Racial disparities in low-risk cesarean birth rates across hospitals

Abstract

Background:

We compared low-risk cesarean birth rates for Black and White women across hospitals serving increasing proportions of Black women and identified hospitals where Black women had low-risk cesarean rates less than or equal to White women.

Methods:

In this cross-sectional analysis of secondary data from four states, we categorized hospitals by their proportion of Black women giving birth from “low” to “high”. We analyzed the odds of low-risk cesarean for Black and White women across hospital categories.

Results:

Our sample comprised 493 hospitals and the 65,524 Black and 251,426 White women at low risk for cesarean who birthed in them. The mean low-risk cesarean rate was significantly higher for Black, compared with White, women in the low (20.1% vs. 15.9%) and medium (18.1% vs. 16.9%) hospital categories. In regression models, no hospital structural characteristics were significantly associated with the odds of a Black woman having a low-risk cesarean. For White women, birthing in a hospital serving the highest proportion of Black women was associated with a 21% (95% CI: 1.01–1.44) increase in the odds of having a low-risk cesarean.

Discussion:

Black women had higher odds of a low-risk cesarean than White women and were more likely to access care in hospitals with higher low-risk cesarean rates. The existence of hospitals where low-risk cesarean rates for Black women were less than or equal to those of White women was notable, given a predominant focus on hospitals where Black women have poorer outcomes. Efforts to decrease the low-risk cesarean rate should focus on (1) improving intrapartum care for Black women and (2) identifying differentiating organizational factors in hospitals where cesarean birth rates are optimally low and equivalent among racial groups as a basis for system-l evel policy efforts to improve equity and reduce cesarean birth rates.

1 |. INTRODUCTION

Cesarean rates vary significantly across hospitals.¹ Black women at low risk for cesarean birth have significantly higher rates of this outcome than White women.² Black women also tend to give birth in different hospitals than White women.^3,4 If racial disparities in the cesarean rate were predicated on comorbidities, we would expect the disparity to narrow among women at low risk for having a cesarean birth. National data⁵ and other research² indicate, however, that the disparity in the low-risk cesarean rate is not due to differences in medical or obstetric comorbidities, suggesting that the difference is due to disparate care. Research on hospital variation in maternal outcome disparities has focused predominantly on severe maternal morbidity,^3,6–8 though the magnitude of the disparity in the cesarean rate is similar and cesareans are far more common.^2,7 Low-risk cesarean birth, therefore, presents an opportunity to evaluate system-based contributors to health disparities for a common outcome.

Decreasing the cesarean birth rate among women at low risk for having a cesarean birth is a major focus of obstetric quality improvement initiatives.^5,9,10 Healthy People 2030 set the national cesarean rate target at 23.6% for women at low risk for having a cesarean birth, where low risk was defined as nulliparous women birthing a term, singleton fetus in the vertex position.⁵ In 2019, 25.6% of women at low risk for having a cesarean birth in the United States had a cesarean.⁵ For Black women, it was 29.7%, the highest of any racial or ethnic group.⁵ Quality initiatives to reduce low-risk cesarean section have not explicitly addressed the racial disparity in the rates or been directly focused on reducing cesarean rates among Black women at low risk for cesarean.^5,9,10 The purpose of this study is to compare the cesarean birth rates of Black and White women at low risk for cesarean birth across hospitals serving low, medium, and high proportions of Black women. A secondary purpose is to identify hospitals where Black women have low-risk cesarean rates equal to or less than those for White women.

2 |. METHODS

2.1 |. Study design

We used a cross-sectional design to conduct a secondary analysis of hospital discharge abstracts linked with American Hospital Association (AHA) Annual Survey data.

