Racial Disparities in Prematurity Persist among Women of High Socioeconomic Status

Abstract

Objectives:

Despite persistent racial disparities in preterm birth (PTB) in the US among non-Hispanic (NH) black women compared to NH white women, it remains controversial whether sociodemographic factors can explain these differences. We sought to evaluate whether disparities in PTB persist among NH black women with high socioeconomic status (SES).

Study Design:

We conducted a population-based cohort study of all live births in the US from 2015–2017 using birth certificate data from the National Vital Statistics System. We included singleton, non-anomalous live births among women who were of high SES (defined as having ≥ 16 years of education, private insurance, and not receiving Women, Infants and Children [WIC] benefits) and who identified as NH white, NH black, or ‘mixed’ NH black and white race. The primary outcome was PTB <37 weeks; secondary outcomes included PTB <34 and <28 weeks. In addition, analyses were repeated considering birthweight <2500g as a surrogate for preterm birth <37 weeks, birthweight <1500g as a surrogate for preterm birth <34 weeks, and birthweight <750g as a surrogate for preterm birth <28 weeks’ gestation. Data were analyzed with chi-square, t-test, and logistic regression.

Results:

2,170,686 live births met inclusion criteria, with 92.9% NH white, 6.7% NH black, and 0.4% both NH white and black race. Overall, 5.9% delivered <37, 1.3% <34, and 0.3 % <28 weeks. In unadjusted analyses of women with high SES, the PTB rate at each gestational age cutoff was higher for women of ‘mixed’ NH white and black race, and highest for women who were NH black only compared to women who were NH white only. In regression models we further adjusted for women with insurance and prenatal care their entire pregnancy, maternal race was associated with higher odds of PTB at each GA cutoff, with the highest odds observed at <28 weeks. Finally, in further adjustement analysis including only the 1,934,912 women who received prenatal care in the first trimester, findings were similar. Rates of preterm birth at each gestational age cutoff remained highest for women who identified as non-Hispanic black, intermediate for women identifying as both non-Hispanic black and white race, and lowest for non-Hispanic white women at <37 weeks (9.9% vs. 6.1% vs. 5.5%, respectively; p<0.001), <34 weeks (3.5% vs. 1.5% vs. 1.1%, respectively; p<0.001), and <28 weeks’ gestation (1.2% vs. 0.4% vs. 0.2%, respectively, p<0.001).

Conclusions:

Even among college-educated women with private insurance who are not receiving WIC, racial disparities in prematurity persist. These national findings are consistent with prior studies that suggest factors other than socio-demographics are important in the underlying pathogenesis of PTB.

Condensation:

Even among women of high socioeconomic status - college-educated women with private insurance who are not receiving WIC - disparities in preterm birth rates persist.

Introduction

Preterm birth continues to be a source of significant neonatal and maternal morbidity in the United States and worldwide¹. Premature infants have both short- and long- term medical and neurodevelopmental morbidity that contribute significant financial and emotional strain on families^2–4. Mothers of premature infants frequently experience medical and surgical morbidity such as classical cesarean deliveries,⁵ and have an increased risk of adverse cardiovascular complications later in life^6–8.

Unfortunately, pathophysiology underlying the multifactorial etiology remains poorly understood. Non-Hispanic black women have a preterm birth rate that is up to 2-fold that of non-Hispanic white women^9,10 in the United States. Despite medical (e.g., progesterone supplementation) and surgical (e.g., cervical cerclage) interventions to prevent recurrent preterm birth, racial disparities in prematurity persist^11–13. Though multiple researchers have hypothesized that neighborhood deprivation^14–17, family history of preterm birth¹⁸, and socioeconomic status¹⁹ all contribute to this disparity, the reasons for the difference in preterm birth rates among women of different racial groups remains unknown.

In the past, studies have had mixed results when investigating whether socioeconomic differences explain the racial disparity in preterm birth rates, however, a growing body of evidence continues to make the implications of structural and societal racism on the health of women of color harder and harder to ignore. Schempf et al showed that women who lived in a more economically-deplete community had an increased risk of preterm birth, and that this burden was borne mostly by women of Non-Hispanic black race¹⁷. Conversely, when looking at a cohort of women with increased psychosocial stress non-Hispanic black women had persistence of low birthweight infants despite having less psychosocial stress²⁰. And in a cohort of women who were all college graduates, another study found that Non-Hispanic black women continued to have a higher overall risk of preterm birth²¹.

