Objective
The objective of the study was to investigate differences in sociodemographic, medical, and obstetric risk factors for primary cesarean delivery between black and white women.
Study Design
We conducted a retrospective cohort study among 25,251 black and white women delivering a live, singleton infant with vertex presentation at a large, regional hospital between 2004 and 2010. Demographic and clinical data were derived from electronic hospital records. Differences in risk factors for primary cesarean delivery were analyzed using a modified Poisson regression approach stratified by race and parity.
Results
Black and white women had a primary cesarean delivery rate of 24.7% and 22.2%, respectively ( P < .001). Black women had an increased risk of cesarean delivery after adjusting for sociodemographic and clinical risk factors (adjusted relative risk [RR], 1.23; 95% confidence interval [CI], 1.17–1.29). Among nulliparas, labor induction had a greater effect on cesarean delivery for black women (adjusted RR, 1.32; 95% CI, 1.20–1.44) than for white women (adjusted RR, 1.13; 95% CI, 1.07–1.20). Among multiparas, labor induction reduced the risk of cesarean delivery for white women (adjusted RR, 0.63; 95% CI, 0.55–0.72), whereas no association was observed for black women (adjusted RR, 1.08; 95% CI, 0.92–1.28). Advanced maternal age was a stronger risk factor for black women (adjusted RR, 1.72; 95% CI, 1.43–2.08) than for white women (adjusted RR, 1.30; 95% CI, 1.11–1.52) among multiparas only. Among nulliparas, delivery at 37-38 weeks’ gestation reduced the risk of cesarean delivery for black women (adjusted RR, 0.82; 95% CI, 0.73–0.92), whereas no association was observed for white women (adjusted RR, 0.96; 95% CI, 0.90–1.04).
Conclusion
Labor induction, among nulliparous women, and advanced maternal age, among multiparous women, are stronger risk factors for primary cesarean delivery for black women than for white women.
Total and primary cesarean delivery rates have been higher for black women than white women in the United States since the mid-1990s. In 2012, the primary cesarean delivery rate for black and white women was 26.1% and 23.1%, respectively. Repeatedly higher rates among black women are surprising, given their younger age distribution and the association between increasing maternal age and cesarean delivery. Studies suggest that black-white differences are largely unexplained by sociodemographic, medical, or obstetric risk factors. Variation in women’s preferences for delivery, patient-provider communication, or physician decision making have been suggested to play a role in these differences.
High rates of cesarean delivery have raised concerns in the United States, especially when considered alongside rates of labor induction, early-term births, and other obstetric practices. Maternal risks associated with cesarean delivery include surgical complications, longer hospitalization, and rehospitalization; newborn risks include mild to severe respiratory problems. Despite efforts to reduce cesarean delivery rates, quality improvement cannot be achieved without addressing persistently higher rates for black women.
In this study, we investigated differences between black and white women in sociodemographic, medical, and obstetric risk factors for primary cesarean delivery using electronic clinical data from a regional hospital. Identifying different risk factors, or factors for which the relation is significantly stronger for black women, through the study of a large community population may suggest future areas of research or intervention to reduce disparities.
Materials and Methods
We conducted a retrospective cohort study of deliveries at a private, not-for-profit hospital from 2004 through 2010 with the following maternal and infant characteristics: live, singleton birth with vertex presentation; birthweight of 500 g or greater; 23 or more completed weeks’ gestation; and no history of cesarean delivery. Although this was a single-institution study, the hospital provides obstetric care to more than 6500 women a year, accounting for more than 85% of births in the surrounding region. Given the high representation of births and the diversity of the patient population, the deliveries analyzed in this study approximate population-level data for the region.
Data were derived from electronic clinical records, including an obstetric database that captures information on labor and delivery and select pregnancy outcomes. Medical history and pregnancy complications are identified on forms submitted by the patient’s prenatal care provider and entered into the clinical record. For a small subset of women without prenatal care, medical history and pregnancy complications are either self-reported or diagnosed during hospitalization for labor. Clinical information during hospitalization is recorded in real time by nursing staff. Logic checking occurs periodically to identify errors in data entry; historical audits show that data validation exceeds 95%.
Race is self-identified and the categories used for institutional statistics are white, black, Hispanic, Asian, American Indian (or Native American), and other. Our sample included white and black women only. The outcome variable was dichotomous, primary cesarean vs vaginal delivery.
