Background
Racial disparities in preterm birth and infant death have been well documented. Less is known about racial disparities in neonatal morbidities among infants who are born at <37 weeks of gestation.
Objective
The purpose of this study was to determine whether the risk for morbidity and death among infants who are born preterm differs by maternal race.
Study Design
A retrospective cohort design included medical records from preterm deliveries of 19,325 black, Hispanic, and white women in the Consortium on Safe Labor. Sequentially adjusted Poisson models with generalized estimating equations estimated racial differences in the risk for neonatal morbidities and death, controlling for maternal demographics, health behaviors, and medical history. Sex differences between and within race were examined.
Results
Black preterm infants had an elevated risk for perinatal death, but there was no difference in risk for neonatal death across racial groups. Relative to white infants, black infants were significantly more likely to experience sepsis (9.1% vs 13.6%), peri- or intraventricular hemorrhage (2.6% vs 3.3%), intracranial hemorrhage (0.6% vs 1.8%), and retinopathy of prematurity (1.0% vs 2.6%). Hispanic and white preterm neonates had similar risk profiles. In general, female infants had lower risk relative to male infants, with white female infants having the lowest prevalence of a composite indicator of perinatal death or any morbidity across all races (30.9%). Differences in maternal demographics, health behaviors, and medical history did little to influence these associations, which were robust to sensitivity analyses of pregnancy complications as potential underlying mechanisms.
Conclusion
Preterm infants were at similar risk for neonatal death, regardless of race; however, there were notable racial disparities and sex differences in rare, but serious, adverse neonatal morbidities.
Black and Hispanic women in the United States are more likely to deliver preterm (<37 weeks of gestation) compared with white women; a higher proportion of black women experience fetal or perinatal losses and infant death compared with other races. The persistent racial disparity in infant death may be driven, in part, by the greater incidence of extremely preterm births (those occurring before the limit of viability) among black women, essentially all of which result in neonatal death. Among viable preterm births however, black births historically have experienced lower risks of neonatal death relative to white births ; however, more recent data suggest that the survival advantage among black early preterm infants diminished during the 1990s and that the racial disparity in mortality rates at >33 weeks of gestation has increased.
Reasons behind these entrenched disparities remain largely unexplained but are thought to be driven by complex and multifaceted risk exposure that arises from generations of social and economic disadvantage that place women and their neonates at risk for morbidity and death. Although disparities in death and preterm birth rates have been well documented, less is known about racial differences in morbidity among preterm neonates, particularly for rare conditions, which may shape the trajectory of infant health and survival. Sparse evidence suggests trends similar to death with risk by racial group inverse between preterm and term neonates. In an analysis based on gestational age–matched birth records from the state of Ohio, Loftin et al reported that, compared with white infants, black infants had lower risk of morbidity during the preterm period (through 36 weeks of gestation), but the risk increased after 37 weeks of gestation. Berman et al found that among neonates who were born at <34 weeks of gestation, white infants were more than twice as likely as black infants to have a Score for Neonatal Acute Physiology of >10, which indicates greater physiologic severity for neonatal intensive care.
Few previous studies have identified sex differences in neonatal morbidities among very preterm births (<29 weeks of gestation); however, to date, none of the cohorts were among the US population. An analysis of data from 797 infants who were born at 23–28 weeks of gestation in the United Kingdom found that female infants were at a significantly lower risk for death and oxygen dependency, pulmonary hemorrhage, and other major cranial abnormalities. An Australian cohort of 2549 very preterm neonates identified increased risk for adverse long-term neurologic outcomes among male infants compared with female infants; sex differences were most pronounced among births at <27 weeks of gestation. Neither study examined the joint effects of sex and race on preterm neonatal health.
The purpose of this analysis was to describe racial/ethnic differences in neonatal morbidities among neonates who were born preterm and to investigate whether racial/ethnic disparities in neonatal morbidity differed by infant sex in a large US medical record–based cohort. Further, we aimed to explore how differences in maternal demographic characteristics, health behaviors, and medical history explain racial differences in preterm neonatal morbidities.