As we continue to explore the implications that one’s life experience has on racial disparities of health outcomes, many discussions have shift to a life course perfective. Specifically, Lu et al poses two pathways – early programming and the implications of life experiences on future health and cumulative pathways suggesting that continuous wear and tear on the body increases one’s allostatic load and subsequently leads to poor health outcomes²². Many researchers in California have looked at the effects of socioeconomic status on black-white health disparities and preterm birth. A study by Culhane et all found that black women who initiate prenatal care in the first trimester do not fare better than white women who initiate care in the third trimester²³ and there were similar findings for education differences,

Interestingly, as we have increasingly granularity on how race is recorded in our country, there is more opportunity to see how women who identify as more than one race compare amongst the black-white perinatal disparities that we see. For example, in previous studies on the effects of paternal race on preterm birth rates, mixed race couples with a black mother or father had a higher odds of preterm birth that a white race couple^24,25. To date, birth certificate data has not been used to investigate both the relationship of socioeconomic status and race in a national cohort.

The objective of our study was to further observe the relationship between maternal race and socioeconomic status on a woman’s risk for preterm birth in a national cohort. Specifically, we sought to evaluate whether disparities in preterm birth persist in the United States among women of high socioeconomic status.

Materials and Methods

We conducted a retrospective population-based cohort study of all live births in the US from 2015–2017 using live birth certificate data from the National Vital Statistics System of the Centers for Disease Control and Prevention’s National Center for Health Statistics. This publicly available database contains more than 99% of all live births that occur within the 50 US states and US territories each year. Because this analysis was performed using a publicly available, de-identified dataset, this research did not meet criteria for human subjects research and following administrative review, was deemed exempt from oversight by the Institutional Review Board at the University of North Carolina at Chapel Hill.

Women within the 50 US states with singleton, non-anomalous live births, who identified as non-Hispanic white, non-Hispanic black, or both non-Hispanic black and white race, and had high socioeconomic status were included. We defined high socioeconomic status for the study population a priori based on prior definitions in the literature when birth certificate data was used^26,27 as having ≥ 16 years of education, private insurance, and not receiving Women, Infants and Children [WIC] benefits.

The primary outcome was preterm delivery less than 37 weeks 0 days gestation. Secondary outcomes included preterm birth less than 34 weeks 0 days and less than 28 weeks 0 days gestation. The best obstetric estimate variable in the live birth record was considered the gestational age at delivery.^28,29 Because some studies report that birthweight may be more accurate than gestational age in states that do not have centralized prenatal care screening systems that validate clinical gestational age with sonographic gestational age³⁰, we performed a sensitivity analysis with low birthweight and very low birthweight as surrogates for preterm birth. Neonatal birthweight was classified as small for gestational age (<10%tile) based on gender and gestational age- specific contemporary national norms.¹⁰ In addition, analyses were repeated considering birthweight <2500g as a surrogate for preterm birth <37 weeks, birthweight <1500g as a surrogate for preterm birth <34 weeks, and birthweight <750g as a surrogate for preterm birth <28 weeks’ gestation.

Data were analyzed with chi-square, t-test, and logistic regression. Pregnancy, intrapartum, and immediate postpartum characteristics were described. Regression analysis was adjusted for covariates that could increase or decrease one’s risk of preterm birth such as marital status, history of preterm birth, chronic hypertension, tobacco use during pregnancy, interpregnancy interval less than 12 months, and male fetus. In regression models, nulliparous women were coded as not having a short interpregnancy interval or a previous preterm birth even though they had no previous pregnancy. Finally, an additional adjustment was done including only the women who initiated prenatal care in the first trimester. All statistical analyses were performed using STATA/SE, Version 15.1 (StataCorp, Inc., College Station, TX); p < 0.05 was considered statistically significant.

Results

From 2015–2017, there were 11,809,599 births in the United States registered in the National Vital Statistics System; of these, 2,170,686 live births met inclusion criteria (Figure 1). Of included live births, 6.7% women were non-Hispanic black, 0.4% of women identified as both non-Hispanic white and black race, and 92.9% women were non-Hispanic white. Overall, 5.9% delivered less than 37 weeks, 1.3% delivered less than 34 weeks, and 0.3% delivered less than 28 weeks’ gestation. Rates of preterm birth less than 37, less than 34, and less than 28 weeks differed by maternal race (Figure 2), with non-Hispanic black women having the highest and non-Hispanic white women having the lowest rates of preterm birth. Women who identified as both non-Hispanic black and white races had a preterm birth rate that was higher than that of non-Hispanic white women and lower than that of non-Hispanic black women at each gestational age cutoff.

Figure 1.

Study population.

Figure 2.

