NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Slaughter-Acey J, Behrens K, Claussen AM, et al. Social and Structural Determinants of Maternal Morbidity and Mortality: An Evidence Map [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2023 Dec. (Comparative Effectiveness Review, No. 264.)
Social and Structural Determinants of Maternal Morbidity and Mortality: An Evidence Map [Internet].
Show details6.1. Overview
This review sought to provide a broad overview of research that examined exposures related to social-structural determinants of health and at least one health or healthcare-related outcome affecting postpartum health. Strengths of our report include a comprehensive search and inclusion of observational studies most relevant to the topic, high-level mapping of the research on social-structural determinants and outcome domains identified from the studies, and suggestions for new research based on our findings.
Our review identified 118 studies categorized to eight outcome domains and 11 domains related to social-structural determinants of health representing 221 specifically named exposures of interest. Identified domains of social-structural determinants of health included race/ethnicity, socioeconomic, violence, trauma, psychological stress, structural/institutional, rural/urban, environment, comorbidities, hospital, and healthcare use factors. Broad outcome domains included maternal mortality, severe maternal morbidity, hypertensive disorders of pregnancy, gestational diabetes, cardio/metabolic disorders, weathering (the physiological effect of premature aging caused by chronic stressful experiences), depression, other mental health or substance use disorders, and cost/healthcare use outcomes. A large proportion of studies examined depression and other mental health outcomes for both pregnant and birthing people, even compared with mortality and other severe maternal morbidity outcomes.
Overall, study exposures broadly covered the social-structural determinants of health for both pregnant and birthing people; however, these included exposures represent only a subset of potential social-structural determinants that may affect the health and care of pregnant and birthing people. Further, no studies examined interdependencies with biologic/medical risk factors. Limited depth and quality of available research within each risk factor domain impeded our ability to understand the pathways connecting social-structural determinants of health and maternal health outcomes. We found an unexpectedly large volume of research in the domain of violence and trauma relative to other domains of social-structural determinants of health for pregnant people. Likely this stems in part from the addition of violence-related questions in the Centers for Disease Control and Prevention’s Pregnancy Risk Assessment Monitoring System (PRAMS). This system, an ongoing collaboration between state, territory, and local departments of health and the Center for Disease Control and Prevention, is a “population-based surveillance system designed to identify groups of women and infants at high risk for health problems, to monitor changes in health status, and to measure progress towards goals in improving the health of mothers and infants.”182
Across all domains related to social-structural determinants of health, an overwhelmingly large number of studies used correlational study designs to describe associations between social-structural determinants of health and outcomes. Fewer studies used analytic approaches that would allow one to try to untangle the causal relationship, such as experimental designs or quasi-experimental designs or analytic methods. Experimental designs were lacking because randomization is difficult if not inappropriate to conduct, since randomizing pregnant and birthing people to different levels of a social-structural determinant of health is unethical. Perhaps not surprisingly, we found all studies to be at high risk for alternative explanations. Therefore, we approached study-reported results from the perspective of supporting future researchers in generating hypotheses for risk factors to test with potential interventions. Only a handful of studies used analytic methods to explore cause-and-effect relationships using approaches such as propensity score methods, difference-in-difference, and instrumental variable methods. Even fewer studies attempted to break differences in a particular outcome into separate risk components for pregnant and birthing people experiencing different levels of a social-structural determinant of health. We note that studies that were more likely to use analytic methods that would allow reporting the excess risk attributable to a specific exposure are mostly from the past three years, and our findings point to the need for more of this. This increased rigor would bolster an evidence base that helps us understand the potential mechanisms through which social determinants of health—including racism— work, and thereby design effective interventions.
6.2. Future Research
Identifying the social-structural determinants of health that affect pregnant and birthing people is of vital importance. Not only do pregnant and birthing people face an unacceptably high risk of maternal morbidity and mortality in the United States,1 but that risk is unevenly distributed, with Black and Indigenous women three to four times as affected as their white counterparts.9 While each pregnant or birthing person will confront their own unique risk factors, individuals can benefit when research identifies themes and patterns at the population level that suggest opportunities to deliver interventions that address the impact of social and structural determinants of health, not just social needs. Our review overall identified a great number of potentially eligible studies. However, even after narrowing the included literature to only the studies better designed to address our Key Questions, we remain unable to draw strong conclusions due to the study design, conduct, and dispersion reasons stated above. Deeper investigation of an individual risk factor and its mechanisms would require more study designs than we included here. Such a mixed studies review would be best approached through targeted reviews of specific scope. And while the literature published in the last three years showed a definite trend toward improved rigor, much remains to be addressed. In concert with standards recently suggested for publishing research on racial health inequities,44 we outline below several future research areas that could inform research, practice, and policy.
In addition, we noted in this literature a widespread “deficit” perspective. That is, researchers often describe disparities and adverse outcomes as expected in individuals and populations who experience structural vulnerability and violence. Future research and approaches to addressing maternal morbidity and mortality would benefit from a shift towards a “strengths-based” approach, wherein researchers intentionally explore what exposures might be health-enhancing and health-promoting even in the face of structural vulnerability. Such a strengths-based approach supports thinking expansively about how to reduce barriers and achieve optimal perinatal outcomes.