Materials and Methods
Study population
Data were obtained from the Consortium on Safe Labor, a retrospective cohort of 228,438 deliveries from 12 US clinical centers between 2002 and 2008 in the following states: California, Delaware, Florida, Illinois, Indiana, Maryland, Massachusetts, New York, Ohio, Texas, and Utah and the District of Columbia. Detailed information was obtained from electronic medical records after Institutional Review Board approval was obtained at all participating institutions. International Classification of Diseases, version 9 (ICD-9) maternal discharge diagnosis codes were recorded for each pregnancy, which included pregnancy complications and medical history. Maternal records were linked to newborn electronic medical records, which included details of neonatal intensive care unit (NICU) admissions, and ICD-9 discharge codes for the infants. Women’s self-reported race and ethnicity were recorded in the medical record and compiled in the central Consortium on Safe Labor database in the following categorizations: non-Hispanic black, Hispanic, non-Hispanic white, Asian/Pacific Islander, multi-racial, other race, or unknown. We limited the analysis to deliveries that occurred at <37 weeks of gestation (n=29,592) and excluded multiple fetuses (n=3462), infants with congenital anomalies (n=4122), and births missing maternal age (n=28). Given limited sample sizes, we excluded Asian/Pacific Islanders (n=658) and all other and unknown race/ethnicities (n=1465), which left a final sample of 19,857 singletons among 19,325 women (black, 6720 [34.8%]; Hispanic=3799 [19.7%]; and white, 8806 [45.6%]).
Outcome and covariates
Neonatal outcomes of interest that were identified by diagnoses in the medical record and were supplemented with ICD-9 codes included asphyxia, NICU admission, transient tachypnea (TTN), respiratory distress syndrome (RDS), sepsis, necrotizing enterocolitis (NEC), peri-/intraventricular hemorrhage (PVH/IVH), intracranial hemorrhage (ICH), retinopathy of prematurity (ROP), neonatal death, and perinatal death (defined as stillbirth or neonatal death). We also examined a composite outcome that included neonates who experienced at least 1 of the aforementioned outcomes, perinatal death and/or any morbidity, compared with those who were liveborn without morbidities. Analyses of all outcomes, with the exception of perinatal death and the composite outcome, excluded stillbirths (n=648).
A priori selected covariates included study site, maternal age (continuous), parity (nulliparous, multiparous), prepregnancy body mass index (BMI, kg/m 2 : underweight, <18.5; normal weight, 18.5–24.9; overweight, 25 -29.9; obese, ≥30; unknown), insurance type (private, public/self-pay, other/unknown), marital status (married, not married or unknown), smoking during pregnancy (yes, no/unknown), alcohol use during pregnancy (yes, no/unknown), maternal history of asthma, depression, hypertension, renal disease, heart disease, diabetes mellitus, or thyroid disease (separate covariates, yes, no/unknown).
Statistical analysis
Descriptive statistics compared covariates between racial/ethnic groups with the use of the chi-square test and analysis of variance for categoric and continuous variables, respectively. We fit a series of sequentially adjusted Poisson regression models with generalized estimating equations to estimate racial/ethnic differences in the risk for adverse neonatal outcomes using white infants as the reference group and to account for multiple deliveries contributed by some women. Model A was adjusted only for study site. Model B was adjusted further for maternal demographics that included age, age-squared, parity, insurance, smoking and alcohol use during pregnancy, and marital status. Model C further included history of chronic diseases and prepregnancy BMI. Prepregnancy BMI was missing in 7107 of deliveries (35.8%) because of a lack of height and/or weight measure recorded in the medical record. Therefore, we used multiple imputations to derive estimates of prepregnancy BMI and retain all observations for modeling. The multiple imputations were based on observed values of prepregnancy weight, parity, insurance status, marital status, smoking, alcohol use, age, race, and study site.
A 6-category variable grouped infants by race/ethnicity and sex to compare relative risk (RR) and 95% confidence interval (CI) for each neonatal outcome and was controlled for all covariates in the fully adjusted Model C. The reference group of white female infants was selected based on the sex/race group with the highest proportion of infants who experienced a live birth with the lowest burden of morbidity. We tested for interactions between race/ethnicity and infant sex and maternal race/ethnicity and age, given documented racial differences in age-patterned risk for adverse birth outcomes.
Finally, in a sensitivity analysis, we examined the robustness of racial/ethnic differences in the composite outcome (death or at least 1 morbidity) after controlling for a range of complications that may have arisen during pregnancy differentially by maternal race/ethnicity and were associated with neonatal outcomes: preeclampsia, preterm premature rupture of membranes, chorioamnionitis, gestational hypertension, and gestational diabetes mellitus. These were included in a model that contained all the covariates in the fully-adjusted Model C.