2.2 |. Setting

Hospital discharge abstracts from 2016 were made available to the researchers from another study, which was conducted in four states (California, New Jersey, Florida, and Pennsylvania) (PI Aiken R01NR014855). These states were selected as representing three geographically distinct areas in the United States and accounting for 20% of annual hospitalizations. These inpatient discharge summaries were obtained from the Office of Statewide Health Planning and Development in California, the Healthcare Cost and Utilization Project’s State Inpatient Database for Florida, the New Jersey Department of Health and Senior Services, and the Pennsylvania Health Care Cost Containment Council.¹

2.3 |. Participants

A subset of inpatients with diagnosis-r elated group codes 765–780 from Major Diagnostic Category 14 (Pregnancy, Childbirth, and the Puerperium) were provided to the authors. These diagnostic-related groups included birth and postpartum, false labor, abortion, and ectopic pregnancy coding and excluded antepartum diagnoses and patients whose principal diagnosis was invalid as a discharge diagnosis.² From this initial sample, women giving birth were identified using a validated methodology that we cross-referenced to the International Classification of Diseases, Tenth Revision (ICD-10).³ Hospital characteristics data were obtained from the AHA’s 2016 Annual Survey data, a voluntary annual survey of US hospitals, and merged with the patient data using the AHA hospital identifier.

Hospitals were excluded if they had less than 100 births in 2016. This exclusion criterion is consistent with other research in which a minimum of 100 births is considered enough births to allow meaningful comparisons across hospitals.⁴ Hospitals were also excluded if race or ethnicity was missing for more than 20% of the women who gave birth in the hospital.⁶ Women were included if they gave birth at one of the 493 hospitals (31 hospitals excluded for having fewer than 100 births; one excluded for missing race/ethnicity on more than 20% of women giving birth), were non-Hispanic Black or White, and met the definition of low risk for cesarean birth.

2.4 |. Variables

Low risk for cesarean birth is a term denoting a lack of pre-existing medical, obstetric, or fetal risk factors that would increase the chance of a woman having a cesarean birth. We defined low risk for cesarean birth status following the Society for Maternal-Fetal Medicine (SMFM) definition.^11,12 The SMFM definition was developed in response to the Joint Commission and Agency for Healthcare Research and Quality (AHRQ) definitions because both of those definitions omitted conditions that (1) preclude vaginal birth (e.g., placenta previa) or (2) make vaginal birth less likely (e.g., HIV infection).¹¹ Armstrong et al.¹¹ recognized the need for a definition that could be used with administrative data, unlike the Joint Commission definition, which required hospital data abstraction. The authors used the two existing definitions and then built on those to exclude additional clinically relevant risk factors for cesarean.¹¹ The final definition included all term, singleton, vertex, and live births to women without prior cesareans and without high-risk diagnoses.¹¹ After building the definition, the authors determined the face validity of their measure and then tested it using the 2011 Nationwide Inpatient Sample of over 863,000 births in 612 hospitals.¹¹ They found that, across all hospitals, the mean low-risk cesarean rate was lowest for SMFM (12.65%) but not significantly different from the Joint Commission and AHRQ measures (13.12% and 13.29%, respectively). Low-risk cesarean birth rate was calculated as the percentage of all births to women at low risk for having a cesarean birth during the study period that were cesareans using the hospital discharge data. Cesarean birth was identified using the ICD-10 code (O82) and the DRG codes 765 and 766, a method validated in prior research.^3,4

Hospital categories were created based on the proportion of all Black women, out of all women, giving birth at a hospital. This classification allowed us to look at hospitals in which Black women were disproportionately birthed and how this was associated with low-risk cesarean rates. It also allowed us to test the hypotheses that (1) where a woman births matters and (2) that Black women at low risk for cesarean were more likely to have the surgery wherever they gave birth. These categories were created before sample exclusion, and included women of all races and ethnicities, as well as high-risk and low-risk status. We ranked hospitals by the proportion of Black women birthing in them. Then we categorized the top 5% of these hospitals as “high”, the following 20% as “medium”, and the remainder as “low”, following prior research.⁶ This categorization, which results in an unequal distribution across categories, assured that the high-category hospitals truly served a greater proportion of Black women. Maternal race was derived from the hospital discharge data.

