Disparities in Screening and Treatment Patterns for Depression and Anxiety During Pregnancy: An Integrative Review

Abstract

Introduction

Symptoms of untreated depression and anxiety during pregnancy are associated with serious adverse effects for the pregnant person, birth outcomes, and child development. However, pregnant persons are less likely to be screened and treated compared with nonpregnant people. In this systematic review, we aimed to explore individual, provider, and systems factors that impact screening, identification, and treatment patterns for depression and anxiety during pregnancy.

Methods

Studies were eligible for inclusion if they were conducted within the United States and published in English between January 2012 and January 2023. Each study included analysis that compared rates of screening, identification, or treatment engagement and explicitly discussed disparities or health equity in marginalized groups. Fifteen articles met full inclusion criteria.

Results

Results demonstrated variation in the screening, identification, and treatment of depression and anxiety during pregnancy among diverse groups of patients. Screening rates ranged from 51.3% in Puerto Rico to 90.7% in Alaska. Among specific clinical populations, rates were as low as 2.0%. Fewer than half of patients were referred to treatment when indicated by screening or diagnoses. Patient characteristics such as age, race, ethnicity, socioeconomic and health factors, mental health history, and obesity were associated with variation in the rates of screening, diagnoses, or treatment engagement. Language factors were the most common factor associated with lower rates of screening and treatment access.

Discussion

Results suggest that many pregnant people are being overlooked and lack appropriate referrals or resources to access treatment. Results are consistent with previous findings that role confusion and lack of time, provider training, and interest contribute to low rates of screening and treatment. Future research must focus on system level factors to address perceived barriers to screening and treating depression and anxiety during pregnancy in a systematic and equitable way.

INTRODUCTION

Despite impacting approximately 20% to 25% of pregnancies, ¹ , ² , ³ the symptoms of mood and anxiety disorders are often overlooked during pregnancy. Untreated mood and anxiety disorders during pregnancy have been associated with a range of adverse birth outcomes that include preeclampsia, prematurity, low birth weight, and birth complications. ² , ⁴ , ⁵ High rates of postpartum depression, substance and alcohol use, low rates of breastfeeding, low parental confidence, and impaired infant attachment have also been connected with untreated depression. ⁶ , ⁷ Notably, mental health conditions are a leading underlying cause of pregnancy‐related death ⁸ and account for 11% of pregnancy‐related deaths between 2008 and 2017. ⁹ Furthermore, perinatal patients with depression have higher overall health care use and medical costs than those without depression. ¹⁰

Screening during pregnancy and the postpartum period can increase the likelihood of receiving or being referred to treatment for depression. ¹¹ , ¹² The US Preventive Services Task Force ¹³ and the American College of Obstetricians and Gynecologists ¹⁴ recommend screening for depression during pregnancy. However, screening policies and practices vary at the state and local level. Despite evidence that a majority of patients with postpartum depression began to have symptoms prior to birth, multiple authors have found that providers are less likely to screen for depression during pregnancy compared with the pre‐ and postpartum period. ¹⁵ , ¹⁶ , ¹⁷

  

Continuing education (CE) is available for this article. To obtain CE online, please visit http://www.jmwhce.org. A CE form that includes the test questions is available in the print edition of this issue.

Treatment use for mood and anxiety disorders is also lower among pregnant compared with nonpregnant women. In a nationally representative sample of women who experienced a major depressive episode between 2011 and 2017, a smaller percentage of women who were pregnant (49%) received mental health treatment compared with nonpregnant women (57%). ¹⁸ Health care providers are also less likely to provide treatment consistent with current guidelines during pregnancy. ¹⁵ , ¹⁶ , ¹⁷ Confusion over who is responsible for screening and treating for mood and anxiety disorders during pregnancy, disagreement over role responsibility, and a lack of knowledge regarding treatment options for mental health concerns discourage many health care providers from screening their patients. ¹⁹

QUICK POINTS

• ✦Patient characteristics such as age, race, ethnicity, socioeconomic factors, mental health history, and obesity were associated with variation in the rates of screening, diagnoses, and treatment engagement.
• ✦Language factors were frequently associated with lower rates of screening and treatment due to a lack of resources available in patients’ preferred language.
• ✦Training is needed to improve providers’ familiarity and comfort with screening and treatment for depression and anxiety during pregnancy.

Consistent with patterns seen in adverse birth outcomes more broadly, the risk for mood and anxiety disorders during the perinatal period is increased among marginalized groups and may vary by ethnicity, ²⁰ , ²¹ age, ²² and environmental factors. ²³ , ²⁴ However, it is unclear whether similar disparities are present in screening and treatment patterns and might contribute to or heighten the risk for underrecognized and untreated mood and anxiety disorders.

The aim of this review was to query the existing literature to identify individual, provider, and systems factors that impact screening, identification, and treatment patterns for depression and anxiety during pregnancy with an emphasis on identifying individuals at elevated risk of being overlooked.

