The role of neighborhood on preterm birth among a high-risk group of birthing people in Boston, MA

Abstract

Background

Preterm birth (PTB) is associated with adverse short- and long-term health. There are known racial, ethnic, and socioeconomic inequities in PTB. Because of historical de jure and modern-day de facto segregation and neighborhood divestment, neighborhoods are a source of structural racism and disenfranchisement and a potential target for policies and interventions to reduce PTB inequity. However, the role of neighborhoods on PTB, specifically among high-risk birthing people, is largely unexplored.

Methods

The Boston Birth Cohort is a longitudinal birth cohort of birthing people-infant dyads at a safety-net hospital in Boston, MA between 2000 and 2018. The primary outcome was preterm birth at 35 weeks or prior. The primary predictor was neighborhood defined by census tract. We used generalized linear mixed effects models to test our hypothesis that neighborhood accounts for a signficiant proportion of PTB risk among socially at-risk birthing people.

Results

In multilevel models, neighborhoods were a significant predictor of preterm birth, yet accounted for only 3% of the variability in outcome. In models stratified by race, individual-level factors such as prior preterm birth, nativity status, and advanced birthing person age were significant predictors of PTB.

Conclusions

Neighborhood is a significant, though small, predictor of preterm birth in a high-risk birthing population. These findings suggest that individual-level interventions, rather than neighborhood-level policies, may be more effective in reducing preterm birth among high-risk birthing populations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-024-06957-3.

Background

Preterm birth (PTB), birth before 37 weeks gestation, has a significant impact on health at each stage of development across the life-course [1]. Preterm infants often require long-term hospitalizations in the Neonatal Intensive Care Unit (NICU) due to physiologic immaturity and can have associated pulmonary, neurologic, gastrointestinal, and cardiovascular complications in the newborn period. This care is associated with significant material and emotional stress on families as well as healthcare costs [2–4]. For example, the care of low-birth-weight infants accounts for 35% of the total health care costs for all infants, and preterm infants incur medical expenditures between approximately 80–600 thousand dollars in the first six months of life [5, 6]. Following their discharge from the NICU, many preterm infants require additional screening and medical, behavioral, and educational support services into adulthood [1, 7–11].

The incidence of PTB is not equal across populations. Non-Hispanic/Latinx Black (NHB) and Hispanic/Latinx birthing people have a significantly increased likelihood of having a preterm infant compared to non-Hispanic/Latinx white (NHW) birthing people [12, 13]. Additionally, individuals in lower socioeconomic strata have a significantly increased risk of PTB compared to birthing people in higher socioeconomic strata [14, 15]. Despite medical advances, the PTB rate has steadily been rising over the past decade, and despite increased awareness and interventions designed to the address inequities in PTB, the racial inequity gap has also been increasing [12, 13]. Thus, PTB, and the associated financial, emotional, and health costs disproportionately impact minoritized individuals and those in lower socioeconomic strata at birth and longitudinally across the life-course. In this way inequities in PTB may further perpetuate, and even exacerbate, inequities across the life-course [16].

Because individual-level interventions have proven ineffective at closing the equity gap in health, identifying population-level targets for health promoting interventions is paramount [17, 18]. Research has identified place-based drivers of PTB, such as residential segregation, neighborhood deprivation, and historical redlining [19–25]. However, results have not been consistent across the literature and effect estimates have varied between racial and ethnic groups.

High-quality neighborhoods, those variably defined as having high opportunity and resources, are associated with improved physical and mental health [26]. However, high-quality neighborhoods are not equally accessible to all individuals, in particular racial and ethnic minoritized individuals and those in the lowest socioeconomic strata, as a result of neighborhood segregation and divestment [26, 27]. Neighborhoods are also modifiable and thus are potential targets for population-level interventions. Studies focused on PTB inequity have suggested that health policy interventions geared at neighborhoods may be effective to reduce the risk of PTB among those at highest risk—minoritized and socioeconomically disenfranchised birthing people [28–31]. Importantly, most of the research examining the impact of neighborhood on PTB has not specifically focused on the highest risk birthing population, racial and ethnic minoritized individuals and individuals of low socioeconomic status. To reduce PTB inequities, drivers of PTB must be specifically examined among those at highest risk.

