If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Correspondence to: Roxanne Mirabal-Beltran, PhD, Georgetown University School of Nursing & Health Studies, 3700 Reservoir Road, NW, Washington, DC 20057. Phone: 202-687-3496.
Despite a lower percentage of primary cesareans than non-Hispanic White and Black women, Hispanic women in the United States had the highest rate of repeat cesarean deliveries (RCD) in 2016; it is unclear if reasons for differences are due to known risk factors. Our study examined the association between ethnicity/race and RCD among women with one previous cesarean and whether demographic (age, marital status, education, language, and delivery year), anthropomorphic (height, prepregnancy body mass index), obstetrical/medical (parity, gestational age, infant birth weight, gestational diabetes, labor induction or augmentation, vaginal birth after cesarean delivery history), or health system (delivery day/time, payer source, provider gender) factors accounted for any observed differences by ethnicity/race.
Methods
Our retrospective cohort study used logistic regression to evaluate the relationship between ethnicity/race and RCD based on data from electronic delivery and prenatal records from 2010 to 2016, including 1800 births to Hispanic and non-Hispanic women with one previous cesarean at a District of Columbia hospital.
Results
Statistically significant differences by ethnicity/race were noted after adjustment for obstetric/medical factors, particularly parity and use of induction or augmentation methods. Hispanic (adjusted odds ratio, 2.48; 95% confidence interval, 1.03–6.01) and Black women (adjusted odds ratio, 2.83; 95% confidence interval, 1.67–4.81) had higher odds of RCD than White women.
Conclusions
Adjustment for parity and use of induction or augmentation methods revealed higher odds of RCD for Hispanic and Black women than White women. Demographic and anthropometric factors did not alter these results. Our work is a first step in creating effective public health policy and programs that target potentially preventable RCD by highlighting the need to evaluate risk factors beyond those included in the literature to date.
Repeat cesarean deliveries (RCDs) account for one-third of all cesarean deliveries (
). Despite a lower percentage of primary cesarean delivery than non-Hispanic Black and White women, Hispanic women in the United States had the highest rate of RCD in 2016 (
An examination of cesarean and vaginal birth histories among Hispanic women entering prenatal care in two California counties with large immigrant populations.
Journal of Immigrant and Minority Health/Center for Minority Public Health.2012; 14: 209-214
). The percentage of vaginal birth after cesarean (VBAC) was lower among Hispanic women than non-Hispanic Black or White women based on 2016 U.S. data (11.5 vs. 12.4 and 12.8 per 100 live births, respectively) (
Although studies of ethnic differences in RCD show variation in the risk of cesarean, the direction and magnitude of this risk has been inconsistent across studies, in part owing to limited adjustment for factors related to cesarean delivery and variation in data sources and study designs (
). There also is limited empirical evidence about the reasons for differences, particularly if they are due to demographic, anthropomorphic, obstetrical/medical, or health system risk factors. Studies have also excluded or underrepresented U.S. Hispanic women with previous cesarean births, resulting in limited understanding about their risk for RCD (
). The current state of the literature on this topic requires rigorous examination of a variety of factors to understand potential interventions for this population.
The current study used electronic medical records data to examine the association between ethnicity/race and RCD among women with one previous cesarean and to evaluate whether known risk factors accounted for any observed differences. Assessing these associations, particularly related to modifiable risk factors, is important in understanding possible maternal health disparities among the growing population of U.S. women of Hispanic origin who may be disproportionately affected by RCD.
Methods
This retrospective epidemiologic cohort study was based on electronic medical data for women with one previous cesarean birth who delivered at Medstar Washington Hospital Center (Hospital Center) in Washington, DC between January 1, 2010 and October 31, 2016. The Hospital Center is a public, urban, Neonatal Intensive Care level III nursery, 912-bed general medical and surgical teaching hospital with an in-house staff of neonatologists and high-risk practitioners, including maternal–fetal medicine physicians, serving a diverse patient population. It also meets the
recommendation for providing safe VBAC: having immediate availability to surgeons and anesthesiologist. The Hospital Center uses an electronic obstetrics database and links prenatal records with delivery records, allowing for standardized electronic medical data. In 2016, 3,564 births occurred at the Hospital Center.