METHODS

Design and Literature Search

This review used the integrative review methodology described by Whittemore and Knafl. ²⁵ Search strategy followed the Preferred Reporting Items for Systematic Review and Meta‐Analyses statement. ²⁶ A systematic search of PsychINFO via Ovid, MEDLINE via PubMed, ProQuest Central, and CINAHL was performed in November 2022 and repeated in January 2023. Three concepts guided the search: pregnancy, depression or anxiety, and screening, identification, or treatment patterns. The search was amended to the specifications of each database; subject headings, medical subject headings, database‐specific headings, keywords, synonyms, truncation, and Boolean phrases were used. Results were limited to articles published between January 2012 and January 2023 to focus on recent clinical practice. The search strategy was developed in consultation with a health sciences librarian. The remainder of the search process was conducted independently by the primary author. A full set of search terms is presented in Table 1.

Table 1.

Search Terms

Database	Search Terms	Results
PubMed	((“Prenatal Care”[Mesh] OR “Pregnancy”[Mesh]) AND ((“Depression”[Mesh] OR “Depressive Disorder”[Mesh]) OR “Anxiety”[Mesh])) AND (((“Mass Screening”[Mesh] OR “diagnosis” [Subheading]) OR “Referral and Consultation”[Mesh]) OR “Health Services Accessibility”[Mesh])	1915
PsychInfo via OVID	(pregnan* or prenatal or perinatal or antenatal).mp. [mp = title, abstract, heading word, table of contents, key concepts, original title, tests & measures, mesh word] OR (pregnancy) OR exp Perinatal Period) OR (exp Prenatal Care) AND (exp Major Depression/ or exp ‘‘Depression (Emotion) OR (exp Anxiety Disorders/ or exp Anxiety) OR (anxiety or depress*)) mp. [mp = title, abstract, heading word, table of contents, key concepts, original title, tests & measures, mesh word] AND (disease screening/ or health screening/ or medical diagnosis/ or psychodiagnosis/ or screening tests/) OR (exp Mental Health Services/ or exp Health Care Utilization/ or exp Health Care Access/ or exp Treatment Barriers/)	318
CINAHL	((MH “Pregnancy”) OR (MH “Perinatal Period”) OR (MH “Pregnancy, Multiple”) OR (MH “Pregnancy Trimesters+”) OR (MH “Prenatal Care”)) OR pregnan* AND ((MH “Depression+”) OR (MH “Anxiety”) OR (MH “Anxiety Disorders+”)) OR (anxiety OR depress*) AND ((MH “Health Screening”) OR (MH “Mental Health Screening (Saba CCC)”) OR (MH “Referral and Consultation”) OR (MH “Remote Consultation”) OR (MH “Secondary Health Care”) OR (MH “Tertiary Health Care”) OR (MH “Health Services Accessibility+”)) OR MW diagnosis	1678

Eligibility and Study Selection

The search strategy resulted in 3911 articles. These were imported to Covidence Systematic Review program. ²⁷ Duplicate articles were removed, with 3118 unique articles remaining. Figure 1 presents a diagram of the results of the search. Given the aim of this review and the robust volume of search results, inclusion criteria were amended in an iterative process based on emerging categories within the original search. First, titles and abstracts were reviewed for relevance to the initial inclusion criteria: (1) English language, (2) original research that explored (3) screening and/or treatment of mood and anxiety disorders (4) during pregnancy (5) conducted within the United States. The search identified 267 articles that were eligible for full‐text review. At this stage, articles were not yet excluded if the title or abstract did not specifically mention an analysis that explicitly explored disparities or health equity in marginalized groups (eg, race, ethnicity, disability status, urbanicity, age) to address the possibility that additional information regarding disparities might be revealed in the full‐text review of articles.

Figure 1.

Study Selection and Search Strategy

During full‐text review, 253 additional articles were excluded: 126 were conducted in international settings, 4 addressed only postpartum and not pregnancy, 4 discussed only offspring‐related outcomes, and 3 had unclear pregnancy status during the study period. One study explored the lifetime mental health treatment history of pregnant people; however, authors did not specify treatment patterns during pregnancy. Several studies examined prevalence, risk factors, or associations between symptoms of mood and anxiety disorders and other health outcomes (n = 16) or evaluated the impact of a training, intervention, treatment program, screening initiative, or tool development or validation (n = 92) but did not describe disparities within the results (eg, did the screening program increase rates of screening or identification of mood and anxiety disorders equally for all participants, or were persons of color screened at lower rates?). Several studies that included discussion of disability status (n = 2), gender identity or sexual orientation (n = 2), preferred language or English proficiency (n = 2), or screening and treatment programs for specific populations such as veterans, imprisoned persons, or deaf persons but did not include analyses of disparities related to these demographic characteristics were excluded. Some articles are referenced more than once and the above numbers do not sum to the number of excluded studies. Fourteen articles were identified as appropriate for inclusion. Hand search of reference lists lead to the inclusion of one additional article, and a total of 15 articles were selected for review.

Data Extraction and Quality Appraisal

Articles within the derived sample were read multiple times in an iterative process to conduct critical appraisal, extract methodological details, and compare findings. The design, aims, sampling strategy and characteristics, methods, and key results were extracted from each article for summary and comparison. These details are presented in Table 2. First, studies were classified by methodology and sample characteristics. Articles were then compared to identify recurring domains and relationships among the results. In this review, the terms pregnant patients and pregnant people are used to be inclusive of persons across the gender spectrum with capacity for pregnancy. However, when discussing specific results, the terminology used by the authors of the included studies was used (eg, women, pregnant women).

Table 2.