Independent variables were dichotomous or categorical. Sociodemographic measures included maternal age (<20, 20–29, 30–34, and ≥35 years), parity (nulliparous and multiparous), insurance type (private and Medicaid/uninsured), and marital status (married and unmarried). Uninsured women constituted less than 1% of deliveries. Unmarried women included single, divorced, and widowed women.
Medical risk factors included chronic hypertension, diabetes (type 1 and type 2), and body mass index (BMI). Prepregnancy BMI was calculated (kilograms per square meter) using self-reported data on height and prepregnancy weight, with the following categories: less than 18.5 kg/m 2 (underweight), 18.5–24.9 kg/m 2 (normal), 25.0–29.9 kg/m 2 (overweight), and 30.0 kg/m 2 or greater (obese).
Obstetric risk factors included gestational diabetes, gestational hypertensive disorders (including preeclampsia), premature rupture of the membranes (PROM), macrosomia (infant birthweight ≥4000 g), excess weight gain, gestational age, and labor induction. Excess weight gain was defined as pregnancy weight gain exceeding 40 pounds if underweight, 35 pounds if normal BMI, 25 pounds if overweight, and 20 pounds if obese. Gestational age was recorded at delivery as completed weeks, measured by last menstrual period and validated by first-trimester ultrasound, with the following categories: less than 34, 34–36 (late preterm), 37–38 (early term), 39–40 (full term), 41 (late term), and 42 weeks or longer (postterm).
Labor induction, as compared with spontaneous labor, was identified by nurses through a review of the obstetrical record. Only oxytocin and Foley bulb for preinduction cervical ripening were used at the study hospital. Bishop score was not documented in the database, but it was general practice at the hospital during the study years that patients with scores less than 6 receive cervical ripening. Use of a Foley bulb was a proxy measure for an unfavorable cervix at the time of induction.
Patient type and year of delivery were control variables. Service patients received care from hospital staff, whereas private patients received care from community obstetricians. Year of delivery adjusted for temporal trends in cesarean delivery.
Indications for cesarean delivery were categorized in the data as labor dystocia, fetal distress, maternal medical condition, and other. The data did not include additional detail on these indications. Multiple indications were recorded for 7.9% of cesarean deliveries; primary indication was not specified and all indications were included for analysis.
Pearson χ 2 tests were used to assess differences in the distribution of independent variables by race and to test the association between cesarean delivery and the independent variables. Race-specific cesarean delivery rates were calculated for each variable, and 2-sample tests of proportions were used to compare rates by race. To clarify the initial findings, the use of a Foley bulb and cesarean delivery rates with labor induction were described by parity and race, and racial differences were assessed using Pearson χ 2 tests and 2-sample tests of proportions. Racial differences in indications for cesarean delivery were evaluated using Pearson χ 2 tests.
Poisson regression with a robust error variance was used to estimate unadjusted and adjusted relative risk (RR). To meet the regression assumption of independence, only one delivery was included in the sample for women who delivered multiple times during the study period. Multiple deliveries were clustered by a common patient identification number. After removing repeat cesarean deliveries, one delivery was randomly selected within each patient cluster.
Missing values ranged from 0.03% to 7.5% for parity, insurance, marital status, gestational age, PROM, macrosomia, BMI, and excess weight gain. We performed multivariate imputation of missing data by chained equations using Stata/IC version 12.0 (StataCorp, College Station, TX). The imputation model included all other independent variables and the dependent variable and 20 imputations were generated. Diabetes, chronic hypertension, gestational diabetes, and gestational hypertensive disorders were recorded as present (1) or null (0); the extent to which the null category included missing values was unknown.
For the regression analysis, we first estimated unadjusted RRs for race and the independent variables. We then estimated the RR for race, adjusting for all variables except diabetes, gestational hypertensive disorders, and PROM because they did not vary by race in bivariate analysis. Marital status was also excluded because it was not associated with cesarean delivery. All regression results were estimated based on the multiple imputation of missing data.
Poisson regression models including all sociodemographic characteristics (maternal age, parity, insurance, marital status), medical risk factors (diabetes, chronic hypertension, BMI), obstetric risk factors (excess weight gain, gestational diabetes, gestational hypertensive disorders, macrosomia, gestational age, labor induction, PROM), and control variables (patient type, year) were then estimated separately for black and white women; t tests were used to assess differences in regression coefficients by race. The t statistic was calculated as follows: t=(biB−biW)/√(S.E.biB)2+(S.E.biW)2
t = ( b iB − b iW ) / ( S . E . biB ) 2 + ( S . E . biW ) 2
, where b iB and b iW are the regression coefficients for the ith variable for black women and white women, respectively. Adjustment for the same variables in the models for black and white women facilitated comparisons of the direct effect of each variable by race.