A total of 21,587 births occurred at the Hospital Center during the study period. The study sample was limited, however, to women over 18 years of age with births at over 37 weeks gestation who had a history of a prior cesarean delivery (Figure 1). Births of multiples, with fetal demise, or that took place outside of the hospital were excluded. Only the first birth at the Hospital Center after the primary cesarean was included for women who had more than one Hospital Center delivery during the study period. These initial inclusion/exclusion criteria resulted in 2,787 births for inclusion. Delivery and prenatal electronic medical records for these births were then linked using an optimized deterministic linking algorithm with three criteria (medical record number, maternal date of birth, and delivery date) (
Among the 2,787 linked birth records, those with missing information about previous cesarean history and ethnicity/race were individually reviewed to obtain this information. Additionally, delivery records with ethnicity/race reported as unknown, declined, do not use, other, two or more races, multiracial, or Asian were individually reviewed to confirm ethnicity/race. Information on previous delivery history was complete after individual review of records, but 64 women were excluded from final analysis owing to missing information on ethnicity/race. The study sample was further restricted to Hispanic women of all races and non-Hispanic Black and White women with one prior cesarean, resulting in a sample for analysis of 1,800 women. Data were de-identified for analysis.
The dependent variable was dichotomous, whether a woman had an RCD after a primary cesarean birth. The primary independent variable was ethnicity/race as reported in the medical record with mutually exclusive categories for Hispanic, non-Hispanic Black, and non-Hispanic White women. Covariates were categorized as demographic, anthropomorphic, obstetrical/medical, and health system factors and defined as categorical or dichotomous for analysis.
Some covariates require further clarification. For marital status, divorced or separated women were included in the single category, because they made up less than 1% of the sample. Elementary maternal education was defined as less than 8 years of education and secondary education as 9–12 years or high school diploma. Maternal languages other than English or Spanish included French and Amharic. Only births after 2010 were sampled to account for American College of Obstetricians and Gynecologists (2010) guidelines for VBAC.
Determination of gestational age was based on best obstetric estimate. The medical record did not distinguish between insulin and non–insulin-dependent diabetes; diabetes was coded as present or absent. Preeclampsia and hypertensive disorders were not included owing to questionable validity; only 0.8% (n = 14) and 1.3% (n = 24) of the sample had documented preeclampsia and hypertension, respectively. It also was not possible to obtain reliable data from the medical records on whether a trial of labor was attempted before an RCD, but prior delivery history was extracted to determine if a woman experienced a previous VBAC.
The year, time of day, and day of the week of the delivery were included in the analysis to adjust for possible temporal trends in delivery preferences. Insurance type was identified as public or private; self-pay/uninsured (8.8%) was included in the public category.
We first conducted a descriptive analysis of the overall sample characteristics, and then analyses for Hispanic, non-Hispanic Black, and non-Hispanic White women separately. A χ2 test was used to compare RCD rates across ethnicity/race groups. Multivariable logistic regression models estimated odds ratios (ORs) with 95% confidence intervals (CIs) for the association of ethnicity/race with RCD. Variables were chosen for the multivariate model based on findings from other studies, results of the exploratory data analysis, and analyses of unadjusted odds for each variable. Covariates were included in models if they significantly changed the odds ratio for ethnicity/race by 10% or more or if they had an independent effect on the odds of RCD.
Missing values for the final sample ranged from 0.1% (insurance) to 9.9% (education) for eight variables: marital status, language, insurance, maternal height, prepregnancy weight, weight at admission, infant birth weight, and education. Multiple imputation by chained equations was conducted with a total of 35 imputations generated for the variables with missing data (
). Variables included in the imputations model were determined after correlation analysis and independent means testing (Wilcoxon Mann-Whitney and t tests) resulted in no significant differences in ethnicity/race and mode of delivery between women with and without missing information.
Sensitivity analyses were performed for obstetrical and medical factors to assess their effect on the ethnicity/race estimates. Two separate analyses also were conducted to address the small cell size for the language covariate because this covariate was of interest to the primary independent variable of ethnicity/race. STATA 14.2 software (
) was used for all analyses, with a p value of .05 to indicate statistical significance.
The Johns Hopkins Bloomberg School of Public Health Institutional Review Board reviewed and approved the study protocol and an Institutional Review Board Authorization Agreement was approved by the Medstar Research Institute.