Study Characteristics and Key Results

Author (Year)	Method	Sample	Outcome or Aim	Key Results
Bauman et al. (2020) 28	Secondary data analysis of retrospective self‐report survey and infant birth certificate data (PRAMS)	32,659 women with recent live birth in 2018 across 21 PRAMS sites	Prevalence of postpartum depressive symptoms, whether a health care provider screened for depression during prenatal and postpartum visits	Assessment for depression was higher during postpartum compared with pregnancy in 68% of sites. 79.1% reported that they had been asked about depression during a prenatal visit, range between 51.3% in Puerto Rico to 90.7% in Alaska. A higher percentage of women who were younger than 25 y; Black, Hispanic, American Indian/Alaska Native, or Hispanic (other); had 12 or fewer y of education, were not married, participated in WIC, had Medicaid at the time of birth, smoked during third trimester of pregnancy, or self‐reported depression prior to pregnancy reported that a provider had screened for depression during their pregnancy.
Bruney et al. (2022) 29	Retrospective observation using health record data	1042 women who received cared at a Federally Qualified Health Center in New York, New York	Examine effectiveness of an onsite screening program and identify barriers to treatment engagement	English‐speaking patients were twice as like to participate in treatment compared with non‐English‐speaking patients.
Felder et al. (2020) 30	Survey	303 pregnant individuals at risk for depression in United States (not otherwise specified)	Explore extent to which providers offered recommendations for preventive treatment for depression during pregnancy and whether recommendations varied by patient characteristics	Less than 15% reported that they had been referred for preventive counseling, 12% had been offered medication to prevent depression, referral rates varied by risk factors and did not vary by patient characteristics. Those with multiple risk factors, history of depression, mild depressive symptoms, and moderate to severe anxiety symptoms were more likely to receive a recommendation for therapy. Variations in treatment use by patient characteristics were not described.
Geier et al. (2014) 31	Retrospective longitudinal cohort analysis of administrative claims data	54,508 women with recent live birth and nonpregnant control group in Medicaid in California	Compared diagnoses of depression prior to conception, pregnancy, and postpartum period, analysis of treatment patterns	Women were 32% less likely to be diagnosed with depression during pregnancy compared with those in nonpregnant control group. Fewer than half of women diagnoses with depression received treatment during pregnancy. Lower rates of detection and treatment were noted among individuals of Hispanic decent, those <25 y old, or living in a rural setting.
Hsieh et al. (2021) 32	Semistructured in‐depth interview; demographic survey	29 women who had been screened for perinatal depression in Illinois	Describe the perception and experience of racially diverse women concerning the quality of depression screening for depression during perinatal period and its impact treatment seeking decisions	Screening was perceived to be ineffective by respondents due to an approach that lacked formality, privacy, follow‐up, and inclusion of fathers; screening process related to timing and frequency of screening, poor explanation of the screening, results not communicated, and negative emotions related to the screening, and perceived lack of support from providers, and macro‐level barriers of racism and mental health stigma.
Ko et al. (2012) 33	Secondary data analysis (NSDUH)	9032 women of reproductive age in a nationally represented sample	Estimate the prevalence of undiagnosed depression and treatment, and treatment barriers among pregnant and nonpregnant women	Depression in pregnant women was more likely to remain undiagnosed compared with nonpregnant women. Undiagnosed depression was more common among women who were younger; African American, Hispanic, or of other racial/ethnic groups; and uninsured. Fewer than half (49.6%) of pregnant women with depression received treatment compared with 53.7% of nonpregnant women. Women with clinical diagnoses were more likely to receive treatment. Treatment barriers did not vary by pregnancy status.
Koire et al. (2022) 34	Retrospective observation using health record data	490 women screened for depressive symptoms in obstetric or pediatric setting in Texas	Assess the relationship between wait time for mental health care and mental health outcomes among perinatal women	Patients who were seen by reproductive psychiatry specialty clinic were less likely to be Black or Asian. Wait time was not significantly associated with patient characteristics; however, pregnant women experience longer wait times compared with postpartum women and women closer to their due dates had longer wait times. All other analysis compared only pregnant and nonpregnant groups.
Kominiarek et al. (2014) 35	Secondary data analysis health record data/Medicaid claim data	107 minority, low‐income women at risk for adverse birth outcomes who had a birth paid by Medicaid in Illinois	Examine provider adherence to prenatal care guideline among patients with low income, with and without obesity	Provider adherence to prenatal care, including screening for depression, was lower among patients with obesity (2.0%) compared with those without obesity (11.4%).
Mestad et al. (2016) 36	Retrospective chart review	245 women screened for depression while receiving treatment at a hospital‐based clinic in Syracuse, New York	Examine rates of screening, referral, and treatment during pregnancy. Identify barriers to referral among women who screened positive for depression	More than half of those who screen positive for depression received referral, follow‐up, or treatment. Non‐English‐speaking patients, those without insurance, and those with multiple stressors were less likely to be referred and treated for depression following a positive screen. 9% of patients were not screened due to language barriers related to lack of validated measure in that language.
Phan et al. (2019) 37	Secondary data analysis with self‐report survey and newborn birth certificate data (PRAMS)	2888 randomly selected women with recent live births in New York, New York	Examine whether prepregnancy mental health concerns are associated with mental health care experience during and after pregnancy	Higher rates of help‐seeking were noted among women who were Asian/Pacific Islander compared with White, less likely among women with at least 16 y of education compared with women with <9 y and among US‐born women compared with non‐US‐born women. Prepregnancy mental health care use was associated with higher rates of consultation during pregnancy. Black and Hispanic women were more likely to receive mental health consultation during pregnancy. Women with <9 y of education were less likely to receive consultation.
River et al. (2019) 38	Interview, health records data, self‐report survey	101 pregnant women who planned to give birth at a safety‐net hospital in San Francisco, California	Examine the prevalence of unmet mental health needs and barriers between need and use among low income pregnant women	Only 54% of women with identified mental health needs received care. Most women with high PTSD symptom severity received mental health care during pregnancy, however, those with low‐grade depression were less likely to receive care. Women of color were even less likely to receive care. Many women named transportation and childcare needs as barriers to treatment engagement. Those that received care were more likely to be White, not partnered with the father of the child, and reported more childhood family dysfunction.
Salameh et al. (2019) 39	Secondary data analysis (NSDUH)	1232 Pregnant women with mental health and/or substance use disorder in a nationally representative sample	Examine racial and ethnic difference in the mental health treatment use	Black and Hispanic pregnant women were less likely to receive mental health treatment, even when adjusting for other factors such as age, income, education, and urbanicity.
Sidebottom et al. (2021) 40	Retrospective cohort analysis using electronic health records data	7548 women with recent birth within identified health system in Minnesota	Rates of depression screening during pregnant and 3 mo postpartum and identify factors that impact screening to illuminate disparities in screening	Clinical cite explained 23%‐30% of the variability in screening for depression and ranged between 34.7% to 100% of patients screened (cite differences were not described). Patients were more likely to be screened when receiving care from family medicine doctors (78.4%) and obstetrician‐gynecologists (64.6%) compared with certified nurse midwives (54.4%) and nurse practitioners (54.5%). Those who attended more prenatal care visits and those with documented history of depression or anxiety were more likely to be screened and women who spoke a language other than English were less likely to be screened. Those excluded from the analysis were more likely to be insurance with Medicaid/Medicare and/or speaking a language other than English.
Stevens et al. (2018) 41	Treatment efficacy study using interview, survey, standardize measures	67 perinatal patients referred for outpatient mental health care by provider within a large urban medical center in a large urban health care center, not otherwise specified	Demonstrate treatment effectiveness and variations in treatment engagement	Non‐Hispanic White women and partnered women were the least likely to return to therapy after birth. Results suggest that Black or African American women were the most engaged in treatment reflected in the highest number of sessions attended and lowest rates of early termination (did not reach statistical significance).
Wu et al. (2014) 42	Retrospective cohort analysis Medicaid claims data	804 low income, insured, pregnant women with major depression in South Carolina	Describe patterns of perinatal antidepressant treatment	Non‐White pregnant women were 36% more likely to have a gap in antidepressant use of >15 days compared with White women. Women with a history of depression prior to pregnancy were 44% more likely to discontinue antidepressants during pregnancy.

