Racial Bias in Cesarean Decision-Making

Abstract

Background

Continuous external fetal monitoring showing category II fetal heart tracing is a frequent indication for cesarean birth in the United States, despite its somewhat subjective interpretation. Black patients have higher rates of cesarean birth, as well as higher rates for this indication. Racial bias in clinical decision-making has been demonstrated throughout medicine, including in obstetrics.

Objective

We sought to examine if racial bias is present in providers’ decisions about cesarean birth due to category II fetal heart tracings.

Study Design

We constructed an online survey study consisting of two clinical scenarios of patients in labor with category II tracings. Patient race was randomized to Black and White; vignettes were otherwise identical. Participants had the option to continue with labor or proceed with cesarean birth at three decision points in each scenario. Participants reported their own demographics anonymously. This survey was distributed to obstetric providers via email, listserv, and social media. Data were analyzed using chi-square tests at each decision point in the overall sample and in subgroup analyses by various participant demographics.

Results

A total of 726 participants contributed to the study. We did not find significant racial bias in cesarean decision-making overall. However, in a scenario of a patient with prior cesarean birth, Fisher’s exact test showed providers < 40 years old (n=322, p= 0.01) and those with < 10 years of experience (n=239, p= 0.050) opted for cesarean birth in Black patients more frequently than for White patients at the first decision point. As labor progressed in this scenario, rates of cesarean birth equalized across patient race.

Conclusion

Younger providers and those with fewer years clinical experience demonstrated racial bias in cesarean decision-making at the first decision point early in labor. Providers did not show racial bias as labor progressed, nor in the scenario without prior cesarean birth. This bias may result from provider training with the Maternal Fetal Medicine Unit Network's Vaginal Birth After Cesarean calculator, developed in 2007 and widely used to estimate the probability of successful vaginal birth after cesarean. This calculator used race as a predictive factor until it was removed in June 2021. Future studies should investigate if this bias persists following this change, while also focusing on interventions to address these findings.

Introduction:

Fetal heart tracing is widely used throughout the United States to monitor the wellbeing of a fetus during labor. Fetal monitoring requires interpretation by healthcare providers, and category II fetal heart tracings are a frequent indication for cesarean birth. Despite the shift in terminology to the category system, fetal heart tracings that are deemed category II represent a wide range of fetal statuses, from benign to concerning.¹ The objective ‘category II’ findings on a fetal heart tracing therefore still require subjective interpretation to guide clinical decision-making; the subjective interpretation guiding these decisions provides a meaningful area for investigation.

At the same time, racial and ethnic disparities persist in rates of cesarean birth in this country, with Black patients having higher rates of cesarean birth than White patients.^2,3 More specifically, Black patients have higher chances than White patients of undergoing cesarean for fetal indications.^4-6 Given that implicit racial bias is present in many areas of medicine,^7-12 including in obstetrics,^13,14 it is possible that some amount of self-reinforcing bias in the decision to perform a cesarean birth is present among obstetric providers, with Black patients in part having higher rates of cesarean because providers expect them to have higher rates of cesarean. Given the increased morbidity associated with cesarean birth,¹⁵ such a racial bias toward cesarean birth for Black patients deserves investigation.

In addition, White patients continue to have significantly higher rates of vaginal birth after Cesarean (VBAC) than Black patients in the United States.¹⁶ The Maternal Fetal Medicine Units (MFMU) Network’s widely used VBAC calculator predicts the chance of successful VBAC based on several clinical factors. The calculator was developed in 2007; race was included in its algorithm until its removal in June of 2021.¹⁷

Therefore, we aimed to investigate whether providers made different decisions regarding the need for cesarean birth for “non-reassuring” fetal status when given otherwise identical vignettes for patients of different races, including in situations of trial of labor after cesarean (TOLAC), due to racial bias. Secondary analyses sought to determine whether responses varied when stratified by various demographics of participating providers. We hypothesized that providers would be quicker to move to cesarean birth with Black patients compared to White patients in situations with “non-reassuring” fetal heart tracing. We additionally hypothesized that providers would be quicker to move to cesarean with Black patients than White patients for patients with history of prior cesarean delivery, perhaps due to a prediction that Black patients might be more likely to deliver via cesarean birth.

