hospital Anxiety and Depression Scale

CES – Center for Epidemiologic Study Depression Scale

MINI – Mini International Neuropsychiatric Interview

Table 2.

Major Characteristics of Studies Included in Meta-Analysis of Anxiety

Article citation	Study Design	Study time, Setting, and Location	n	Instrument	Cut-off	Elevated score (%)
Byatt et al. (2014)42	Prospective cohort	June, 2011 – October, 2012. University Hospital, Massachusetts, USA.	62	GAD	≥10	12.9%
Cornsweet Barber et al. (2015)59	Cross-sectional	July, 2009 – April, 2010; Regional Hospital, New Zealand.	118	STAI	≥41	47.5%
Doty et al. (2020)46	Cross-sectional	Not reported (abstract)	60	STAI	≥40	55.0%
Hanko et al. (2020)43	Case-control	August 2015–2017; University Hospital, Dresden, Germany.	75	STAI	≥47	18.7%
Nagle-Yang et al. (2019)58	Cross-sectional	January – November, 2016; University Hospital, Cleveland, Ohio, USA.	98	STAI	≥40	46.9%
Thiagayson et al. (2013)47	Cross-sectional	September, 2011 – March, 2012; KK Women’s and Children’s Hospital, Singapore	200	MINI	n/a	12.5%
Weidner et al. (2010)45	Randomized-controlled trial	September 2003–2004. University Hospital, Dresden, Germany.	238	HADS	≥11	15.5%

GAD – Generalized Anxiety Disorder – 7 Scale

STAI – State-Trait Anxiety Inventory

MINI – Mini International Neuropsychiatric Interview

HADS – Hospital Anxiety and Depression Scale

Composite mean subject age was 30.1 years (SD 5.3)^{42,45,47,48,50–55,57–59}. Most subjects were multiparous (59.3%)^{43,46,50,51,55,57,58}, married or cohabitating (75.1%)^{42,43,46,48,49,53–55,57,58}, unemployed (56.2%)^{43,48,50,53–55,58}, and 59.6% had ≥ high school educational attainment^{42,43,48,50,51,53–55,58}. Body mass index (BMI) was most often < 30 (kilograms/meters²)(75.1%)^46,53,54 and most subjects were non-smokers (88.5%)^48,51,53,54. Mean gestational age at admission to antepartum unit was 33 weeks (SD 4.1)^{42–44,46,48,53,55,57} and mean antepartum length of stay was 9.8 days (SD 1.4)⁴². Indications for antepartum admission were preterm labor (43.3%), fetal growth problems or placental problems (30.5%) preterm prelabor rupture of membranes (12.3%), and hypertensive disorders of pregnancy or diabetes in pregnancy (13.5%). See Appendix 4, available online at http://links.lww.com/AOG/C263 for detailed aggregate demographic information of included studies.

For details of the screening scales and diagnostic interviews used by investigators, see Appendix 5, available online at http://links.lww.com/AOG/C263. Timing of survey administration relative to exposure was not reported in the majority of studies^{44–46,50–52,56–58}.

The estimated prevalence of the depression outcome was 34% (95% CI 27–41%) (Figure 2). Included studies demonstrated statistically significant heterogeneity as detected by χ² test (I² 92%, p<0.01). There were no observable differences among studies in the random-effects meta-regression for study design type, type of depression scale used, year of publication, country of origin, or quality assessment score and therefore heterogeneity was not arising from these systematic differences. A sub-analysis examining only studies using EPDS to report depression demonstrated very similar results (pooled prevalence of depression outcome 36% (95% CI 30–41%)) with a modest improvement in observed study heterogeneity (I²=76%, p<0.01) (Figure 3a). Visual examination of a funnel plot of included studies does not demonstrate publication bias (Appendix 6, available online at http://links.lww.com/AOG/C263) and Egger’s test for funnel plot asymmetry is not significant (p=0.05).

Figure 2.