Other hospital variables, including hospital teaching, rural or urban and technology status, bed size, and ownership, were obtained from the AHA Annual Survey data. Hospital birth volume, the total number of births in 1 year, was obtained from the hospital discharge data. Hospital rural or urban status was based on the United States Census Core-Based Statistical Area (CBSA). Bed size categories followed the Agency for Healthcare Research and Quality definitions.⁸ High technology status was defined as the capacity for performing open-heart surgery or major organ transplant.^9–11 Hospital teaching status, based on the AHA Annual Survey report of medical resident to bed ratio, was categorized as major (one trainee per four beds), minor (less than one trainee per four beds), and none.⁹

2.5 |. Statistical methods

We examined Black and White women at low risk for cesarean who gave birth in three different categories of hospital defined by the proportion of Black women served. We used descriptive statistics (mean, median, standard deviation, and interquartile range) to describe the sample and low-risk cesarean birth rates for Black and White women across hospital categories. We used chi-square tests to examine how hospitals in different categories varied across structural characteristics and ANOVAs to compare mean cesarean birth rates across categories (see Table 1). We estimated odds ratios (OR) and 95% confidence intervals (95% CI) from robust multilevel logistic regression models with women clustered within hospitals. For the mixed regression models, the outcome was low-risk cesarean at the patient level. We ran unadjusted models and then ran a series of adjusted models. In the first adjusted model, we controlled for hospital structural characteristics. In the second, we controlled for patient race and hospital structural characteristics. Finally, we stratified by patient race and controlled for hospital structural characteristics. Confounding at the patient level was controlled for by restricting the sample to women who met the SMFM definition for low risk for cesarean birth. Using the SMFM definition as an exclusion criteria results in the most conservative control of medical or obstetric risk factors for cesarean section.¹¹

TABLE 1.

Characteristics of sample hospital categories based on the proportion of Black women birthing at the hospital.

	Overall (n = 493)	Low (n = 371)	Med (n = 98)	High (n = 24)	p-Values
State, n (%)
California	234 (47.5)	214 (57.7)	19 (19.4)	1 (4.2)	<0.0001
New Jersey	50 (10.1)	33 (8.9)	15 (15.3)	2 (8.3)
Pennsylvania	93 (18.9)	75 (20.2)	12 (12.2)	6 (25)
Florida	116 (23.5)	49 (13.2)	52 (53.1)	15 (62.5)
Technology, n (%)
High technology	236 (48.9)	169 (46.4)	49 (51.6)	18 (75)	0.02
Low technology	247 (51.1)	195 (53.6)	46 (48.4)	6 (25)
Teaching, n (%)
Non-teaching	215 (44.7)	182 (50)	30 (31.9)	3 (13)	<0.001
Minor teaching	225 (46.8)	162 (44.5)	51 (54.3)	12 (52.2)
Major teaching	41 (8.5)	20 (5.5)	13 (13.8)	8 (34.8)
Bed size, n (%)
Small	190 (41)	161 (45.9)	26 (29.2)	3 (12.5)	<0.001
Medium	229 (49.4)	165 (47)	49 (55.1)	15 (62.5)
Large	45 (9.7)	25 (7.1)	14 (15.7)	6 (25)
Location, n (%)
Rural	43 (9.1)	39 (10.9)	4 (4.4)	0 (0)	0.045
Metro	431 (90.9)	320 (89.1)	88 (95.7)	23 (100)
Annual birth volume, mean (SD)	1783.6 (1535.3)	1732.1 (1462.7)	2015 (1879.2)	1635.1 (902.8)	0.2

Missingness on hospital variables was minimal and was addressed by listwise deletion. In the final sample, 1.61% of women were missing race or ethnicity. Women missing race or ethnicity were not included in the denominator for the creation of the hospital categories, as this would assume that none of the women missing information on race or ethnicity were Black.

Analyses were conducted using Stata Version 15.0.¹² This study used deidentified data and, as a result, did not meet the definition of human subject research. This determination was confirmed by the Institutional Review Board.