Finally, based on our initial findings for labor induction, we estimated the regression models stratified by parity and race because the relationship between labor induction and cesarean delivery has been shown to vary by parity. The model for the stratified groups (white-nulliparous, black-nulliparous, white-multiparous, black-multiparous) included the same variables as previously listed, except parity; t tests were used to assess differences in regression coefficients by race.
The study was approved by the Christiana Care Health System Institutional Review Board. Data were analyzed using Stata/IC version 12.0 (Stata Corp).
Results
We identified 48,696 deliveries from 2004 through 2010; 308 were excluded for fetal demise, 161 for birthweight less than 500 grams or gestational age less than 23 weeks, 1795 for multiple gestations, 1453 for breech presentation, and 7108 for history of cesarean delivery. Additionally, 6909 deliveries were randomly excluded to avoid multiple births to the same women, 25 were missing data on delivery method, and 5686 were to women of other races. The final sample included 25,251 deliveries to white and black women.
The primary cesarean delivery rate was 23.0%; the respective rates for black and white women were 24.7% and 22.2% ( P < .001). Table 1 shows the distribution of characteristics by race and race-specific primary cesarean delivery rates for each characteristic. Only 11.2% of black women were 35 years old or older compared with 18.0% of white women; older black women had a significantly higher cesarean delivery rate than older white women (35.7% vs 22.9%). Black women were significantly less likely to undergo labor induction (32.1% vs 41.4%) but were more likely to have a cesarean delivery following an induction (32.0% vs 23.9%).
| Characteristic | Distribution, n (%) | Cesarean delivery, % | ||||||
|---|---|---|---|---|---|---|---|---|
| White | Black | P value a | Total | P value b | White | Black | P value c | |
| Total | 17,698 (100.0) | 7553 (100.0) | — | 23.0 | — | 22.2 | 24.7 | < .001 |
| Maternal age, y | < .001 | < .001 | ||||||
| <20 | 1107 (6.3) | 1346 (17.8) | 23.5 | 22.6 | 24.3 | .32 | ||
| 20-29 | 8211 (46.4) | 4106 (54.4) | 22.5 | 22.1 | 23.3 | .14 | ||
| 30-34 | 5185 (29.3) | 1257 (16.6) | 22.0 | 21.9 | 22.5 | .62 | ||
| ≥35 | 3194 (18.0) | 844 (11.2) | 25.6 | 22.9 | 35.7 | < .001 | ||
| Parity | < .001 | < .001 | ||||||
| Nulliparous | 8754 (49.5) | 3436 (45.5) | 35.1 | 34.4 | 36.9 | .01 | ||
| Multiparous | 8905 (50.3) | 4096 (54.2) | 11.6 | 10.2 | 14.6 | < .001 | ||
| Insurance type | < .001 | < .001 | ||||||
| Private | 13,788 (77.9) | 3144 (41.6) | 24.1 | 23.2 | 28.3 | < .001 | ||
| Medicaid/uninsured | 3902 (22.1) | 4408 (58.4) | 20.7 | 19.0 | 22.2 | < .001 | ||
| Marital status | < .001 | .73 | ||||||
| Married | 12,510 (70.7) | 2147 (28.4) | 23.1 | 22.4 | 27.1 | < .001 | ||
| Unmarried | 5002 (28.3) | 5305 (70.2) | 22.9 | 22.0 | 23.8 | .03 | ||
