[Skip to Content]
[Skip to Content Landing]
Figure.
Hospital Rankings for Risk-Adjusted Neonatal Morbidity and Mortality, New York City, NY, 2010-2014
Hospital Rankings for Risk-Adjusted Neonatal Morbidity and Mortality, New York City, NY, 2010-2014

Error bars indicate 95% CI; horizontal line, the sample mean (0.28; 95% CI, 0.27-0.29).

Table 1.  
Sample Characteristics and Risk of Neonatal Morbidity or Mortality Among Very Preterm Birth Infants, New York City, NY, 2010-2014
Sample Characteristics and Risk of Neonatal Morbidity or Mortality Among Very Preterm Birth Infants, New York City, NY, 2010-2014
Table 2.  
Patient Characteristics and Odds of Neonatal Morbidity or Mortality Among Very Preterm Birth Infants, New York City, NY, 2010-2014
Patient Characteristics and Odds of Neonatal Morbidity or Mortality Among Very Preterm Birth Infants, New York City, NY, 2010-2014
Table 3.  
Adjusted Risk for Neonatal Morbidity and Mortality by Race/Ethnicity in Hospitals of Low–, Middle–, and High–Morbidity and Mortality Rank
Adjusted Risk for Neonatal Morbidity and Mortality by Race/Ethnicity in Hospitals of Low–, Middle–, and High–Morbidity and Mortality Rank
Table 4.  
Contributing Factors to Racial/Ethnic Disparities in Neonatal Morbidity and Mortality Among Very Preterm Birth Infants
Contributing Factors to Racial/Ethnic Disparities in Neonatal Morbidity and Mortality Among Very Preterm Birth Infants
1.
Horbar  JD, Edwards  EM, Greenberg  LT,  et al.  Variation in performance of neonatal intensive care units in the United States.  JAMA Pediatr. 2017;171(3):e164396.PubMedGoogle ScholarCrossref
2.
Lorch  SA.  A decade of improvement in neonatal intensive care: how do we continue the momentum?  JAMA Pediatr. 2017;171(3):e164395.PubMedGoogle ScholarCrossref
3.
Horbar  JD, Carpenter  JH, Badger  GJ,  et al.  Mortality and neonatal morbidity among infants 501 to 1500 grams from 2000 to 2009.  Pediatrics. 2012;129(6):1019-1026.PubMedGoogle ScholarCrossref
4.
Lapcharoensap  W, Gage  SC, Kan  P,  et al.  Hospital variation and risk factors for bronchopulmonary dysplasia in a population-based cohort.  JAMA Pediatr. 2015;169(2):e143676.PubMedGoogle ScholarCrossref
5.
Matthews  TJ, MacDorman  MF, Thoma  ME.  Infant mortality statistics from the 2013 period linked birth/infant death data set.  Natl Vital Stat Rep. 2015;64(9):1-30.PubMedGoogle Scholar
6.
Gephart  SM, Spitzer  AR, Effken  JA, Dodd  E, Halpern  M, McGrath  JM.  Discrimination of GutCheck(NEC): a clinical risk index for necrotizing enterocolitis.  J Perinatol. 2014;34(6):468-475.PubMedGoogle ScholarCrossref
7.
Wallace  ME, Mendola  P, Kim  SS,  et al.  Racial/ethnic differences in preterm perinatal outcomes.  Am J Obstet Gynecol. 2017;216(3):e301-e306, e312.Google Scholar
8.
Rowley  DL, Hogan  V.  Disparities in infant mortality and effective, equitable care: are infants suffering from benign neglect?  Annu Rev Public Health. 2012;33:75-87.PubMedGoogle ScholarCrossref
9.
Saigal  S, Doyle  LW.  An overview of mortality and sequelae of preterm birth from infancy to adulthood.  Lancet. 2008;371(9608):261-269.PubMedGoogle ScholarCrossref
10.
Howell  EA, Hebert  P, Chatterjee  S, Kleinman  LC, Chassin  MR.  Black/white differences in very low birth weight neonatal mortality rates among New York City hospitals.  Pediatrics. 2008;121(3):e407-e415.PubMedGoogle ScholarCrossref
11.
Morales  LS, Staiger  D, Horbar  JD,  et al.  Mortality among very low-birthweight infants in hospitals serving minority populations.  Am J Public Health. 2005;95(12):2206-2212.PubMedGoogle ScholarCrossref
12.
Synnes  AR, Chien  LY, Peliowski  A, Baboolal  R, Lee  SK; Canadian NICU Network.  Variations in intraventricular hemorrhage incidence rates among Canadian neonatal intensive care units.  J Pediatr. 2001;138(4):525-531.PubMedGoogle ScholarCrossref
13.
New York State Department of Health. Statewide Planning and Research Cooperative System (SPARCS). https://www.health.ny.gov/statistics/sparcs/. Revised June 2016. Accessed November 26, 2017.
14.
Jensen  EA, Lorch  SA.  Effects of a birth hospital’s neonatal intensive care unit level and annual volume of very low-birth-weight infant deliveries on morbidity and mortality.  JAMA Pediatr. 2015;169(8):e151906.PubMedGoogle ScholarCrossref
15.
Lain  SJ, Hadfield  RM, Raynes-Greenow  CH,  et al.  Quality of data in perinatal population health databases: a systematic review.  Med Care. 2012;50(4):e7-e20.PubMedGoogle ScholarCrossref