Results
Table 1 describes the characteristics of the final study sample of 1,800 women after exclusion criteria were applied. Most women in the sample were non-Hispanic Black women (74.6%) and Hispanic women (18.7%). Hispanic and non-Hispanic Black women had similar distributions of age, marital status, parity, and insurance. Hispanic women and non-Hispanic Black women were younger, more likely to be single, less educated, of higher parity, more likely to undergo induction of labor and be on public insurance than non-Hispanic White women. Hispanic women were generally shorter in stature than non-Hispanic White and Black women.
Table 1Characteristics of the Sample (January 1, 2010 through October 31, 2016)
Characteristic
Distribution, n (%)
Hispanic (n = 337; 18.7%)
Non-Hispanic Black (n = 1,342; 74.6%)
Non-Hispanic White (n = 121; 6.7%)
Total (N = 1,800; 100.0%)
Demographic factors
Maternal age, y
Mean
30.9
29.5
33.6
30.0
18–24
47 (14.0)
317 (23.6)
4 (3.3)
368 (20.4)
25–29
90 (26.7)
380 (28.3)
18 (14.9)
488 (27.1)
30–34
107 (31.8)
357 (26.6)
48 (39.7)
512 (28.4)
35–55
93 (27.6)
288 (21.5)
51 (42.2)
432 (24.0)
Marital status
Single
234 (69.6)
1,009 (75.3)
37 (30.6)
1,280 (71.2)
Married
102 (30.4)
331 (24.7)
84 (69.4)
517 (28.8)
Maternal education
Elementary
113 (37.3)
57 (4.7)
0 (0.0)
170 (10.5)
Secondary
155 (51.2)
743 (61.3)
24 (22.4)
922 (56.8)
College or above
35 (11.6)
413 (34.1)
83 (77.6)
531 (32.7)
Maternal language
English
54 (16.6)
1,287 (96.3)
120 (100.0)
1,461 (81.9)
Spanish
272 (83.4)
0 (0.0)
0 (0.0)
272 (15.3)
Other
0 (0.0)
50 (3.7)
0 (0.0)
50 (2.8)
Year of delivery
2010
65 (19.3)
294 (21.9)
20 (16.5)
379 (21.0)
2011
58 (17.2)
239 (17.8)
24 (19.8)
321 (17.8)
2012
45 (13.4)
206 (15.4)
12 (9.9)
263 (14.6)
2013
59 (17.5)
193 (14.4)
10 (8.3)
262 (14.6)
2014
29 (8.6)
151 (11.3)
18 (14.9)
198 (11.0)
2015
42 (12.5)
131 (9.8)
16 (13.2)
189 (10.5)
2016
39 (11.6)
128 (9.5)
21 (17.4)
188 (10.4)
Anthropomorphic factors
Maternal height, cm
<145.0
10 (3.1)
10 (0.8)
1 (0.8)
21 (1.2)
145.0–149.9
33 (10.1)
46 (3.4)
2 (1.7)
81 (4.5)
150.0–154.9
91 (27.9)
159 (11.9)
18 (15.0)
268 (15.0)
155.0–160.0
99 (30.4)
321 (24.0)
20 (16.7)
440 (24.7)
>160.0
93 (28.5)
802 (60.0)
79 (65.8)
974 (54.6)
Prepregnancy BMI, kg/m2
Mean
28.0
30.5
26.1
29.7
<25.0
106 (33.8)
341 (27.0)
58 (49.6)
505 (29.8)
25.0–29.9
118 (37.6)
355 (28.1)
38 (32.5)
511 (30.2)
≥30.0
90 (28.7)
566 (44.9)
21 (18.0)
677 (40.0)
Obstetrical/medical factors
Parity
2
218 (64.7)
969 (72.2)
115 (95.0)
1,302 (72.3)
3
72 (21.4)
201 (15.0)
3 (2.5)
276 (15.3)
≥4
47 (14.0)
172 (12.8)
3 (2.5)
222 (12.3)
Gestational age at delivery, wk
37.0–40.0
281 (83.4)
1,114 (83.0)
95 (78.5)
1,490 (82.8)
40.1–41.0
51 (15.1)
172 (12.8)
16 (13.2)
239 (13.3)
>41.0
5 (1.5)
56 (4.2)
10 (8.3)
71 (3.9)
Infant birth weight, g
Mean
3,421.4
3,345.0
3,497.5
3,369.5
Range
2,248–4,911
1,614–5,741
2,255–4,834
1,614–5,741
<3.499
197 (59.5)
838 (63.8)
64 (54.2)
1,099 (62.4)
3,500–4,000
101 (30.5)
354 (27.0)
35 (29.7)
490 (27.8)
>4,000
33 (10.0)
121 (9.2)
19 (16.1)
173 (9.8)
Gestational diabetes
No
294 (87.2)
1,264 (94.2)
118 (97.5)
1,676 (93.1)
Yes
43 (12.8)
78 (5.8)
3 (2.5)
124 (6.9)