Boston Medical Center (BMC) is the largest safety-net hospital in New England and serves a predominantly low-income, minoritized, and immigrant population. Safety-net hospitals provide care to a large proportion of poor, underserved, and minoritized communities and this has not changed even after Medicaid expansion under the Affordable Care Act [32, 33]. Thus, safety-net hospitals are a unique setting to study qualities of high-risk populations. We therefore set out to investigate drivers of PTB among the high-risk birthing population. In contrast to most previous studies focusing on either individual or neighborhood factors, the Boston Birth Cohort, with both neighborhoods and individual-level risk factors, provides a unique opportunity to understand the role of neighborhood-based drivers of PTB, accounting for individual-level factors among high-risk birthing people. We hypothesized that neighborhoods account for a significant proportion of the risk of PTB among high-risk birthing people and that neighborhoods have a differential impact on the risk of PTB for white compared to non-white birthing people.

Methods

Study design and population

We conducted a secondary analysis of a large longitudinal birth cohort to determine the role of neighborhood on the risk of PTB. The study population was derived from the Boston Birth Cohort (BBC), a well characterized, prospective birth cohort study of birthing people who delivered at Boston Medical Center, a large safety-net hospital in urban Boston, Massachusetts. For this study, we included individuals who delivered a baby between 1 January 2000 and 31 March 2018. The BBC enrolled parent-infant dyads at birth and longitudinally followed those individuals who continued to receive healthcare at BMC. A description of BBC recruitment and follow up has been previously described [34, 35]. Briefly, birthing people were eligible for participation after delivering a singleton live birth at BMC. Pregnancies resulting from in vitro fertilization, multiple-gestation pregnancies, deliveries induced by maternal trauma, or newborns with significant congenital anomalies were excluded. At birth, baseline sociodemographic, clinical, and biomarker data were collected for both the birthing person and child. Race and ethnicity was self-reported. Because of our interest in the impact of neighborhood on PTB, as well as within- and between-racial and ethnic place-based drivers in PTB, we excluded individuals with missing race or ethnicity data or those with race or ethnicity listed as “other.” Given all participants are socially high-risk birthing people seeking care at a safety-net hospital, it was necessary to ensure that there was adequate variability in the resources and qualities of the neighborhoods. To do this, we examined the state-normed Child Opportunity Index (COI) 2.0, a multi-dimensional census tract-level measure of neighborhood-based opportunities and resources conducive to health child development, to assess adequate variability in the neighborhoods represented [36, 37].

This study was approved by the Institutional Review Board (IRB) at BMC. The IRB at Beth Israel Deaconess Medical Center determined that use of data from the BBC for secondary analysis was not human subjects’ research.

Exposure and outcome

Our primary exposure was neighborhood, defined by the census tract at time of delivery. As part of enrollment procedures in the BBC, address was recorded during the index hospitalization. Street address was then used to identify the U.S. Census Bureau census tract. Census tracts are small generally contiguous areas with between 1,200 and 8,000 people and are often used in research to approximate a neighborhood [38].

Our primary outcome was preterm birth defined as birth before 35 weeks gestation. While infants born at a gestational age less than 37 weeks are considered preterm, infants born prior to 35 weeks present a unique population. At BMC, infants born at prior to 35 weeks are automatically admitted to the NICU thus representing increased healthcare costs [4]. Furthermore, NICU admission interrupts infant-parent bonding which impacts short- and long-term health [2, 3]. Finally, compared to infants born prior to 37 weeks but after 35 weeks, this patient population is at an increased risk of prematurity-associated medical morbidity [39–43].

Variables

We selected a set of control variables that have been previously reported in the literature to be associated with PTB [13, 44]. These included: highest achieved birthing person education (none or elementary school; high school or GED; college or above), advanced maternal year of age at delivery (age greater than or equal to 35 years), early prenatal care (at least one visit to prenatal care in the first trimester), and history of prior PTB (birth prior to 37 weeks completed gestation).