16.
Dietz  P, Bombard  J, Mulready-Ward  C,  et al.  Validation of selected items on the 2003 U.S. standard certificate of live birth: New York City and Vermont.  Public Health Rep. 2015;130(1):60-70.PubMedGoogle ScholarCrossref
17.
Phibbs  CS, Baker  LC, Caughey  AB, Danielsen  B, Schmitt  SK, Phibbs  RH.  Level and volume of neonatal intensive care and mortality in very-low-birth-weight infants.  N Engl J Med. 2007;356(21):2165-2175.PubMedGoogle ScholarCrossref
18.
Olsen  IE, Groveman  SA, Lawson  ML, Clark  RH, Zemel  BS.  New intrauterine growth curves based on United States data.  Pediatrics. 2010;125(2):e214-e224.PubMedGoogle ScholarCrossref
19.
Ash  AS, Normand  ST, Stukel  TA, Utts  J; Committee of Presidents of Statistical Societies (COPSS).  Statistical Issues in Assessing Hospital Performance. Washington, DC: Centers for Medicare & Medicaid Services; 2012.
20.
Hollander  M, Wolfe  DA.  Nonparametric Statistical Methods. 2nd ed. New York, NY: John Wiley & Sons; 1999.
21.
Fairlie  RW.  An extension of the Blinder–Oaxaca decomposition technique to logit and probit models.  J Econ Soc Meas. 2005;30:305-316.Google Scholar
22.
Johnson  TJ, Patel  AL, Jegier  BJ, Engstrom  JL, Meier  PP.  Cost of morbidities in very low birth weight infants.  J Pediatr. 2013;162(2):243-49.e1. doi:10.1016/j.jpeds.2012.07.013PubMedGoogle ScholarCrossref
23.
Braveman  P, Barclay  C.  Health disparities beginning in childhood: a life-course perspective.  Pediatrics. 2009;124(suppl 3):S163-S175.Google ScholarCrossref
24.
Braveman  P.  What is health equity: and how does a life-course approach take us further toward it?  Matern Child Health J. 2014;18(2):366-372.PubMedGoogle ScholarCrossref
25.
Elder  GH.  Perspectives on the Life Course: Life Course Dynamics: Trajectories and Transitions. Ithaca, NY: Cornell University Press; 1985:23-49.
26.
Lu  M, Halfon  N.  Racial and ethnic disparities in birth outcomes: a life-course perspective.  Matern Child Health J. 2003;7(1):13-30.Google ScholarCrossref
27.
Halfon  N, Hochstein  M.  Life course health development: an integrated framework for developing health, policy, and research.  Milbank Q. 2002;80(3):433-479, iii.PubMedGoogle ScholarCrossref
28.
Barnato  AE, Lucas  FL, Staiger  D, Wennberg  DE, Chandra  A.  Hospital-level racial disparities in acute myocardial infarction treatment and outcomes.  Med Care. 2005;43(4):308-319.PubMedGoogle ScholarCrossref
29.
Stansbury  JP, Jia  H, Williams  LS, Vogel  WB, Duncan  PW.  Ethnic disparities in stroke: epidemiology, acute care, and postacute outcomes.  Stroke. 2005;36(2):374-386.PubMedGoogle ScholarCrossref
30.
Howell  EA, Egorova  NN, Balbierz  A, Zeitlin  J, Hebert  PL.  Site of delivery contribution to black-white severe maternal morbidity disparity.  Am J Obstet Gynecol. 2016;215(2):143-152.PubMedGoogle ScholarCrossref
31.
Hebert  PL, Chassin  MR, Howell  EA.  The contribution of geography to black/white differences in the use of low neonatal mortality hospitals in New York City.  Med Care. 2011;49(2):200-206.PubMedGoogle ScholarCrossref
32.
Bonet  M, Smith  LK, Pilkington  H, Draper  ES, Zeitlin  J.  Neighbourhood deprivation and very preterm birth in an English and French cohort.  BMC Pregnancy Childbirth. 2013;13:97.PubMedGoogle ScholarCrossref
33.
Piper  JM, Mitchel  EF  Jr, Snowden  M, Hall  C, Adams  M, Taylor  P.  Validation of 1989 Tennessee birth certificates using maternal and newborn hospital records.  Am J Epidemiol. 1993;137(7):758-768.PubMedGoogle ScholarCrossref
34.
DiGiuseppe  DL, Aron  DC, Ranbom  L, Harper  DL, Rosenthal  GE.  Reliability of birth certificate data: a multi-hospital comparison to medical records information.  Matern Child Health J. 2002;6(3):169-179.PubMedGoogle ScholarCrossref
35.
Yasmeen  S, Romano  PS, Schembri  ME, Keyzer  JM, Gilbert  WM.  Accuracy of obstetric diagnoses and procedures in hospital discharge data.  Am J Obstet Gynecol. 2006;194(4):992-1001.PubMedGoogle ScholarCrossref
36.
Lorch  SA, Enlow  E.  The role of social determinants in explaining racial/ethnic disparities in perinatal outcomes.  Pediatr Res. 2016;79(1-2):141-147.PubMedGoogle ScholarCrossref
Original Investigation
March 2018

Differences in Morbidity and Mortality Rates in Black, White, and Hispanic Very Preterm Infants Among New York City Hospitals

