Interventions Addressing Social Needs in Perinatal Care: A Systematic Review

Introduction: Health is impacted by a wide range of nonmedical factors, collectively termed the social determinants of health (SDoH). As the mechanisms by which these factors influence wellness and disease continue to be uncovered, health systems are beginning to assess their roles in addressing patient's social needs. This study seeks to identify and analyze clinic-based interventions aimed at addressing patients' social needs in perinatal care, including prenatal, antepartum, and postpartum care. Methods: We conducted a search of six databases through May 2020 for articles describing screening or intervention activities addressing social needs in at least one SDoH domain as defined by Healthy People 2020. We required that studies include pregnant or postpartum women and be based in a clinical setting. Results: Thirty-one publications describing 26 unique studies were identified. Most studies were either randomized-controlled trials (n=10) or observational studies (n=7) and study settings were both public and private. The mean age of women ranged from 17.4 to 34.1 years. Most studies addressed intimate partner violence (n=19). The next most common need addressed was social support (n=5), followed by food insecurity (n=3), and housing (n=2). Types of interventions varied from simple screening to ongoing counseling and case management. There was wide heterogeneity in outcomes investigated. Most IPV interventions that included counseling or ongoing support resulted in reduced IPV recurrence and severity. No intervention with only screening showed a reduction in rate of IPV. Conclusion: This systematic review shines light on several avenues to support pregnant and postpartum women through interventions that embed acknowledgment of social needs and actions addressing these needs into the clinical environment. The results of this review suggest that interventions with counseling or ongoing support may show promise in alleviating social risk factors and improving some clinical outcomes. However, the strength of this evidence is limited by the paucity of studies. More rigorous research is imperative to augment the knowledge of social needs interventions, especially in domains outside of IPV.


Introduction
Evidence indicates that the social determinants of health (SDoH) shape health in powerful ways. 1,2 The SDoH have been defined as the conditions in which people are born, grow, live, work, and age. 1 SDoH in-clude a wide range of social, structural, and economic factors that relate to housing, access to healthy food, transportation, education, crime and violence, and the environment, among other conditions that contribute to a person's quality of life. Fundamental to the study of SDoH is understanding that differences in how population groups experience these conditions are key drivers of disparities in health and health care. Structural racism exacerbates the differential experience of these social and economic conditions, especially among minority communities. 3 While SDoH refer to social conditions at the level of communities, social needs refer to the acute social and economic challenges of individuals, which are often direct consequences of those broader social conditions. 4 The health system has long recognized that SDoH are intricately tied to the development of disease and efforts to address SDoH have been described as looking upstream. However, the extent to which the health system can intervene upon SDoH in the setting of clinical care is limited because community-level conditions are often governed by entities outside the health system such as local and national policies. An alternative approach to improving social conditions is addressing the social needs of individual patients, targeting a ''middle stream'' that is further upstream than traditional medical care, but downstream from the community-level SDoH. Traditionally, addressing individuals' acute social needs has been the role of social service agencies, but a growing number of health systems are beginning to incorporate practice-based interventions for social needs into clinical care. 5, 6 In the primary care literature, some studies have shown that addressing social needs in the clinic increased patients' use of preventive health care. 7,8 In response to this growing area of practice, the National Academies of Medicine published a consensus report in which they describe activities of health systems that facilitate the integration of social needs care into clinical care. 9 Using transportation as an example, they suggest that health systems can identify social risks and assets of patients by asking their patients about access to transportation. They also suggest that health systems can reduce social risk by assisting in connecting patients with relevant resources, for example, by providing transportation vouchers so that patients can reliably travel to their health care appointments. These are not novel interventions, but integrating these actions into the clinical care environment is a newer focus of practice.
The idea of unmet social needs as a barrier to accessing health care is of special interest in perinatal care. In perinatal care, visits are frequent and occur in a concentrated period of time; to illustrate, the recommended American Congress of Obstetrics and Gyne-cology (ACOG) prenatal visit schedule consists of visits every 4 weeks until 28 weeks, every 2 weeks until 36 weeks, and every week until delivery. 10 The prenatal period is a critical time in pregnancy where essential health screening, pregnancy education, and counseling are performed. 11 Inadequate prenatal care, defined in one study as fewer than or equal to four antenatal visits, or first visit in the third trimester, has been linked to adverse neonatal outcomes. 12 Unfortunately, only 77.1% of women initiate pregnancy care in the first trimester. 13 This number is lower among certain minority groups; only 66.5% of African American and 63.0% of American Indian/Alaska native women receive prenatal care in the first trimester. 13 These disparities in prenatal care utilization are hypothesized to be linked to social needs, as women have cited lack of transportation, psychosocial barriers, and low health literacy as barriers to care. [14][15][16] The objective of this systematic review was to summarize the evidence on interventions that address social needs in perinatal care settings. For the purposes of this review, we use the term ''intervention'' to describe any clinic-based activity that addresses patients' social needs, including screening, education, counseling, and referral to support services. In this study, we define ''perinatal care'' as all pregnancy-related medical care, including prenatal, antenatal, and postpartum care.