3 |. RESULTS

The sample of women and hospitals was constructed according to the methods. There were 880,153 women potentially eligible for inclusion in the study in the data set. Of these, 409 women were excluded because they gave birth in hospitals that had fewer than 100 births per year. A furthermore 417 women were excluded because they gave birth in the one hospital, missing greater than 20% on race or ethnicity. There were 526 hospitals with births in the sample. Of these, 32 were excluded for having fewer than 100 births, and one was excluded for missing greater than 20% on race or ethnicity. Our final sample comprised 493 hospitals and the 879,327 women who gave birth in those hospitals. Of the overall sample of women, 606,484 were at low risk for cesarean, including 65,524 Black and 251,426 White women. That is, of all women at low risk for cesarean, 10.8% were Black and 41.5% were White. These numbers are similar to the overall demographics of women who gave birth in the four states in 2016, of whom 11.5% were Black and 40.7% were White.¹³

The sample hospitals were described overall and according to the proportion of Black women who gave birth in the hospital. Of the 493 hospitals, 382 served low (0%–15.3%, mean = 5%), 98 medium (15.4%–40.8%, mean = 23.5%), and 24 high (41.2%–82.2%, mean = 53.6%) proportions of Black women. These three hospital categories are referred to as “low”, “medium”, and “high”, respectively. Hospitals serving high proportions of Black women were mostly in Florida (62.5%% vs. 23.5% overall, p < 0.001). They were more likely to be high technology (75% vs. 48.9% overall, p = 0.02), major teaching (34.8% vs. 8.5% overall, p < 0.001), urban (100% vs. 90.9% overall, p = 0.045), and large (25% vs. 9.7%, p < 0.001). Furthermore detail is presented in Table 1.

Variation in the low-risk cesarean rate was described overall and according to hospital category (see Table 2). The mean hospital-level low-risk cesarean rate was 16.1% (SD 5.8) and increased across hospital categories from low to high, though not significantly (15.8% vs 18.3%, p = 0.05). The hospital-l evel mean low-risk cesarean rates for Black (19.6%, SD 12.4) and White (16.2%, SD 5.6) women varied significantly (p < 0.001). The mean cesarean rate for Black (20.1% vs. 18.1% vs. 18.4%, p = 0.3) and White (15.9% vs 16.9% vs. 17.7%, p = 0.1) women at low risk for cesarean birth decreased and increased across the hospitals categories from low to high, respectively, though not significantly. Mean low-risk cesarean rates varied significantly between Black and White women in the low (20.1% vs. 15.9%, p < 0.001) and medium (18.1% vs. 16.9%, p = 0.02) hospital categories, though not in the high (18.4% vs. 17.7%, p = 0.5) hospital category. In a little over one-third of the hospitals (36.6%), Black women had low-risk caesarean rates equivalent to or less than those of White women (referred to as “equivalent or lower” moving forward). More than half (58.3%) of the hospitals serving high proportions of Black women had equivalent or lower cesarean rates, compared with 42.9% of medium and 33.4% of low hospitals (p = 0.02). Hospitals with equivalent or lower low-risk cesarean rates could be found across all the structural variables listed in Table 1, with statistically significant variation in only three categories (birth volume, state, and rurality). They were significantly more likely to not be high-volume hospitals with 2500 or more births per year (13.1% vs. 25.4% overall, p < 0.001). While 18.9% and 23.5% of the overall hospitals were in PA and FL, 22.7% and 33% of the equivalent or lower hospitals were in those states, respectively (p < 0.001). Finally, there was significant variation in rurality, with 7.8% of hospitals being rural overall, compared with 13% of the equivalent or lower hospitals (p < 0.001). The hospital average low-risk cesarean rate for Black women at these equivalent or lower hospitals was 11.2% (SD 7.7), compared with 24.5% (SD 12.1) at hospitals that did not have equivalent or lower rates (p < 0.001). The hospital average low-risk cesarean rate for White women at these equivalent or lower hospitals was 17.5% (SD 5.9) and 15.5% (SD 5.2) at non-equivalent or lower hospitals (p = 0.002).