Author Affiliations
  • 1Department of Population Health Science and Policy, Icahn School of Medicine at Mount Sinai, New York, New York
  • 2Women’s Health Research Institute, Icahn School of Medicine at Mount Sinai, New York, New York
  • 3Department of Obstetrics, Gynecology, and Reproductive Science, Icahn School of Medicine at Mount Sinai, New York, New York
  • 4Department of Health Services, University of Washington School of Public Health, Seattle, Washington
  • 5INSERM Joint Research Unit 1153, Obstetrical, Perinatal, and Pediatric Epidemiology Research Team (Epopé), Center for Epidemiology and Biostatistics Sorbonne Paris Cité, University Hospital Department Risks in Pregnancy, Paris Descartes University, Paris, France
JAMA Pediatr. 2018;172(3):269-277. doi:10.1001/jamapediatrics.2017.4402
Key Points

Question  Do differences in where non-Hispanic black, Hispanic, and non-Hispanic white very preterm infants are born contribute to racial and ethnic disparities in morbidity and mortality?

Findings  In a population-based cohort study including 7177 very preterm infants born in 39 New York City hospitals, hospital rates of risk-standardized very preterm birth morbidity and mortality varied widely, and non-Hispanic black and Hispanic infants were more likely than non-Hispanic white infants to be born in hospitals with the highest risk-standardized rates of morbidity and mortality. These differences in hospital of birth explained 39.9% of the black-white disparity and 29.5% of the Hispanic-white disparity in outcomes.

Meaning  Poor performance at hospitals where non-Hispanic black and Hispanic mothers deliver is an important and modifiable cause of racial disparities in neonatal outcomes.

Abstract

Importance  Substantial quality improvements in neonatal care have occurred over the past decade yet racial and ethnic disparities in morbidity and mortality remain. It is uncertain whether disparate patterns of care by race and ethnicity contribute to disparities in neonatal outcomes.

Objectives  To examine differences in neonatal morbidity and mortality rates among non-Hispanic black (black), Hispanic, and non-Hispanic white (white) very preterm infants and to determine whether these differences are explained by site of delivery.

Design, Setting, and Participants  Population-based retrospective cohort study of 7177 nonanomalous infants born between 24 and 31 completed gestational weeks in 39 New York City hospitals using linked 2010 to 2014 New York City discharge abstract and birth certificate data sets. Mixed-effects logistic regression with a random hospital-specific intercept was used to generate risk-adjusted neonatal morbidity and mortality rates for very preterm infants in each hospital. Hospitals were ranked using this measure, and differences in the distribution of black, Hispanic, and white very preterm births were assessed among these hospitals. The statistical analysis was performed in 2016-2017.

Exposure  Race/ethnicity.

Main Outcomes and Measures  Composite of mortality (neonatal or in-hospital up to 1 year) or severe neonatal morbidity (bronchopulmonary dysplasia, severe necrotizing enterocolitis, retinopathy of prematurity stage 3 or greater, or intraventricular hemorrhage grade 3 or greater).

Results  Among 7177 very preterm births (VPTBs), morbidity and mortality occurred in 2011 (28%) and was higher among black (893 [32.2%]) and Hispanic (610 [28.1%]) than white (319 [22.5%]) VPTBs (2-tailed P < .001). The risk-standardized morbidity and mortality rate was twice as great for VPTB infants born in hospitals in the highest morbidity and mortality tertile (0.40; 95% CI, 0.38-0.41) as for those born in the lowest morbidity and mortality tertile (0.16; 95% CI, 0.14-0.18). Black (1204 of 2775 [43.4%]) and Hispanic (746 of 2168 [34.4%]) VPTB infants were more likely than white (325 of 1418 [22.9%]) VPTB infants to be born in hospitals in the highest morbidity and mortality tertile (2-tailed P < .001; black-white difference, 20%; 95% CI, 18%-23% and Hispanic-white difference, 11%; 95% CI, 9%-14%). The largest proportion of the explained disparities can be attributed to differences in infant health risks among black, Hispanic, and white VPTB infants. However, 40% (95% CI, 30%-50%) of the black-white disparity and 30% (95% CI, 10%-49%) of the Hispanic-white disparity was explained by birth hospital.

Conclusions and Relevance  Black and Hispanic VPTB infants are more likely to be born at hospitals with higher risk-adjusted neonatal morbidity and mortality rates, and these differences contribute to excess morbidity and mortality among black and Hispanic infants.

Introduction

Substantial improvements in neonatal care have been documented over the past decade.1 From 2005 to 2014, rates of death before discharge and serious morbidities decreased among neonatal intensive care units (NICUs) in the United States; by 2014, most NICUs achieved rates of performance equivalent to the best quartile of the 2005 benchmark.1,2 In fact, 98% of NICUs achieved the risk-adjusted rate of mortality of the best 10% of units in 2005.2 Despite these improvements, wide variation in mortality and serious morbidities exist across NICUs.1,3,4

Although rates of death have decreased, persistent racial/ethnic neonatal mortality disparities exist, with non-Hispanic black neonates dying at more than twice the rate of non-Hispanic white neonates and Puerto Rican neonates also having an elevated risk of death.5 Limited research also suggests that, compared with non-Hispanic white preterm infants, non-Hispanic black and Hispanic preterm infants are at increased risk for severe neonatal morbidities, such as necrotizing enterocolitis,6 intraventricular hemorrhage,7 retinopathy of prematurity,7 and bronchopulmonary dysplasia.8 Disparities in these morbidities are associated with later neurodevelopmental disabilities, placing very preterm birth (VPTB) infants at a disadvantage over the life course.3,9

Data from the early 2000s document wide variation in very preterm neonatal outcomes across hospitals and showed that non-Hispanic black infants were more likely to be born in hospitals with poorer neonatal outcomes.10,11 Risks of morbidity have also been related to hospital factors and suggest that quality of care contributes to variation across hospitals.12 The extent to which these variations contribute to racial/ethnic disparities in neonatal mortality and morbidity is not known.