Forest plot of random-effects (RE) meta-analytic estimate of prevalence of depression among women hospitalized on antepartum unit for obstetric complication.

Figure 3.

Forest plot (A) and meta-regression (B) of random-effects (RE) meta-analytic estimate of prevalence of anxiety among women hospitalized on antepartum unit for obstetric complication.

The estimated pooled prevalence of the anxiety outcome was 29% (95% CI 16–43%) (Figure 3a) with moderate-high heterogeneity among included studies (I²=75%, p=0.01). Study design type and study quality score were associated with significant differences in study heterogeneity in random-effects meta-regression (Figure 3b). A sub-analysis examining only studies using STAI to report anxiety demonstrated pooled prevalence of 42% (95% CI 26–58%) with an increase in observed study heterogeneity (I²=90%, p<0.01). Increase in heterogeneity likely related to small number of included studies as well as a single study outlier⁴³ using a STAI cut-off value of 47 as a positive screen, which likely contributed to under-ascertainment of prevalence compared to other investigators. Visual examination of a funnel plot of included studies does not demonstrate publication bias (Appendix 7, available online at http://links.lww.com/AOG/C263). Egger’s test was not performed given included study number < 10.

Study quality assessment using the NIH quality assessment tool demonstrated 4 studies^42,43,46,49 in the top quartile (rating ≥ 75%), 13 studies^{45,47,48,50–59} in the third quartile (with rating ≥ 50–74%), and a single study⁴⁴ in the second quartile (rating < 50%) (Table 2). All studies reported important quality measures such as clearly defined study objective, patient population and inclusion or exclusion criteria. Although sample size calculations with power justifications were not reported by 13 investigators^{42–47,50,51,53–56,58} for their various primary outcomes, this does not diminish the validity of the reported prevalence of included studies. Selection bias was possible in 7 studies^{43,49,51,52,55–57} which did not report the participation rate of eligible subjects and one study⁴⁴ which reported < 50% participation rate. Detection bias was possible in all studies, given the heterogeneity in validated screening measures and their selected cut-off values for a positive screen by the various investigators. Fifteen investigators^44,46–59 did not measure effect of exposure over time, limiting the conclusions which can be draw regarding changes in prevalence of anxiety or depression over the duration of hospitalization. Finally, of those investigators who did measure exposure over time, loss to follow-up was reported as > 20% in 5 studies^{44,47,50–52}, introducing the possibility of systematic differences in drop-outs impacting interpretation of longitudinal results.

Discussion:

Symptoms of antenatal depression or anxiety are common among women admitted to antepartum units with pregnancy complications. Our meta-analysis estimates that one in three women will have a positive depression or anxiety screening test or diagnostic interview during their hospitalization. Compared to previously reported estimates of the prevalence of antenatal mood disorders in the general obstetric population, our meta-analysis demonstrates twice the prevalence of symptoms of depression and anxiety among women hospitalized in the antepartum period with pregnancy complications.

There was significant heterogeneity identified among studies included in our meta-analysis, even after planned a priori subgroup and sensitivity analyses were performed. Included studies had significant clinical heterogeneity. Subjects were admitted for a wide variety of antepartum complications for varying durations of hospitalization. Presence of pre-pregnancy or antepartum pre-hospitalization mood or anxiety symptoms or disorders were not measured or adjusted for. These differences among subjects could affect both initial and cumulative prevalence of mood and anxiety symptoms reported in the studies. Subjects were also drawn from studies performed in a wide variety of countries with differing cultural, ethnic, social and economic characteristics. This could influence baseline rates of depression and anxiety and could also affect symptom reporting due to differences in universal screening rates (as related to subject familiarity with screening scales) or presence of societal mental health stigmas affecting comfort in truthful reporting. Studies also had significant methodological diversity which was not able to be accounted for in sensitivity analysis, including heterogeneity in study objectives, subject eligibility, inclusion or exclusion criteria, timing of administration of depression or anxiety screening scales relative to hospital length of stay, and differing cut-off values defined as positive screening test.