TABLE 2.

Mean low-risk cesarean rates across hospital categories based on the proportion of Black women birthing at the hospital.

	Overall (n = 493)	Low (n = 370)	Med (n = 99)	High (n = 24)	p-Values
Overall, mean (SD)	16.1 (5.8)	15.8 (6.1)	16.8 (4.6)	18.3 (4.9)	0.05
Black, mean (SD)	19.6 (12.4)a	20.1 (14)b	18.1 (5.7)c	18.4 (4.4)	0.3
White, mean (SD)	16.2 (5.6)a	15.9 (5.7)b	16.9 (4.7)c	17.7 (6.3)	0.1
Low-risk cesarean rates for Black women less than or equal to that for White, n (%)	176 (36.6)	120 (33.4)	42 (42.9)	14 (58.3)	0.02

Note: p-values here result from testing the variation across all three hospital categories.

^ap-value < 0.001;

^bp-value < 0.001;

^cp-value = 0.02; superscripts represent tests of Black vs. White means within hospital categories.

Results from the mixed regression models, unadjusted as well as controlling for hospital and patient characteristics are presented in Tables 3 and 4. In the unadjusted model, there were significantly increased odds of low-risk cesarean in middle and high hospital categories compared with low, consistent with Table 2 (OR 1.1 [95% CI 1.01–1.19] and OR 1.2 [1.05–1.4], respectively). When we controlled for hospital structural characteristics, with women being clustered by hospital, the effect of the hospital category (proportion of Black women served) was not significant. The state in which the hospital was located, however, was significant. Women giving birth in Florida and New Jersey had a 50% and 21%, respectively, increased odds of a low-risk cesarean compared with those giving birth in California (OR 1.5, 95% CI 1.3–1.7; OR 1.21, 95% CI 1.05–1.39). When we adjusted for whether the patient was Black or White, however, the effect size of the state in which the hospital was located decreased and only remained significant for Florida compared with California (OR 1.19, 95% CI 1.08–1.3). In the fully adjusted regression model, Black women still had an 18% increased odds of having a low-risk cesarean compared with White women (OR 1.18, 95% CI 1.13–1.24). We then stratified by race and found that none of the hospital structural characteristics significantly increased a Black woman’s odds of having a low-risk cesarean. Giving birth in a hospital in PA compared with CA, however, decreased a Black woman’s odds of having a low-risk cesarean by 20% (OR 0.8, 95% CI 0.71–0.9). Meanwhile, giving birth in a hospital serving a high proportion of Black women increased a White woman’s odds of having a low-risk cesarean by 21% (OR 1.21, 95% CI 1.01–1.44), as did giving birth in Florida compared with California (OR 1.21, 95% CI 1.1–1.33).

TABLE 3.

Robust multilevel logistic regression model of the low-risk cesarean rate.

Variables	Unadjusted	Adjusted for hospital characteristics	Adjusted for hospital characteristics and patient race
Minority-serving hospital category
Low: Ref	Ref	Ref	Ref
Medium	1.1 (1.01–1.19)	0.93 (0.84–1.03)	1.01 (0.93–1.09)
High	1.2 (1.05–1.4)	1.02 (0.86–1.19)	1.04 (0.9–1.2)
State
CA: Ref		Ref	Ref
FL		1.5 (1.3–1.7)	1.19 (1.08–1.3)
NJ		1.21 (1.05–1.39)	1.11 (0.95–1.3)
PA		1.01 (0.94–1.08)	0.95 (0.88–1.02)
Teach
Non-teaching: Ref		Ref	Ref
Minor teaching		1.07 (0.99–1.16)	1.03 (0.96–1.1)
Major teaching		1.00 (0.83–1.2)	0.92 (0.75–1.13)
Tech
High tech: Ref		Ref	Ref
Low tech		1.05 (0.96–1.14)	1.05 (0.97–1.15)
Bed
Small: Ref		Ref	Ref
Medium		1.01 (0.93–1.1)	1.00 (0.92–1.1)
Large		0.98 (0.86–1.12)	1.01 (0.89–1.14)
Rural
Rural: Ref		Ref	Ref
Metro		0.89 (0.76–1.04)	0.93 (0.8–1.08)
Patient race
White: Ref			Ref
Black			1.18 (1.13–1.24)

TABLE 4.