Data sources
We developed our search strategy in consultation with a medical librarian, which we based on previously published reviews conducted on interventions for social needs in health care settings. 17,18 Search terms included key words in four domains: (1) pregnant or postpartum women; (2) SDoH; (3) screening and interventions; and (4) outcomes. The complete description of our search strategy is available in the Supplementary Appendix. We searched six databases (PubMed, Embase, Cochrane, CINAHL, Scopus, and Web of Science) for studies describing health care-based interventions published through June 27, 2019, in English. We repeated the search on May 6, 2020, to identify articles published in the time since we began the review. We did not place lower limits on the date range of the search.

Study selection
Search results were imported into the Rayyan QCRI web app for systematic reviews. 19 Articles were screened by title and abstract for relevance by a team of reviewers (A.M.R., P.W.A., M.R.Y., R.D., and A.E.). To increase inter-rater reliability, a calibration between reviewers was conducted. The first 100 articles were independently screened by each reviewer and the decision for inclusion was compared. Discrepancies were solved by open discussion between reviewers. The remaining articles were divided evenly among the team and screened by two independent reviewers, with ties broken by a third reviewer.
To be eligible for inclusion, studies had to describe interventions that (1) involved pregnant or postpartum women; (2) addressed a social need in one or more Healthy People 2020 SDoH domains (economic stability, education, social, and community context, health and health care, or neighborhood and built environment); 20 and (3) took place in whole or in part in a health care setting that we defined as a physical location where women received medical care outside of the home. This included medical care that was given by physicians, physician assistants, nurses, or nurse midwives. Studies describing home visiting, case management, and community health worker programs were only included if the programs were part of an intervention that was clinic based. For example, a study describing an intervention in which a patient is referred to a home visiting program after screening positive for transportation needs during a prenatal appointment would meet the inclusion criteria, while a study describing a home visiting program alone would not.
The Healthy People 2020 framework gives examples of specific social needs in each SDoH domain. 20 We used these examples to guide study selection but did not limit the selection to only these examples. Both observational and experimental studies were included. When conducting reviewer calibration, we identified areas of misalignment between readers' interpretations of study inclusion. These were settled through group discussion before continuing with study reviews. Although some definitions of SDoH include diet and exercise, substance use, and mental health, 21 the Healthy People 2020 definition does not. 20 Thus, we only included studies describing screening/interventions for substance use (including smoking and alcohol use), mental health (depression, post-traumatic stress disorder [PTSD]), and nutrition if they also addressed a Healthy People 2020 domain (economic stability, education, etc.). We included interventions for intimate partner violence (IPV) as IPV is part of crime and violence, which lies under the larger Healthy People 2020 category, neighborhood and the built environment. We did not consider novel methods of prenatal care (e.g., group prenatal care) because we wanted to identify interventions that can be translated to routine clinical care.
We chose to limit studies to those in high-income countries upon recognizing that the research questions could best be answered by evaluating interventions in countries where patients' social needs were most similar to those experienced by U.S. patients. We used World Bank categories to define country income level. 22 Data extraction A standard template was constructed in Microsoft Excel to extract the following data from included studies: year, country or region, study design, setting, participant inclusion/exclusion criteria, population characteristics, social need addressed, description of intervention including who carried out the activities, screening tool used, conceptual framework, and if applicable, whether there was a referral process and/or mention of specific community partners. Studies were assessed for risk of bias and assigned quality ratings based on criteria for randomized-controlled trials (RCTs) and observational studies in the Grading Recommendations Assessment Development and Evaluation (GRADE) approach. 23,24 Data from each article were extracted independently by two reviewers and discrepancies were discussed between reviewers. Given the heterogeneity of interventions and outcomes across studies, no meta-analysis was performed.

