The Relationship Between Provider and Patient Racial Concordance and Receipt of Postpartum Care

Abstract

Access to postpartum care (PPC) varies in the US and little data exists about whether patient factors may influence receipt of care. Our study aimed to assess the effect of provider-patient racial concordance on Black patients’ receipt of PPC. We conducted a cross-sectional study analyzing over 24,000 electronic health records of childbirth hospitalizations at a large academic medical center in Alabama from January 2014 to March 2020. The primary outcome variable was whether a Black patient with a childbirth hospitalization had any type of PPC visit within 12 weeks after childbirth. We used a generalized estimating equation (GEE) logistic regression model to assess the relationship between provider-patient racial concordance and receipt of PPC. Black patients with Black main providers of prenatal or childbirth care had significantly higher adjusted odds of receiving PPC (adj. OR 2.26, 95% CI 1.65–3.09, p < .001) compared to Black patients with non-Black providers. White patients who had White providers did not have statistically significantly different odds of receiving PPC compared to those with non-White providers after adjustment (adj. OR 0.88, 95% CI 0.68–1.14). Although these results should be interpreted with caution given the low number of Black providers in this sample, our findings suggest that in one hospital system in Alabama, Black birthing people with a racially concordant main prenatal and delivery care provider may have an increased likelihood of getting critical PPC follow-up.

Introduction

The United States (US) is experiencing a maternal mortality crisis with stark and unjust racial disparities in maternal outcomes [1–3]. The majority (51%) of women who die of pregnancy-related causes in the US do so from one to 365 days after childbirth, with 64% of those deaths occurring seven or more days after birth [4]. Infections, cardiovascular conditions, and cerebrovascular accidents are the leading causes of death from 7 to 42 days after birth, and cardiomyopathy and other cardiovascular conditions make up more than 50% of the causes of death from 43 to 365 days postpartum [4].

Postpartum care (PPC) is critical to the health and survival of birthing people, presenting an opportunity to manage these and other chronic or pregnancy-related conditions that could become life-threatening [5]. Current guidance from the American College of Obstetricians and Gynecologists (ACOG) recommends a PPC “check” (by phone or in person) within 3 weeks and another more comprehensive visit within 12 weeks postpartum [6]. Along with screening for and addressing specific chronic or pregnancy-related conditions, this comprehensive visit should include an assessment of physical, social, and emotional well-being and reproductive life planning [6].

A recent systematic review found high variability in people’s receipt of PPC across US studies, ranging from 24.5 to 90% [7]. Receipt of PPC has been shown to vary by patient factors such as chronic conditions or clinical complications and sociodemographic characteristics, including education, insurance type, geographic residence, and race or ethnicity [7–10].

Barriers to accessing high-quality PPC include the acceptability of care, that is, the patient’s assessment of the adequacy, suitability, or effectiveness of care [11–13]. Perceived discrimination and disrespect from healthcare providers during childbirth and the prenatal period can make the care experience unacceptable and may influence whether people then return for postpartum visits [14–17]. The relationship between discrimination and non-return to postpartum care has been shown to be more pronounced when the discrimination is reported to be related to race or insurance status when compared to other factors [18–20].

For birthing people who identify as Black, this experience of discrimination can be understood as an aspect of institutional racism, which has been shown to impact healthcare use and health outcomes [21–24]. The experience of racism, and the awareness of the long history of racism in American healthcare, can inform how people feel about their care and their providers [25]. For Black and other racially minoritized people, the influence of the lived experience and collective memory of racism may be mitigated by a shared racial or cultural identity between provider and patient [26, 27]. Provider-patient racial concordance refers to a shared racial identity between a healthcare provider and a patient. Racial discordance refers to a patient and a healthcare provider having different racial identities [28]. While the evidence is mixed, some studies suggest that racial concordance between healthcare providers and patients, especially among racially minoritized people, is associated with positive health-related outcomes, including increased use of care, decreased expenditures, better communication, and decreased mortality [12, 28–31].

This retrospective cohort study aimed to assess the effect of provider-patient racial concordance on Black patients’ receipt of PPC, adjusting for other provider and patient characteristics. We hypothesized that patients who identified as Black who had a racially concordant provider during prenatal and intrapartum care would have greater odds of returning for PPC than Black patients whose providers were racially discordant.