Additionally, we included categorical maternal self-reported race and ethnicity in the following categories: non-Hispanic white (reference), non-Hispanic Black, and Hispanic. We use the term Hispanic, and not Latina/o/x/e, as per the categorizations initially defined by the BBC survey procedures. All other race and ethnicity categories were excluded. Race-specific stratified models allowed us to explore the differential role of neighborhood on racial and ethnic populations.

We recognize that race is a social construct reflecting hierarchies of power. Racial and ethnic inequities in health are not reflective of biology, genetics, or individual behavior. Rather, these differences represent the effects of racism [45, 46].

Statistical approach

We performed bivariate analyses using Student’s t-tests for continuous variables and Chi-square tests for categorical variables between all individual-level variables and the primary outcome. We derived the list of variables for our multivariable model from bivariate analyses with threshold for inclusion of p < 0.10.

The data were structured in two levels, with individual birthing people at level one and neighborhoods, defined by census tract, at level two. The clustering of individuals within their neighborhoods informed the analytic approach, generalized linear mixed effects modeling, affording the opportunity to quantify the relative contributions of each level to the outcome.

We first created a null model with random neighborhood-level intercepts using census tracts to calculate the interclass correlation coefficient (ICC). The ICC is a ratio of the between-neighborhood variance to the total variance and is used to determine the proportion of the total variance attributable to neighborhood only. We then added race and ethnicity to the model to estimate the effect of race on preterm birth. We recognize that the race is a proxy for racism and that differences in health by race or ethnicity category do not reflect biologic or genetic risks, but instead highlight the effect of racism—structural, institutional, interpersonal, and internalized—on preterm birth. Because we were interested in inequity in preterm birth compared to white birthing people, we coded white as the reference category. We then included all individual-level variables from the candidate list. To explore the differential impact of factors on racial and ethnic populations, race-specific stratified models were estimated.

All statistical analyses were performed using SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

During the study period from 1 January 2000 to 31 March 2018, there were a total of 7,980 birthing person-infant dyads enrolled in the BBC, 7,345 that met inclusion criteria.

There were 768 census tracts represented among the 7,345 birthing people included in the study. Among birthing parents, 4,120 (56%) were non-Hispanic Black (NHB), 925 (13%) were non-Hispanic white (NHW), 2,300 (31%) were Hispanic, and 635 were excluded for having a race defined as Asian, Pacific Islander, Mixed Race, or Other. Among participants included in the study, 989 (13.5%) gave birth at less than 35 weeks GA. Most birthing people received early prenatal care (80.6%, defined as at least one prenatal care visit in the first trimester of pregnancy), had a high school or GED level education (54.2%), and were born outside the U.S. (60.8%). In our sample, 3,136 (43.1%) were nulliparous, and 4,180 (56.9%) had at least one prior live birth and nearly 10% of birthing people had a prior preterm birth (Table 1). 2% of individuals lived in Very High COI neighborhoods, 4% lived in High Opportunity Neighborhoods, 7.5% lived in Moderate COI neighborhoods, 24.4% lived in Low and 62.3% lived in Very Low COI neighborhoods (Supplemental Table 1).

Table 1.

Boston Birth Cohort demographics

		All birthing people (N = 7345)		Non-Hispanic white birthing people (N = 925)		Non-Hispanic Black birthing people (N = 4120)		Hispanic birthing people (N = 2300)
		N	%	N	%	N	%	N	%
Advanced birthing age		1217	16.6	143	15.5	789	19.2	285	12.4
Prior preterm birth		623	10.1	91	10.7	369	10.7	163	8.7
Early prenatal care		4871	80.6	629	78.8	2611	78.2	1631	85.4
Birthing person education
College graduate or above		2387	32.7	212	23	996	24.3	1215	53.2
High school graduate		3953	54.2	532	57.8	2487	60.8	934	40.9
Less than high school completion		956	13.1	176	19.1	609	14.88	135	5.9
Birthing person birth in the U.S.		2856	39.2	757	82.6	1677	41.1	422	18.4
Preterm birth		989	13.5	123	13.3	628	15.3	238	10.4