Robust multilevel logistic regression model for the low-risk cesarean rate stratified by patient race.

Variables	Black women	White women
Minority-serving hospital category
Low: Ref	Ref	Ref
Medium	0.94 (0.86–1.04)	1.02 (0.94–1.11)
High	0.97 (0.84–1.11)	1.21 (1.01–1.44)
State
CA: Ref	Ref	Ref
FL	1.08 (0.96–1.21)	1.21 (1.1–1.33)
NJ	1.11 (0.96–1.27)	1.11 (0.93–1.33)
PA	0.8 (0.71–0.9)	0.99 (0.91–1.08)
Teach
Non-teaching: Ref	Ref	Ref
Minor teaching	1.07 (0.96–1.19)	1.02 (0.95–1.09)
Major teaching	1.01 (0.85–1.2)	0.9 (0.71–1.16)
Tech
High tech: Ref	Ref	Ref
Low tech	1.00 (0.91–1.1)	1.07 (0.97–1.16)
Bed
Small: Ref	Ref	Ref
Medium	0.96 (0.86–1.08)	1.01 (0.91–1.11)
Large	1.00 (0.85–1.18)	1.02 (0.89–1.16)
Rural
Rural: Ref	Ref	Ref
Metro	0.94 (0.69–1.29)	0.93 (0.8–1.08)

4 |. DISCUSSION

We compared low-risk cesarean birth rates of Black and White women across hospitals serving low, medium, and high proportions of Black women. Racial differences in low-risk cesarean birth rates are uniquely informative about the quality of care because maternal clinical factors vary minimally. Therefore, racial differences in the low-risk cesarean rate across hospitals likely reflects disparities at the system and interpersonal level.¹⁴ Overall, the hospital mean low-risk cesarean rate for Black women was higher than that for White women (19.6% vs. 16.2%, p < 0.001). These rates are well below the Healthy People 2030 target of 23.6%, but the definition of low risk being used here is much more conservative.⁵ Our findings are similar to a recent paper that used the 2018 National Inpatient Sample data and the SMFM low-risk definition and reported a 17.4% low-risk cesarean rate for Black women and a 14.2% rate for White women.¹⁵ We found that the low-risk cesarean rate decreased for Black, and increased for White, women at low risk for cesarean birth across categories of hospitals serving low to high proportions of Black women, though this was not statistically significant. Black women had higher mean rates of cesarean birth than White women, and these differences were statistically significant, in hospitals serving low and medium proportions of Black women.

Black women at low risk for cesarean birth experienced higher rates of cesarean birth than their White counterparts at almost two-thirds (63.4%) of hospitals in the sample, and had an 18% increased odds (95% CI 1.13–1.24) of low-risk cesarean compared with White women after controlling for hospital structural variables. suggesting differential labor care.^2,5 Valdes found that, in a national sample of nearly all the women who gave birth in the United States in 2016 (1.2% excluded for missing data on birth certificates), Black women in Robson categories 1–4 (i.e., term, singleton, cephalic pregnancies with no history of cesarean, and differentiated by parity and labor onset type) were more likely to receive a cesarean than any other racial or ethnic group.² She also found that Black women at high risk, whose diagnostic codes indicated a likely need for a cesarean, were less likely to receive a cesarean than any other racial or ethnic group.² These findings suggested poorer quality intrapartum care for Black women, no matter their medical risk.² Similarly, Ouyang et al.¹⁵ reported a 23% increased odds (95% CI 1.19–1.27) of low-risk cesarean for Black compared with White women after adjusting for age, payer, rurality, region, admission day, obesity, diabetes, and hypertension. Hamm et al.¹⁶ found that using a standardized induction protocol was associated with a reduced cesarean rate in Black women. This suggests that implementing standardized care decreases racially disparate care and results in more equitable outcomes.¹⁶ We believe the differences highlighted in our study are also likely due to racially disparate care because (1) these are all women who meet the same stringent “low risk” criteria, (2) sociodemographic variation does not explain racially disparate clinical outcomes,¹⁴ and (3) research has documented racially disparate care and outcomes within hospitals.¹⁶