Social needs
The vast majority of interventions addressed IPV (n = 19). 25 Risk Overview by Harrison et al., patients were asked about access to a telephone, transportation, IPV, social support, housing instability, legal system involvement, and food insecurity. 51 Food security and access to healthy foods were addressed in three studies. 45 Housing stability and quality were addressed in only two studies, the same multi-item screening surveys as mentioned above. 50,51 Education, employment, and poverty were only addressed in the 10-item brief risk factor survey by Price et al. 50 It should be noted that no intervention went beyond screening for housing, education, employment, or poverty. One study screened for workplace exposures. 49 Two studies addressed health literacy: Sayakhot et al. offered a web-based education module for gestational diabetes that was issued during clinic visits and Backonja et al. incorporated reproductive health education into regular clinic-based counseling sessions. 31,36 No studies addressed civic participation, discrimination, or incarceration history, which fall under the social and community context domain of Healthy People 2020.

Interventions and outcomes
To facilitate the reporting of results, we classified interventions into four broad categories: screening (n = 11), brief (n = 6), counseling (n = 6), and ongoing support (n = 6). One study compared three different interventions, so we cross-counted this intervention in three categories. 35 We also classified outcomes into four categories: process measures, social needs outcomes, health behaviors, and health impacts, which we based on previous work by Gottlieb et al. 17 Figure 2 provides a visual display of interventions according to type of intervention, social need(s) addressed, and outcomes reported. Table 2 provides a more detailed breakdown of study outcomes and how they were classified.
The studies in our review were heterogeneous in terms of intervention type, social need(s) addressed, and outcomes reported. Thus, we did not perform a meta-analysis or attempt to make formalized comparisons across studies. Instead, in this narrative section, we summarize the results by type of intervention and outcome, which we report fully in Tables 1 and 2. Study quality based on GRADE criteria is also reported in Table 1. In some studies, a specific referral process to outside organizations was outlined as part of the intervention. Other interventions were carried out in partnership with community organizations. We describe these fully in Table 3.
Nine studies in this review were categorized as screening interventions. Four of these studies addressed solely IPV, one addressed IPV and social support, one addressed solely social support, and one addressed workplace exposures. The interventions in Harrison et al. and Price et al., the multi-need screening interventions mentioned above, were also categorized in this group. In terms of outcomes, most studies reported process measures such as patient-reported helpfulness, ease of use, or internal validation of the screening tool. Price et al. compared the results of their 10-item brief risk factor survey that they administered in the waiting room to private interviews and found that their screener concurred with the private interviews in identifying social needs. Zachek et al. also issued their screening tool as a waiting room survey and found that closed-ended questions identified more workplace exposures than open-ended questions did. Patients who screened positive for workplace exposures were then referred to an occupational and environmental medicine clinic for follow-up. The only other screening intervention that utilized a referral process was that in Harrison et al., which used an algorithmic approach with a risk score to refer patients to specialized services based on their reported risk severity. No screening studies reported health outcomes, health behavior outcomes, or measures of social risks. Overall, patients found screening interventions acceptable, helpful, and felt comfortable discussing social needs with clinicians. 32,41,47 We defined brief interventions as screening with brief education or advice related to the social needs disclosed. We considered the web-based gestational diabetes education tool to be a brief intervention for health literacy. 36 Of the six studies describing brief interventions, five addressed IPV and one addressed health literacy. The brief interventions for IPV varied in approach and outcomes. One intervention consisted of a tablet-based screening tool with motivational interviewing prompts that were individually generated based on reported risks. 48 Women interviewed in this study reported feeling that they could better adhere to the treatment guidelines for IPV after using the tablet-based screener. Another intervention was a computer-based screener with a 15-min informational video and prompting card that was given to the provider. 34 Women in this   32,34,38,43,45,48 Length of administration 30,45,49,51 Referral/patient use of referrals 33,35,45,46,53,54 Screening rate 54 Appointment adherence 45 Health knowledge 31,36 Social support 52,53 ETSE 25,26 IPV (episodes of violence, severity, behaviors) 29,11,35,[37][38][39]55 Physical activity 52,53 Healthy diet 52 Substance use (smoking, alcohol use) 25,52,53 Help-seeking behavior 38,43 Safety planning 54 Infant birthweight 26,27,29 Gestational age at birth 26,27,29 Miscarriage, perinatal death, nonlive birth, neonatal hospitalization days 26 Maternal weight gain 52 Infant development 52 Mental health status (depression, PTSD, stress, selfesteem) 25,33,38,52,53 Pain and physical functioning 37 Breastfeeding 52 ETSE, environmental tobacco smoke exposure; IPV, intimate partner violence. intervention group had significantly more discussions about IPV with their provider compared with women receiving usual care. The intervention in Price et al. utilized a screening intake form plus brief IPV information and guided referral. 