Advanced birthing age: age greater than or equal to 35years old

Prior preterm birth: having a prior live birth born at gestational less than 37 weeks

Early prenatal care: having at least one prenatal care visit in the first trimester

Bivariate analyses

The preterm birth rate (< 35 weeks) was 15.3% for non-Hispanic Black birthing people, 13.3% for white birthing people, and 10.4% for Hispanic birthing people (p < 0.001). Compared to infants born after 35 weeks, infants born prior to 35 weeks gestation were more likely to be born to birthing people who were born in the United States (46.6% v. 38.1% p < 0.001), with advanced maternal age (21.2% v. 15.8%, p < 0.001), who had a prior preterm birth (21.3% v. 8.4%, p < 0.001), and who did not have early prenatal care (78.5% v. 80.9%, p < 0.001). (Table 2)

Table 2.

Bivariate analysis of birthing people factors associated with preterm birth

		Term birth (N = 6356)		Preterm births (N = 989)
		N	%	N	%	p-value
Birthing person race
Non-Hispanic white		802	12.62	123	12.44	< 0.001
Non-Hispanic Black		3492	54.94	628	63.50
Hispanic		2062	32.44	238	24.06
Advanced birthing age		1007	15.84	210	21.23	< 0.001
Prior preterm birth		450	8.40	173	21.25	< 0.001
Early Prenatal Care		4244	80.88	627	78.47	< 0.001
Birthing person education
College graduate or above		835	13.22	121	12.35	0.2001
High school graduate		3396	53.77	557	56.84
Less than high school completion		2085	33.01	302	30.82
Born in the United States		2403	38.08	453	46.60	< 0.001

Advanced birthing age: age greater than or equal to 35years old

Prior preterm birth: having a prior live birth born at gestational less than 37 weeks

Early prenatal care: having at least one prenatal care visit in the first trimester

Generalized linear mixed effects modeling

In our null random intercepts the ICC was 0.032 (covariance parameter p = 0.03), indicating that 3.2% of the variability in the outcome was due to census tract alone. After including race and ethnicity into the model along with all covariables, we found that census tract was no longer a significant predictor of the outcome (covariance parameter p = 0.06).

Exploratory analyses

Given the very small contribution of census tract to the risk of preterm birth before 35 weeks gestation in our sample, we explored the individual-level contributors to preterm birth among a high-risk birthing population in logistic regression models of the total sample and then stratified by race and ethnicity.

In the total sample, NHB birthing people had a 1.3 increased odds of giving birth before 35 weeks compared to NHW birthing people (95% Confidence interval [CI] 1.00-1.66). The difference in the odds of PTB between Hispanic and NHW birthing people was not significant (aOR 1.03, 95% CI 0.77–1.39). Prior PTB, advanced birthing age, and nativity status were significantly associated with the odds of PTB among the entire population (Table 3).

Table 3.

Logistic regression of preterm birth among all birthing people

		aOR	95% Confidence Interval	p-value
Birthing person race
Non-Hispanic white		ref		0.025
Non-Hispanic Black		1.29	1.00-1.66
Hispanic		1.03	0.77–1.39
Advanced birthing age		1.33	1.07–1.64	0.01
Early preterm birth		2.77	2.23–3.44	< 0.0001
Adequate Prenatal Care		0.94	0.77–1.15	ns
Born in the United States		1.39	1.16–1.67	< 0.001

In race and ethnicity stratified models, each model found different covariates that were associated with the risk of PTB. For NHB birthing people, advanced birthing age (aOR 1.5 95% CI [1.1-2.0]), prior preterm birth (aOR 2.7 95% CI [2.0-3.6]), and U.S. nativity (aOR 1.4 95% CI [1.1–1.7] were all adversely associated with the risk of preterm birth. For Hispanic birthing people, similarly being born in the U.S. was associated with an increased aOR of 1.4 (95% CI [0.95–2.1]) that trended toward significance. Only a history of prior PTB was a significant predictor of subsequent PTB across all populations (Table 4).

Table 4.