Pregestational diabetes
No
331 (98.2)
1,305 (97.2)
121 (100.0)
1,757 (97.6)
Yes
6 (1.8)
37 (2.8)
0 (0.0)
43 (2.4)
Labor induction
No
305 (90.5)
1,210 (90.2)
120 (99.2)
1,635 (90.8)
Yes
32 (9.5)
132 (9.8)
1 (0.8)
165 (9.2)
Labor augmentation
No
281 (83.4)
1,155 (86.1)
107 (88.4)
1,543 (85.7)
Yes
56 (16.6)
187 (13.9)
14 (11.6)
257 (14.3)
Previous VBAC
No
287 (85.2)
1,178 (87.8)
118 (97.5)
1,583 (87.9)
Yes
50 (14.8)
164 (12.2)
3 (2.5)
217 (12.1)
Health system factors
Delivery day of the week
Weekday
264 (78.3)
1,120 (83.5)
105 (86.8)
1,489 (82.7)
Weekend
73 (21.7)
222 (16.5)
16 (13.2)
311 (17.3)
Time of delivery
Day shift (6 am–5:59 pm)
206 (61.1)
876 (65.3)
84 (69.4)
1,166 (64.8)
Night shift (6 pm–5:59 am)
131 (38.9)
466 (34.7)
37 (30.6)
634 (35.2)
Payer source
Public
249 (73.9)
867 (64.7)
21 (17.4)
1,137 (63.2)
Private
88 (26.1)
474 (35.4)
100 (82.6)
662 (36.8)
Provider gender
Female
241 (71.5)
898 (66.9)
86 (71.1)
1,225 (68.1)
Male
96 (28.5)
444 (33.1)
35 (28.9)
575 (31.9)
Abbreviations: BMI, body mass index; VBAC, vaginal birth after cesarean.
Division of Vital Statistics (DVS)U.S. Department of Health and Human Services (USDHHS)Centers for Disease Control and Prevention (CDC)National Center for Health Statistics (NCHS) Natality public use natality file. 2015 update.
), comparable to the VBAC rate among women with a prior cesarean at the Hospital Center in 2013 of 25.2%. The unadjusted RCD rates differed significantly by ethnicity/race (p < .001) (Table 2); rates were lowest for Hispanic women. Educational attainment and maternal language were also significantly associated with RCD (p < .001). Women with more than 12 years of education had a higher RCD rate than those with secondary education (78.5% vs. 72.2%), and those who spoke English had a higher rate than those who spoke Spanish (75.9% vs. 60.7%). Only prepregnancy body mass index was significantly related to RCD (p < .001) among the anthropomorphic factors; women with a prepregnancy body mass index of 30.0 or above had higher rates of RCD.
Table 2Repeat Cesarean Delivery Rates (January 1, 2010, through October 31, 2016)
Among the obstetrical/medical factors, parity, gestational age, and labor induction or augmentation were significantly associated with RCD (p < .001). Women whose current birth was their second had the highest RCD rate and women with parities of four or more the lowest RCD. Women who delivered at 41.1 weeks gestation or more had the lowest RCD rate. Infant birth weight was also significantly associated with RCD. Among women with birth weights greater than 4,000 grams, Hispanic women had higher rates of RCD (87.9%) than non-Hispanic White (57.9%) women; non-Hispanic Black women had similar rates to Hispanic women (85.1%). Women who underwent labor induction or augmentation or had a history of a previous VBAC had lower rates of RCD than those who did not. All health systems factors were statistically significant (p < .001). Women who had a birth during the weekend were significantly less likely to deliver by RCD. Publicly insured Hispanic women had the lowest RCD rates (58.6%) and privately insured non-Hispanic Black women the highest (82.9%).