46 Uptake of referral in this study was low and most women reported being uninterested in specialized services. The investigators found the biggest predictor of successful referral to be women's level of engagement during screening. Two studies addressed IPV using a screening questionnaire combined if a referral card with resources and safety tips. Results showed that the referral cards did not decrease violence scores, IPV victimization, or symptoms of depression, and did not improve help seeking behaviors. 35,38 The webbased health literacy intervention did improve women's knowledge of gestational diabetes topics compared with women in the control group. 36 Counseling interventions consisted of interventions where one or more counseling sessions were provided in concert with clinical prenatal care. Six studies on counseling interventions were reported in 11 publications. All counseling interventions addressed IPV, with one also addressing health literacy and one also addressing tobacco smoke exposure. 31 Six publications reported on the DC-HOPE study, which was an RCT that tested a screening and counseling intervention for IPV, ETSE, depression, and smoking. Screening took place at the patients' first prenatal care appointment and counseling took place over four to eight prenatal care visits. Counseling sessions were individually tailored to patients' reported risks. The studies reported a wide range of outcomes, including reduction in ETSE, recurrent episodes of IPV victimization, cigarette smoking, infant birthweight, gestational age at birth, miscarriage, and symptoms of maternal depression. Women were recruited from six prenatal care clinics serving primarily minority women and all women included in the study identified as black or African American. The study found that the intervention group had significantly fewer very preterm births ( < 34 weeks) (odds ratio [OR] = 0.22, 95% confidence interval [CI] = 0.07-0.68) and very low birthweight babies ( < 1500 g) (OR = 0.11, 95% CI = 0.01-0.86) compared with control. 26,27,29 There were no significant differences in preterm births ( < 37 weeks), low birthweight ( < 2500 g), miscarriage, perinatal death, nonlive birth, or neonatal hospitalization days. Among women initially reporting ETSE, women in the intervention group were significantly less likely to report ETSE before delivery when controlling for other covariates, including other psychobehavioral risks (OR = 0.50, 95% CI = 0.35-0.71). 26 Among women initially reporting IPV, intervention group women were less likely to have recurrent episodes of IPV victimization (OR = 0.48, 95% CI = 0.29-0.80). 29 The study by Backonja et al. was adjunct to the DC-HOPE study and incorporated reproductive health education into four to eight counseling sessions. Women receiving the health education scored higher on reproductive knowledge assessments compared with women receiving usual care. Aside from the DC-HOPE studies, there were three other counseling interventions for IPV. 37,39,55 All consisted of empowerment counseling sessions during the medical visit and all except for the intervention in Zlotnick et al. showed a significant reduction in measures of violence for women in the intervention groups. 35,37,39,55 While the intervention in Zlotnick et al. did not show a significant reduction in IPV, they did find a moderate reduction in symptoms of depression and PTSD during pregnancy.
Interventions that provided women with regular support outside of the clinic were categorized as interventions with ongoing support. There were seven of these interventions in total, with four for IPV, 33,35,43,54 one for social support, 53 and two for food insecurity. 45 In this study, pregnant teenagers were given the support of a nurse midwife who assisted them in connecting with various community services. While results showed that teenagers in the intervention group had increased uptake of certain resources, there were no differences in measures of social support, self-esteem, physical activity, or smoking behavior at 4 weeks postpartum. There were also no differences in uptake of National Health Service services between the intervention and control groups.
We grouped the food insecurity interventions with ongoing support interventions; however, they were structured differently than the interventions for IPV and social support. The interventions in both Trapl et al. and Watt et al. consisted of nutritional counseling given at regular clinic visits. They both partnered with local farmer's markets to issue food vouchers, which were also distributed at regular clinic visits. The intervention in Watt et al. also included a 1.5-h nutrition and cooking class that was taught in Spanish and lactation counseling that was offered at the clinic and over the phone. The interventions in Trapl et al. and Watt et al. did not include ongoing mentoring or case management. Trapl et al. reported appointment adherence and patient use of referrals. Providers felt that the intervention increased appointment adherence, and objectively, 43% of patients kept 100% of appointments and another 31% of patients kept 50-99% of appointments. Using redemption logs from the farmer's market, the investigators found that 56% of participants redeemed at least one voucher. Watt et al. reported a wider range of outcomes. Women in the intervention group had significant improvements in diet, exercise, and depression ( p £ 0.05) compared with women receiving usual care. In addition, women were more likely to breastfeed ( p = 0.07) and their infants were more likely to pass the ages and stages developmental screen ( p = 0.06) than women in the usual care group.

Discussion
The 26 studies summarized in this review provide an overview of all clinic-based social needs interventions for pregnant and postpartum women that have been reported in the literature to date. They also provide a summary of the evidence for such interventions in improving maternal and infant health outcomes and mitigating social risks and behaviors that have been associated with poor health outcomes.
