Disparities by race/ethnicity in unplanned cesarean birth among healthy nulliparas: a secondary analysis of the nuMoM2b dataset

The majority of individuals in this sample (N = 5095) reported being perceived by others in the United States as white (67.2%), while 15.8% were perceived as Black, and 17.0% presented as Hispanic (Fig. 1). Interestingly, some 18.7% -7.2% of participants did not self-identify as the same race/ethnicity they were most frequently perceived as by others in the U.S. (Supplemental Table 1). At the time of hospital admission for labor, 10.9% of individuals in this sample had a BMI in the normal or underweight range, 39.1% were in the overweight range, 27.7% classified as Obese I, 13.0% Obese II, and 9.3% were in the Obese III category. The median maternal age among participants was 27 years (Interquartile range, IQR: 22.0, 31.0) and gestational age was 39.6 weeks (IQR: 39.0, 40.4). Very few nulliparas in this sample developed gestational diabetes during their pregnancy (3.5%), while nearly a quarter developed some form of hypertensive disorder (22.7%).

Unplanned cesarean birth occurred in 1,001 (19.6%) of the low risk nulliparas in this sample (Table 1). Rates of unplanned cesarean birth were significantly higher among Black-presenting (24.1%) and Hispanic-presenting participants (24.7%) compared to white-presenting participants (17.4%). Cesarean birth also occurred more frequently among those with higher BMI at hospital admission, more advanced gestational age at labor onset, and older maternal age. Unplanned cesarean birth also varied by the mode of labor onset. For example, among the nulliparas in this sample with a cesarean, 53.2% started labor with an induction, compared to only 30.7% of nulliparas ending labor with a vaginal birth. In bivariate logistic regression models, white-presenting participants had significantly lower odds of experiencing an unplanned cesarean compared to Black-presenting participants 0.66 (97.5% CI [0.55–0.79], p < 0.001), while Hispanic-presenting participants had similar odds as Black-presenting participants (Table 2).

In the final multivariate model, unplanned cesarean was associated with maternal age, maternal BMI at or above 25.0 kg/m² (reference group: BMI < 25 kg/m²), gestational age at birth between 40.0–41.6 (reference group: 39.0–39.6 weeks), development of hypertensive complications during pregnancy, and labor augmentation or induction (reference group: spontaneous labor without augmentation). These predictor variables were assessed for multicollinearity, and none included in the adjusted model exceeded the recommended GVIF (to the power of ½ degrees of freedom (DF) cut-off point 2.24, range: 1.01–1.08). Therefore, no predictor variables were excluded from subsequent models. In the final adjusted models, compared to Black-presenting participants, white-presenting participants had 0.57 (95% CI [0.45–0.73], p < 0.001) lower odds of unplanned cesarean birth, while Hispanic-presenting participants had similar odds of cesarean birth (OR 1.17, 95% CI [0.88–1.55], p = 0.28) (Table 2).

It is well-documented that BMI is associated with both race/ethnicity and unplanned cesarean birth [19, 26]. Therefore, BMI at time of labor was assessed as a potential moderator or mediator of the association between participant-identified presenting race/ethnicity and unplanned cesarean birth. We saw no evidence of significant two-way interaction between presenting race/ethnicity and BMI overall (as either a categorical or continuous variable) in predicting unplanned cesarean birth. However, in examining different combinations of BMI and presenting race/ethnicity, we found that BMI was a moderator of the odds for unplanned cesarean among Black-presenting women. Relative to the lowest BMI category (< 25.00 kg/m²), higher BMI categories carried higher odds for unplanned cesarean birth among Black-presenting, but not white- or Hispanic-presenting race/ethnicity (Supplemental Fig. 1).