We found that the inclusion of patient race in the regression model attenuated the association of state with the odds of low-risk cesarean. Race is not a biological construct and has no effect on one’s medical need for a cesarean. That the inclusion of patient race decreased the association between state and the odds of a low-risk cesarean, therefore, suggests that patient race is associated both with the state where she lives as well as her odds of having a low-risk cesarean. Both findings make sense in the context of the racialized history of where people live (e.g., redlining, the Great Migration) and racially disparate maternity care. In our stratified regression models, the only significant association for Black women was giving birth in PA compared with CA, which resulted in a 20% decrease in the odds of cesarean (95% CI 0.71–0 .9). As hospital structural variables do little to explain racially disparate low-risk cesarean rates, this suggests to us that hospital organizational characteristics and racially disparate care likely explain more of the variation than structural variables. Ouyang et al.¹⁵ list similar reasons for their findings, including facility and healthcare practitioner practices, patient-practitioner communication, and racial discrimination. This is excellent news in the sense that organizational variables and care quality are amenable to intervention, where structural variables are not.

That the odds of low-risk cesarean increased by 21% (95% CI 1.01–1.44) in high compared with low hospital categories for White women suggests that hospitals which serve higher proportions of Black women may be more likely to perform cesareans in women at low risk for having a cesarean birth, compared with other hospitals. This finding suggests that hospitals serving higher proportions of Black women provide lower quality care to women at low risk for having a cesarean birth overall. It is possible that hospitals specializing in providing “higher levels of care”,¹⁷ that is, care to women at high risk for complications due to pre-existing medical or obstetric conditions, are not specialized, or well suited in terms of obstetric culture, to providing care for women at low risk for complications. For instance, systems put in place to support and manage high-risk pregnancies, may not support physiologic birth practices. Another possibility is that a unit culture that is primed to intervene for high-risk pregnancies may not benefit low-risk pregnancies that do not require similar rates of intervention. Heelan-Fancher and Edmonds, in a study of labor and delivery nurses, found that maternity units have different cultures around birth.¹⁸ Some unit cultures are more supportive of physiologic birth and other unit cultures favor medical intervention in labor.¹⁸ This lends credence to the idea that some hospitals might be more supportive of the labors of women at low-risk for cesarean and others geared to be more supportive of women at higher risk. A study of 53 maternity unit management practices across the United States found that proactive unit management, units that actively tried to preempt challenges like patient bottlenecks, was associated with a significantly higher risk of cesarean birth among women at low risk, compared with units with reactive management.¹⁹ Plough et al.¹⁹ found that women at low risk for having a cesarean birth who birthed in maternity units with proactive physician and nursing cultures, compared with reactive, had a greater risk of primary cesarean birth (relative risk [RR] 1.30, 95% CI 1.02–1.66). This finding was contrary to their expectation¹⁹ but makes sense if we consider that some maternity units specialize to care for women at higher risk and that proactive unit management, which might be associated with greater intervention in labors, would be a benefit to women at high risk and lead to higher cesarean rates among women at low risk.¹⁹ The finding also makes sense in the context of a larger literature and clinical understanding that decreasing the low-risk cesarean rate is best accomplished by means of decreasing labor interventions when medically unnecessary and increasing high touch intrapartum care.^9,20–24 Given that low-risk cesarean birth reflects one aspect of quality, our findings also suggest that Black women may be more likely to access care in hospitals that provide lower quality care for women at low risk for cesareans.^3,6,25