Results
In the cohort, white women were older and more likely to be married and to have private insurance compared with other racial/ethnic groups ( Table 1 ). Smoking and alcohol use during pregnancy was more frequent among white women compared with blacks or Hispanic women. Rates of asthma, hypertension, and obesity were highest among black women; white women had a higher rate of depression and thyroid disease.
| Characteristic | White (n=9024), n (%) | Black (n=6950), n (%) | Hispanic (n=3883), n (%) | P value |
|---|---|---|---|---|
| Maternal age, y a | 28.3 ± 6.2 | 26.3 ± 6.7 | 27.3 ± 6.8 | <.001 |
| Parity | <.001 | |||
| Nulliparous | 3811 (42.2) | 2437 (35.1) | 1444 (37.2) | |
| Multiparous | 5213 (57.8) | 4513 (64.9) | 2439 (62.8) | |
| Marital status | <.001 | |||
| Married | 6289 (69.7) | 1554 (22.4) | 1798 (46.3) | |
| Not married/unknown | 2735 (30.3) | 5396 (77.6) | 2085 (53.7) | |
| Insurance type | <.001 | |||
| Private | 6029 (66.8) | 2187 (31.5) | 1125 (29.0) | |
| Public/self-pay | 2514 (27.9) | 4006 (57.6) | 2012 (51.8) | |
| Other/unknown | 481 (5.3) | 757 (10.9) | 746 (19.2) | |
| Smoking during pregnancy | 1121 (12.4) | 724 (10.4) | 224 (5.8) | <.001 |
| Alcohol use during pregnancy | 283 (3.1) | 173 (2.5) | 52 (1.3) | <.001 |
| Prepregnancy body mass index | <.001 | |||
| Underweight (<18.5 kg/m 2 ) | 439 (4.9) | 207 (3.0) | 119 (3.1) | |
| Normal weight (18.5–24.9 kg/m 2 ) | 3174 (35.2) | 1627 (23.4) | 1295 (33.4) | |
| Overweight (25–29.9 kg/m 2 ) | 1244 (13.8) | 1005 (14.5) | 689 (17.7) | |
| Obese (≥30 kg/m 2 ) | 1180 (13.1) | 1187 (17.1) | 584 (15.0) | |
| Chronic medical condition | ||||
| Asthma | 889 (9.5) | 837 (12.1) | 283 (7.3) | <.001 |
| Depression | 783 (8.7) | 261 (3.8) | 167 (4.3) | <.001 |
| Hypertension | 225 (2.5) | 412 (5.9) | 121 (3.1) | <.001 |
| Renal disease | 124 (1.4) | 117 (1.7) | 65 (1.7) | .22 |
| Heart disease | 209 (2.3) | 171 (2.5) | 79 (2.0) | .36 |
| Diabetes mellitus | 256 (2.8) | 323 (4.7) | 183 (4.7) | <.001 |
| Thyroid disease | 300 (3.3) | 73 (1.1) | 65 (1.7) | <.001 |
| Pregnancy complications | ||||
| Preeclampsia (mild+severe) | 1213 (13.4) | 952 (13.7) | 488 (12.6) | .24 |
| Preeclampsia (mild+severe+superimposed) | 1447 (16.0) | 1385 (19.9) | 660 (17.0) | .24 |
| Chorioamnionitis | 316 (3.5) | 514 (7.4) | 271 (7.0) | <.001 |
| Preterm premature rupture of membranes | 1839 (20.4) | 1301 (18.7) | 707 (18.2) | <.01 |
| Gestational hypertension | 342 (3.8) | 197 (2.9) | 88 (2.3) | <.001 |
| Gestational diabetes mellitus | 591 (6.6) | 420 (6.0) | 346 (8.9) | <.001 |
| Mode of delivery | <.05 | |||
| Vaginal | 5751 (63.7) | 4345 (62.5) | 2380 (61.3) | |
| Cesarean section | 3273 (36.3) | 2605 (37.5) | 1503 (38.7) | |
| Infant sex | <.0001 | |||
| Male | 4765 (53.1) | 3388 (49.4) | 2040 (52.7) | |
| Female | 4202 (46.9) | 3466 (50.6) | 1829 (47.3) | |
| Neonatal outcome | ||||
| Respiratory distress syndrome | 1736 (19.2) | 1182 (17.0) | 613 (15.8) | <.001 |
| Sepsis | 820 (9.1) | 948 (13.6) | 450 (11.6) | <.001 |
| Periventricular/intraventricular hemorrhage | 231 (2.6) | 232 (3.3) | 126 (3.2) | .01 |
| Transient tachypnea | 936 (10.4) | 838 (12.1) | 364 (9.4) | <.001 |
| Asphyxia | 63 (0.7) | 63 (0.9) | 33 (0.9) | .32 |
| Neonatal death | 93 (1.0) | 153 (2.2) | 62 (1.6) | <.001 |
| Perinatal death | 313 (3.5) | 405 (5.8) | 194 (5.0) | <.001 |
| Perinatal death or any morbidity | 3115 (34.5) | 259 (37.3) | 1270 (32.7) | <.001 |
| Necrotizing enterocolitis | 86 (1.0) | 64 (0.9) | 44 (1.1) | .54 |