Logistic regression of preterm birth among birthing people stratified by race and ethnicity

	Non-Hispanic white		Non-Hispanic Black		Hispanic
	aOR	95% Confidence Interval	aOR	95% Confidence Interval	aOR	95% Confidence Interval
Advanced birthing age	0.89	0.48–1.64	1.47	1.12–1.92	1.26	0.79–2.01
Prior preterm birth	2.06	1.12–3.79	2.71	2.05–3.58	3.29	2.17–5.01
Early Prenatal Care	1.00	0.59–1.69	1.02	0.79–1.32	0.76	0.50–1.15
Born in the United States	1.81	0.91–3.62	1.37	1.10–1.71	1.41	0.95–2.09

Discussion

We identified that among a high-risk group of birthing people seeking care in the largest safety-net hospital in New England, neighborhood was a statistically significant predictor of PTB. However, the effect of neighborhood was relatively small, accounting for only 3% of the variability in the outcome. We identified significant racial and ethnic inequities in PTB consistent with existing literature.

Race is a social construct, and inequity in health outcomes, including preterm birth, are the result of racism in its many forms—structural, institutional, interpersonal, and internalized—and not due to biologic or genetic etiologies [45–49]. We identified that racial and ethnic inequities exist even among a high-risk birthing population seeking care at a safety-net hospital, further supporting prior research that racial and ethnic inequity in PTB is not solely the result of social or economic factors or delivery hospital quality.

Contrary to our hypothesis, in our multilevel model we found that neighborhood effects account for only 3% of the variance in the outcome and thus are not the primary driving force of PTB among high-risk birthing people. These findings suggest that among a high-risk birthing population, other factors are responsible for the variable, and inequitable, risk of PTB. Potential drivers supported by our analysis include individual-level risk factors, such as advanced maternal age and prior PTB. Notably a prior PTB may indicate additional upstream drivers that continue to increase the risk of preterm birth across multiple pregnancies. Furthermore, given the low contribution of neighborhood on PTB we wonder whether there is are unmeasured individual-level factors such as individual-level exposure to the adverse social determinants of health, or upstream unmeasured exposures such as as exposure to institutional or structural racism that may contribute to the risk of PTB.

Our findings complement existing studies on the effects of neighborhood on health. Often, accounting for individual-level factors attenuates any observed effects of neighborhoods. For example, adjusting for individual-level covariates attenuated the effect of crime rates on preterm birth inequities [30]. In addition, moving individuals from lower- to higher-quality neighborhoods has generally limited effects on health, education, employment, and income, as seen in the Moving to Opportunity trial. Such interventions may even lead to adverse impacts on health [26, 50–52]. There are many reasons this may be the case given the multi-level ecological drivers of health across the life-course. Those with the greatest individual-level risk for preterm birth may have already accumulated and embodied substantial insults over time. Furthermore, cross-sectional explorations do not account for the intergenerational transmission of individual-, community-, and sociopolitical- level exposures.

While neighborhood-based interventions may improve the risk of PTB across the entire population, the lack of neighborhood-level effect in our study suggests that among the highest-risk birthing people, those who should be focused on, neighborhood-based interventions and policies may not have the highest efficacy. In fact, because of the known neighborhood-based risk of PTB among the general population, interventions at the neighborhood level may disproportionately benefit white and socioeconomically advantaged individuals, but not those individuals similar to our cohort. This may inadvertently widen the PTB equity gap. Instead, our study identified that individual-level factors, such as nativity status, were significant drivers of PTB and may represent meaningful targets to help reduce PTB among the high-risk birthing population. Notably, the higher risk of PTB among U.S.-born individuals suggeststhat there is an adverse experience for minoritized individuals living in the U.S., which has been previously demonstrated in the literature [53, 54]. This may be due to the effect of U.S.based structural racism and discrimination. Future individual-level interventions might include a nativity-based risk-stratified approach to providing obstetric care.