Abbreviations: NSDUH, National Survey on Drug Use and Health; PRAMS, Pregnancy Risk Assessment Monitoring System; PTSD, posttraumatic stress disorder; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.

Articles were critically appraised using an assessment tool appropriate to the study design. Joanna Briggs Institute Critical Appraisal tools ⁴³ were used for cross‐sectional observational studies and quantitative aspects of mixed‐methods designs. Validity criteria described by Whittemore et al. ⁴⁴ were used to assess qualitative designs. The critical evaluation of each article is summarized in Table 3.

Table 3.

Critical Appraisal: Key Strengths and Weaknesses for 15 Included Articles

First Author (Year)	Strengths	Weaknesses
Bauman (2020) 28 , a	Clear description of design Clearly stated objective Balanced summary of methods and results Brief but clear rationale for study Statistical analyses well described	Retrospective data collection Inclusion and exclusion criteria not well described Unable to address nonresponders and missing data, high likelihood of bias in results Language not included in analysis
Bruney et al. (2022) 29 , a	Clear description of design Clearly stated objective Use of reliable and valid tools for data collection, outcomes clearly defined	Retrospective data collection Inclusion and exclusion criteria not well described Unable to address nonresponders and missing data
Felder et al. (2020) 30 , a	Clear description of design Clearly stated objective Acknowledgment of potential for bias Results discussed in the context of previous finding and national policy Potential practice, policy, education, and research implications	Development, piloting, and modification of questionnaire not well described No exploration of researchers’ relationship with participants Limited generalizability due to recruitment strategy Language not included in analysis
Geier et al. (2014) 31 , a	Balanced summary of methods and results Clear description of recruitment, analysis and variables Statistical analyses well described Comparison made between relevant subgroupings	Retrospective data collection Unable to address missing data Language not included in analysis
Hsieh et al. (2021) 32 , b	Clear description of design Clearly stated objective Balanced summary of methods and results Results discussed in the context of previous finding and national policy	No exploration of researchers’ relationship with participants Little evidence of recursively or checking
Ko et al. (2012) 33 , a	Clear description of design Clearly stated objective	Unable to address missing data Language not included in analysis
Koire et al. (2022) 34 , a	Clear description of design Clearly stated objective	Retrospective data collection Unable to address missing data Homogenous sample Unclear pregnancy status
Kominiarek et al. (2014) 35 , a	Clear description of design Clearly stated objective Balanced summary of methods and results Results discussed in the context of previous finding and national policy	Retrospective data collection Unable to address missing data
Mestad et al. (2016) 36 , a	Clear description of design Clearly stated objective Balanced summary of methods and results	Retrospective data collection Unable to address missing data
Phan et al. (2019) 37 , a	Clear description of design Clearly stated objective Statistical analyses well described	Retrospective data collection Unable to address missing data Language not included in analysis
River et al. (2019) 38 , a , b	Clear description of design Clearly stated objective Clear description of recruitment, analysis and variables	No exploration of researchers’ relationship with participants Qualitative analysis lacks vividness, thoroughness, sensitivity Details for data integration limited
Salameh et al. (2019) 39 , a	Clear description of design Clearly stated objective Balanced summary of methods and results Brief but clear rationale for study Clear description of recruitment, analysis and variables	Unable to address missing data Language not included in analysis
Sidebottom et al. (2021) 40 , a	Clear description of design Clearly stated objective Balanced summary of methods and results	Retrospective data collection Unable to address missing data Clinical site characteristics not described
Stevens et al. (2018) 41 , a , b	Clear description of design Clearly stated objective Balanced summary of methods and results Use of theoretical framework	Unclear comparison of pregnancy and postpartum periods
Wu et al. (2014) 42 , a	Clear description of design Clearly stated objective	Retrospective data collection Unable to address missing data Some variables used in analysis poorly described