| Intracranial hemorrhage | 56 (0.6) | 128 (1.8) | 61 (1.6) | <.001 |
| Retinopathy of prematurity | 90 (1.0) | 182 (2.6) | 94 (2.4) | <.001 |
| Neonatal intensive care unit admission | 4107 (45.5) | 3235 (46.6) | 1613 (41.5) | <.001 |
The Figure illustrates the proportion of infants who experienced perinatal death or at least 1 of the morbidities of interest by race/ethnicity and gestational week (for morbidity-specific figures, see the Supplemental Figure ). Empiric rates of death or morbidity varied among white, black, and Hispanic neonates until 29 weeks of gestation and onward, with white neonates the highest, followed by black neonates, then Hispanic neonates (with the exception of 32 weeks of gestation when the Hispanic rate was higher than the black rate).
Associations between maternal race/ethnicity and neonatal outcomes are shown in Table 2 . Overall, adjustment for covariates across Models A–C did not change the magnitude of associations appreciably between race and risk for each outcome. Across all 3 models, black neonates were at 41–43% increased risk for sepsis, 31–33% increased risk for PVH/IVH, 52–67% increased risk for ICH, 41–60% increased risk for ROP, and 45–48% increased risk for perinatal death compared with white neonates. Hispanic neonates were at 11–12% decreased risk for RDS and 13–15% increased risk for sepsis. Site-only adjusted Model A indicated a lower risk for TTN among Hispanic neonates compared with white neonates (RR, 0.87; 95% CI, 0.77–0.99) and a 5% decreased risk for NICU admission, but the differences were not statistically significant in subsequent models. There were no racial/ethnic differences in a combined perinatal death or any morbidity in the final fully adjusted Model C.
| Variable | Model A a | Model B b | Model C c | |||
|---|---|---|---|---|---|---|
| Relative risk | 95% Confidence interval | Relative risk | 95% Confidence interval | Relative risk | 95% Confidence interval | |
| Respiratory distress syndrome | ||||||
| White | Reference | Reference | Reference | |||
| Black | 0.99 | 0.91–1.08 | 1.01 | 0.93–1.10 | 0.99 | 0.91–1.08 |
| Hispanic | 0.89 | 0.81–0.98 | 0.88 | 0.80–0.96 | 0.88 | 0.79–0.96 |
| Sepsis | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.41 | 1.27–1.57 | 1.43 | 1.29–1.61 | 1.42 | 1.27–1.60 |
| Hispanic | 1.13 | 1.00–1.27 | 1.15 | 1.01–1.30 | 1.15 | 1.02–1.31 |
| Peri-/intraventricular hemorrhage | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.32 | 1.04–1.69 | 1.33 | 1.02–1.72 | 1.32 | 1.02–1.72 |
| Hispanic | 0.99 | 0.76–1.28 | 1.01 | 0.77–1.33 | 1.01 | 0.77–1.33 |
| Transient tachypnea | ||||||
| White | Reference | Reference | Reference | |||
| Black | 0.98 | 0.89–1.09 | 1.00 | 0.90–1.11 | 1.01 | 0.91–1.13 |
| Hispanic | 0.87 | 0.77–0.99 | 0.89 | 0.79–1.01 | 0.90 | 0.80–1.02 |
| Asphyxia | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.40 | 0.92–2.12 | 1.48 | 0.96–2.28 | 1.46 | 0.94–2.26 |
| Hispanic | 1.11 | 0.70–1.75 | 1.08 | 0.68–1.72 | 1.09 | 0.69–1.73 |
| Necrotizing enterocolitis | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.28 | 0.82–1.98 | 1.13 | 0.70–1.81 | 1.14 | 0.70–1.83 |
| Hispanic | 1.11 | 0.72–1.70 | 1.01 | 0.64–1.59 | 1.02 | 0.65–1.61 |
| Intracranial hemorrhage | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.52 | 1.11–2.11 | 1.67 | 1.18–2.37 | 1.59 | 1.11–2.28 |
| Hispanic | 1.36 | 0.94–1.97 | 1.31 | 0.87–1.98 | 1.27 | 0.84–1.93 |
| Retinopathy of prematurity | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.41 | 1.06–1.89 | 1.60 | 1.17–2.18 | 1.59 | 1.16–2.17 |