Given the quality-improvement trajectory over the past decade, we wondered whether disparate patterns of care by race and ethnicity were still apparent, whether hospital performance varied for both serious very preterm morbidities and mortality, and about the extent to which variations in hospital performance contributed to disparities in care for non-Hispanic black, Hispanic, and non-Hispanic white very preterm infants. Our objectives were to (1) measure risk-adjusted neonatal morbidity and mortality rates for VPTBs in New York City hospitals; (2) assess whether non-Hispanic black, Hispanic, and non-Hispanic white births occur at hospitals with different risk-adjusted neonatal morbidity and mortality rates; and (3) determine whether differences in site of care by race/ethnicity contribute to excess morbidity and mortality events.

Methods
Data Source

We used vital statistics birth records linked with New York State discharge abstract data, the Statewide Planning and Research Cooperative System (SPARCS),13 for all live births in New York City hospitals in years 2010 through 2014. The New York State Department of Health linked the data by using a probabilistic linking method, and 99.9% of infant discharge abstracts were linked with infant birth certificates. The linkage rate with maternal discharge abstracts was 96%. The linked data set included birth data, death data, and SPARCS record for all infants born between 2010 and 2014 who were discharged by December 31, 2014, and all deaths from 2010 to 2015. The linked data set comprised 596 295 births in 41 hospitals. We excluded births from 1 hospital that was not open for most of the study period, leaving 595 835 births in 40 hospitals. Additional maternal procedures and diagnosis codes were merged from a separate linked birth-maternal SPARCS data set after selecting the study cohort described below. Institutional review board approvals were obtained from the New York City Department of Health and Mental Hygiene, the New York State Department of Health, and the Icahn School of Medicine at Mount Sinai. We received a waiver of informed consent from the Icahn School of Medicine at Mount Sinai.

Cohort Selection

We created a VPTB cohort of 8311 infants born from 24 to 31 completed weeks of gestation. We excluded infants with congenital anomalies and infants born in 1 hospital with a total volume of VPTB infants that did not meet our inclusion criterion of more than 25 for 2010 through 2014 (n = 11). The final cohort for analysis was 7177 infants in 39 hospitals. For infants transferred before death (n = 522), mortality was attributed to the hospital of birth. Neonatal morbidities after transfer were not included in our data set.

Outcome Variable

Quiz Ref IDOur outcome variable was a combined measure of neonatal mortality and severe neonatal morbidity. Neonatal mortality was defined as death up to but not including 28 days, or within 1 year if continuously hospitalized. Severe neonatal morbidity was defined by the presence in the infant hospital record of the International Classification of Diseases, Ninth Revision code for any of the following diagnoses or procedures: bronchopulmonary dysplasia, necrotizing enterocolitis (unspecified, stage 2 or 3, laparotomy), retinopathy of prematurity (stage 3, 4, or 5), and intraventricular hemorrhage (grade 3 or 4) (eTable 1 in the Supplement).14

Covariates

We obtained maternal sociodemographic characteristics from the birth certificate, including race, ethnicity, maternal educational level, nativity, and age. We created a race/ethnicity variable by combining the race and information on Hispanic ethnicity into the following categories: non-Hispanic black, Hispanic, non-Hispanic white, Asian, and other. For this analysis, we refer to non-Hispanic black as black and non-Hispanic white as white. Maternal morbidities were ascertained using a combination of the mother’s SPARCS record and birth certificate to maximize sensitivity of our measures (eTable 2 in the Supplement).15,16 We used the clinical estimate of gestational length in weeks and birth weight from the birth certificate. Congenital anomalies were identified using diagnosis codes from the infant’s SPARCS record (eTable 3 in the Supplement).17 Similar to other models on neonatal mortality rates,10 we adjusted for the mother’s sociodemographic characteristics (age, race, nativity, educational level, and insurance), maternal behaviors (tobacco, alcohol, and drug use during pregnancy and prenatal care visits), maternal medical risk factors (parity, comorbidities, pregnancy complications, and body mass index), and infant factors (multiple birth, sex, birth weight, gestational age, Apgar score at 1 minute, and delivery method). Missing values for each variable were included as dummy variables in the model and ranged from 0.3% to 3.7%. The exception was delivery type; 7 observations were missing and therefore were excluded from the analysis.

