Keywords
racial disparities - maternal morbidity - neonatal morbidity - education
While medical technology advances at unprecedented rates, severe obstetric morbidity
and mortality present a significant challenge to the U.S. health care system, with
a particularly high burden among black and Hispanic women.[1] Racial disparities in obstetrics remain a major public health challenge as research
consistently demonstrates that non-Hispanic black and Hispanic women experience higher
adverse obstetric outcomes than non-Hispanic white women.[2]
[3] These disparities are often attributed to differences in socioeconomic status, access
to care, or level of education.[4] However, conflating race and ethnicity with other social determinants of health
can distract from, or at worst neglect, understanding the true underlying etiologies
of racial disparities.
While race and ethnicity have already been well documented as independent risk factors
for obstetric morbidity and mortality, this phenomenon must be clarified specifically
in a population of women with a higher level of education to understand whether a
true disparity exists independent of education. As there is often overlap between
minority race and disadvantaged socioeconomic or educational background, it is possible
that racial disparities may not exist among women with college degrees. The purpose
of this research, therefore, is to investigate associations between race and maternal
and neonatal morbidities in a population of college-educated women.
Materials and Methods
This is a retrospective cohort study of a multicenter prospective cohort of women
undergoing cesarean delivery (Eunice Kennedy Shriver National Institute of Child Health
and Human Development Maternal-Fetal Medicine Units Network Cesarean Registry, 1999–2002).
Demographic information, medical history, and obstetric data for the patients were
collected from the medical chart by trained research staff; full details of the design
have been described by prior studies.[5]
[6]
[7] Nulliparous women with live, nonanomalous singleton gestations who underwent primary
cesarean section were included, and women were defined as “college-educated” if they
reported completion of a 4-year college degree. Race and ethnicity were determined
from the patient's chart and categorized by the parent trial as non-Hispanic white,
non-Hispanic black, Hispanic, Asian, Native American, or unknown. The analysis focused
on non-Hispanic white, non-Hispanic black, and Hispanic women, as the other groups
had limited sample sizes.
The primary outcome was a composite of maternal complications including hysterectomy,
uterine atony, blood transfusion, surgical injury, arterial ligation, infection, wound
complication, and ileus. We also evaluated a composite of neonatal morbidity as a
secondary outcome, including death, adverse respiratory outcomes, neonatal intensive
care unit (NICU) admission, and other severe complications (adverse respiratory outcomes,
NICU admission, sepsis, treated hypoglycemia, seizure, hypoxic ischemic encephalopathy,
necrotizing enterocolitis, cardiopulmonary resuscitation in the first 24 hours of
life, 5-minute Apgar score < 3, and hospitalization ≥ 5 days). Both these composites
have been previously utilized and described in studies from the Cesarean Registry.[8]
[9]
[10]
Baseline characteristics by race and ethnicity were assessed in univariable analyses
using the Student's t-test, Mann–Whitney's U test, chi-square test, and Fisher's exact test, as appropriate. We then fit a multivariable
logistic regression model adjusting for selected demographic and obstetric variables
that may influence the likelihood of the primary outcome. This included maternal age,
body mass index (BMI), marital status, toxic exposures (smoking, illicit drug use,
and alcohol use), diabetes, preeclampsia-spectrum disorders (gestational hypertension,
preeclampsia, eclampsia, and HELLP syndrome), and chorioamnionitis. All statistical
tests were two-tailed, and p-value less than 0.05 was considered significant. The sample size was fixed from the
original registry study. All analyses were performed using SAS 9.4 (SAS Institute
Inc., Cary, NC). This analysis was considered exempt by the Institutional Review Board
at Columbia University Medical Center as these data are deidentified and publicly
available.
Results
Of 73,257 women in the parent trial, 2,540 women met inclusion criteria ([Fig. 1]). The total incidence of the primary outcome of composite maternal morbidity was
11.9% (303 women). The total incidence of the secondary outcome of composite neonatal
morbidity was 25.8% (655 women). The population was composed of 1,766 non-Hispanic
white women (69.5%), 343 non-Hispanic black women (13.5%), 206 Hispanic women (8.4%),
5 Native American women (0.2%), 121 Asian women (4.9%), and 99 women who reported
unknown race/ethnicity (4.0%). [Table 1] demonstrates significant demographic and obstetric differences by racial/ethnic
group. More specifically, compared with non-Hispanic white women, non-Hispanic black
and Hispanic women were younger (mean age 29.5 and 30.0, respectively, vs. 30.6; p < 0.01) and less likely to be married (51.0 and 65.0%, respectively, vs. 92.9%; p < 0.01). Non-Hispanic black women also had higher prepregnancy BMI compared with
non-Hispanic white women (27.3 vs. 24.1; p < 0.01). Non-Hispanic white women had higher rates of smoking (6.2%, compared with
5.2 and 2.4% of non-Hispanic black and Hispanic women, respectively; p < 0.01) and alcohol use (4.0%, compared with 2.3 and 1.5% of non-Hispanic black and
Hispanic women, respectively; p < 0.01), while non-Hispanic black women had a higher rate of illicit drug use (1.5
vs. 0.2%; p = 0.04). Non-Hispanic black and Hispanic women also had significantly higher incidence
of diabetes (11.7 and 12.6%, respectively, vs. 6.6%; p > 0.01), whereas there were no significant differences in other baseline comorbidities.