Materials and Methods:

After IRB approval (2020-1379), we used expert opinion from providers within our Obstetrics department to develop two clinical vignettes in which the decision to perform cesarean birth was controversial. We randomized each participant to receive a version of these vignettes in which the patients were identified as either Black/African-American or White/Caucasian; vignettes were otherwise identical. One vignette involved a patient with a history of cesarean birth, and one involved a patient without that history. These two vignettes were each accompanied by three segments of fetal heart tracings, which we selected from a collection of category II tracings used in another IRB-approved study at our institution. Each vignette had three decision points at which we provided the participant a fetal heart tracing and asked if they would recommend a cesarean birth; at each successive decision point the fetal heart tracing would appear closer to category III (more concerning for abnormal fetal acid-base status) than the one before it. We informed participants that the goal of the study was to investigate cesarean delivery decision-making without mentioning racial bias, so participants were not consciously alerted to the subject of race.

Following survey draft creation, we performed a pilot study within our department to ensure there was consensus in an increasing likelihood of opting for cesarean birth at each of the three decision points in the vignettes rather than a unanimous decision towards or against cesarean birth at any decision point. We adjusted the language of the vignettes after collecting pilot information from 11 providers.

We distributed our survey online via email to department chairs, residency directors, and community hospital contacts, listserv, and social media platforms. Participants completed the survey anonymously and were incentivized to participate with the chance to win one of 6 $50 Amazon gift cards. We also collected demographic information from participants. Data were analyzed using chi-square tests at each decision point looking first at the overall sample and then in subgroup analyses by various participant demographics. Fisher’s exact test was used when there were fewer than 5 participants per cell. Survival analyses were also performed; however, the results did not enhance our understandings of the findings. All statistical analyses were performed using SPSS 27.

Results:

726 providers completed the survey over the course of 4 months. Of these, 576 reported their demographic information (Table 1). As presented in Table 1, a majority of the responders were female (89.4%), non-Hispanic (95.0%), White (81.3%), and the most common age was 30-39 years old (49.0%). The sample was almost equally split between those who had been practicing for less than 10 years (42.5%) and those practicing between 10 and 24 years (43.6%). Responders were primarily obstetrics and gynecology (OBGYN) attending physicians (81.8%) as opposed to residents or fellows, and working in university or teaching hospital settings (43.1%).

Table 1 –

Participant Demographics

Category	Sub-category	n	%
Gender	Male	61	10.6%
	Female	515	89.4%
Age	<30	40	6.9%
	30-39	282	49.0%
	40-49	156	27.1%
	50-59	51	8.9%
	60+	45	7.8%
	NA*	2	0.3%
Years in Practice	<10	245	42.5%
	10-24	251	43.6%
	25+	80	13.9%
Ethnicity	Non-Hispanic	549	95.3%
	Hispanic	27	4.7%
Race	White/Caucasian	468	81.3%
	Asian	51	8.9%
	Black/African-American	32	5.6%
	All Other	25	4.3%
Provider Type	OB-GYN Attending	471	81.8%
	OB-GYN Resident	72	12.5%
	OB-GYN Fellow	29	5.0%
	Other	4	0.7%
Practice Setting	Medical school/University/Teaching Hospital	248	43.1%
	Ob-Gyn Group	205	35.6%
	Community Clinic or Hospital	64	11.1%
	Multispecialty	34	5.9%
	Staff model HMO	7	1.2%
	Military/Government	5	0.9%
	Solo	5	0.9%
	Other	8	1.4%

^*Note: NA = Not applicable, two participants wrote in “over 50” and “decline to state.”

Providers opted for cesarean birth at increased rates as the clinical scenarios progressed and comparing across vignettes that were identical except for race, survey responders opted for cesarean births at similar rates at each decision point for both races (Table 2).

Table 2.

Decision for Cesarean (All Participants)^*

	Vignette 1 (TOLAC)				Vignette 2
Decision stage and decision	Patient race, n (%)		p		Patient race, n (%)		p
1	Black	White	0.11		Black	White	0.51
Yes	15 (4.6)	7 (2.3)			21 (6.8)	16 (5.5)
No	311 (95.4)	300 (97.7)			288 (93.2)	275 (94.5)
2	Black	White	0.73		Black	White	0.23
Yes	61 (19.9)	62 (21.0)			35 (12.3)	25 (9.1)
No	246 (80.1)	233 (79.0)			250 (87.7)	249 (90.9)
3	Black	White	0.74		Black	White	0.97
Yes	201 (82.0)	193 (83.2)			211 (85.4)	213 (85.5)
No	44 (18.0)	39 (16.8)			36 (14.6)	36 (14.5)

^*Note: participant numbers decrease due to survey structure; those opting for cesarean were not shown subsequent pieces of vignette

In subgroup analysis of the scenario of a patient with prior Cesarean birth (Vignette 1), Fisher’s exact two-tailed test demonstrated that providers less than 40 years old (n=322, p=0.01) opted for cesarean for Black/African-American patients more frequently than for White/Caucasian patients at the first decision point (Table 3). Providers with less than 10 years experience (n=239, p=0.050) demonstrated the same bias (Table 4) using a Fisher’s exact two-tailed test. As labor progressed in this scenario, rates of cesarean birth equalized across patient race. We did not note differences in cesarean decision-making in subgroup analysis by provider race, ethnicity, or setting (university hospital, community hospital, etc.).