Preterm birth rates in the high socioeconomic status population by race (%).

Demographic and prior pregnancy characteristics of women by maternal race are presented in Table 1. Overall, the study population had low rates of pre-gestational hypertension and diabetes, however, these co-morbidities were more common among non-Hispanic black women, who were also more likely to have a higher pre-pregnancy body mass index, a short interpregnancy interval, and a prior preterm birth. Non-Hispanic black women were less likely to smoke compared to non-Hispanic white women. When considering pregnancy, intrapartum, and postpartum characteristics, significant differences were also observed by race (Table 2). Rates of overall pregnancy complications including hypertensive disorders of pregnancy and gestational diabetes mellitus were highest among non-Hispanic black women compared to non-Hispanic white women; those who identified as both non-Hispanic black and white races had lower rates of pregnancy complications compared to those who identified solely as non-Hispanic black.

Table 1.

Demographic characteristics by maternal race. Data are n(%) unless specified.

Characteristic	NH black race N=144,612	Both NH black and NH white race N=8,604	NH white race N=2,017,470	p-value
Maternal age, mean years ± SD	32.6 ± 4.6	31.6 ± 4.6	32.0 ± 4.1	<0.001
Married	107,407 (74.3)	6,501 (75.6)	1,866,436 (92.5)	<0.001
Pre-pregnancy body mass index, median kg/m2 (IQR)	27.1 (22.6, 30.2)	25.7 (22.6, 30.2)	23.8 (21.5, 27.5)	<0.001
Tobacco use during pregnancy	534 (0.4)	81 (0.9)	12,357 (0.6)	<0.001
Previous preterm birth less than 37 weeks’ gestation	4,220 (2.9)	204 (2.4)	41,560 (2.1)	<0.001
Interpregnancy interval <12 months since last live birth*	527 (0.73)	14 (0.35)	5,300 (0.50)	<0.001
Chronic hypertension	5,580 (3.9)	158 (1.8)	28,119 (1.4)	<0.001
Pregestational diabetes mellitus	1,486(1.0)	62 (0.72)	10,448 (0.52)	<0.001

^*Of 1,137,495 multiparous women with a previous live birth

Abbreviations:

IQR = interquartile range

NH = non-Hispanic

Table 2.

Pregnancy, intrapartum, and immediate postpartum characteristics, by maternal race. Data are n(%) unless specified.

Characteristic	Women of NH black race N=144,612	Women Of both NH black and NH white race N=8,604	Women of NH white race N=2,017,470	p-value
No prenatal care	999 (0.7)	32 (0.4)	6,215 (0.3)	<0.001
Initiated prenatal care in the first trimester	118,355 (85.0)	7,533 (89.6)	1,809,024 (91.5)	<0.001
Gestational hypertension and/or preeclampsia	10,928 (7.6)	515 (6.0)	120,289 (6.0)	<0.001
Gestational diabetes mellitus	8,531 (5.9)	414 (4.8)	98,785 (4.9)	<0.001
Male fetus	73,292 (50.7)	4,414 (51.3)	1,035,115 (51.3)	<0.001
Birthweight, median grams (IQR)	3260 (2920,3572)	3395 (3076,3695)	3444 (3144,3750)	<0.001
Birthweight less than 10th centile for gestational age and gender*	10,426 (7.2)	400 (4.7)	69,448 (3.4)	<0001

^*for 2,169,527 fetuses delivered ≥ 23 weeks’ gestation

Abbreviations:

IQR = interquartile range

NH = non-Hispanic

When considering neonatal birthweight as a surrogate for preterm birth, we found that differences by race persisted similar to those that were observed when using gestational age. Birthweight was available for 2,169,116 of 2,170,686 neonates (99.99%). Of those with birthweight available, 87,223 (4.02%) weighed <2500g, including 9.07% of infants born to non-Hispanic black mothers, 4.78% of infants born to non-Hispanic black and white mothers, and 3.66% of infants born to non-Hispanic white mothers. 13,728 (0.63%) infants weighed <1500g at birth, which included 2.31% of the infants born to non-Hispanic black mothers, 0.94% of the infants born to non-Hispanic black and white mothers, and 0.51% of the infants born to non-Hispanic white mothers (p<0.001). Finally, 3,386 (0.16%) weighed <750g at birth, which included 0.77% of infants born to non-Hispanic black mothers, 0.17% of infants born to non-Hispanic black and white mothers, and 0.11% of infants born to non-Hispanic white mothers (p<0.001).