Methods

Data Sources

We used a dataset drawn from electronic health records (EHRs) for childbirth hospitalizations from the University of Alabama at Birmingham Health System (UABHS). Within the UABHS’s flagship hospital, UAB Medicine, the Women and Infants’ Center WIC) in Birmingham provides comprehensive obstetric and gynecological care across the reproductive life span, serves as the obstetric referral hospital for the state, and accounts for 7 to 8% of births in Alabama. The EHRs for childbirth hospitalizations were identified through the health system’s Informatics for Integrated Biology and the Bedside (i2b2) platform and selected using methods developed for earlier studies [32, 33]. The general methods for patient database extraction are detailed in Tipre et al. [33].

The main maternity care provider for each childbirth hospitalization was determined based on the physician or nurse practitioner with the highest number of pregnancy-related encounters (that pertained to the pregnancy) with the patient during childbirth and over the ten months preceding it, as recorded in the EHR. We note that nurse practitioners provided prenatal care but did not attend childbirths, and certified nurse-midwives did not give any pregnancy or childbirth-related care during this period within the UABHS. We integrated information on the providers noted in the electronic health records (EHRs) with information on provider demographic characteristics drawn from the UABHS’s human resources (HR) database. We used the following procedures for identifying the main provider: The main provider was identified for the pregnant patient; the primary provider’s name was searched using the UABHS Human Resources database; demographic data (race/ethnicity, gender, year of birth) was identified and years of practice was estimated from the date of initial license to practice and date of delivery for the pregnant patient.

Finally, we merged data on postpartum visits into this dataset drawn from the University of Alabama Department of Obstetrics and Gynecology’s Maternity Electronic Database System (MATES), a specialized database for obstetric patients with information taken from UAB’s electronic data warehouse. The MATES information allowed us to identify whether a patient was seen by a maternity care provider within the UABHS within 12 weeks of their childbirth hospitalization. Ethical approval to conduct the study was provided by the University of Alabama at Birmingham Institutional Review Board.

The dataset included EHR data for 33,104 deidentified childbirth hospitalizations with live births at the WIC between January 1, 2010, and December 31, 2020. MATES became fully operational only in January 2014. Thus, we excluded delivery hospitalizations that occurred earlier than January 1, 2014. Because we assumed the COVID-19 pandemic would significantly decrease PPC access, we also excluded records after March 31, 2020 (n = 8,530). We then excluded 464 records for patients with residences outside of Alabama. Twenty-one patients included in this dataset died during delivery or postpartum within the UAB Health System. Within that group, we were able to identify nine birthing people who died before being discharged, whom we excluded from analyses of the PPC outcome for a final analysis dataset of 24,101 records. We retained the 12 birthing people who died postpartum based on a detailed record-by-record review of their dates of death and whether they attended a PPC or MEU visit.

Outcome Measures

The primary outcome variable was a composite measure identifying that a patient with a childbirth hospitalization had a PPC visit in the UABHS within 12 weeks after childbirth. The measure included a standard postpartum visit at one of the 11 obstetrics and gynecology clinics in the UABHS and/or a Maternity Evaluation Unit (MEU) visit within 12 weeks of childbirth. The Maternity Evaluation Unit is a special emergency unit in the WIC dedicated to the evaluation and triage of pregnant (16 weeks or more) and postpartum UABHS patients. A person entering the MEU with an urgent concern has direct access to obstetrics and gynecology medical teams and equipment. Integrating the standard PPC and MEU visits into one composite measure made it possible to ascertain if patients had any contact with the UABHS within 12 weeks after childbirth.

Study Populations

Patients

The following information was drawn from the EHRs for each childbirth hospitalization. We included patient demographic characteristics: self-identified race/ethnicity (Black, White, other), age in years (11–17, 18–25, 26–30, 31–35, 36–40, 41–45, 46–59), and insurance status (private/commercial, Medicaid/Medicare/uninsured). We also identified patient residence as rural or urban, based on whether the county where they resided was metro or non-metro [34]. We included characteristics that were related to health conditions or clinical details of the childbirth: presence of a chronic condition (diabetes or chronic hypertension), type of delivery (vaginal or cesarean section), or having a singleton or multiple gestation birth. We included the number of living children (none, one or two, or more). Finally, we included whether the patient experienced severe maternal morbidity or death during the 10 months up to and including childbirth. Severe maternal morbidity was defined as a childbirth hospitalization for which the EHR included one or more of the 21 procedure or diagnostic codes that the CDC identifies as related to an adverse maternal health event [35]. We also identified and included cases where the EHR contained information that the pregnant person died after discharge but within 1 year of that discharge for any reason.

