TABLE 1.
Systematic Review Structure
| Variable, Author, and Reference | Demographics of Study (If Available) | Outcome(s) of Interest | Summary of Main Findings | Relative Risk or Measures of Effect (If Applicable) | Suggested Mechanism |
|---|---|---|---|---|---|
| Nursing characteristics | |||||
| Lake et al10 | 8252 infants with VLBW | Relationship between nursing characteristics and hospital-level disparities in VLBW outcomes | Higher infection rates, higher rates of discharge without breast milk, more nurse understaffing, and poorer practice environments in hospitals with a high proportion of black patients (“high-black”) compared with hospitals with a low proportion (“low-black”); 70% of black infants with VLBW are born at high-black hospitals, so these disparities strongly affect them. Black infants are particularly disadvantaged even in high-black hospitals when it comes to breastfeeding. Black infants and white infants no different within hospitals overall. | aOR of nosocomial infection in high-black versus low: 1.64. aOR for discharge without BM high versus low: 1.61. Differences not significant when controlling for nursing characteristics. Practice Environment (scale of 1–4): high-black 2.95, low-black 3.16, P = .004. Understaffing (measured in fraction of a nurse per infant needed to meet guidelines): high-black 0.22, low-black 0.14, P = .004. | Nursing characteristic differences likely due to financial constraints and poor management. Nursing characteristics likely drive infection and discharge without breast milk. |
| Lake et al11 | Random sample survey of licensed nurses in 4 large US states: 1037 staff nurses in 134 NICUs. | Average patient load, individual nurses’ patient load, professional nursing characteristics, nurse work environment, and nursing care missed on the last shift. | More care activities are missed in hospitals serving a higher proportion of black infants (>31%) than in hospitals that serve a smaller proportion of black infants (<11%). This indicates that quality of care may be lower in high-black hospitals. | Nurses in high-black NICUs missed nearly 50% more care activities than those in low-black NICUs (average 1.51 vs 1.05 activities missed per shift, P = .03). | Disparities in missed care were likely due to higher patient-to-nurse ratios in high-black hospitals. High-black hospitals were more likely to have larger proportion of Medicaid patients, which can lead to financial strain and therefore fewer staffing resources. |
| Appropriate setting | |||||
| Bronstein et al12 | Infants with VLBW; 1118 white and 1478 infants of color. | Delivery in hospitals with NICUs. | Mothers of color with early prenatal care were the group most likely to deliver their infants with VLBW in hospitals with NICUs. White women with Medicaid were more likely to be transferred before birth than white women with private insurance. | OR: 1.353 (1.119–1.636) for mothers of color versus white mothers. OR: 1.733 (1.316–2.283) for mothers of color with Medicaid and early prenatal care versus white mothers with no Medicaid and early prenatal care. | Women with early prenatal care are likely able to contact their care providers sooner if they go into labor prematurely or are part of care systems with established referral relationships for emergency situations. Some hospitals may selectively retain privately insured women for high-risk deliveries, refer less Medicaid-insured women to subspecialty regional centers. |
| Gould et al13 | Data from 1 state (analyzed by region) from 1989 to 1993 of 24 094 live-born infants with VLBW. | Factors associated with likelihood of VLBW birth at a Level 1 hospital (hospital without a NICU or 24-h on-call neonatologist) | 10.5% (24 094) infants with VLBW were delivered in Level 1 hospitals. Significant regional variation: from 3.1% to 24.3%. The odds were decreased for African Americans and Southeast Asians and increased in Hispanic women as compared with white non-Hispanic women. For all women, less than adequate prenatal care, living in a 50%–75% urban zip code, and living >25 miles from the nearest NICU significantly increased the odds of VLBW delivery at a Level 1 hospital. | Odds of inappropriate delivery site ranged from 0.37 to 2.75 across California’s 9 geographic perinatal regions. OR for African Americans: 0.65. OR for Southeast Asians: 0.54. OR for white 1.00. OR for Hispanic 1.16. | Common to all 3 racial ethnic groups: risks of Level 1 VLBW delivery associated with (1) less than adequate prenatal care, (2) living in a zip code that is only 50%–75% urban, and (3) living at a distance to a hospital with a NICU. However, regional differences in the odds for inappropriate VLBW birth remained after adjusting for contributing socio-demographics and geographical risk factors. Hispanic teenage pregnancies were at particular risk of inappropriate birth setting. |
| Gortmaker et al14 | Data from 4 states for 1978 and 1979 used to estimate survival curves for first 24 h of life. | Survival by hospital setting (high-technology centers versus urban or rural centers). | Among 750–1000 g infants: cumulative probability of survival at Level III center was ∼0.63 among white infants and ∼0.70 among black infants. Among 1000–1500 g: ∼0.90 among white infants and ∼95 among black infants (Level III). | Black infants had better access to specialized services. In Washington and Tennessee, >50% of black infants with VLBW were born in specialized centers. For all birth weight categories, survival during the first 96 h was greater among black than white infants at the same level of hospital care. Birth in a Level III center conferred the highest survival rates. | Birth in a Level III center greatly improves infant outcomes. Differential access to specialized Level III centers between black infants and white infants may explain the higher observed survival rates in black infants. |