In mediation analyses, we found partial mediation of the direct relationship between presenting race/ethnicity and unplanned cesarean birth by maternal BMI at the time of labor. In the final structural equation model, which included all covariates from the main model (maternal age, BMI, gestational age at birth, development of hypertensive complications during pregnancy, and mode of labor onset), BMI explained 24% of the association between presenting race/ethnicity and unplanned cesarean birth (analyses not shown). Mediation analyses assume, in this scenario, that presenting-race influences differences in BMI or body size, which in turn cause cesarean birth; however, the relationship between body mass index and racial identity is complex, and this study was not designed to allow for causality conclusions. Thus, from these analyses was conclude that 1) even after accounting for maternal BMI, presenting race/ethnicity was associated with unplanned cesarean birth and 2) the relationship between higher BMI and likelihood of cesarean birth was noted specifically among Black-presenting participants.

Given our findings that the odds of unplanned cesarean birth were different by presenting race/ethnicity, we were interested to see if there were differences by presenting race/ethnicity in the primary clinical indication for cesarean. For this analysis, we focused on a subsample of participants who ended labor with a cesarean (n = 1,001). Four conditions identified in the dataset explained nearly all primary indications for unplanned cesarean births: non-reassuring fetal heart rate (FHR), arrest of descent, arrest of dilation, and failed induction of labor (IOL). More Black- and Hispanic-presenting participants who experienced an unplanned cesarean birth were diagnosed with non-reassuring FHR (Black-presenting: 43.80%, Hispanic-presenting: 32.6%) compared to white-presenting participants (28.20%, P < 0.001). By contrast, white-presenting participants were more often diagnosed with arrest of descent (28.5%) compared to Black- (8.2%) or Hispanic-presenting nulliparas (16.2%, P < 0.001) (Fig. 2).

Among the other primary indications for unplanned cesareans in this sample, we saw no difference by presenting race/ethnicity for either arrest of dilation (labor dystocia) or failed labor induction. To investigate whether intrapartum providers had different tolerances for diagnosing arrest of dilation in nulliparas with different presenting race/ethnicity, we compared duration of labor from hospital admission to birth by presenting race/ethnicity, stratified by mode of labor onset (Table 3). However, we found no evidence of statistical significance between participant-identified presenting race/ethnicity and labor duration, regardless of the mode of labor onset.

There were significantly worse maternal outcomes following cesarean birth by presenting race/ethnicity, regardless of the indication for the surgery. For example, Black- and Hispanic- presenting nulliparas had roughly double the rates of postpartum maternal morbidity (15.5% and 17.2%, respectively) following unplanned cesarean birth as did white-presenting nulliparas (8.8%, p < 0.001, Supplemental Table 2). Compared to white-presenting nulliparas with cesarean birth, Black- and Hispanic- presenting nulliparas had higher rates of SGA neonates (14.9% among Black, 9.8% among Hispanic, and 5.9% among white-presenting people in this sample, P = 0.003). However, neonatal outcomes were similar by presenting race/ethnicity; rates of Apgar scores < 7 at 5 min of life and NICU admission were not different by presenting race/ethnicity (Supplemental Table 2).

In this secondary analysis of a large, multi-site prospective study, we found evidence of a racial disparity in the use of cesarean birth following a trial of labor. Differences were identified among healthy nulliparas, with Black- and Hispanic- presenting nulliparas having roughly double the odds compared to white-presenting nulliparas of ending labor with an unplanned cesarean in analyses adjusting for clinical characteristics. This finding is not new, as several other groups of investigators have seen similar disparities in cesarean birth rates [27‐30]. For example, Williams and colleagues recently published evidence of higher cesarean birth rates in Black, Hispanic, and Asian compared to white people in a large retrospective cohort study from two academic medical centers in the same geographic area (N = 18,946). They found that Black women’s odds of cesarean were 1.68 (95% CI 1.45–1.96) those of white women in adjusted analyses [27]. Similarly, Stark and colleagues saw higher likelihood of cesarean birth among non-Hispanic Black women compared to non-Hispanic white women in retrospective cohort analysis (N = 9,865) of women laboring at a single tertiary care center (OR in analysis adjusting for maternal, perinatal, and systems-level factors 1.58, 95% CI: 1.31–1.91) [29].