^aAppraised with Joanna Briggs Institute Critical appraisal tools. ⁴³

^bAppraised with Whittemore et al. validity criteria for qualitative designs. ⁴⁴

Theoretical Framework

King's Theory of Goal Attainment and Transaction Process ⁴⁵ , ⁴⁶ guided the analysis of the included articles. Although King's theory was developed specifically within nursing practice, it remains relevant to other disciplines of clinical practice through several of its underlying assumptions. First, patients and families have the right to knowledge about their health. Second, health occurs as an interaction between human beings and their environment. The third important assumption is that the perceptions, judgements, and actions of both patients and providers impact health interactions and their outcomes. King's conceptual framework provides an excellent point of entry for researchers to identify barriers to screening, identification, and treatment of depression and anxiety during pregnancy.

The findings of this review are organized and discussed among 3 interactive systems described by King's framework: personal (individuals), interpersonal (groups), and social (society). Here, personal system is used to refer to traits and behaviors of patients, interpersonal system refers to interactions or relationships between patients and others including providers and other individuals, and social system is used to describe groups norms, cultural values, and structural or systems level factors. By considering the personal, interpersonal, and social systems involved in mental health care, King's framework is used to identify approaches to mitigate barriers to care that involve patients, providers, and social systems.

RESULTS

Data Analysis

Study Characteristics

Fifteen articles met the full criteria for inclusion in this review. Eligible articles included 13 quantitative, ²⁸ , ²⁹ , ³⁰ , ³¹ , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁹ , ⁴⁰ , ⁴² qualitative, ³² and mixed‐methods ³⁸ , ⁴¹ research designs. The most common design was secondary retrospective cohort analysis using electronic chart review, ²⁹ , ³¹ , ³⁴ , ³⁶ , ⁴⁰ Medicaid claims data, ³¹ , ³⁵ , ⁴² or an existing data set. ²⁸ , ³⁷ , ³⁹ Sample size ranged from 29 to 101 in qualitative studies and 67 to 54,508 in quantitative studies. Data were collected from local health care systems and state and national databases. Studies examining state or national samples used the National Survey on Drug Use and Health (NSDUH) ³³ , ³⁹ and the Pregnancy Risk Assessment Monitoring System (PRAMS) multistate ²⁸ and New York City ³⁷ data.

Aims varied by study. These included examining adherence to screening guidelines and rates of screening or identification of depression, treatment use, symptom prevalence or need, patient perceptions of care, health care provider behavior, and several specifically named identifying factors that impact these patterns among their aims. One study presented results as health care providers’ likelihood of adhering to screening or treatment recommendations. ³⁵ All others reported results in terms of patients’ likelihood of being screened, diagnosed, or treated. All studies examined the phenomena from patient data, although several reported on provider behavior (eg, adherence to screening recommendations) as extrapolated from patient data.

All included studies use the words woman or mother to refer to individuals included in the analysis. Participant age ranged from 14 to 44. However, several studies limited participation to those older than 18 years or did not report on age. Two articles specified that assessment was possible for monolingual Spanish‐language speakers to participate. ²⁹ , ³⁸ The remainder required English proficiency for participation or did not comment on language. All studies reported on depression or depressive symptoms in the perinatal period. Some also included discussion of posttraumatic stress disorder (PTSD) ³⁸ or anxiety. ³⁰ Mental health conditions were assessed via self‐report, International Classification of Diseases, Tenth Revision codes extracted from electronic health records or Medicaid claims data, Edinburgh Postpartum Depression Scale, PTSD Checklist for DSM‐5, Patient Health Questionaire‐2, Patient Health Questionnaire‐9, or clinical interview. Key variables used in the analysis of the included articles are presented in Supporting Information: Table S1. Results are organized within the personal, interpersonal, and social systems of King's framework. Due to the interactive and interrelated nature of these systems, many possible categorizations are possible. We attempt to organize results by patient related factors, provider or patient‐provider relationship factors, and factors that reflect social systems.