Specific obstetric complications, including preeclampsia-spectrum disease and chorioamnionitis
also had significantly higher incidence among college-educated non-Hispanic black
and Hispanic women compared with non-Hispanic white women.
Fig. 1 Derivation of the study population.
Table 1
Patient characteristics by race/ethnicity
|
Non-Hispanic white
N = 1,766
|
Non-Hispanic black
N = 343
|
Hispanic
N = 206
|
Native American
N = 5
|
Asian
N = 121
|
Unknown
N = 99
|
p-Value
|
|
Maternal age, mean (SD)
|
30.6 (4.6)
|
29.5 (5.1)
|
30.0 (4.7)
|
32.4 (6.7)
|
31.2 (4.1)
|
30.0 (4.3)
|
< 0.01
|
|
Prepregnancy BMI, median (IQR)
|
24.1 (21.4–28.1)
|
27.3 (23.0–32.2)
|
24.0 (20.6–26.8)
|
23.3 (21.3–24.8)
|
21.6 (19.9–24.1)
|
23.4 (21.2–26.6)
|
< 0.01
|
|
Married, N (%)
|
1,640 (92.9)
|
175 (51.0)
|
134 (65.0)
|
5 (100.0)
|
107 (88.4)
|
83 (83.8)
|
< 0.01
|
|
Smoking, N (%)
|
109 (6.2)
|
18 (5.2)
|
5 (2.4)
|
0 (0)
|
0 (0)
|
2 (2.0)
|
< 0.01
|
|
Illicit drug use, N (%)
|
4 (0.2)
|
5 (1.5)
|
1 (0.5)
|
0 (0)
|
0 (0)
|
1 (1.0)
|
0.04
|
|
Alcohol use, N (%)
|
70 (4.0)
|
8 (2.3)
|
3 (1.5)
|
1 (0.2)
|
1 (0.8)
|
3 (3.0)
|
0.03
|
|
Diabetes, N (%)
|
117 (6.6)
|
40 (11.7)
|
26 (12.6)
|
0 (0)
|
10 (8.3)
|
6 (6.1)
|
< 0.01
|
|
Asthma, N (%)
|
134 (7.6)
|
36 (10.5)
|
12 (5.8)
|
0 (0)
|
6 (5.0)
|
4 (4.0)
|
0.13
|
|
Hypertension, N (%)
|
39 (2.2)
|
15 (4.4)
|
2 (1.0)
|
0 (0)
|
2 (1.7)
|
1 (1.0)
|
0.10
|
|
Renal disease, N (%)
|
21 (1.2)
|
5 (1.5)
|
2 (1.0)
|
0 (0)
|
0 (0)
|
0 (0)
|
0.69
|
|
Heart disease, N (%)
|
44 (2.5)
|
3 (0.9)
|
4 (1.9)
|
0 (0)
|
1 (0.8)
|
3 (3.0)
|
0.40
|
|
Preeclampsia-spectrum disease[a], N (%)
|
147 (8.3)
|
41 (12.0)
|
13 (6.3)
|
0 (0)
|
3 (2.4)
|
18 (18.2)
|
0.02
|
|
Placental abruption, N (%)
|
33 (1.9)
|
2 (0.6)
|
2 (1.0)
|
0 (0)
|
5 (4.1)
|
4 (4.0)
|
0.07
|
|
Chorioamnionitis, N (%)
|
152 (8.6)
|
49 (14.3)
|
53 (25.7)
|
2 (0.4)
|
21 (17.4)
|
8 (8.0)
|
< 0.01
|
Abbreviations: BMI, body mass index; IQR, interquartile range; SD, standard deviation.
a Preeclampsia spectrum disorders including gestational hypertension, preeclampsia,
eclampsia, and HELLP.