Statistical Analysis

We conducted a bivariate analysis of maternal and infant covariates with neonatal morbidity and mortality using χ2 tests. We used mixed-effects logistic regression with a random hospital-specific intercept to generate risk-standardized neonatal morbidity and mortality rates for each hospital. The covariates for risk adjustment were maternal sociodemographic characteristics, including race/ethnicity, medical, and obstetric factors, as well as infant factors occurring before birth (Table 1). Because birth weight and gestational age are collinear, we calculated a z score of birth weight for gestational age using the 2010 US birth weight standard by sex.18 We then included gestational age and z score as risk-adjustment covariates. Hospital risk-standardized rates were computed from this model by using methods recommended by Hospital Compare.19 These rates were the ratio of predicted to expected neonatal morbidity and mortality rates multiplied by the sample mean neonatal morbidity and mortality rate. For each hospital, the numerator of the ratio is the predicted number of neonatal morbidity cases and deaths based on the hospital’s performance with its case mix, and the denominator is the expected number of cases and deaths based on the New York City performance with that hospital’s case mix. We then ranked hospitals from lowest to highest risk-standardized neonatal morbidity and mortality. We conducted sensitivity analyses of hospital rankings by repeating the analysis as follows: (1) with neonatal mortality only as the outcome; (2) within the cohort of very low-birth-weight infants based on 6594 infants with a birth weight from 500 to 1499 g; and (3) after excluding the Apgar score from the risk-adjustment model. Because our data did not include morbidities after an infant was transferred to another hospital or care facility and because transferred infants differed from nontransferred infants with regard to morbidity and mortality, we also conducted a fourth sensitivity analysis after excluding hospitals with level 2 nurseries because the hospitals that transferred a substantial proportion (>20%) of infants were all level 2 nurseries.

To assess racial/ethnic disparities in the use of hospitals, we used the Kolmogorov-Smirnov test to assess whether the distributions of VPTB infants among hospitals differed by racial and ethnic group.20 We used χ2 tests to compare the percentage of black, Hispanic, and white VPTB infants in the highest morbidity and mortality tertile of hospitals (defined by the number of hospitals in each group). We then explored interactions between maternal race/ethnicity and hospital morbidity and mortality tertile by comparing risk-adjusted neonatal morbidity and mortality rates by race/ethnicity in hospitals in the low–, middle–, and high–morbidity and mortality tertiles.

We also assessed how maternal and infant characteristics and choice of delivery hospital contributed to the observed ethnic disparities in neonatal morbidity and mortality rates. We grouped mothers’ characteristics into those representing health risks (age, parity, delivery type, smoking status, body mass index, pregnancy complications, and comorbidities) and socioeconomic status (educational level, insurance, and an indicator for non–US born). Infant characteristics included birth weight, gestational age, sex, and multiple pregnancies. We used methods described by Fairlie et al21 to assess the contribution of each of these 4 factors to disparities in neonatal morbidity and mortality. Briefly, these methods involved (1) taking a random sample of black mothers equal in number to the sample of white mothers; (2) matching the sampled black mothers to white mothers based on the predicted probability of morbidity and mortality from the first logit model described above, with dummy variables included for each hospital; (3) sequentially replacing each black mother’s characteristics, including choice of hospital, with the matched white mother’s characteristics for each of the mother, infant, and hospital choice characteristics; and (4) predicting morbidity and mortality for each black mother with these new sets of variables. Because the results of these steps depend on the sample of black mothers chosen and the sequence by which white mother, infant, and hospital characteristics are substituted for black mother characteristics, this process was repeated 500 times, taking a different sample of black mothers and a different sequence for substituting variables with each iteration and calculating the mean of the results.

A P value of .05 was set a priori to denote statistical significance. All P values were 2-sided. Statistical analyses were performed using SAS software, version 9.4 (SAS Institute Inc) or Stata, version 14 (StataCorp).

Results
Maternal and Infant Demographic Characteristics

Of 595 835 births, black infants accounted for 20.8% (n = 123 739), Hispanic infants for 30.3% (n = 180 340), and white infants for 32.0% (n = 190 676). Quiz Ref IDVery preterm births accounted for 1.5% of total births for these years: Quiz Ref ID2.8% of black births (n = 2775), 1.5% of Hispanic births (n = 2168), and 1.0% of white births (n = 1418). Black infants accounted for 38.4%; Hispanic infants, 30.1%; and white infants, 20.1% of all VPTB infants.Table 1 shows the sociodemographic and clinical characteristics of the VPTB infants in our study sample. Of the 7177 infants, 2011 (28.0%) died or experienced serious morbidity. There were significant differences in VPTB morbidity and mortality rates for some maternal and infant characteristics. As given in Table 1, 570 infants (7.9%) died and 1597 (22.5%) experienced a serious morbidity; morbidity rates differed significantly by race/ethnicity. Table 2 presents the results of our model fitting for neonatal morbidity and mortality using maternal and infant variables. Birth weight, gestational age, Apgar score at 1 minute, maternal body mass index, and maternal diabetes were significantly associated with neonatal morbidity and mortality composite outcome.

Hospital Risk-Standardized Morbidity and Mortality Rates

Twenty-seven of the 39 hospitals (69.2%) were private, 33 (84.6%) had level 3 or 4 nurseries, and all were teaching hospitals. Hospitals were ranked according to risk-standardized morbidity and mortality rates. Five-year risk-standardized rates ranged from 9.7% (95% CI, 3.8-20.0) to 57.7% (95% CI, 44.7-72.1) (Figure). The cumulative distribution of deliveries among hospitals ranked from lowest to highest morbidity and mortality rates differed for black vs white mothers (P < .001) and for Hispanic vs white mothers (P < .001) (eFigure 1 in the Supplement). We divided the hospitals in tertiles based on hospital ranking, and the adjusted morbidity and mortality risk for very preterm infants differed for infants born in low– (0.16; 95% CI, 0.14-0.18), middle– (0.25; 95% CI, 0.22-0.27), and high– (0.40; 95% CI, 0.38-0.41) morbidity and mortality hospitals. There was no racial/ethnic difference in adjusted risk for morbidity and mortality composite outcome within the low–, middle–, or high–morbidity and mortality tertiles as shown in Table 3. Quiz Ref IDHowever, 1204 of 2775 black (43.4%), 746 of 2168 Hispanic (34.4%), and 325 of 1418 white (22.9%) VPTB infants were born in the highest morbidity and mortality tertile of hospitals (P < .001). The risk difference for black compared with white very preterm infants was 20% (95% CI, 18%-23%); for Hispanic very preterm infants compared with white very preterm infants, it was 11% (95% CI, 9%-14%).