Unadjusted odd ratios (ORs) and adjusted ORs (AORs) for RCD are reported in Table 3. Hispanic women had lower unadjusted odds of an RCD than non-Hispanic White women (unadjusted OR, 0.74; 95% confidence interval [CI], 0.47–1.15) whereas non-Hispanic Black women had higher unadjusted odds of RCD than non-Hispanic White women (unadjusted OR, 1.35; 95% CI, 0.90, 2.04). Neither difference was statistically significant in the unadjusted model (p > .15). Hispanic women, however, had more than twice the odds of RCD than non-Hispanic White women (AOR, 2.48; 95% CI, 1.03–6.01; p < .05) after adjusting for all factors in the full regression model, as did non-Hispanic Black women (AOR, 2.83; 95% CI, 1.67–4.81; p < .001) in the full model. Women aged 25 and older, with prepregnancy body mass indexes of greater than 30.0, or with gestational diabetes had significantly increased odds of RCD. Birth weight of more than 4,000 grams and provider gender (male) also were associated with increased odds of RCD. Married women, Spanish-speaking women, women taller than 160.0 cm, and women with higher parities had decreased odds of RCD. Other factors significantly associated with lower odds of RCD were use of induction and augmentation methods, a history of a previous VBAC, and delivering on the weekend or during the night shift.
Table 3Unadjusted and Adjusted Odds Ratios of Repeat Cesarean Delivery (N = 1,800) Using Multivariable Logistic Regression
The logistic regression results revealed that the odds of RCD were significant for Hispanic and non-Hispanic Black women only when obstetrical/medical factors were added to the model. Analyses were conducted to examine the impact of individual obstetrical and medical factors on the adjusted odds for ethnicity/race (Table 4). Addition of gestational age, infant birth weight, and presence of gestational diabetes improved the fit of the model, but resulted in little change in the odds for Hispanic or non-Hispanic Black women; the change in odds was less than 5%. Adjustment for parity and induction or augmentation methods, in contrast, resulted in increased odds of RCD for both Hispanic and non-Hispanic Black women, regardless of adjustment for other variables. The effect estimate was similar even after adjustment for the presence of a previous VBAC. Adjustment for both induction or augmentation methods and history of a previous VBAC also resulted in significantly increased odds of RCD for both Hispanic (AOR, 2.49; 95% CI, 1.04–5.99; p < .05) and non-Hispanic Black women (AOR, 2.46; 95% CI, 1.47–4.11; p < .001).
Covariates included maternal age, marital status, maternal education, maternal language, year of delivery, maternal height, maternal prepregnancy body mass index, day of the week of the delivery, time of delivery, payer source, and delivering provider gender.
∗ Covariates included maternal age, marital status, maternal education, maternal language, year of delivery, maternal height, maternal prepregnancy body mass index, day of the week of the delivery, time of delivery, payer source, and delivering provider gender.
Language also had a significant effect on the association between ethnicity/race and odds of RCD; sensitivity analysis was conducted to examine this effect (Table 5). The odds for Spanish-speaking Hispanic women did not differ significantly from those for English-speaking non-Hispanic White women when the sample was restricted to women who only spoke English or Spanish (p = .771). The odds of an RCD were significant only for English-speaking Hispanic women (AOR, 2.77; 95% CI, 1.12–6.85; p < .05) and non-Hispanic Black women (AOR, 2.79; 95% CI, 1.64–4.74; p < .001) when compared with non-Hispanic White women in the restricted sample.
Table 5Sensitivity Analysis, Language (n = 1,733), Using Multivariable Logistic Regression
We examined the association between ethnicity/race and RCD and whether demographic, anthropomorphic, obstetrical/medical, and health system factors explained observed differences in mode of delivery among 1,800 Hispanic and non-Hispanic Black and White women who delivered at a large hospital in the District of Columbia with an RCD rate lower than the national average (73.4% vs. 89.5%, respectively) (
Division of Vital Statistics (DVS)U.S. Department of Health and Human Services (USDHHS)Centers for Disease Control and Prevention (CDC)National Center for Health Statistics (NCHS) Natality public use natality file. 2015 update.