In regression models adjusted for marital status, history of preterm birth, chronic hypertension, smoking during pregnancy, interpregnancy interval less than 12 months, and male fetus, maternal race remained significantly associated with preterm birth at each gestational age cutoff, with the highest odds at the earliest delivery gestational ages (Figure 2). Specifically, for women who identified as both non-Hispanic black and white races, though the odds for preterm birth <37 weeks did not remain statistically significant [adjusted odds ratio (aOR) 1.09, 95% confidence interval (95% CI) 1.00–1.19], both non-Hispanic black and white race was significantly associated with delivery <34 weeks (aOR 1.34, 95% CI 1.14–1.58) and <28 weeks (aOR 1.92, 95% CI 1.40–2.64) compared to non-Hispanic white women (referent group, OR 1.0). In contrast, for those of only non-Hispanic black race, for each gestational age cutoff, the odds of preterm birth <37 weeks (aOR 1.69, 95% CI 1.66–1.73), <34 weeks (aOR 2.69, 95% CI 2.60–2.77), and <28 weeks (aOR 4.99, 95% CI 4.70–5.28) was even more pronounced compared to non-Hispanic white women. Notably, in each gestational age model, the magnitude of risk conferred by non-Hispanic black race was similar to or greater than other predictors in the model. For example, for regression models evaluating delivery less than 28 weeks, non-Hispanic black race conferred an aOR of 4.99 (95% CI 4.7–5.28), whereas a history of a prior preterm birth conferred an aOR of 2.81 (95% CI 2.54–3.12), chronic hypertension an aOR of 2.84 (95% CI 2.55–3.17), and interpregnancy interval less than 12 months an aOR of 4.47 (95% CI 3.58–5.58).

Finally, in a final adjustment including only the 1,934,912 women who received prenatal care in the first trimester, findings were similar. Rates of preterm birth at each gestational age cutoff remained highest for women who identified as non-Hispanic black, intermediate for women identifying as both non-Hispanic black and white race, and lowest for non-Hispanic white women at <37 weeks (9.9% vs. 6.1% vs. 5.5%, respectively; p<0.001), <34 weeks (3.5% vs. 1.5% vs. 1.1%, respectively; p<0.001), and <28 weeks’ gestation (1.2% vs. 0.4% vs. 0.2%, respectively, p<0.001). In adjusted analyses, both non-Hispanic black and white race was significantly associated with delivery less than 34 weeks (aOR 1.23, 95% CI 1.02–1.48) and less than 28 weeks’ gestation (aOR 2.01, 95% CI 1.42, 95% CI 1.42–2.83). There was no statistically significant association observed for preterm birth <37 weeks among women who identified as both non-Hispanic black and white (aOR 1.05, 95% CI 0.95–1.16). Results for those who identified as non-Hispanic black were similar to the findings in the larger cohort that included all women, even those who did not receive prenatal care in the first trimester, as the odds of preterm birth less than 37 weeks (aOR 1.72, 95% CI 1.68–1.75), less than 34 weeks (aOR 2.75, 95% CI 2.65–2.85), and less than 28 weeks’ gestation (aOR 5.11, 95% CI 4.79–5.45) remained statistically significant and similar in magnitude even when limiting the analysis to women who first sought prenatal care in the first trimester.

Comment

Principle Findings

Even among college-educated women with private insurance who are not receiving WIC services, racial disparities in prematurity persist in a national birth cohort. Notably, we found that the black-white disparities in prematurity persisted for preterm birth less than 37, less than 34, and less than 28 weeks’ gestation. The relationship between maternal race and preterm birth risk was greatest for the earliest gestational age cutoffs. These results were consistent when birthweight was used as surrogate for preterm birth, and when we controlled for factors consistently associated with either an elevated (e.g., prior preterm birth, pre-gestational hypertension, short inter-pregnancy interval, male fetus, maternal smoking) or reduced (e.g., marital status) risk of preterm birth. Finally, although more than 91% of women initiated prenatal care in the first trimester, when we limited our analysis to only those who sought early prenatal care, results were consistent with the findings in the larger cohort.

As with prior studies that have examined the impact of socioeconomic status, we feel that creating a definition of high socioeconomic status with more than one variable – such as including medical insurance type and use of women-infant-children supplementation in addition to education level – helped to refine our cohort in an effort to capture more women who have financial stability. This definition of high socioeconomic status was determined a priori. It is possible that some women who reported private insurance at the time of delivery may not have had insurance early in pregnancy and that this was a financial barrier to seeking early prenatal care, and thus inappropriately classified some individuals as being of high socioeconomic status. For that reason, we performed an additional analysis limiting the cohort to those who sought prenatal care in the first trimester, in order to increase the likelihood of limiting our analysis to women of high socioeconomic status, and confirmed our findings.