Sensitivity analyses using mortality only as the outcome produced very similar rankings of hospitals and racial/ethnic distributions among hospitals (eFigures 2 and 3 in the Supplement). When we excluded 6 hospitals with level 2 nurseries, the racial/ethnic differences in distributions were similar, with 1174 of 2717 black (43.2%), 695 of 2060 Hispanic (33.7%), and 321 of 1392 white (23.1%) VPTB infants (P < .001) born in the highest morbidity and mortality tertile of hospitals (eFigure 4 in the Supplement). Sensitivity analysis excluding Apgar score from the risk-adjustment model produced similar results (eFigure 5 in the Supplement).

Table 4 shows that 49.8% of the explained black-white disparity can be attributed to differences in infants’ health risks between black and white VPTB infants, and 39.9% can be attributed to the fact that white mothers received care at better-performing hospitals than black mothers. The contributions of black mothers’ health and socioeconomic status were negligible. Similarly, 48.4% of the explained Hispanic-white disparity can be attributed to differences in infant health risks between Hispanic and white VPTB infants, and 29.5% can be attributed to the fact that white mothers receive care at better-performing hospitals than Hispanic mothers.

Discussion

Black and Hispanic VPTB infants are more likely to be born in New York City hospitals with higher risk-adjusted neonatal morbidity and mortality rates than are white VPTB infants. Our data document that these disparate patterns of utilization explain 40% of the black-white racial disparity and nearly one-third of the Hispanic-white disparity in VPTB neonatal morbidity rates in New York City. Despite more than 10 years of quality improvement, poor performance at hospitals where black and Hispanic mothers deliver is an important and likely modifiable cause of racial and ethnic disparities in neonatal outcomes.

Our study is one of few to investigate hospital quality and racial and ethnic disparities in severe neonatal morbidity. In our analyses, observed racial and ethnic disparities in VPTB neonatal morbidity and mortality rates were driven by higher rates of severe neonatal morbidities among black and Hispanic VPTB infants. These morbidities are important adverse neonatal outcomes and are associated with additional diagnostic, therapeutic, and surgical interventions; increasing length of stay; risk for rehospitalization; and costs.3,22 In addition, they have major implications for disparities over the life course, as they are associated with long-term neurodevelopmental disabilities.3,9 These early disparities set a trajectory for later life, as health differences in the short term can persist and be amplified over the long term.23-27

Our results that black, Hispanic, and white VPTB infants are born in different hospitals are similar to findings by previous studies from more than a decade ago documenting that differences in where black and white babies were born contributed to black-white disparities in mortality rates of very low-birth-weight neonates.10,11 The study by Howell et al10 concluded that improving outcomes at the lowest-performing hospitals would result in a significant narrowing of disparities. During the past few decades, a large body of research from pediatrics and other areas of medicine has documented that black patients often receive care in a concentrated set of hospitals and that these hospitals have poorer quality of care and higher mortality rates.28,29

Why black and Hispanic women deliver their infants in poorer-performing hospitals is not known. Factors including distance to the hospital, insurance, hospital structural characteristics, patterns of racial segregation, community factors, physician referral, risk perception, patient choice, access, and the management of medical emergencies during pregnancy likely all contribute to why women deliver at specific hospitals.10,30 Previous research found that distance did not explain why black women were more likely to deliver at hospitals in New York City with higher risk-adjusted neonatal mortality.31

In the present study, we used a common decomposition tool from the literature21 and found that a significant proportion of both the black-white and the Hispanic-white VPTB disparities in neonatal morbidity and mortality could be attributed to infant factors, whereas maternal social and demographic factors did not explain these differentials. Although these maternal factors are strongly related to the probability of preterm birth, previous research has not found them to be predictors of very preterm outcomes in the NICU.32 In contrast, a significant proportion of the explained disparity could be attributed to birth hospital, suggesting that hospital quality does contribute to racial and ethnic VPTB morbidity and mortality gaps. Given the expanding body of literature associating hospital-level variables, such as volume of deliveries and nurse staffing ratios, with neonatal outcomes,17 more research is needed to elucidate the mechanisms by which these hospital characteristics may contribute to disparities.

Limitations

Our analysis has some limitations. Both vital statistics and SPARCS data have limitations with the reliability of specific variables.33-35 However, we combined both sources as recommended to optimize validity.16 We could not identify morbidity after transfer to another hospital, but sensitivity analyses removing hospitals with a higher proportion of transfers did not affect our results. We conducted a decomposition analysis and estimated the extent to which hospital of birth may contribute to observed racial disparities. However, we were unable to account for unmeasured factors that are associated with both race and VPTB neonatal mortality. Although we adjusted for some social factors likely associated with infant outcomes, others (eg, crime, neighborhood poverty, and environmental toxins) were not available in our data set.36 Furthermore, our cross-sectional data could not evaluate trends and did not include fetal deaths. Nevertheless, we conducted a population-based study based on linked data and were able to construct a robust risk-adjustment model that included important social and clinical confounders and documented that racial/ethnic disparities in hospital of birth continue. Our results may be generalizable to other large diverse urban areas.