Providers

We integrated characteristics drawn from the HR database of the main care providers during the pregnancy and childbirth periods. The information was drawn from EHR based on the names of the providers listed as physicians or nurse practitioners providing care during that encounter. This includes all outpatient and inpatient visits related to the pregnancy. Characteristics, including self-reported demographic information, included provider type (nurse-practitioner or physician), self-identified gender (male, female), age in years (21–30, 31–40, 41–50, 51–60, 61 or more), years of healthcare experience (1–4, 5–10, or 10 or more), and race (Black, White, other). The race categories drawn from the original HR database included Black, White, Hispanic, Asian, or multi-racial. However, due to very low counts in Hispanic, Asian, and multi-racial groups, we merged these categories into “other” to protect against the potential loss of anonymity as required by the IRB.

Statistical Analyses

Using the SPSS statistical package (Version 29), we developed descriptive statistics to summarize the characteristics of patients with delivery hospitalizations and their main providers by frequencies and proportions. Using chi-squared tests, we also compared distributions of these characteristics by receipt of different types of postpartum visits during the 12 weeks post-childbirth. We computed variables for the following combinations of visits: receiving MEU visit(s) only, receiving standard PPC visit(s) only, receiving both MEU and standard PPC visits, and receiving either a MEU or a PPC visit (any PPC). To examine the association between provider race and the composite outcome variable of any postpartum visit (either standard PPC or MEU visit) receipt and to account for the fact that some patients had more than one childbirth hospitalization during the period (n = 3897 or 16.2% of the cases), we fit a generalized estimating equation (GEE) model with an exchangeable matrix using the SAS statistical package (Version 9.4, Cary, NC) [36]. First, we computed unadjusted and adjusted odds ratios (OR) with 95% confidence intervals (CI) for main providers grouped as Black, White, or other and patient receipt of any postpartum visit, with White providers as the reference group. To examine the effect of provider-patient racial concordance on the outcome, we then computed separate models for patients of each race to obtain unadjusted and adjusted ORs with 95% CIs for provider-patient race pairs: Black provider-Black patient compared to the non-Black provider-Black patient and White provider-White patient compared to non-White provider-White patient. Because our research focus was understanding the potential effect of racial concordance and discordance on Black patients’ receipt of postpartum care, we made the non-Black provider and Black patient pair the reference group for this analysis. For comparison, we also analyzed the effect of racial concordance on White patients. For this analysis, the reference group was the non-White provider and White patient. For both analyses, we adjusted for covariates, including insurance, maternal age, number of living children, severe maternal morbidity or mortality, rural–urban residence, chronic conditions, mode of delivery, multiple gestation, provider years practiced, and provider type. Provider age and provider gender were highly correlated with years practiced, so we dropped these characteristics from the covariates for which we adjusted. Because visiting the MEU is not a substitute for receiving a comprehensive PPC visit, and might have different predictors, we also performed a sensitivity analysis excluding those with any postpartum MEU admission.

Results

Demographic and Other Characteristics of Alabamian Patients with Delivery Hospitalizations