We also found evidence that maternal BMI at the time of labor accounted for about a quarter of the influence of maternal presenting race/ethnicity on unplanned cesarean, and that the relationship between BMI and higher odds for unplanned cesarean birth was unique to the Black-presenting race group. In other words, in this sample of healthy nuMoM2b participants it appears that Black-presenting people in larger bodies had the highest risks for unplanned cesarean—higher than white- or Hispanic-presenting people in similar sized bodies. It is possible that there were differences by presenting race and body size during intrapartum care processes that placed this group of people at particular risk for unplanned cesarean birth. This and other possible mechanisms for this finding should be addressed in future research.

This study also shows evidence that non-reassuring FHR was the primary indication for more cesarean births in Black- and Hispanic-presenting, compared to white-presenting people. This finding has also been noted by other investigators [27‐31]. In recently published studies, Williams and colleagues found that Black women had roughly double the rates of non-reassuring FHR than white women (37% of cesareans in white women, compared to 64% of Black women and 42% of Hispanic women, p < 0.001) [27]. Similar findings were also seen in older studies by several other teams [28, 30, 31]. In a 2022 retrospective cohort study (N = 16,687 women), Okinwandu and colleagues found that Black women were 50% more likely to have a cesarean for fetal intolerance (aOR = 1.51, 95% CI: 1.10–2.07) [32]. Those authors theorized that racial differences may be attributed to factors that influence the shared decision-making process for Black women—some of which may be rooted in fear.

Interestingly, we found that fewer cesareans occurred among Black or Hispanic- compared to white-presenting individuals for the primary indication of arrested descent. This means that the higher odds for unplanned cesarean birth we see in Black- and Hispanic-presenting people compared to white-presenting people in the overall sample are even more compelling, as people of color were relatively less likely than white-presenting nulliparas to experience unplanned cesarean for arrested descent. This finding raises the question of whether fewer Black-presenting individuals reached the 2^nd stage of labor, when arrest of descent was diagnosed. However, nuMoM2b original data abstraction limits us in testing this consideration. We can presume that some of the cesareans for non-reassuring FHR may have occurred during 2^nd stage despite adequate progress in descent. It is also possible that the smaller size of newborns from Black- and Hispanic-presenting compared to white-presenting nulliparas contributed to easier second stage descent.

In this analysis, mode of labor onset was an important predictor of racial disparities in non-reassuring FHR cesarean birth. While racial disparities existed in this outcome among nulliparas who began laboring spontaneously, they were not evident among those with labor induction. Given that induced labor involves use of medications like prostaglandins and higher-duration synthetic oxytocin, which are associated with increased incidence of fetal intolerance [33], it is surprising that racial disparities in non-reassuring FHR cesareans were not found among those with induced labor. To our knowledge, this is the first time that this link between non-reassuring FHR cesarean racial disparity and spontaneous labor onset has been published.

There are two possible explanations for the observation of increased odds of unplanned cesarean for the indication of non-reassuring FHR in Black- and Hispanic- compared to white-presenting nulliparas. First, it is possible that the fetus in nulliparas of color from this sample did not tolerate uterine contractions during labor as well as the fetus in white-presenting nulliparas. Second, there could be differences in the quality of intrapartum care received by Black- and Hispanic- compared to white-presenting nulliparas which predispose them to end labor with a cesarean birth for fetal intolerance. Each of these options is considered in the following sections.

Fetal intolerance of labor has been linked to placental insufficiency secondary to various maternal and fetal factors [34, 35]. However, in sub-analyses of cases from this study with unplanned cesarean birth (Supplemental Tables 2 and 3), we saw few differences in maternal/fetal factors by presenting maternal race/ethnicity. Although maternal age and gestational age were different by race/ethnicity in this group, neither of these were different in Black- presenting compared to white-presenting nulliparas in a direction expected to predict poor outcomes.