About 18% of included subjects (358/1998) were screened on the day of hospital admission and 34% within 72 hours of admission (680/1998), whereas timing of survey administration relative to hospital admission was not reported in over half of the studies (58%; n=1168 subjects). This points to the possibility that our prevalence estimates reflect increased depression and anxiety prior to hospitalization, rather than resulting from hospitalization itself. There were also reported conflicting trends in the course of symptoms over hospitalization periods in longitudinal studies, highlighting the uncertainty present in the current literature regarding cause and effect. The current evidence limits our ability to draw consistent conclusions about the relationships between pre-hospitalization depression and anxiety symptoms or diagnoses, and the possibility of a transient or enduring change (improvement or worsening) in depression or anxiety symptoms resulting from the acute stressor of hospitalization itself. In future studies, there would be benefit to standardizing timing of screening administration and longitudinal administration to help clarify the trajectory and persistence of these symptoms during hospitalization. Moreover, most of the studies relied on screening tools of symptom severity and do not include clinical diagnoses of depression and anxiety disorders. Because symptoms may fluctuate tremendously throughout hospitalization, instability is a challenge when monitoring symptoms alone, and future research should include measurement of the diagnosis of depression or anxiety through diagnostic interview (such as the MINI) in addition to use of screening scales, both when characterizing pre-hospital depression or anxiety and trends over hospitalization course.

There are additional limitations of the current meta-analysis. The primary outcome measures of depression and anxiety represent an aggregation of depression or anxiety screening scales and diagnostic interviews. This was necessary given the very limited number of studies assessing depression and anxiety in our study population using diagnostic interview. Though authors acknowledge that inclusion of positive screening tests may over-estimate the true prevalence of anxiety or depression in this analysis, large validation studies have noted high diagnostic accuracy of positive EPDS screen correlating to diagnosis of depression by diagnostic interview⁶⁸. An additional limitation of this review is the lack of included studies reporting on prevalence of other important mood disorders in pregnancy, such as bipolar disorder or post-traumatic stress disorder.

Screening rates for antenatal mood disorders, on antenatal units and in the outpatient setting, are low. Studies show that as many as 50% of patients are not screened or are misdiagnosed prior to delivery^69,70. Because evidence suggests that postpartum depression and anxiety disorders are often a continuum of underdiagnosed or undertreated antenatal mood disorders, antenatal screening and referral is recommended by the American College of Obstetricians and Gynecologists, American College of Nurse Midwives, United States Preventative Services Task Force^71–73, as well as other international societies^74–76. Based on our review, universal screening on admission to antepartum unit should be investigated as potential tool to better identify and treat women at risk of perinatal mood disorders. Moreover, clinical assessment in response to positive screens will discern if the elevated symptoms are a transient stress response or warrant active intervention.

Our results highlight the importance of education, support and interventions for women on antenatal units. There is a small body of literature on interventions to improve in-hospital depressive or anxiety symptoms in hospitalized obstetric patients. Interventions including acceptance-based therapy³², yoga²¹, relaxation techniques^26,49,77,78, psychosomatic interventions⁴⁵, music therapy^38,79, pet therapy²⁷, bed exercise programs³¹, aromatherapy⁸⁰, educational interventions^81,82, a stress-coping app⁸³ and other integrative modalities⁸⁴ have been studied. It is important to recognize that in addition to pharmacologic treatment, interpersonal therapy (IPT) and cognitive behavioral therapy (CBT) interventions are the most well studied and most effective treatment^85,86 and prevention strategies⁸⁷ for perinatal mood disorders. Adapting evidence-based, low cost and low intensity interventions designed to be self-administered or administered by non-specialists should be investigated in future research to better support all women on antepartum units and address the results of positive screening.

In conclusion, the current study demonstrates the high prevalence of perinatal mood and anxiety disorders identified in women hospitalized on antepartum units for obstetric complications of pregnancy and highlights the need for future research regarding screening protocols and implementation strategies of interventions for this at-risk population.