Despite our rigorous methodological approach, our study has several limitations. First, the study population represents a circumscribed population, in that they are a high-risk group of birthing people seeking care at a safety-net hospital. This limits the generalizability of our study to the general U.S. population but provides a unique opportunity to study a systematically disenfranchised and often understudied and excluded population from research. Additionally, this approach “centers at the margins,” a process that makes the “perspectives of socially marginalized groups, rather than those of the dominant race or culture, the central axis around which discourse revolves.” [55] This process provides meaningful results for the most high-risk and disenfranchised communities in their own right, as opposed to solely in comparison to the majority. Second, race and ethnicity categories are limited and mutually exclusive. This approach presents race and ethnicity as a monolith, overlooking the importance of sub-populations within racial or ethnic categories and the intersectional social construction of race and identity, thus inadvertently reenforcing racial stereotypes and upholding white supremacy, a limitation of many quantitative approaches to health equity research. Nonetheless, we accept these limitations for the overall pursuit of health equity and justice. We also acknowledge that census tract may not be a perfect definition of neighborhood and accept the limitations of using census tract as a proxy for neighborhood. Finally, compared to national COI distribution, the BBC’s COI distribution was skewed towards low and very low COI, which may limit our statistical power to detect the benefit of high opportunity. Finally, we recognize that the street address data provided by individuals may have been inaccurate; or may reflect a mailing address that does not accurately represent the environmental exposures of individuals if there was housing instability, change of domicile, or if significant time was spent at another location prior to, or during, the pregnancy.

It is important to note that individual-level factors do not represent individual or population choices or culpability. Instead, systematic racism, discrimination, and disenfranchisement leads to the inequitable distribution of resources and opportunities, coupled with an increased exposure to adverse experiences [25, 56]. Our epidemiologic study explored associations and was not intended to test causal relationships. Further study, utilizing a critical race public health praxis oriented to identifying the modifiable drivers of PTB among the highest risk birthing people is imperative [57]. Potential future studies should include a combination of quantitative and qualitative approaches that “center at the margins” to better understand specific strategies, focused on structural change, that will be effective at reducing the risk of PTB among the highest risk birthing population.

Conclusions

Neighborhood was a significant, but not large, contributor to PTB risk among a high-risk birthing population seeking care at a large, urban, safety-net hospital and significant racial and ethnic inequities exist. This lack of a large of effect of neighborhood on the risk of PTB in this population, suggests that interventions to address PTB in this population may have minimal impact if focused solely at the neighborhood. Further study is needed to better understand the drivers of PTB in the high-risk birthing populations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

PTB

preterm birth

NICU

Neonatal Intensive Care Unit

NHB

Non-Hispanic/Latinx Black

NHW

non-Hispanic/Latinx white

BMC

Boston Medical Center

BBC

Boston Birth Cohort

COI

Child Opportunity Index

ICC

Interclass correlation coefficient

Author contributions

YSF and JSL conceptualized and designed the study. YSF drafted the initial manuscript, reviewed and revised the manscript, supported the analysis and interpretation of data. XW and SR had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.XW designed the data collection, supervised all aspects of the analysis and interpretation of the data, reviewed and critically revised the manuscript. SAR conducted all analyses and supported the interpretation of the results and initial drafting of the manuscript. JRE critically reviewed the manusciprt and provided important intellectual contributions. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Funding

Dr Fraiman was supported by the Agency for Healthcare Research and Quality, grant T32HS000063, as part of the Harvard-wide Pediatric Health Services Research Fellowship Program. Boston Birth Cohort (the parent study) is supported in part by the Health Resources and Services Administration (HRSA) of the U.S. Department of Health and Human Services (HHS) under grant number UT7MC45949, Autism Single Investigator Innovation Program; and by the National Institutes of Health (NIH) grants (R01HD086013, 2R01HD041702, R01HD098232, R01ES031272, R01ES031521, U01ES034983). This information or content and conclusions are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by HRSA, HHS or the U.S. Government.

Data availability

The dataset analyzed during the current study is not publicly available to ensure the privacy of study participants but are available from co-author, Dr. Xiaboin Wang, upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board (IRB) at Boston Medical Center. The IRB at Beth Israel Deaconess Medical Center determined that use of data from the Boston Birth Cohort for secondary analysis was not human subjects’ research.

Consent to publish

Not applicable.

Conflict of interest

None of the authors have any conflicts of interest to disclose.

Competing interests

The authors declare no competing interests.