Results

Our results further underscore previous data that reported socioeconomic factors only partially explain preterm birth disparities^19–21. Additionally, the diminished risk of preterm birth observed in high socioeconomic status women who identified as non-Hispanic ‘mixed’ race compared to those who identified as non-Hispanic black race only in our study provides supporting evidence that there are genetic influences on preterm birth that could be potential targets for future personalized, preventative therapies.

Clinical Implications

As we continue to examine the implications of one’s race as a risk factor for preterm birth, it is important to have obstetrical providers that acknowledge patient race as a surrogate marker for structural and societal racism that can negatively or positively affect one’s risk of preterm birth.

Research Implications

Future studies on the epidemiology of prematurity would be remiss to ignore the impact of non-clinical, but significantly different experiences related to one’s race that have generational effects on the health of mothers and their families.

Strengths and Limitations

There are multiple strengths of this study. The population-level design provides one of the largest cohorts to date to examine differences in preterm birth rates by self-identified non-Hispanic black or non-Hispanic white race. In an era of increasing admixture in the United States, our study design also allowed us to evaluate women who identify as belonging to both non-Hispanic black and non-Hispanic white race. In addition, we were able to control for multiple modifiable preterm birth risk factors, including tobacco use during pregnancy and short interpregnancy interval.

Our study is not without limitations. First, National Vital Statistics data are subject to reporting inaccuracies and are unable to be manually verified. This includes underreporting of maternal characteristics and inaccuracies in gestational age. Because some studies report that birthweight may be more accurate than gestational age, we repeated our analysis with low birthweight and very low birthweight as surrogates for prematurity, and confirmed our findings²³. Second, though prior studies suggest the risk factors and etiologies underlying medically-indicated and spontaneous preterm birth are similar, our study design did not permit us to evaluate underlying etiologies of preterm birth. Third, coding nulliparous women has not having a short interpregnancy interval could potentially confound the effects of parity and interpregnancy interval, however, in a clinical setting, nulliparous women would not be expected to have a risk of preterm birth similar to a woman who has had a short interval pregnancy. Fourth, our definition of high socioeconomic status does not guarantee that all women in this analysis have financial stability and does not account for factors such as wealth - which can differ markedly between black and white populations with similar education. Finally, despite the detailed race data available, we are limited in our interpretation of the mechanism of the lived experience of one’s self-identified race, and how that experience- or, possibly just their identification with a particular racial group- may positively or negatively affect their clinical risk of preterm birth.

Conclusions

In conclusion, our results within this national cohort further underscore previous state-level data that differences in socioeconomic factors at the time of pregnancy do not fully explain preterm birth disparities. Even among women with high socioeconomic status, there is a persistently higher rate of preterm birth among women of self-identified non-Hispanic black race and both non-Hispanic black and white race compared to women of non-Hispanic white race. As stated previous, in this study, as in others in medicine, race is a social construct - a surrogate for structural and societal racism that disproportionately affects the birth outcomes of women of color. Focused efforts should continue to evaluate how factors such as life experiences including racism, resilience, and allostatic load or weathering – in combination with traditional prematurity risk factors influence an individual’s risk of preterm birth. Intentional efforts are needed to eliminate racism alongside translational research in the fight against preterm birth.

Figure 3.

Adjusted odds ratio of preterm birth at <37, <34, and <28 weeks’ gestation by race. The figure highlights that maternal race remains significantly associated with preterm birth at each gestational age cutoff, with the highest odds at the earliest delivery gestational ages.

*Adjusted for marital status, history of preterm birth, chronic hypertension, tobacco use during pregnancy, interpregnancy interval less than 12 months, and male fetus

Abbreviations:PTB = Preterm Birth NH = non-Hispanic

At a Glance:

A. Why was this study conducted?

To evaluate whether disparities in preterm birth persist among non-Hispanic black women with high socioeconomic status.

B. What are the key findings?

The PTB rate at each gestational age cutoff was higher for women of ‘mixed’ non-Hispanic white and black race, and highest for women who were non-Hispanic black race only compared to women who were non-Hispanic white race only. Adjusted analysis for women with insurance and prenatal care their entire pregnancy, maternal race was associated with higher odds of PTB at each GA cutoff, with the highest odds observed at <28 weeks.

C. What does the study add to what is already known?

These findings are consistent with prior studies that suggest factors other than socio-demographics are important in the underlying pathogenesis of PTB.