Conclusions

We found that black and Hispanic VPTB babies are more likely than white VPTB babies to be born at hospitals with worse neonatal outcomes, and these differences in site of care contribute to disparities in VPTB neonatal morbidity and mortality in New York City. Disparities in severe neonatal morbidity are consequential, as these morbidities are major determinants of severe impairments, such as cerebral palsy, neurocognitive developmental delays, and mental health problems, and therefore perpetuate health and socioeconomic disparities across the life course.9 These disparities, coupled with the wide variation in hospital NICU performance across New York City hospitals, are alarming. Given the significant improvements in neonatal care over the past decade, it is time to direct these quality efforts to reduce racial and ethnic disparities in neonatal outcomes.

Back to top
Article Information

Corresponding Author: Elizabeth A. Howell, MD, MPP, Icahn School of Medicine at Mount Sinai, One Gustave L. Levy Place, Box 1077, New York, New York 10029 (elizabeth.howell@mountsinai.org).

Accepted for Publication: October 3, 2017.

Published Online: January 2, 2018. doi:10.1001/jamapediatrics.2017.4402

Author Contributions: Dr Janevic had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Howell, Janevic, Hebert, Egorova, Zeitlin.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Howell, Janevic, Hebert, Balbierz, Zeitlin.

Critical revision of the manuscript for important intellectual content: Janevic, Hebert, Egorova, Zeitlin.

Statistical analysis: Janevic, Hebert, Egorova, Zeitlin.

Obtained funding: Howell, Balbierz.

Administrative, technical, or material support: Egorova, Balbierz.

Study supervision: Howell, Balbierz, Zeitlin.

Conflict of Interest Disclosures: None reported.