Among the 24,101 delivery hospitalizations, the majority, or 92.3%, of birthing people lived in urban areas. The mean age of patients was 27.6 years (SD = 6.0) at delivery, with 65.3% of the delivery hospitalizations occurring among patients 18 to 30 years old. Most of the patients with these delivery hospitalizations identified as Black (47.3%) or White (33.0%), and 66.4% were insured through government programs such as Medicaid or were uninsured. At the time of each birth, 63.7% had one or more living children. Vaginal delivery was the most common mode of delivery at 64.4%, with 35.6% of patients experiencing a cesarean section. Among these patients, 2.9% had twins, triplets, or more. The prevalence of chronic hypertension and diabetes was 6.3% and 12.4%, respectively. Severe maternal morbidity was experienced before or during childbirth by 9.2% or 2221 patients. Overall, 13,281 (55.1%) of patients had either a standard PPC visit, an MEU visit, or both, and 44.9% had no standard or MEU visit. Twelve patients died after being discharged but within 365 days of their childbirth. All three patients who died within 12 weeks of their first day of childbirth hospitalization had one or more PPC (standard PPC or MEU) visits. Of the nine patients who died 13 or more weeks after the first day of their childbirth hospitalization, three received a standard postpartum visit, and six did not receive any type of postpartum visit within the UABHS. Results of bivariate comparisons of the entire study population showed statistically significant differences in the composite outcome measure (a receipt of MEU and/or standard PPC) by all patient factors examined, except for severe maternal morbidity and post-discharge mortality (Table 1).

Table 1.

Demographic and Maternal characteristics of study population^* January 2014 and March 2020 and receipt of maternity evaluation unit or standard postpartum visits^{^}

Demographic and maternal Characteristics	Total delivery hospitalizations		Standard PPC and/or MEU visit		No standard PPC or MEU visit		p-value#
	n	%	n	%	n	%
Total delivery hospitalizations (N)	24,101	100	13,281	55.1	10,820	44.9
Age categories at delivery (years)							< .001
11–17	590	2.5	260	44.1	330	55.9
18–25	8822	36.7	4249	48.2	4573	51.8
26–30	6892	28.7	3937	57.1	2955	42.9
31–35	5145	21.4	3247	63.1	1898	36.9
36–40	2163	9.0	1336	61.8	827	38.2
41–45	368	1.5	236	64.1	132	35.9
46–59	41	0.2	16	39.0	25	61.0
Insurance							< .001
Private/employer-based	8087	33.6	5550	68.6	2537	31.4
Medicaid/government/uninsured	16,014	66.4	7731	48.3	8283	51.7
Race							< .001
White	7958	33.4	4562	57.3	3396	42.7
Black	11,409	47.8	5869	51.4	5540	48.6
Latino/Hispanic	3738	15.7	2252	60.2	1486	39.8
Asian or Pacific Islander	589	2.5	371	63.0	218	37.0
American Indian/Alaskan Native	40	0.2	28	70.0	12	30.0
Multi-racial	119	0.5	65	54.6	54	45.4
Delivery type							< .001
Vaginal	15,514	64.4	8051	51.9	7463	48.1
C-section	8587	35.6	5230	60.9	3357	39.1
Living children at time of delivery							< .001
None	8323	36.4	5209	62.6	3114	37.4
One	6719	29.4	3924	58.4	2795	41.6
Two or more	7851	34.3	4126	52.6	3725	47.4
Single or multiple gestation							.006
Singleton	23,412	97.1	12,866	55.0	10,546	45.0
Multiple &Chronic conditions	689	2.9	415	60.2	274	39.8
Chronic hypertension	1511	6.3	932	61.7	579	38.3	< .001
$Diabetes	2984	12.4	1743	58.4	1241	41.6	< .001
Severe morbidity/mortality							.894
None	21,868	90.7	12,058	55.1	9810	44.9
Severe morbidity	2221	9.2	1217	54.8	1004	45.2
Post-discharge mortality	12	0.0	6	50.0	6	50.0
Rural/urban							< .001
Metro counties	22,240	92.3	12,534	56.4	9706	43.6
Non-metro counties	1855	7.7	745	40.2	1110	59.8

^*Study population is people with delivery hospitalizations at the UAB Women & Infants Center between January 2014 and March 2020 who reside in Alabama

^{^}Frequencies for PPC and MEU visits are row (and not column) percent values

^&Diabetes and chronic hypertension were analyzed separately and thus have separate chi-square p-values

^$Pre-gestational or gestational

^#Chi-square

Demographic Characteristics of Main Healthcare Providers Giving Care

Among the 275 healthcare providers of pregnancy and maternity care for these patients from January 1, 2014, to March 31, 2020, 93.1% were physicians, and 6.9% were nurse practitioners. The average age of the main provider cohort was 44.5 years (SD = 10.7), with equal numbers of providers (137) identifying as male and female. Most providers, 82.2% (226), identified as White, 8.7% (24) identified as Black, and 4% of providers were in the “other” group (Hispanic, Asian, or two or more races). These providers’ average healthcare work experience was 28.6 years (SD = 12), with 48.9% of them reporting more than 10 years of experience (Table 2).