Data Presentation

The results of this review demonstrate variation in the screening, identification, and treatment of depression and anxiety during pregnancy among diverse groups of patients. However, patient‐, provider‐, and systems‐level factors were not consistent and were frequently contradictory among studies. Analyses often compared only pregnant with nonpregnant patients, and few provider characteristics were discussed. One study presented results as providers’ likelihood of adhering to screening or treatment recommendations. ³⁵ All others reported results in terms of patients’ likelihood of being screened, diagnosed, or treated. Factors associated with higher and lower rates of screening or detection and treatment engagement are presented in Figure 2.

Figure 2.

Factors Associated with Variations in Rates of Screening, Detection, and Treatment Engagement for Depression and Anxiety During Pregnancy

Abbreviation: PTSD, posttraumatic stress syndrome.

Patient (Personal) Level Factors Associated With Screening, Identification, and Treatment Engagement

Most studies included in this review presented their results in terms of patient characteristics or behaviors. Patient characteristics of interest included age, race and ethnicity, language, nativity, mental health history, history of trauma, substance use, health status, disability, income, education, marital status or other social support, involvement with social services, and health insurance factors. Individuals not speaking English were consistently less likely to be screened, with language barrier named as the most predictive element for lack of screening across studies. ²⁹ , ³⁶ , ³⁸ , ⁴⁰ In 2 studies, higher rates of undiagnosed depression were associated with Black or Hispanic race or ethnicity and lack of insurance. ³¹ , ³³ However, Bauman et al. found higher percentages of patients screened if they were Black, Hispanic, American Indian/Alaskan Native, or non‐Hispanic (other) compared with White or Asian/Pacific Islander. ²⁸

Women who had fewer than 12 years of education, were unmarried, received Special Supplemental Nutrition Program for Women, Infants, and Children or Medicaid at the time of their newborn's birth, had a history of smoking during pregnancy, and had a self‐reported history of depression were more likely to have had a screening for depression during pregnancy compared with those without the listed characteristics. Age under 25 years was also associated with both lower rates of depression detection ³¹ , ³³ and higher rates of screening ²⁸ in separate studies. Sidebottom et al. found no difference in screening rates during pregnancy related to patient characteristics. ⁴⁰

Treatment engagement refers to both the provider behavior of providing a referral for mental health services including further assessment, counseling, psychotherapy, and psychopharmacology treatment modalities and use of care by patients. Felder et al. found that less than 15% of patients reported that they had been referred for preventive counseling, and 12% had been offered medication to prevent depression. ³⁰ Many authors reported low rates of treatment referral. For example, more than half of the women diagnosed with depression were found to lack referral or follow‐up. ³⁶ Likelihood of receiving a recommendation for mental health services was not found to vary by patient characteristics but was associated with risk factors for depression including a documented history of depression, mild depression symptoms, and moderate to severe anxiety symptoms. ³⁰

Only 50% of women with identified mental health care needs received treatment during pregnancy. ³¹ , ³⁸ One study reported that non‐Hispanic White women were the least likely to return to therapy after the birth of their child, whereas Black women appeared to be the most engaged as measured through low rates of early termination and highest number of sessions attended, although these findings did not reach statistical significance. ⁴¹ Black and Hispanic women were reported more likely to have had mental health service consultation during pregnancy in New York City PRAMS 2009‐2011 data ³⁷ and least likely to receive mental health services according to NSDUH 2008‐2014. ³⁹ Authors found that women who followed up with perinatal mental health specialty care were less likely to be Black or Asian, ³⁴ and women of color, defined as non‐White by the authors, were especially unlikely to receive mental health care. ³⁸ Non‐White women were 36% more likely to discontinue antidepressants during pregnancy or have a 15‐day or greater gap in treatment. ⁴²

Severity of depressive symptoms and history of depression were associated with treatment patterns. Low‐grade depressive symptoms were also associated with less treatment engagement. ³⁸ Women who had a history of antidepressant use prior to pregnancy were 44% more likely to discontinue this treatment during pregnancy compared with those who had initiated antidepressants during pregnancy. ⁴² Prepregnancy mental health service use was associated with higher rates of perinatal mental health consultation during pregnancy in PRAMS 2009‐2011 data. ³⁷

Provider (Interpersonal) Level Factors Associated With Screening, Identification, and Treatment Engagement

Screening and detection patterns varied greatly across locations and by provider type. In an analysis of multistate PRAMS 2018 data, screening for depression during pregnancy varied by state or territory, ranging from 51.3% in Puerto Rico to 90.7% in Alaska. ²⁸ Rural setting was associated with lower rates of depression detection. ³¹ Clinical site was found to explain 23% to 30% of variation in screening rates, ⁴⁰ signifying that receiving health care from specific health care locations had a meaningful impact on whether patients would be screened for depression and anxiety or receive care. However, this variation was not further described. Patients were more likely to be screened when receiving care from family medicine physicians (78.4%) and obstetrician‐gynecologists (64.6%) compared with certified nurse‐midwives (54.4%) and nurse practitioners (54.5%). ⁴⁰ Providers were less likely to adhere to depression screening for pregnant patients who were obese (2.0%) compared with those who were not obese (11.4%). ³⁵