| Hispanic | 1.18 | 0.85–1.63 | 1.24 | 0.88–1.75 | 1.27 | 0.90–1.79 |
| Neonatal intensive care unit admission | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.03 | 0.99–1.07 | 1.04 | 1.00–1.09 | 1.04 | 1.00–1.08 |
| Hispanic | 0.95 | 0.91–1.00 | 0.96 | 0.92–1.01 | 0.96 | 0.91–1.01 |
| Neonatal death | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.32 | 0.94–1.86 | 1.37 | 0.96–1.95 | 1.25 | 0.87–1.80 |
| Hispanic | 0.97 | 0.65–1.45 | 1.05 | 0.69–1.58 | 1.06 | 0.71–1.60 |
| Perinatal death d | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.46 | 1.23–1.74 | 1.48 | 1.22–1.78 | 1.45 | 1.20–1.77 |
| Hispanic | 1.15 | 0.94–1.41 | 1.17 | 0.95–1.44 | 1.18 | 0.96–1.47 |
| Perinatal death or any morbidity | ||||||
| White | Reference | Reference | Reference | |||
| Black | 1.05 | 1.00–1.10 | 1.06 | 1.01–1.12 | 1.06 | 1.00–1.11 |
| Hispanic | 0.92 | 0.87–0.98 | 0.93 | 0.87–0.99 | 0.93 | 0.87–0.98 |
b Adjusted for study site, maternal age, squared maternal age, parity, insurance type, smoking and alcohol use during pregnancy, and marital status
c Adjusted for all covariates in Model B and maternal history of chronic diseases (asthma, depression, hypertension, renal disease, heart disease, diabetes mellitus, and thyroid disease) and prepregnancy body mass index
d Includes stillbirths (n=648) that were excluded from all other outcomes.
There was evidence of an interaction between infant sex and race but not between maternal age and race/ethnicity in these data. Stratification by both race/ethnicity and sex revealed more nuanced trends ( Table 3 ). Risk of perinatal death was highest among black infants overall and was not specific to infant sex; both black female and male infants had 55% and 70% increase in risk, respectively, compared with white female infants. Risk of perinatal death was also elevated among Hispanic female infants relative to white female infants (RR, 1.47; 95% CI, 109–1.98). Likewise, Hispanic infants were at a lower risk for RDS overall, and there was no difference in risk between male or female infants compared with white female infants. Conversely, although there were no racial/ethnic differences in RDS risk, sex differences emerged with male infants of either white or black race who were at increased risk relative to white or black female infants (black male infants: RR, 1.19 [95% CI, 1.06–1.33]; white male infants: RR, 1.25 [95% CI, 1.15–1.36]; black female infants: RR, 1.06, [95% CI, 0.95–1.19]). Similarly, TTN risk was not differential by race/ethnicity but by sex with both black and white male infants at 17–18% increased risk compared with white female infants and no difference between black, white, and Hispanic female infants. The combined influence of both sex and race/ethnicity was apparent for asphyxia, NICU admission, and the composite outcome of death or any morbidity. There were no racial/ethnic differences in these outcomes overall; however, after further stratification, black male infants were more likely to experience asphyxia than were white female infants (RR, 2.08; 95% CI, 1.19–3.68); white male infants and black neonates of either sex were 10–13% more likely to be admitted to the NICU than white female infants and were 11–22% more likely to experience perinatal death or at least 1 morbidity, with a stronger association in male infants of either race. Finally, relative to white female infants, black female infants were at increased risk for PVH/IVH (RR, 1.47; 95% CI, 1.06–2.03).