Adjusted analysis with a multivariable logistic regression model demonstrated that
composite maternal morbidity was significantly higher for college-educated non-Hispanic
black (adjusted odds ratio [aOR] 1.77, 95% confidence interval [CI] 1.12–2.80, p < 0.01) compared with non-Hispanic white women ([Table 2]). Neonatal morbidity was significantly higher in both non-Hispanic black women (aOR
1.91, 95% CI 1.31–2.79, p < 0.01) and Hispanic women (aOR 3.34, 95% CI 2.23–5.01, p < 0.01) compared with non-Hispanic white women.
Table 2
Association of race with maternal and neonatal morbidities among college-educated
women
|
Maternal morbidity
|
Neonatal morbidity
|
|
aOR[a]
|
95% CI
|
p-Value
|
aOR[a]
|
95% CI
|
p-Value
|
|
Non-Hispanic white
|
Reference
|
Reference
|
|
Non-Hispanic black
|
1.77
|
1.12–2.80
|
0.01
|
1.91
|
1.31–2.79
|
< 0.01
|
|
Hispanic
|
1.20
|
0.70–2.07
|
0.50
|
3.34
|
2.23–5.01
|
< 0.01
|
|
Native American
|
2.59
|
0.25–26.55
|
0.42
|
20.47
|
1.93–217.61
|
0.01
|
Abbreviations: CI, confidence interval; aOR, adjusted odds ratio.
a Model adjusted for maternal age, body mass index, marital status, toxic exposures
during pregnancy (tobacco, alcohol, and illicit drugs), maternal diabetes, preeclampsia,
and chorioamnionitis.
Discussion
This cohort study of college-educated women demonstrates that even among women with
higher education, non-Hispanic black and Hispanic women experience significantly more
maternal and neonatal complications compared with non-Hispanic white women. As the
statistical models employed also controlled for several key sociodemographic and clinical
variables, there is a strong implication of a fundamental disparity in morbidity by
race that is not attributable to other factors. College education serves as a proxy
in this study both for higher social and income status as well as a higher level of
health literacy. In fact, in prior research investigating disparities in self-esteem
and social resources by race in pregnant women, all differences between black, Hispanic,
and white women were explained by differences in income and level of education.[11] These, in turn, should function as equalizers, translating into improved access
to care and a better ability to understand and advocate for one's medical needs. College
education, however, does not seem to mitigate the racial disparities in outcomes observed
in the general obstetric patient population.
There is an emerging body of literature dedicated to the problem of racial disparities
in obstetrics which has documented increased morbidity in general as well as in specific
contexts such as postpartum hemorrhage and advanced maternal age.[1]
[12]
[13]
[14] This study examines a specific subpopulation of women with advanced education to
better understand whether racial and ethnic differences are present even among women
with college degrees, suggesting a true underlying disparity based solely on race.
As maternal morbidity and mortality remain significant problems in our health care
system, with disproportionate manifestation among women of minority race and ethnicity,
research that elucidates these deeply rooted disparities is an important step in beginning
to combat them. In fact, a recent study of U.S. vital statistics data demonstrated
small but statistically significant differences in maternal and neonatal morbidities
by race among women with at least a bachelor's degree.[15] If socioeconomic or health literacy differences do not account for racial disparities
in morbidity among society's educated elite, it is probable that bias at the provider,
hospital, and societal level may play an important role in propagating differential
adverse outcomes.[16]
[17] Every effort should be made, including institutional implicit bias training and
individualized patient risk assessment that incorporates race and ethnicity into counseling
and decision-making, to attempt to alleviate these disparities.[18]
Strengths of this study include the unique ability to evaluate a focused cohort of
college-educated women with composite maternal and neonatal morbidities outcomes based
on research quality data, as opposed to administrative data generated from diagnosis
codes. Information on patient educational status is often not available in large datasets,
and this cohort offers the opportunity to gain important insight into the dynamics
of race and obstetric morbidity in this subpopulation defined by higher education.
The study is limited by inclusion of primarily academic medical centers in the original
study and its use of data collected from 1999 to 2002, which may render it less generalizable
to the obstetric patient population today. Generalizability is further limited by
its inclusion only of women undergoing cesarean delivery. Prior research has already
demonstrated differential rates in cesarean delivery by race, and this may impact
interpretation of these findings.[19] The study is also confined to the variables included for analysis in the parent
study as well as potential confounding by the individual racial and ethnic, sociodemographic,
and educational backgrounds of the patient mix at each participating hospital.
Conclusion
Ultimately, these data demonstrate that college-educated obstetric patients of minority
race and ethnicity are still more likely to suffer significant morbidity than their
non-Hispanic white counterparts. This points to a deeply entrenched and poorly understood
disparity in how care is delivered to patients of color in the United States. Clinical
and research efforts must target systems-level bias to tackle this pervasive problem
in our health care system and effect population-wide reduction in maternal and neonatal
morbidities and mortalities.