Results may also reflect the quality of the patient‐provider relationship and are considered within the interpersonal systems. For example, variation was noted in patients’ perspectives and experience of screening for depression. Some patients reported low confidence in the value of screening. ³² Some participants did not understand the significance of self‐report depression assessments and did not receive the results of these screenings ³² or did not understand the screening questions. ²⁹ Patients reported that providers appeared disinterested and unsupportive during screening. ³²

Qualitative inquiry found that Black patients were fearful of reporting depressive symptoms related to concerns about custody or child protective services involvement. ³² Screening was also perceived to be ineffective by respondents due to an approach that lacked formality, privacy, follow‐up, sharing of results, or the inclusion of fathers. Participants reported uncertainty related to the timing and frequency of screening, poor explanation of the screening, a perceived lack of support from providers, and included macro‐level barriers of racism and mental health stigma. ³²

Systems (Social) Level Factors Associated With Screening, Identification, and Treatment Engagement

The most frequently reported barrier to screening, ³⁶ , ⁴⁰ or referral and treatment ²⁹ , ³⁶ was non‐English language. ³⁶ , ³⁸ , ⁴¹ For example, English‐speaking patients were twice as likely to participate in treatment. ²⁹ Among patients receiving perinatal care at an urban hospital‐based clinic, 9% of patients were not screened for depression due to language barriers. ³⁶ Although this location had access to translators who were able to communicate in many languages that reflected the community, health care providers lacked validated measures to assess for depression in these languages. ³⁶ Furthermore, individuals who did not speak English were more likely to be excluded from the analysis or from participation in research. ³⁸ , ⁴⁰ Insurance status and having multiple concurrent life stressors were associated with low treatment engagement. ³⁶ Additionally, women with at least 9 years of education were less likely to engage with care. ³⁷ Transportation and childcare needs were cited as barriers to participating in treatment or attending follow‐up appointments. ³⁸

Critical Appraisal

Several studies included in this review shared methodological limitations (see Table 3). For example, although many studies included large sample sizes, these relied largely on preexisting data sets or existing health records. Researchers could not account for missing data, and their results may be impacted by incomplete documentation. Several studies lacked a clear distinction between pregnant or postpartum status among participants at the time of data collection. Many of the included studies used retrospective data collection regarding experiences during pregnancy from postpartum samples. For example, PRAMS data are collected several months postpartum, and the survey asks whether a health care provider assessed for depression during any prenatal visits. ²⁸ It is possible that some patients did not remember or did not realize they were being asked about depression during these visits. Random samples were not possible in the designs of most studies, and individuals who did not respond or declined to participate may have had different answers or experiences than those who did.

Several personal characteristics were not included or described, and the perspectives of these individuals are therefore missing. For example, none of the included studies reported on factors such as sexual orientation or gender identity, and only one mentioned disability. The NSDUH data set excluded persons experiencing homelessness, in active military duty, or residing in specific institutions such as incarcerated individuals. Persons reporting severe depressive symptoms, suicidal ideation, and self‐harm behavior were also often excluded from participation in research or excluded from the analysis. ³³ , ³⁹

DISCUSSION

In this systematic review, we aimed to explore current literature to identify individual, provider, and systems factors that impact screening, identification, and treatment patterns for depression and anxiety during pregnancy. Results of this review are consistent with previous studies that demonstrate lower rates of screening, identification, and treatment of depression and anxiety among pregnant patients compared with preconception and postpartum periods. ¹⁵ These results expand on previous literature and provide additional clarity regarding disparities that may further decrease the frequency of screening and treatment use for depression and anxiety for some patients during their pregnancy.

Patient characteristics such as age, race, ethnicity, socioeconomic factors, mental health history, and obesity were associated with variation in the rates of screening, diagnoses, and treatment engagement. Language was the most common factor associated with lower rates of screening and treatment due to a lack of health care providers who were able to communicate in patients’ preferred language or validated screening tool in that language. Overall, fewer than half of patients were referred to treatment when indicated by screening or diagnoses. These findings are not consistent with national recommendations and highlight that screening is necessary but not sufficient to address depression and anxiety during pregnancy, and further research is needed to determine best strategies for widespread implementation of existing guidelines.

Results of this review are also consistent with a reluctance among patients to disclose mental health symptoms or risk factors due to internal and external stigma, concern that parental custody may be impacted, or the perceived threat of psychiatric hospitalization. ⁴⁷ , ⁴⁸ Patients also describe feeling dismissed by family and providers who attribute symptoms to normal pregnancy ⁴⁹ or maintain the outdated belief that pregnancy is protective against depression. ⁵⁰ , ⁵¹

The conflicting and contradictory findings among the included articles may mirror local variation in screening and treatment practices. Although site characteristics were found to explain much of the variation, these differences were not described. However, the importance of site difference is consistent with previous literature. For example, health care provider interests and motivation, training, and available resources have been found to impact rates of screening and treatment decisions for depression during pregnancy. ⁵² Clinic priorities, familiarity with mental health treatment, and provider motivation were associated with higher rates of screening during pregnancy. ¹⁵ Some providers reported screening for depression only when they observed symptoms or risk factors. ¹⁵ , ⁵² However, this reliance on provider discretion disproportionately places the burden on patients to request assistance and may contribute to disparities if Black individuals or other marginal groups are fearful of disclosure.