Table 2.

Demographic characteristics of main providers^* giving care during pregnancy and childbirth for childbirth hospitalizations January 2014 to March 2020

Main providers characteristics	n	%
All providers	275	100
Provider type
Physician	256	93.1
Nurse practitioner	19	6.9
Gender
Male	137	49.8
Female	137	49.8
Race
White	226	82.2
Black	24	8.7
Other	11	4.0
Age
Mean	44.5 (SD = 10.7)
21–30	2	1.0
31–40	97	48.5
41–50	49	24.5
51–60	38	19.0
61 +	14	7.0
Experience, years categorized
Mean	28.6 (SD = 28.6)
1–2	2	0.9
3–4	12	5.3
5–10	101	44.9
10 +	110	48.9

^*Main provider is the maternity care provider noted most frequently in the EHR for each case for the ten months up to and including the delivery hospitalization

Among the patients experiencing these childbirth hospitalizations, 98.5% were cared for by a main provider who was a physician, and only 1.5% were cared for by a nurse practitioner. Among patients whose main provider was a physician, 55.2% received any PPC (standard PPC, MEU care, or both) versus 63.0% of patients whose main provider was a nurse practitioner. When comparing male and female providers, 52.4% of patients with male providers received any PPC, and 63.8% of patients with female providers received any PPC. Most patients were cared for by main providers aged 61 or older (62.1%) and those with ten or more years of experience (84.0%). Most patients were cared for by a main provider who identified as White (97.4%). Only 2.2% of patients were cared for by a main provider who identified as Black, and 0.5% by a provider in the “other” group. Among patients with a main provider who was White, 55.3% received any PPC compared to 74.2% of patients with a main provider who was Black. The relationships between all these provider factors and the composite outcome measure for PPC were statistically significant except for provider age (Table 3).

Table 3.

Relationships between postpartum visits and main maternity providers’^* characteristics in study population^& January 2014 to March 2020^{^}

Postpartum visits by provider characteristics	Total childbirth hospitalizations		Standard PPC and/or MEU visit		No standard PPC or MEU visit		p-Value#
	n	%	n	%	n	%
Type							.003
Physician	23,641	98.5	13,049	55.2	10,592	44.8
Nurse practitioner	359	1.5	226	63.0	133	37.0
Gender							< .001
Male	16,874	71.9	8843	52.4	8031	47.6
Female	6604	28.1	4211	63.8	2393	36.2
Race							< .001
White	22,851	97.4	12,631	55.3	10,220	44.7
Black	507	2.2	376	74.2	131	25.8
Other	107	0.5	42	39.3	65	60.7
Age							< .001
21–30	20	0.1	10	50.0	10	50
31–40	3419	14.6	1848	54.1	1571	45.9
41–50	2716	11.6	1705	62.8	1011	37.2
51–60	2290	9.8	1573	68.7	717	31.3
61 +	14,959	63.9	7887	52.7	7072	47.3
Years of experience							.149
1–4	112	0.5	55	49.1	57	50.9
5–10	3631	15.5	1984	54.6	1647	45.4
10 +	19,681	84.0	10,995	55.9	8686	44.1

^*Main provider is the maternity care provider noted most frequently in the EHR for each case for the ten months up to and including the delivery hospitalization. Percentages are based on cases with non-missing values for provider characteristics

^{^}Frequencies for PPC and MEU visits are row (and not column) percent values

^&Study population is women with delivery hospitalizations at the UAB Women & Infants Center between January 2014 and March 2020 who reside in Alabama

^#Chi-square

Analysis of Provider Race, Provider‑Patient Racial Concordance, and Patient Receipt of PPC

Results of the statistical analysis for the race of the main provider and patients of any race receiving at least one PPC visit (standard PPC, MEU visit, or both) are provided in Table 4. Compared to patients with main providers who were White, patients whose main providers were Black had more than twice the odds of receiving any PPC, even when adjusting for other characteristics, including patient race (crude OR 2.31, 95% CI 1.89–2.83; adjusted OR 2.02, 95% CI 1.64–2.49). Patients whose main providers identified as Asian, Latino/Hispanic, or two or more races (categorized together as “other”) had lower odds of receiving any PPC compared to patients with White providers (crude OR 0.51, 95% CI 0.34–0.75, adjusted OR 0.44, CI 95% 0.29–0.67).