Funding/Support. This study was supported by award R01HD078565 from the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References
1.
Horbar  JD, Edwards  EM, Greenberg  LT,  et al.  Variation in performance of neonatal intensive care units in the United States.  JAMA Pediatr. 2017;171(3):e164396.PubMedGoogle ScholarCrossref
2.
Lorch  SA.  A decade of improvement in neonatal intensive care: how do we continue the momentum?  JAMA Pediatr. 2017;171(3):e164395.PubMedGoogle ScholarCrossref
3.
Horbar  JD, Carpenter  JH, Badger  GJ,  et al.  Mortality and neonatal morbidity among infants 501 to 1500 grams from 2000 to 2009.  Pediatrics. 2012;129(6):1019-1026.PubMedGoogle ScholarCrossref
4.
Lapcharoensap  W, Gage  SC, Kan  P,  et al.  Hospital variation and risk factors for bronchopulmonary dysplasia in a population-based cohort.  JAMA Pediatr. 2015;169(2):e143676.PubMedGoogle ScholarCrossref
5.
Matthews  TJ, MacDorman  MF, Thoma  ME.  Infant mortality statistics from the 2013 period linked birth/infant death data set.  Natl Vital Stat Rep. 2015;64(9):1-30.PubMedGoogle Scholar
6.
Gephart  SM, Spitzer  AR, Effken  JA, Dodd  E, Halpern  M, McGrath  JM.  Discrimination of GutCheck(NEC): a clinical risk index for necrotizing enterocolitis.  J Perinatol. 2014;34(6):468-475.PubMedGoogle ScholarCrossref
7.
Wallace  ME, Mendola  P, Kim  SS,  et al.  Racial/ethnic differences in preterm perinatal outcomes.  Am J Obstet Gynecol. 2017;216(3):e301-e306, e312.Google Scholar
8.
Rowley  DL, Hogan  V.  Disparities in infant mortality and effective, equitable care: are infants suffering from benign neglect?  Annu Rev Public Health. 2012;33:75-87.PubMedGoogle ScholarCrossref
9.
Saigal  S, Doyle  LW.  An overview of mortality and sequelae of preterm birth from infancy to adulthood.  Lancet. 2008;371(9608):261-269.PubMedGoogle ScholarCrossref
10.
Howell  EA, Hebert  P, Chatterjee  S, Kleinman  LC, Chassin  MR.  Black/white differences in very low birth weight neonatal mortality rates among New York City hospitals.  Pediatrics. 2008;121(3):e407-e415.PubMedGoogle ScholarCrossref
11.
Morales  LS, Staiger  D, Horbar  JD,  et al.  Mortality among very low-birthweight infants in hospitals serving minority populations.  Am J Public Health. 2005;95(12):2206-2212.PubMedGoogle ScholarCrossref
12.
Synnes  AR, Chien  LY, Peliowski  A, Baboolal  R, Lee  SK; Canadian NICU Network.  Variations in intraventricular hemorrhage incidence rates among Canadian neonatal intensive care units.  J Pediatr. 2001;138(4):525-531.PubMedGoogle ScholarCrossref
13.
New York State Department of Health. Statewide Planning and Research Cooperative System (SPARCS). https://www.health.ny.gov/statistics/sparcs/. Revised June 2016. Accessed November 26, 2017.
14.
Jensen  EA, Lorch  SA.  Effects of a birth hospital’s neonatal intensive care unit level and annual volume of very low-birth-weight infant deliveries on morbidity and mortality.  JAMA Pediatr. 2015;169(8):e151906.PubMedGoogle ScholarCrossref
15.
Lain  SJ, Hadfield  RM, Raynes-Greenow  CH,  et al.  Quality of data in perinatal population health databases: a systematic review.  Med Care. 2012;50(4):e7-e20.PubMedGoogle ScholarCrossref
16.
Dietz  P, Bombard  J, Mulready-Ward  C,  et al.  Validation of selected items on the 2003 U.S. standard certificate of live birth: New York City and Vermont.  Public Health Rep. 2015;130(1):60-70.PubMedGoogle ScholarCrossref
17.
Phibbs  CS, Baker  LC, Caughey  AB, Danielsen  B, Schmitt  SK, Phibbs  RH.  Level and volume of neonatal intensive care and mortality in very-low-birth-weight infants.  N Engl J Med. 2007;356(21):2165-2175.PubMedGoogle ScholarCrossref
18.
Olsen  IE, Groveman  SA, Lawson  ML, Clark  RH, Zemel  BS.  New intrauterine growth curves based on United States data.  Pediatrics. 2010;125(2):e214-e224.PubMedGoogle ScholarCrossref
19.
Ash  AS, Normand  ST, Stukel  TA, Utts  J; Committee of Presidents of Statistical Societies (COPSS).  Statistical Issues in Assessing Hospital Performance. Washington, DC: Centers for Medicare & Medicaid Services; 2012.
20.
Hollander  M, Wolfe  DA.  Nonparametric Statistical Methods. 2nd ed. New York, NY: John Wiley & Sons; 1999.
21.
Fairlie  RW.  An extension of the Blinder–Oaxaca decomposition technique to logit and probit models.  J Econ Soc Meas. 2005;30:305-316.Google Scholar
22.
Johnson  TJ, Patel  AL, Jegier  BJ, Engstrom  JL, Meier  PP.  Cost of morbidities in very low birth weight infants.  J Pediatr. 2013;162(2):243-49.e1. doi:10.1016/j.jpeds.2012.07.013PubMedGoogle ScholarCrossref
23.
Braveman  P, Barclay  C.  Health disparities beginning in childhood: a life-course perspective.  Pediatrics. 2009;124(suppl 3):S163-S175.Google ScholarCrossref
24.
Braveman  P.  What is health equity: and how does a life-course approach take us further toward it?  Matern Child Health J. 2014;18(2):366-372.PubMedGoogle ScholarCrossref
25.
Elder  GH.  Perspectives on the Life Course: Life Course Dynamics: Trajectories and Transitions. Ithaca, NY: Cornell University Press; 1985:23-49.
26.
Lu  M, Halfon  N.  Racial and ethnic disparities in birth outcomes: a life-course perspective.  Matern Child Health J. 2003;7(1):13-30.Google ScholarCrossref
27.
Halfon  N, Hochstein  M.  Life course health development: an integrated framework for developing health, policy, and research.  Milbank Q. 2002;80(3):433-479, iii.PubMedGoogle ScholarCrossref
28.
Barnato  AE, Lucas  FL, Staiger  D, Wennberg  DE, Chandra  A.  Hospital-level racial disparities in acute myocardial infarction treatment and outcomes.  Med Care. 2005;43(4):308-319.PubMedGoogle ScholarCrossref
29.
Stansbury  JP, Jia  H, Williams  LS, Vogel  WB, Duncan  PW.  Ethnic disparities in stroke: epidemiology, acute care, and postacute outcomes.  Stroke. 2005;36(2):374-386.PubMedGoogle ScholarCrossref
30.
Howell  EA, Egorova  NN, Balbierz  A, Zeitlin  J, Hebert  PL.  Site of delivery contribution to black-white severe maternal morbidity disparity.  Am J Obstet Gynecol. 2016;215(2):143-152.PubMedGoogle ScholarCrossref
31.
Hebert  PL, Chassin  MR, Howell  EA.  The contribution of geography to black/white differences in the use of low neonatal mortality hospitals in New York City.  Med Care. 2011;49(2):200-206.PubMedGoogle ScholarCrossref
32.
Bonet  M, Smith  LK, Pilkington  H, Draper  ES, Zeitlin  J.  Neighbourhood deprivation and very preterm birth in an English and French cohort.  BMC Pregnancy Childbirth. 2013;13:97.PubMedGoogle ScholarCrossref
33.
Piper  JM, Mitchel  EF  Jr, Snowden  M, Hall  C, Adams  M, Taylor  P.  Validation of 1989 Tennessee birth certificates using maternal and newborn hospital records.  Am J Epidemiol. 1993;137(7):758-768.PubMedGoogle ScholarCrossref
34.
DiGiuseppe  DL, Aron  DC, Ranbom  L, Harper  DL, Rosenthal  GE.  Reliability of birth certificate data: a multi-hospital comparison to medical records information.  Matern Child Health J. 2002;6(3):169-179.PubMedGoogle ScholarCrossref
35.
Yasmeen  S, Romano  PS, Schembri  ME, Keyzer  JM, Gilbert  WM.  Accuracy of obstetric diagnoses and procedures in hospital discharge data.  Am J Obstet Gynecol. 2006;194(4):992-1001.PubMedGoogle ScholarCrossref
36.
Lorch  SA, Enlow  E.  The role of social determinants in explaining racial/ethnic disparities in perinatal outcomes.  Pediatr Res. 2016;79(1-2):141-147.PubMedGoogle ScholarCrossref
×