6.2.1. Methodological Rigor
As noted, the overwhelming majority of included studies were designed to answer whether two or more things were associated with each other rather than whether one thing could cause another. Because we grouped studies according to their stated purpose and approach, we constrained our review to a likewise hypothesis-generating approach. Indeed, it would be unethical to use traditional gold-standard experimental methodologies that would randomly assign study participants to factors that could harm their health. A few recent studies used quasi-experimental approaches to address the selection bias problems of nonrandomized studies, such as instrumental variables, propensity score matching, or decomposition analysis, to try to identify important drivers of poor maternal health outcomes and maternal health disparities. Future research can emphasize such techniques that improve the ability of observational research to estimate causal impacts.
Assessing methodological rigor is a challenging task. We were struck by a lack of adherence to basic reporting standards for observational studies within these publications. Incompletely reported details, in particular about analytic approaches, further hindered our ability to assess study rigor. It would be easier to assess the rigor of exposures studies if we knew the ideal study design as the standard against which to measure the conduct of studies included in the review. To design studies ideally suited to produce reliable results in this field would require understanding the critical co-exposures and confounders to include in the analysis. Confounders would likely be things along the mediating pathway, and incorporating such things into analyses is difficult to do without inserting bias, especially with incomplete understanding of the mediating pathway. In addition, each study purpose, from the broad array present in the included studies, could require its own enumeration of co-exposures and confounders. Organized and curated catalogues of maternal health exposures and their presumed mechanisms could facilitate future examinations of exposures. More widespread adoption of these approaches could improve rigor of the conduct and reporting of future maternal health exposure studies, increasing the overall quality of the literature.
6.2.2. Populations and Data Sources
During the topic refinement phase of this review, stakeholders very much wanted an inclusive approach to pregnant and birthing populations. A few notable studies focused on specific populations of concern, such as disaggregating social identities within broader racial and ethnicity categories,173 or groups situated at intersections of social-structural determinants of health, such as race/ethnicity and rurality for Indigenous pregnant women.177 However, the majority of included studies were constrained by the available demographic labels used in established datasets. In addition, many studies used enrollment or inclusion criteria that by design excluded the most vulnerable populations, resulting in under-representation of groups such as women who were HIV positive or incarcerated. Indigenous women continue to be grouped in “other” categories in studies because of “small numbers,” rendering them invisible in the literature. Similarly, we did not identify studies that examined the risk factors specifically related to trans- and gender diverse populations and their experiences of pregnancy and birth. Considerable opportunities exist for supporting research infrastructure that ensures these groups are accurately accounted for in future studies.
In addition to the inexact demographics, the data sources themselves can be a source of bias. These structural research resources were generally created under structurally biased conditions, raising concerns for where and how the population sampling was done, as well as the choice of individual, family, community, and social-structural constructs included. The selection and measurement error in the datasets contribute to the under-specification of “disparities” and exploration of causal mechanism through which social-structural drivers of maternal health work. In the near term, research programs and publishing guidelines can encourage analytic approaches discussed earlier, such as instrumental variable or propensity score methods, to try to address the bias inherent in the conditions under which the data was collected. Researchers can name the form of racism being examined, the mechanisms by which it may work, and other intersecting factors that may compound its effect.44 Future research programs can also take up longer-term solutions and create datasets designed to more fully capture the data needed to robustly examine racism and other social determinants of health.
6.2.3. Exposures
While we cannot discuss in detail the wide array of exposures our review covered, we note a few of particular interest. We identified few studies that attempted to measure reported racism or racial discrimination as an exposure. No included study used a measure for intersectionality or approached exposure research from a position of intersectionality. In fact, a recent systematic review of intersectionality in quantitative research noted that researchers need to better understand key features that define quantitative intersectionality analyses.183 Future research would benefit from incorporating approaches and measurement tools explicit to racism and intersectionality. One approach of particular interest is the recent development of a measurement tool that captures the multidimensionality of structural racism.184 Other examples of measurement tools do exist.185, 186
Closer to healthcare delivery, much remains to be understood about how aspects of healthcare delivery contribute to health disparities. Limited work has examined this relationship, and the majority of reported findings that noted attributable risk examined healthcare delivery-related risk factors. For example, no study examined continuity of care or access to prenatal care provided from sources beyond obstetrics as risk or protective factors for maternal outcomes. Rural locations received some attention, but considerable work remains to be done to understand the underlying drivers, such as distance to prenatal or specialized care, delivery centers (or transitions to final delivery location), or deliveries at home or enroute to a delivery center. Because rural health remains a resource-challenged issue, this may also be an area where collaborations to improve data collection may be vital.
6.2.4. Outcomes
Considerable opportunities exist for future research to improve the outcomes measured and captured. We were dismayed to note the amount of research excluded because it captured neonatal but not maternal outcomes. The Centers for Disease Control and Prevention created the Severe Maternal Morbidity measure to track changes over time in the immediate perinatal period that might contribute to maternal mortality. However, the ubiquity of the measure in research can focus attention away from the common postpartum challenges and outcomes most important to birthing people across a wider time period. Our review identified eight outcome domains, including depression and other mental health concerns such as anxiety, which overlap with some of the most common postpartum challenges,187 but the literature did not examine all of these challenges.