Table 4.

Crude and adjusted odds ratio and 95% Confidence limits for relationships between the main maternity providers’ race and any postpartum visit

Providers’ race	Crude OR (95% CI)	Adj. OR (95% CI)*
White	1.00	1.00
Black	2.31 (1.89–2.83) **	2.02 (1.64–2.49)
Others†	0.51 (0.34–0.75)	0.44 (0.29–0.67)

N = 22,049 for adjusted analysis

^†Others include providers who identify as Hispanic, Asian, and multiracial

^*Adjusted for the patient race, insurance, maternal age, number of living children, severe maternal morbidity or mortality, rural–urban, chronic conditions (CHTN or DM), mode of delivery, multiple gestations, provider years practiced, and provider type

^**Font bolded to indicate statistical significance, p < .05

For the second analysis related to the effect of racial concordance on Black patients’ receipt of PPC, we calculated the crude and adjusted odds of having PPC for Black patients with Black providers (concordant), compared to Black patients with a non-Black provider (discordant, reference group). We also analyzed the effect of racial concordance on White patients as a comparison. For White patients, in a separate model, the reference group was White patients with a non-White provider, which we compared to White patients with a White provider. Our analysis revealed significant associations between patient-provider racial concordance and the likelihood of receiving PPC. Black patients with Black providers had significantly higher odds of receiving PPC compared to Black patients with non-Black providers (adjusted OR 2.26, 95% CI 1.65–3.09, p < 0.05). This indicates that Black patients were more likely to receive PPC when their provider was Black. For White patients, racial concordance with their provider did not significantly influence the odds of receiving PPC after adjustment for other factors. While the crude odds ratio showed a significant association (OR 0.76, 95% CI 0.59–0.98, p < 0.05) for this latter relationship, the effect was attenuated and became non-significant after adjustment (adjusted OR 0.88, 95% CI 0.68–1.14, p < 0.05). The analysis adjusted for factors including insurance status, maternal age, number of living children, severe maternal morbidity or mortality, rural–urban classification, chronic conditions, mode of delivery, multiple gestations, provider years of practice, and provider type. See Table 5. Results from a sensitivity analysis which examines returning for standard PPC excluding patients who visited the MEU can be found in the Supplementary Materials. In this analysis, the overall percent of patients receiving a PPC visit was lower (49.7% versus 55.1%). Single versus multiple gestation was no longer a significant predictor and severe maternal morbidity/mortality became a significant predictor. Other results were unchanged in terms of significance and direction.

Table 5.

Crude and adjusted odds ratio and 95% confidence limits for relationships between primary provider*patient pairs by race and any postpartum visit

Patient-provider race concordance	Received PPC (n%) (n = 12,878)	No standard PPC or MEU visit (n%) (n = 10,247)	Crude OR (95% CI)	Adj. OR (95% CI)*
Provider non-Black*patient Black	5601 (97.1)	5317 (98.9)	1.00	1.00
Provider Black*patient Black	168 (2.9)	62 (1.2)	2.55 (1.90–3.43)	2.26 (1.65–3.09)
Provider non-White*patient White	4273 (96.2)	3097 (97.1)	1.00	1.00
Provider White*patient White	170 (3.8)	93 (2.9)	0.76 (0.59–0.98)	0.88 (0.68–1.14)

N = 22,049 for adjusted analysis

^*Adjusted for insurance, maternal age, number of living children, severe maternal mortality or morbidity, rural–urban, chronic conditions (CHTN or DM), mode of delivery, multiple gestations, provider years practiced, and provider type