There were differences by presenting race/ethnicity in newborn weight for nulliparas ending labor with non-reassuring FHR after spontaneous labor. Smaller infants may be influenced by compromised placental perfusion, which may have many consequences for the birth process [34, 35]. However, there were no differences by presenting race/ethnicity in neonatal Apgar scores < 7 at 5 min or in neonatal intensive care unit admission in the first 28 days of life. Greater antenatal depression has been linked to racial disparities in lower newborn weight; [36‐38] however, we did not see differences in pregnancy depression diagnosis by presenting race/ethnicity among those with non-reassuring FHR cesarean and spontaneous labor (analyses not shown).

It is possible that the stress of laboring in an environment where structural racism and/or race/ethnic discordance with the intrapartum care team may place strain on the maternal/fetal unit such that non-reassuring FHR is more likely. In previous investigations of perinatal outcomes among Black presenting populations, experiences of discrimination, racism and general stress were found to increase the risk for adverse perinatal outcomes, such as preterm delivery and low birth weight [36‐38]. The nuMoM2b study collected the Experience of Discrimination scale from each participant at their 2^nd prenatal visit [14], but these scores were not significantly associated with unplanned cesarean birth when added to adjusted models (analyses not shown). Given that most U.S. pregnant people have a choice about the people who will provide their antepartum care, but not their intrapartum care, patient-reported experiences of discrimination affecting intrapartum outcomes should ideally be collected specifically referencing the intrapartum timeframe.

A second option to explain racial disparities in non-reassuring FHR cesarean birth in this study is differences in the quality of intrapartum care received by Black- and Hispanic- compared to white-presenting nulliparas. For the most part, it is impossible to address this theory due to limitations of the nuMoM2b dataset. For example, nuMoM2b does not contain information on the use of amnioinfusion, nor on other intrapartum interventions commonly used during a concerning FHR period (maternal position changes, intravenous bolus, etc.) Among nuMoM2b variables that could be used to address this theory, there was no difference by maternal presenting race/ethnicity in the use of synthetic oxytocin augmentation or artificial rupture of membranes following spontaneous labor in this sample.

Non-reassuring FHR diagnoses largely explain the massive increase in U.S. cesarean rates over the past four decades [28, 39]. During this time, widespread use of continuous fetal monitoring during labor was not associated with significant improvement in perinatal death or cerebral palsy rates but did increase rates of cesarean and instrumental vaginal birth. Although intrapartum resuscitation for non-reassuring FHR tracings is effective, reverting approximately 63.7% of category II to category I tracings within 60 min [40], there is little standardization of intrapartum resuscitation across the United States. Possibly, this lack of standardization of intrapartum resuscitation guidelines allows systemic racism and other drivers of poor care to affect outcomes. In a 2017 secondary analysis of a multicenter observational obstetric cohort (N = 115,000 births), investigators did not find evidence of racial disparities in the utilization of labor management strategies intended to reduce unplanned cesareans performed for failed induction, labor dystocia, and arrest of descent [30]. However, they did not evaluate differences by race in labor management preceding non-reassuring FHR cesareans. More work is needed to evaluate possible racial disparities in the quality of intrapartum care preceding non-reassuring FHR cesarean birth.

A key strength of this study is our focus on the healthiest labors. By using a large, multi-site dataset for secondary analyses, we were able to focus on a large sample of healthy, term nulliparas with a trial of labor. In previous investigations, researchers attributed increased use of cesarean for non-reassuring FHR in Black- compared to white women to maternal issues, including obesity, medical/obstetric complications, and fetal macrosomia [28]. However, we found that the racial disparity in non-reassuring FHR cesarean birth existed even when the sample was limited to healthy nulliparas.