Another theme (outside the scope of our review) that arose during topic refinement was an eagerness among key informants and content experts to press beyond maternal or infant mortality and continue research on downstream effects on maternal and infant/young children’s health problems.
Similar to concerns about dataset impact on populations available to study, longer-term solutions will require datasets designed to more fully capture the outcome data needed. Increasing the availability of longitudinally linked datasets is vital. We find examples in the datasets being supported by the Office of the Secretary’s Patient-Centered Outcomes Research Trust Fund (OS-PCORTF) in the Office of the Assistant Secretary for Planning and Evaluation at the Department of Health and Human Services.188, 189 Another needed advance would involve improving the ability to link and use electronic health records to enrich the available clinic-based variables and link parent and infant records over time. Community-relevant variables and outcomes likely require expansion of data collection beyond the medical encounter; planning new survey-based data designed to improve the ability to link the data to other existing or planned datasets.
6.2.5. Other Research Approaches
Qualitative research fell outside the scope of this review, but our screening process suggested it would be valuable to explore this subset of the literature. Sophisticated analytic approaches can help researchers investigate how segregation, as well as location relative to neighborhood and environmental exposures, affect access to care.190 Qualitative research provides rich data based on listening to the experiences of birthing people. In one example drawn from our screening process, qualitative researchers explored the experiences that pregnant and birthing women of color had while interacting with healthcare providers.191 Supporting this, stakeholders during the topic refinement and protocol development phases noted the importance of qualitative research as counter-argument to traditional philosophy of science approaches and systematic review methodology. They noted that prioritizing only quantitative research that investigates causation hinders and devalues the ability to move a research field forward through hypothesis generating activities. Ultimately, all inquiry begins with direct observation and curiosity, which form the foundation of “good science.”
6.3. Strengths and Limitations of the Review
The methods we selected for this review provided a detailed map of the research connecting racism and social determinants of health exposures to maternal health and morbidity. We purposefully focused on risk factors that operated interpersonally to capture literature most likely to address this intersection. Such high-level mapping will help provide researchers, who are often still siloed in their particular areas of expertise or interest, a wider perspective on the breadth of literature within which their specific practice and advocacy resides.
Our inclusion criteria required studies to examine the impact of a social determinant of health. As such, many studies that only examined comorbidities and medical risk factors were ultimately excluded. Most of these excluded studies not only used patient demographics as control or confounder variables, but also lacked exposures indicative of social determinants of health. This review does not address the large literature exploring many biomedical conditions as risk factors for maternal health. Even more regrettably, this siloed approach to risk factor research meant that the interdependencies, intersections, and feedback loops that can compound risks remain generally unaddressed.
Because of our wide scope, we focused on quantitative epidemiologic studies and similar research. We cannot escape the possibility of publication bias in our review. Not only would papers with statistically significant results be viewed as interesting to publish, but also, registering a protocol prior to conducting a secondary analysis of a dataset remains an uncommon practice. The included studies did not fit cleanly into discrete groups, which required us to categorize exposures subjectively. Likewise, the extreme heterogeneity of the studies in exposures and designs led to a subjective risk of bias assessment; however, we tested our approach by identifying the most rigorous study designs and analytic approaches for deeper assessments in order to confirm that subjecting the full literature set to formal assessment lacked value. Further, the included studies only addressed observed pregnancies.
6.4. Conclusion
Identifying the risk factors faced by pregnant and birthing people is vitally important. Limited depth and quality of available research within each social-structural determinant of health impeded our ability to understand underlying risk pathways. While the most recently published literature showed a definite trend toward improved rigor, future research can emphasize techniques that improve the ability to estimate causal impacts. Improved reporting in studies, along with organized and curated catalogues of maternal health exposures and their recognized mechanisms, could make it easier to examine exposures in the future. In the longer term, future research needs datasets designed to more fully capture the data required to robustly examine racism and other social-structural determinants of health and their intersections with other biologic/medical risk factors.
- Discussion - Social and Structural Determinants of Maternal Morbidity and Mortal...Discussion - Social and Structural Determinants of Maternal Morbidity and Mortality: An Evidence Map
- List of Excluded Studies - Treatment Strategies for Women With Coronary Artery D...List of Excluded Studies - Treatment Strategies for Women With Coronary Artery Disease
- Mus musculus BAC clone RP23-3L20 from 3, complete sequenceMus musculus BAC clone RP23-3L20 from 3, complete sequencegi|19909481|gb|AC098732.3||gnl|wugs 3-3L20Nucleotide
- Discussion - Management of Infantile EpilepsiesDiscussion - Management of Infantile Epilepsies
- Discussion - Safety of Vaccines Used for Routine Immunization in the United Stat...Discussion - Safety of Vaccines Used for Routine Immunization in the United States: An Update
Your browsing activity is empty.
Activity recording is turned off.
See more...