^**Font bolded to indicate statistical significance, p < .05

Discussion

We analyzed an electronic health record dataset of more than 24,000 childbirth hospitalizations from January 1, 2014, to March 31, 2020, at UABHS in Birmingham, Alabama, that included demographic information about each patient and each patient’s main maternity care provider. Overall, receipt of PPC within the UABHS was low: only slightly more than half of the study population was seen by a PPC provider within 12 weeks of childbirth While the percentage of providers who identified as Black was small, we found higher odds of patients of any race attending PPC if their main healthcare provider was Black. We recognize that, because patients of any race had higher odds of returning for postpartum care when they had a Black provider, it may be the quality of care provided by these Black providers in general—not racial concordance with the patient—that is driving our higher odds of PPC for Black patients who had Black providers. We found that Black patients whose main provider was Black had more than twice the odds of receiving any PPC compared to Black patients whose main provider was not Black. We did not find a statistically significant effect of racial concordance on the odds that White patients would return for care after adjusting for covariates.

Our study confirms findings from prior studies using administrative data such as electronic health records or insurance claims that have reported low rates of postpartum visit attendance [7]. This is particularly concerning given the high rate of severe morbidities (9.2%) and chronic conditions such as diabetes (12.4%) among this study population and the fact that many patients may have been high-risk patients, as the UABHS WIC is the main obstetric referral hospital in Alabama. We also found that among the 12 birthing people who died, nine died 13 or more weeks after their childbirth hospitalization, and all three people who died 12 or fewer weeks after their childbirth hospitalization had been seen for PPC (either a standard postpartum or MEU visit) in the UABHS. These data reinforce ACOG’s guidance that PPC is not a one-off visit but a process with ongoing care as needed and a dedicated transition to primary care for everyone [6].

While significant research examines patient factors and their relationship to attending postpartum care visits, relatively little evidence exists about the relationship between provider factors and receipt of postpartum care [7, 10]. Different studies have identified patient race as a predictor of attending postpartum care [9, 37, 38]. But to our knowledge, this is the first study to explore an association between provider-patient racial concordance and receipt of PPC among birthing people who identify as Black, which allows for a discussion about how perceived racial discrimination and disrespect may play a role in this relationship.

Given the small number of Black providers caring for this patient cohort, our results should be interpreted cautiously, especially since the evidence is mixed on the relationship between patient-provider concordance and health outcomes. However, there is a scientific premise for the contention that having a racially concordant provider, especially if you are from a historically marginalized group, during the pregnancy, birth, and postpartum journey, may be beneficial. Racial concordance can provide a sense of shared identity, aligned values, and less social distance, factors that may be important in settings like hospitals with histories of institutional and structural racism [14, 27, 39, 40]. The experience of poor-quality communication during racially discordant prior care may also be a factor. In Shen et al.’s [28] systematic review of patient-provider racial concordance and communication, a majority of studies found that Black patients, compared to White patients, experienced worse communication with their providers, including less participatory decision-making and information-giving, and that racial concordance was associated with better communication across a range of communication domains. In qualitative studies, Black and Latinx people describe childbirth experiences that included communication gaps, racial-ethnic discrimination, racism, provider bias, providers withholding information, barriers to building trusting relationships, and inconsistent social support [5, 41–43].

Against the backdrop of the unjust burden of preventable pregnancy-related death among Black people, understanding and addressing structural barriers that block access to lifesaving and life-affirming postpartum care is critical. Our findings point to policy and clinical strategies for improvement. The low number of Black healthcare providers in a healthcare system with a high proportion of Black patients is striking. The obstetrics and gynecology physician workforce in the US is more diverse than other physician specialties [44]. However, decades of structural racism mean that diversity in obstetrics and gynecology residencies still lags behind the diversity of the US population [44]. Diversifying the maternity and pregnancy care workforce, which includes not only physicians but also nurse practitioners, nurses, nurse-midwives, and other birth workers such as doulas, can increase racial concordance, improve trust, and increase satisfaction among Black birthing people [39, 45, 46].