). Findings indicate that Hispanic and non-Hispanic Black women had higher odds of RCD than non-Hispanic White women after adjusting for demographic, anthropomorphic, obstetrical/medical, and health system factors. These findings are consistent with studies evaluating ethnic/racial differences in cesarean delivery using samples of women that included Hispanic women with prior cesareans (
). AORs in these studies were lower and estimates more precise than for our study, likely owing to our smaller sample size and differences in exclusion criteria. Sensitivity analysis suggests that the variables of parity and induction or augmentation of labor were of significance in explaining the higher AOR for Hispanic and non-Hispanic Black women.
Hispanic and non-Hispanic Black women in our sample were more likely to have parities of greater than two, which was associated with lower odds of RCD. Having one prior cesarean and no vaginal births (parity of two) was an important covariate in our final model. This finding is likely related to limiting our sample to women with one previous cesarean birth but, nevertheless, provides some insight about the role of a primary cesarean birth and differences by ethnicity/race in RCD. Women with parities of greater than two had experienced a previous vaginal delivery or VBAC, both of which may result in a better chance of a successful trial of labor (
, however, indicated no differences in trial of labor rates between Hispanic and non-Hispanic women, although Hispanic and non-Hispanic Black women had significantly greater odds of a failed trial of labor than non-Hispanic White women; the greater odds of RCD are consistent with our findings.
In our study, Hispanic and non-Hispanic Black women were more likely to experience induction or augmentation of labor than non-Hispanic White women, a factor noted to be protective against RCD. A systematic review and meta-analysis of randomized controlled trials of induction of labor versus expectant management identified 157 eligible studies (
.). Results showed that the risk of cesarean after an induction of labor was significantly lower than the risk following expectant management. Studies from 1975 to 2010 were included in the review but study sites were not specified; variability in recommendations over time and practice settings are important in interpreting these results. However, the role of induction or augmentation in mediating the risk of cesareans remains unclear (
Factors explored in this study did not explain why Hispanic and non-Hispanic Black women had increased odds of RCD after adjustment for covariates. In fact, adjusting for parity and the use of induction and augmentation methods actually revealed differences in the odds for these two groups of women relative to non-Hispanic White women. Provider bias has been proposed as a factor that may directly and indirectly influence medical decisions. There is evidence in the literature to suggest that increases in cesarean births are associated with more subjective indications for these births, such as labor arrest and nonreassuring fetal heart tracings (
). Variation in labor practices and management related to use of induction and augmentation methods may be associated with provider perceptions and a lower threshold for intervention. Consideration of provider bias in medical decision-making is a growing focus of research with unclear results (
). We were unable to explore provider bias or cesarean indications in this study.
English-speaking Hispanic women had significantly higher odds of RCD than Spanish-speaking Hispanic women. Our finding supports the importance of considering within group differences when categorizing women by ethnicity/race (
). Reasons for this finding after adjustment for parity and induction or augmentation methods are not clear, but may be related to RCD indications, women's preferences, or patient–provider communication. Additional work is needed, particularly in the area of provider communication about mode of delivery after a previous cesarean with Hispanic women living in the United States.
Strengths and Limitations
Several strengths of the study addressed limitations of prior research. The single-site cohort study design is a strength. By limiting our study to one urban hospital that offered a trial of labor for women with one prior cesarean birth, we controlled for hospital type, state/district liability differences, and institutional policies—all factors that have been identified in the literature as associated with delivery route. The inclusion and exclusion criteria were clearly defined for our study sample and only women with one prior cesarean were included. Electronic medical data captured maternal risks and pregnancy complications not otherwise available from other data sources, and prenatal records were used to address otherwise missing data. The thorough review of the data, with extensive comparisons between data in the electronic delivery and prenatal records for completeness and accuracy, resulted in very complete data. Also, the study site serves a diverse population of women, which contributed to the ability to obtain a sample for the three ethnic/racial groups. The large volume of deliveries at the Hospital Center also assisted in attaining a sufficiently large sample of women with one previous cesarean.
Our study also has several limitations. First, it was not possible to account for all confounders that may play a role in explaining the association between ethnicity/race and mode of delivery, particularly RCD indications. RCD indications were missing from a large percentage of records, which constrained our ability to determine if an RCD was elective or medically indicated. This distinction is important when exploring factors to explain the increased odds of RCD among Hispanic and non-Hispanic Black women, although it could not be included as risk factor; it is only relevant to RCD. Knowing the indication for the RCD is particularly important regarding our findings related to the use of induction or augmentation methods, which are unlikely to occur with elective procedures; that is, neither would be implemented if an RCD were scheduled.