| Diabetes | .10 | < .001 | ||||||
| No | 17,492 (98.8) | 7446 (98.6) | 22.8 | 22.0 | 24.4 | < .001 | ||
| Yes | 206 (1.2) | 107 (1.4) | 40.9 | 38.8 | 44.9 | .30 | ||
| Chronic hypertension | < .001 | < .001 | ||||||
| No | 17,230 (97.4) | 7224 (95.6) | 22.5 | 21.8 | 24.1 | < .001 | ||
| Yes | 468 (2.6) | 329 (4.4) | 38.4 | 38.2 | 38.6 | .92 | ||
| BMI, kg/m 2 | < .001 | < .001 | ||||||
| <18.5 | 667 (3.8) | 255 (3.4) | 12.0 | 11.1 | 14.5 | .15 | ||
| 18.5–24.9 | 9258 (52.3) | 2878 (38.1) | 18.3 | 18.0 | 19.1 | .18 | ||
| 25.0–29.9 | 3937 (22.3) | 1913 (25.3) | 24.4 | 23.5 | 26.4 | .01 | ||
| ≥30.0 | 3199 (18.1) | 2089 (27.7) | 34.2 | 35.5 | 32.2 | .01 | ||
| Excess weight gain | < .001 | < .001 | ||||||
| No | 7224 (40.8) | 3270 (43.3) | 17.7 | 16.4 | 20.6 | < .001 | ||
| Yes | 9360 (52.9) | 3651 (48.3) | 27.3 | 26.7 | 28.6 | .03 | ||
| Gestational diabetes | .04 | < .001 | ||||||
| No | 16,792 (94.9) | 7212 (95.5) | 22.4 | 21.7 | 24.1 | < .001 | ||
| Yes | 906 (5.1) | 341 (4.5) | 33.6 | 31.8 | 38.4 | .03 | ||
| Gestational hypertensive disorders | .63 | < .001 | ||||||
| No | 16,321 (92.2) | 6952 (92.0) | 21.5 | 20.7 | 23.3 | < .001 | ||
| Yes | 1377 (7.8) | 601 (8.0) | 41.0 | 40.8 | 41.8 | .69 | ||
| Macrosomia | < .001 | < .001 | ||||||
| No | 15,846 (89.5) | 7215 (95.5) | 21.8 | 20.8 | 23.9 | < .001 | ||
| Yes | 1833 (10.4) | 326 (4.3) | 36.3 | 34.9 | 44.2 | .001 | ||
| Gestational age, wks | < .001 | < .001 | ||||||
| <34 | 394 (2.2) | 325 (4.3) | 42.7 | 42.9 | 42.5 | .91 | ||
| 34-36 | 1113 (6.3) | 624 (8.3) | 27.3 | 27.1 | 27.6 | .85 | ||
| 37-38 | 4418 (25.0) | 1853 (24.5) | 19.8 | 20.0 | 19.2 | .46 | ||
| 39-40 | 10,382 (58.7) | 4014 (53.1) | 21.3 | 20.5 | 23.1 | < .001 | ||
| 41 | 1289 (7.3) | 698 (9.2) | 33.6 | 31.4 | 37.5 | .01 | ||
| ≥42 | 86 (0.5) | 22 (0.3) | 46.3 | 46.5 | 45.5 | .93 | ||
| Labor type | < .001 | < .001 | ||||||
| Not induced | 10,371 (58.6) | 5126 (67.9) | 21.2 | 21.1 | 21.3 | .74 | ||
| Induced | 7327 (41.4) | 2427 (32.1) | 25.9 | 23.9 | 32.0 | < .001 | ||
| PROM | .24 | .01 | ||||||
| No | 17,432 (98.5) | 7418 (98.2) | 22.9 | 22.1 | 24.6 | < .001 | ||
| Yes | 245 (1.4) | 123 (1.6) | 29.1 | 27.8 | 31.7 | .43 | ||
| Patient type | < .001 | < .001 | ||||||
| Service | 2437 (13.8) | 3172 (42.0) | 20.9 | 19.5 | 21.9 | .03 | ||
| Private | 15,261 (86.2) | 4381 (58.0) | 23.6 | 22.7 | 26.8 | < .001 | ||
| Year of delivery | < .001 | < .001 | ||||||
| 2004 | 2862 (16.2) | 985 (13.0) | 22.1 | 21.0 | 25.2 | .01 | ||
| 2005 | 2783 (15.7) | 980 (13.0) | 20.7 | 20.0 | 22.4 | .11 | ||
| 2006 | 2567 (14.5) | 1053 (13.9) | 22.2 | 21.8 | 23.4 | .30 | ||
| 2007 | 2437 (13.8) | 1115 (14.8) | 23.2 | 22.9 | 23.9 | .48 | ||
| 2008 | 2440 (13.8) | 1198 (15.9) | 24.3 | 23.4 | 26.1 | .08 | ||
| 2009 | 2355 (13.3) | 1156 (15.3) | 24.8 | 24.2 | 26.0 | .24 | ||
| 2010 | 2254 (12.7) | 1066 (14.1) | 23.9 | 23.1 | 25.6 | .11 | ||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