While complex, the mechanisms by which provider-patient racial discordance may affect access to care are modifiable. Provider communication skills can improve, and patient perceptions of shared identity and values can shift. Care paradigms such as the reproductive justice framework can help providers build empathy for the racialized experiences of Black birthing people and center them as experts in their bodies and choices [47]. Patient and relationship-centered models and anti-bias training can build providers’ communication skills [27, 48–51]. These types of provider training must also build critical consciousness of how racism (interpersonal and institutional) impacts the health of their patients. Poor provider communication, including tacit “mother blame narratives” (assuming that individual characteristics or choices of the mother are the reason for adverse maternal outcomes), must be understood as sustaining and reproducing institutional racism [39, 52, 53]. Measurement tools that assess patient experiences of racism, such as Prem-OB Scale^™ [54, 55] and frameworks, such as the “Presence 5 for Racial Justice Framework,” can assist providers to strengthen their communication practices [56]. Large academic medical centers such as UAB Medicine that teach and train large numbers of physicians, nurses, and other healthcare providers can be at the forefront of the efforts to diversify the healthcare workforce and to pilot, test, and scale innovative practices and tools that dismantle institutional racism.

Given our findings on overall low postpartum visit attendance and the relationship between racial concordance and receiving care, future research should also focus on removing barriers to high-quality care for Black birthing people. Interventions that improve both the experience and accessibility of postpartum care must be tested and scaled, such as incorporating community support persons and doula care across the pregnancy-childbirth continuum, with a focus on support people and doulas that represent the communities they serve [57, 58]. One of our team’s primary projects, P³OPPY, is investigating whether digital health and/or community health worker support can improve perinatal patient outcomes. Research should also include testing ways to scale evidence-based strategies to bring care to people and their families, such as the nurse-family partnership [59]. Black birthing communities must lead the definition of respectful, welcoming, and explicitly anti-racist pregnancy and childbirth care [60]. Communities and researchers could also partner in US regions other than the Deep South to investigate whether the relationship between patient-provider racial concordance and postpartum care exists in those settings. We found the intriguing relationship between having a Black main provider and increased receipt of PPC by patients of any race warrants further exploration. Finally, studies with higher percentages of Black providers are necessary.

There are several limitations to our study. We cannot infer causal effects, only associations, as our study was a retrospective cross-sectional analysis. Since Black providers represented only 8.7% of the total providers, we interpret our results cautiously. Only 2.2% of the patients were seen by Black providers and 0.5% by “other race” providers during this time period in the UABHS. If there is any misclassification (random) in how we have assigned a provider to a patient, it may have led to bias. Given these small numbers of Black providers, this may have affected the significance of the results. We note that patients of any race had higher odds of returning for postpartum care when they had a Black provider, and thus there may be unmeasured variables other than racial concordance driving Black patients’ odds of returning for PPC. We also faced limitations in how we defined the main provider, which was based on the highest number of encounters between provider and patient noted in the EHR database. Because these data were drawn from an EHR database in a large teaching hospital, there may be errors in the notation of which provider gave care most frequently. For example, more senior attending physicians may be noted in the EHR, whereas resident physicians provided birth or prenatal care, or nurse practitioners provided prenatal care. We also analyzed childbirth hospitalizations that occurred within the first three months of 2020 and the incipient COVID-19 pandemic may have affected the number of people returning for postpartum care. Further, we created a composite indicator by combining standard postpartum care visits with MEU visits for our outcome of interest because we felt that any contact with a provider was important for this group of patients. However, we recognize that combining two different types of care may have decreased the sensitivity of our analyses. We only captured returns for PPC within the UABHS. Although patients are encouraged to come back to this system for PPC after childbirth, they may have sought PPC outside of this system, particularly those patients who resided in rural areas. This dataset is also not a state-wide representative sample: it contains fewer than 10% of the total births per year in Alabama, which had 57,643 live births in 2020 [61]. Because the UABHS WIC is a referral hospital for the state, with higher risk patients and complicated cases, the results of our analyses are not representative of Alabamian birthing people as a group. Finally, the percentage of people attending postpartum care in our analysis is low (55.1%) compared to other measures of postpartum care such as national PRAMS data (89.4%) so our findings may not be generalizable to people outside of Alabama[62].

Conclusion

In a cross-sectional analysis of electronic health records of more than 24,000 births in Alabama, we found that for Black birthing people, having a Black provider compared to having a White provider resulted in approximately twice the odds of receiving postpartum care. Given the stark racial inequities in perinatal outcomes in the US, these findings underscore the call to design and test care models that create welcoming, inclusive, and explicitly anti-racist perinatal services and break down the unjust systemic and structural barriers that Black birthing people and other racially minoritized communities experience.