Another strength of this study is our use of maternal presenting race/ethnicity, rather than self-identified or medical record-categorized race, as the key predictor in all analyses. Racism is based on perceived differences by other members of a person’s community; thus, presenting race/ethnicity better captures the influence of person- and system-level racism on intrapartum outcomes in this analysis than self-reported race. Although it is important to consider how differences in individual social determinants of health may worsen health outcomes in some populations of people, distribution of the social determinants of health in the United States is largely defined by race/ethnicity [13]. In this way, racism is the structural determinant of the social determinants, and as such it is a key factor to incorporate in studies such as this one that focus on health inequities.

Like any retrospective cohort study, this investigation was limited to demonstrating associations, not causality conclusions. Although nuMoM2b is a rich dataset for secondary analyses, there were multiple variables not collected by the original study’s team which limited this secondary analysis. For example, nuMoM2b contained no information on amnioinfusion, maternal position changes, or other intrapartum interventions commonly used during periods of non-reassuring FHR. We were unable to characterize the extent to which nulliparas in this dataset received interventions to resuscitate their fetus prior to an unplanned cesarean birth. Although nuMoM2b captured the primary and secondary indications for cesarean births as they were documented in participants’ medical records, it is possible that documented indications were not always the truest reasons in the providers’ minds at the time the decision to proceed with cesarean birth was made. Also, nuMoM2b did not contain information describing the intrapartum care received by participants with enough detail to examine the influence of patient-provider racial congruence or professional training of the provider (for example, obstetrician vs. midwife) on unplanned cesarean birth disparities by presenting race/ethnicity.

Finally, this study is limited by how presenting race was identified in nuMoM2b. The gold standard for presenting race/ethnicity as it relates to possible structural racism in healthcare systems/providers would be to collect this information about patients from their healthcare providers (i.e. to collect the healthcare providers’ perception of each research participants’ race/ethnicity). However, presenting race/ethnicity from the perspective of the research participant is as close as this study could get to this ideal.

More research is needed to better elucidate the mechanisms of racial disparities in pregnancy outcomes, including the diagnosis of non-reassuring FHR during labor. These studies should include prospectively collected information on the use of intrauterine resuscitation for non-reassuring FHR changes. If continuous electronic fetal monitoring is used, more information for these studies should be collected from FHR tracings, including variability and presence of decelerations, as well as categorization over the course of labor.

Maternal presenting race/ethnicity, in addition to other variables reflecting structural racism [12], should be incorporated in future analyses of birth outcomes. These studies should also include measures of inequity and mistreatment during labor/birth from the perspective of the birthing person [41‐43], as well as maternal/fetal stress hormone levels. The complexity of the experiences of racism among Black presenting pregnant people are poorly captured in many research investigations. This despite the fact that key manifestations of racism during labor/birth have been identified for incorporation in birth outcome investigations [41, 44, 45]. For example, based on their findings from focus groups with African-American women, Nuru-Jeter and colleagues recommended that birth outcomes research capture measures of women’s childhood experiences of racism, as well as their experiences of racism directed at their children and over their life course—all of which could be ‘sources of stress with potentially serious implications for birth outcomes’[44]. In a 2022 publication, Chambers and colleagues identified several system-level variables described by perinatal clinicians as evidence of racism which could also be measured in birth outcome investigations. These include provision of inequitable care, increased surveillance of Black women and families, and the extent to which structural care issues are present in a system (i.e., history of medical racial experimentation) [45]. There are also several validated questionnaires available for incorporation into birth outcome investigations that measure patient-reported experiences of discrimination [41], respect [42], safety, autonomy, communication and other aspects of obstetric racism during the intrapartum course [43]. Such measures expand the focus of intrapartum research to include not only physical outcomes but also the quality of the birth experience. Finally, although racial/ethnic disparities in birth outcomes persist even when social determinants of health like education [46] and income [47] are optimal, more work is needed to understand if intersectional social determinants of health conditions exist that are protective for birth outcomes, including unplanned cesarean birth for the indication of non-reassuring FHR [16].