Table 3.

Participants Under 40 Years Old – Decision for Cesarean^*

	Vignette 1 (TOLAC)				Vignette 2
Decision stage and decision	Patient race, n (%)		p		Patient race, n (%)		p
1	Black	White	0.01**		Black	White	0.74
Yes	6 (3.8)	0 (0)			3 (1.9)	4 (2.5)
No	152 (96.2)	163 (100)			154 (98.1)	159 (97.5)
2	Black	White	0.66		Black	White	0.63
Yes	31 (20.4)	30 (18.4)			16 (10.4)	14 (8.8)
No	121 (79.6)	133 (81.6)			138 (89.6)	145 (91.2)
3	Black	White	0.40		Black	White	0.27
Yes	100 (82.6)	115 (86.5)			124 (89.9)	124 (85.5)
No	21 (17.4)	18 (13.5)			14 (10.1)	21 (14.5)

^**Note: statistically significant at p≤0.01

^*Note: participant numbers decrease due to survey structure; those opting for cesarean were not shown subsequent pieces of vignette

Table 4.

Participants with <10 Years Experience – Decision for Cesarean^*

	Vignette 1 (TOLAC)				Vignette 2
Decision stage and decision	Patient race, n (%)		p		Patient race, n (%)		p
1	Black	White	0.050*		Black	White	0.47
Yes	4 (3.4)	0 (0.00)			2 (1.7)	4 (3.1)
No	113 (96.6)	128 (100.0)			115 (98.3)	124 (96.9)
2	Black	White	0.63		Black	White	0.64
Yes	23 (20.4)	23 (18.0)			10 (8.7)	13 (10.5)
No	90 (79.6)	105 (82.0)			105 (91.3)	111 (89.5)
3	Black	White	0.39		Black	White	0.18
Yes	73 (81.1)	90 (85.7)			96 (91.4)	95 (85.6)
No	17 (18.9)	15 (14.3)			9 (8.6)	16 (14.4)

^*Note: statistically significant at p≤0.05

^*Note: participant numbers decrease due to survey structure; those opting for cesarean were not shown subsequent pieces of vignette

Comment:

a. Principal Findings

As a whole, participants did not demonstrate significant racial bias in cesarean decision-making in either scenario of category II fetal heart tracing in that opting for a cesarean delivery occurred at similar rates at every decision point for Black and White patient vignettes. However, in subgroup analysis of the vignette with a patient with history of cesarean birth, providers less than 40 years old did demonstrate significant racial bias, opting for cesarean birth for Black/African-American patients more than White/Caucasian patients at the first clinical decision point. In addition, those with less than 10 years experience demonstrated this tendency, with a p value of 0.050.

b. Results in the Context of What is Known

Prior studies have demonstrated increased rates of cesarean birth for Black/African- American patients,² as well as increased rates for fetal indications.^4-6 It is worth noting that studies have also demonstrated that these differences are not attributable to differing patient attitudes toward route of delivery between races (i.e. differing levels of desire for one particular delivery method),¹⁸ and that in fact Black/African-American patients who deliver via cesarean birth have particularly low chances of feeling a sense of shared decision-making about their deliveries.¹⁹

At the same time, studies have demonstrated and continue to demonstrate implicit racial bias across healthcare, including in emergency medicine,⁷ surgery,^8-9 primary care,¹⁰ among nurses,¹¹ and among medical students.¹² A prior study also demonstrated a significant effect of racial bias on physician decision-making regarding thrombolysis treatment: physicians who had an implicit preference for White patients were more likely to treat White patients using thrombolysis than Black patients.²⁰

The field of obstetrics is not immune to this bias, with recent studies demonstrating that Black and Hispanic patients receive fewer opioids for postpartum pain management compared to White patients,¹³ and that they are more likely to have severe postpartum pain.¹⁴ While our study does not outwardly show this bias in cesarean decision-making, in subgroup analyses we did find that providers less than 40 years old and with less than 10 years experience demonstrated racial bias in clinical decision-making in a situation of TOLAC. The American College of Obstetricians and Gynecologists has affirmed the continued presence of racial bias and the need to reduce such bias.²¹ Given the degree to which a fetal heart tracing is deemed non-reassuring is inherently subjective, it is possible that a certain amount of self-reinforcing bias exists regarding the likelihood of cesarean versus vaginal birth for patients of different races – i.e., Black/African-American patients have more cesarean births because we as providers expect them to have more cesarean births. Overall, however, this study did not demonstrate racial bias that significantly affected clinical decision-making when considering cesarean birth for a category II fetal tracing.