Limitations

This review has several limitations. A single researcher implemented the search, developed the exclusion and inclusion criteria, and assessed articles for inclusion. This process was completed by the lead author as an aspect of her doctoral dissertation in close collaboration with the coauthor. Articles were restricted to research conducted within the United States to facilitate comparison. It is possible that the observed patterns do not generalize to a global health care perspective. Furthermore, collectively very few participants lived within rural or suburban settings, and the urban‐centric results of this review may not be applicable to these locations. The review was also constrained by limitations of the available research, which obscures disparities through omission of relevant characteristics. Data were often extracted from chart review or retrospective report from patients about their experiences. Therefore, health care provider type or setting was not always clear. Further exploration is needed to better understand the screening and treatment patterns of midwives and advance practices nurses. Many of the studies, and this review itself, are limited by English language inclusion criteria as non‐English speakers were also excluded from several studies, demonstrating a disparity within the research itself.

Implications

The results of this review have several important implications for practice, research, and policy. Broadly, they diminish confidence that health care providers are able to identify patients in need of depression screening without a screening guideline in place and highlight opportunities to improve referrals and treatment engagement. Through the lens of King's framework, interventions to improve screening, identification, and treatment engagement for depression and anxiety during pregnancy must engage patients, providers, and social systems.

Practice

Health care providers require confidence and familiarity with screening tools and must select tools that are appropriate for their patient population. Given the role of language‐related barriers to screening and treatment, practices may need to expand the language proficiency of staff or use of interpreters to ensure they are able to meet patient needs. Barriers to treatment, such as childcare and transportation needs, may also be important in the screening process. Patients request clear rationale for screening, including what will be done with the information they share, thoughtful explanation of results, and timely follow‐up as part of their treatment encounter. Providers can incorporate patient preference into their care by ensuring adequate privacy or consider including family or other support based on patient preferences.

Health care providers may be reluctant to screen if they are not certain of next steps. Therefore, providers may increase their confidence by becoming more familiar with local resources and referral options following a positive screening. Several continuing education programs for perinatal mental health screening and treatment practices are available, and health care organizations can make these available to their workforce. When possible, mental health resources can be integrated into primary, antenatal, and pediatric care settings, allowing for handoff within a practice or by establishing relationships with a clinic or set of clinicians in their communities. Providers can also be reassured that patients generally find screenings to be acceptable and many screening tools require only a few minutes to administer or review.

Policy

Universal screening policies have been shown to successfully increase rates of screening and are seen as acceptable and feasible by patients and providers. ¹² , ²⁹ , ⁵³ Medicaid reimbursement incentives have been associated with increases in screening and treatment engagement during pregnancy and postpartum. ⁵⁴ In many states, Medicaid will reimburse for maternal depression screenings during a well‐child visit; however, specific clinical requirements, required or recommended tools, and reimbursement amounts vary by state. ⁵⁵

Policies that support universal screening must be accompanied by attempts to mitigate barriers to referral and treatment following a positive screen. Expansion of telehealth resources accompanied by interstate license reciprocity may help expand resources to underserved areas, such as rural communities or areas that lack mental health providers. Although some clinicians may choose to enhance their education through dual credentialing in advanced practicing nursing or other specialized continuing education programing, it may be necessary to include more mental health education in prelicensure training for midwives, advanced practice nurses, and physicians to improve competency in the screening and treatment of depression and anxiety during pregnancy. Funding will also be needed to expand training opportunities and availability of educational resources for providers to meet patient needs and increase provider confidence in addressing mental health concerns during pregnancy. For example, psychiatric consultation programs have been successful in improving rates of treatment or referral among clinicians in perinatal care, family medicine, and pediatrics. ⁵⁶

Research

Future researchers must work to identify and dismantle systems‐level barriers to screening and treatment engagement. This research may explore gender identity, disability, and other factors missing from the overall analysis provided in articles reviewed here. Researchers have found that deaf women ⁵⁷ and sexual minority women ⁵⁸ were also less likely to receive recommended screening or treatment and further investigation is necessary to understand and address these disparities. To address language barriers, screening tools must be translated and validated for cultural and linguistic congruence. Existing tools must be disseminated and used by adequately trained providers. Understanding the barriers to treatment will also be enhanced by a deeper understanding of an observed clinical reluctance to treat pregnant clients reported by some clinicians. ⁴⁹

CONCLUSION

The results of this review expand on observed patterns that reflect less frequent screening and lower rates of treatment for mood and anxiety disorders during pregnancy compared with the postpartum period. The current review highlights several additional factors that exacerbate the risk for missed opportunities to screen and therefore identify, refer, and treat mood and anxiety disorders. Individuals who do not speak English are at the most consistent risk, given the frequency of reported language barriers, lack of adequate or appropriate screening tools, and common exclusion from participating in research by teams that are only able to speak English. As site characteristics were found to explain much of the variance in screening, future research must focus on systems‐level factors to address perceived barriers to screening and treating depression and anxiety during pregnancy in a systematic and equitable way.

CONFLICT OF INTEREST

The authors have no conflicts of interest to disclose.

Supporting information

Table S1. Key Variables Included in Analysis^a,b