The single-site study design also presented some limitations. Generalizability is limited because Hispanic women delivering at the Hospital Center may reflect populations primarily from Central America and, thus, results may not be applicable to Hispanic women from other countries. Additionally, comparisons across ethnic/racial groups may have been affected by inadequate cell sizes for specific factors, especially among non-Hispanic White women whose demographic characteristics were markedly different than those of Hispanic and non-Hispanic Black women; the small cell sizes may have led to wide CIs and imprecise estimates.
Implications for Practice and/or Policy
Our study was based on electronic medical records, which provided a rich source of data. Limitations found with this particular data source, however, could have important practice and policy implications. First, ethnicity was less accurately recorded on delivery records compared with self-reports on prenatal records and obtaining complete information about ethnicity required a thorough review of data through linkage of delivery records with prenatal records. Second, although we were unable to assess indications for RCD or induction or augmentation indications owing to missing data, these data may be important in any multifaceted approach to understanding the association between ethnicity/race, obstetrical/medical factors, and RCD.
The need for more precise and standardized documentation in delivery records presents an important policy implication. Studies using electronic medical records data at the local level in particular may have an impact on provider practices. In prior studies, obstetricians have remarked that local evidence is necessary to effectively change their practices because it validates “the transferability of the guidelines in their own practices” (
). Additionally, our findings suggest that hospital-level policies that create a favorable environment for VBAC may be insufficient to address increased risk for unnecessary RCD. Our study identified that Hispanic and non-Hispanic Black women experiencing birth at a site with favorable policies towards VBAC still experienced disparities in odds of RCD. The need to determine more granular reasons for these disparities provides direction for further research.
Conclusions
Our findings suggest that Hispanic and non-Hispanic Black women may face similar disparities in odds of RCD even in settings with policies favorable to VBAC. After adjusting for obstetrical/medical factors, Hispanic and non-Hispanic Black women had higher odds of RCD than non-Hispanic White women; adjustment for demographic and anthropomorphic factors did not appear to alter the association of ethnicity/race with RCD, whereas adjustment for parity and labor augmentation and induction revealed noticeable disparities. Further work is needed to fully understand the reasons for these differences, particularly the extent to which patient–provider interaction and provider bias play a role in explaining the differences. Our work is a necessary first step in recognizing the need to create effective public health policy and programs that target potentially preventable RCD to decrease the associated risk and health care costs for Hispanic and non-Hispanic women, particularly by highlighting the need to evaluate additional factors than those included in the literature thus far.
Acknowledgments
We would like to acknowledge Brian Caffo, PhD, Biostatistics, Bloomberg School of Public Health, for his feedback.
An examination of cesarean and vaginal birth histories among Hispanic women entering prenatal care in two California counties with large immigrant populations.
Journal of Immigrant and Minority Health/Center for Minority Public Health.2012; 14: 209-214
Roxanne Mirabal-Beltran, PhD, is an Assistant Professor at the Georgetown University School of Nursing & Health Studies, Department of Professional Nursing Practice. Her interests include Hispanic health disparities, patient–provider communication, and health care access, quality, and equity for women.
Donna M. Strobino, PhD, is Professor and Vice Chair of Education, Johns Hopkins Bloomberg School of Public Health Department of Population, Family and Reproductive Health. Her research centers on understanding why disadvantaged women have unfavorable pregnancy outcomes and interventions to improve them.
Article info
Publication history
Published online: October 14, 2019
Accepted:
September 6,
2019
Received in revised form:
August 26,
2019
Received:
January 21,
2019
Footnotes
The authors report no conflict of interest.
Presented at the 2018 CityMatCH Leadership & MCH Epidemiology Conference, Portland, Oregon, September 12–14, 2018.
Funding: Dr. Mirabal-Beltran was supported in part through the Centers of Excellence in Maternal and Child Health in Education, Science and Practice Program, funded through Maternal and Child Health Bureau at the Health Resources and Services Administration, U.S. Department of Health and Human Services (T76MC00003). Additional support provided by the Sylvia and Eddie Brown Scholars and the American Association of University Women American Fellowship. The funding sources had no involvement in the conduct of the research or the preparation of the article.
30472-4&id=gr1.jpg){kind=link}