Additional factors are present when examining rates of VBAC, which continue to be higher among White patients than Black patients in the United States.¹⁶ MFMU’s widely-used VBAC calculator was developed in 2007 with race as a predictive factor, predicting a lower chance of successful VBAC for Black patients.¹⁷ The study used to create this calculator found that marital status and insurance type, among other factors, were also correlated with success, and yet of these only the racial variable was incorporated into the algorithm.²² As the predicted chance of successful VBAC is often used to counsel patients about their decision to undergo TOLAC versus repeat cesarean, the widespread use of this algorithm may have, in a self-perpetuating fashion, caused Black patients to have lower rates of VBAC because they are predicted to have lower rates of VBAC.

This study seems to support this hypothesis, as younger providers and those most recently in training were the only subgroups identified to demonstrate racial bias in cesarean decision-making during TOLAC. These providers may demonstrate this bias, more than older providers or providers with more years experience, due to training with MFMU’s VBAC calculator. This bias clearly can have significant impact on patients and their future reproductive lives, as repeat cesarean deliveries are associated with increased risks of numerous maternal morbidities.¹⁵ Of note, race was removed from MFMU’s VBAC calculator as a predictive factor in June of 2021, following demonstration that race’s inclusion underestimated chance of successful TOLAC for patients of color.²³ Our study highlights how racial bias, not race, may be the true risk factor for cesarean in these scenarios.

c. Clinical Implications

This study demonstrates apparent racial bias in cesarean decision-making in situations of TOLAC, specifically among providers less than 40 years old and with less than 10 years experience. This bias may be in part attributable to training of these providers with MFMU’s VBAC calculator, which used race as a predictive factor from its creation in 2007 until race’s removal from the calculator in June of 2021. Additional research is required to investigate whether this bias persists among providers following the calculator’s modification.

d. Research Implications

This study demonstrates apparent racial bias among some groups within the field of OB-GYN using a novel and anonymous study design. Future studies should investigate whether this bias persists following modification of MFMU’s VBAC calculator in June of 2021. In addition, studies are needed to evaluate interventions to address these findings and eliminate this existent bias.

e. Strengths and Limitations

Strengths of this study include a novel and randomized design and relatively large study size. We posit that its unique design allowed for study of the inexact science of fetal heart tracing interpretation in a controlled fashion, which could be used to study clinical decision-making more precisely in the future. Limitations of this study include that bias was evaluated using written description of race, rather than image or video. Evaluating for bias using image or video could potentially alter results, eliciting more or less bias. In addition, participants’ survey responses may not reflect actual practice patterns. Additionally, we were unable to calculate response rate, as we are unable to determine how many providers viewed our survey on various social media websites, etc., and whether non-obstetric providers were able to complete the survey. Further, our anonymous participants often did not report their own demographic characteristics (79.3% [576 of 726 total participants] reported most factors, such as race, gender, and age [Table 1]). Our participant population lacked diversity in several aspects, such as the majority of respondents being White, non-Hispanic, female, and attending physicians. This might have resulted in undetected selection bias if certain demographic groups were less likely to respond to the demographic portion of the survey. In addition, our original objective had been to assess for implicit racial bias, but study design did not allow for discernment of implicit vs. explicit bias. Both types of bias are harmful and important to identify for the sake of patients and the practice of equitable, high-quality medical care.

f. Conclusions

Our study demonstrated racial bias in cesarean decision-making for patients with a history of prior cesarean birth, but only for providers less than 40 years old and with less than 10 years experience. This bias may be due to training with MFMU’s VBAC calculator, which used race as a predictive factor until it was removed in June of 2021. Future studies should investigate whether this bias persists following VBAC calculator modification, and focus on interventions to address these findings.

Condensation:

Younger providers and those with less experience exhibited racial bias in cesarean decision-making for patients with category II fetal tracing and history of cesarean birth.

AJOG at a Glance:

A. Why was this study conducted?

• To evaluate if racial bias was present among OB-GYN providers when making decisions regarding cesarean birth for category II fetal heart tracing

B. What are the key findings?

• Providers less than 40 years old or with less than 10 years experience demonstrated racial bias in cesarean decision-making when presented with a patient with a history of prior cesarean birth, opting for cesarean more often for Black/African-American patients than for White/Caucasian patients.

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

• Younger and less experienced providers demonstrated racial bias in cesarean decision-making for TOLAC patients with category II fetal heart tracings. This finding may indicate effects of training with MFMU’s VBAC calculator prior to removal of race as one of its predictive components.