We choose to discuss interventions for IPV separate from those addressing other social needs domains as IPV interventions represented the majority of interventions identified in our review.
Among interventions addressing IPV, the evidence in our review points to counseling interventions and interventions with ongoing support as the most ef-fective interventions for improving health and healthrelated outcomes. Overall, most of these studies found significant reductions in recurrent IPV 28,55 and severity of IPV 35 or significant improvements in healthy behaviors such as safety planning in Kramer et al. Two studies reported objective health outcomes; the DC-HOPE studies showed a significant reduction in very preterm births and very low birthweight infants and Zlotnick et al. demonstrated a moderate effect of their intervention in reducing symptoms of depression during pregnancy. On the contrary, no screening or brief intervention for IPV showed a reduction in IPV rates and evidence was mixed for increasing safety planning rates. 34,37,54 These findings are consistent with those of the US Preventive Service Task Force (USPSTF), 56 which concluded that interventions providing women with ongoing support have moderate benefit, but brief interventions in the absence of ongoing services are not shown to have benefit. Of note, in both our review and that of the USPSTF, the DC-HOPE study was the only study reporting birth outcomes, illustrating the paucity of evidence for such interventions.
Interventions addressing social needs other than IPV were few. Among these, the type of intervention did not correspond to effectiveness of intervention as seen among interventions for IPV. For example, both brief and counseling interventions for health literacy were effective at improving health knowledge. 31,36 Meanwhile, an ongoing support intervention for social support did not result in women reporting increased social support. 53 Finally, the counseling intervention for ETSE and the ongoing support interventions for food insecurity resulted in a reduction in tobacco smoke exposure in postpartum women and higher rates of appropriate development among infants, respectively. 25,52 These results suggest that different social needs may be best addressed by different types of interventions, unique to the need being addressed. Overall, the evidence for social needs interventions in perinatal care beyond those for IPV is lacking and more research is needed to understand the effectiveness of such interventions.
We were surprised by how few interventions for food insecurity were identified in our review given that federal food assistance programs such as the Special Supplemental Nutrition Program for WIC are widely used by pregnant women, with nearly 4 in 10 expectant mothers enrolled in the program. 57 Numerous studies have concluded that participation in WIC is associated with improved birth outcomes such as higher birthweight, lower likelihood of neonatal intensive care unit admission, lower likelihood of preterm birth, and lower odds of 1-year infant mortality. [58][59][60][61][62] While several studies mention referring women to food assistance programs, only one specifically mentions WIC. Designing interventions that facilitate enrollment in WIC and implementing them in clinical care settings may be one approach to addressing food insecurity among pregnant and postpartum women, particularly those who are low income. Given that there are numerous governmental and community programs that address other social needs such as transportation and housing, designing interventions that leverage these existing programs might facilitate the study of clinic-based social needs interventions in other domains.
It is worth noting that no study in our review addressed refugee status or racism. Several studies in the immigrant health literature have shown that Asian and African refugees experience higher rates of pregnancy complications compared with women in the local population. 63,64 Although not unique to immigrant health, the literature discusses how immigrant health is not negatively impacted by one or two factors, but instead by a multitude of inequities, including dealing with day-to-day discrimination and racism.
Our review has several limitations. First, the scope was broad and captured a range of studies describing heterogenous interventions and outcomes, which prevented comparisons across studies. However, this breadth was intentional and necessary to examine such a broad subject matter. Second, while populations were diverse, the highest quality studies in this review were in African American populations and results may not generalize to other populations. In addition, very few Native American/indigenous women were included in the studies in our review, which is concerning given that indigenous women experience similarly elevated maternal health disparities. 16,65 Our review also does not account for nontraditional care settings or alternative models of prenatal care. These alternative models may be of benefit to women who desire to carry out their pregnancies outside the traditional health care system, for example, on ancestral lands, which often have less access to specialty care. 66 Finally, we limited our review to peer-reviewed publications and health care settings in high-income countries. We may have excluded gray literature that could have implications for this rapidly growing area of research.

Conclusion
This review provides an overview of all published interventions for social needs in perinatal care that have been reported in the literature to date. The results of this review suggest that most interventions have been for IPV, and among these, interventions with counseling or ongoing support show the most promise in alleviating social risk factors and improving some clinical outcomes. Few interventions address social needs other than IPV. More rigorous research should be conducted that expands beyond IPV to other social needs domains. We find it worth noting that while some social needs interventions may be beneficial, they do not represent long-term solutions to the underlying social conditions that are the root of many health disparities, including those in maternal and child health.