A little over one-third of sample hospitals had equivalent or lower low-risk cesarean rates between Black and White women. Hospitals in the high category had the highest percentage of equivalent rates (58.3%) compared with medium (42.9%) or low (33.4%; p = 0.02). We surmise that these hospitals may have care that is more consistent, with less variability in common obstetric interventions such as inductions or augmentations among women of different races and ethnicities, perhaps reflecting adherence to protocols, as exemplified by the study by Hamm et al.¹⁶ The greater frequency of equivalent or lower hospitals in the high category, however, might also be explained by the increased low-risk cesarean rate experienced by White women in these hospitals. As we noted previously, our findings suggest that hospitals that serve higher proportions of Black women may provide lower-quality care to women at low risk for having a cesarean birth overall. Certainly, adherence to protocols that standardize care across racial and ethnic groups might also coexist with care that is not optimized for supporting low-risk labors and births. These hospitals with equivalent or lower rates were found across all hospital structural characteristics, suggesting that such low-risk cesarean rates are possible in any kind of hospital. Hospitals with equivalent or lower rates were more frequently rural and not in the highest birth volume category, suggesting that these hospitals might be more likely to be providing care for women with low-risk pregnancies, which aligns with the suggested explanations in the preceding paragraph.^18,19

Our findings suggest that policy efforts to decrease the low-risk cesarean rate should focus on system change to improve intrapartum care for Black women, specifically. Identifying organizational characteristics of hospitals where cesarean birth rates are equivalent or lower among racial groups and optimally low would provide the basis for system-level policy efforts to improve equity and reduce cesarean birth rates.

4.1 |. Limitations

Births in the sample states represented about a quarter of the births in the United States in 2016.¹³ This sample does not fully represent all US regions or more rural areas. As a result, generalizations about the larger population should be made cautiously. The number of hospitals in the high group is small because we wanted these hospitals to represent those serving larger proportions of Black women, consistent with prior research. This may have reduced the power of statistical comparisons with that sub-g roup. Secondary race-ethnicity data has been criticized for reflecting how others identify a person and not necessarily how the individual personally identifies. Racially disparate care, however, reflects how a person is identified by others, not how they themselves identify. As a result, for the purpose of this study, we believe secondary race and ethnicity data is more appropriate for studying discrimination. We did not control for confounding beyond limiting the patient sample to women who met the SMFM criteria for low-risk cesarean birth. While the algorithm does not contain all of the comorbidities that might influence decision-making in labor management, it is the most comprehensive of the three most commonly used lists and was developed by subject-matter experts. We did not control for social determinants of health, apart from the hospital where a woman accessed care to give birth. For this study, we agree with the recommendation of the Institute of Medicine¹⁴ that adjusting for social determinants of health when examining racial disparities is inappropriate.

5 |. CONCLUSIONS

Low-risk cesarean rates for Black women were higher than those for White women in our sample. These rates were significantly higher in hospitals serving low and medium proportions of Black women. A Black woman’s odds of having a low-risk cesarean persist despite controlling for the hospital where she births and its structural characteristics. This suggests, in accord with other literature, that organizational characteristics (e.g., unit culture) and racially disparate care in labor likely explain the disparate outcome. The increasing low-risk cesarean rate for White women across hospitals serving low to high proportions of Black women suggests that hospitals serving greater proportions of Black women may be less focused on optimizing care for women at low risk for cesarean who present to give birth. One explanation for this is that these hospitals might be focused on providing care for women with high-risk pregnancies requiring more medically necessary intervention and therefore having a culture of care less suited for women at low risk of cesarean. The presence of hospitals where Black women do not have higher cesarean rates than White women suggests that care in these hospitals may be more consistent among women of different races and ethnicities. To move the needle on the low-risk cesarean rate in the United States, our findings support (1) focusing efforts on providing equitable intrapartum care to Black women, (2) examining hospital organizational characteristics for opportunities for system-level intervention, and (3) for hospitals focused on high-risk obstetric care to consider optimizing care for women at low risk for cesareans.

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no new data were created or analyzed in this study.