| Geography | |||||
| Hebert et al15 | VLBW deliveries in New York City from 1996 to 2001 to non-Hispanic black and non-Hispanic white mothers. | The role of geographic distribution of hospitals in the racial disparity in the use of top-tier hospitals (those in the lowest tertile of hospitals ranked by ratio of observed to expected deaths). | Black mothers less likely to deliver in top-tier hospitals (white = 44%, black = 28%; P < .001). Top-tier hospitals less likely to be in black mothers’ neighborhoods (white = 40%, black = 33%; P < .001). Distance, however, did not contribute to the disparity in use of top-tier hospitals: mothers of both races often bypassed their neighborhood hospital (black = 62% bypassed, white = 71%; P < .001). | OR of relative risk black or white of probability that a mother of a neonate with VLBW used a top-tier hospital: 0.6. | The influence of geography on the use of top-tier hospitals for mothers of neonates with VLBW is complex. Other personal and hospital characteristics, not just distance or geography, also influenced hospital use in New York City. Researchers provide insights to those who would report risk adjusted hospital mortality rankings to foster top-tier selection. However, improving quality at poorly performing facilities could make greater impact than encouraging choice of better-performing hospitals. |
| Featherstone et al16 | The linked birth and death records of singleton infants with VLBW born between 2010 and 2012 (n = 2030). | Impact of travel time from maternal residence to delivery hospital on neonatal mortality rate. | No significant association between travel time to delivery hospital and neonatal mortality after adjusting for confounders. 1-wk increase in GA and non-Hispanic black mothers (versus non-Hispanic white mothers) were associated with lower odds of neonatal death, whereas non- NICU admission at birth was associated with increased odds of death. | 1-wk increase in GA associated with lower odds of mortality: (OR: 0.61); non-Hispanic black mothers (OR: 0.68) associated with lower death; non-NICU admission at birth (OR: 5.9) increased odds of death. | A high proportion of neonatal deaths occurred within 24 h of birth. Underlying causes of death, particularly in the first 24 h, are not sensitive to access to care but are more closely aligned with other maternal, neonatal, or hospital-level factors. |
| Minority-serving hospitals | |||||
| Morales et al17 | 74 050 infants with VLBW treated by 332 VON hospitals between 1995 and 2000 | Whether there is an association between proportion of black infants with VLBW treated at a hospital and neonatal mortality for black and white infants with VLBW. “Minority-serving” defined as >35% of infants with VLBW treated were black; other categories were <15% and 15%–35%. | Minority-serving hospitals had significantly higher risk-adjusted neonatal mortality rates than <15%. Differences were not explained by either hospital or treatment variables. | Risk-adjusted neonatal mortality rates in >35% black hospitals compared with <15%: white OR: 1.30, black OR: 1.29, pooled OR: 1.28 | Minority-serving hospitals may be providing lower quality of care to infants with VLBW. Results were not explained by other hospital characteristics looked at in this study (location, teaching status, % admissions covered by Medicaid), but there could be other characteristics not looked at in this study that might have an effect. |
| Howell et al18 | 11781 infants 500–1499 g in New York City born between 1996 and 2001. | Risk-adjusted neonatal mortality rates for each New York City hospital and racial and ethnic distribution among those hospitals. | White infants with VLBW more likely to be born in lowest mortality tertile of hospitals (49%) compared with black infants with VLBW (29%). Estimated that if black women delivered in same hospitals as white women, black VLBW mortality rates would be reduced by 6.7 per 1000 VLBW births or disparity would be reduced by 34.5%. | — | — |
| Military vs civilian care | |||||
| Kugler et al19 | Black or white singleton live births in Pierce County delivered between 1982 and 1985. | The effect of system of care (military care) on differences in low birth weight and neonatal mortality between black infants and white infants. | (Note: not including findings on low birth weight findings because not quality of care.) Civilian black infants had approximately twice the neonatal death rates of civilian white infants. The neonatal mortality rates for military black infants, however, did not differ significantly from either group of white infants. Disparity did not apply to birth weight <2500 g. | Civilian black infants versus civilian white infants crude risk ratio: 2.33; civilian black infants versus military black infants crude risk ratio: 3.08; military black infants versus military white infants crude risk ratio: 1.04. | Military care appeared to be associated with a protective effect for neonatal mortality for black infants. This effect was not due to differences in birth weight distribution or to the quantity of prenatal care received. The effect was most prominent for normal weight black infants, especially for those from low income census tracts. |
| Composite quality | |||||
| Profit et al9 | 18 616 infants with VLBW in 134 California NICUs between January 1, 2010, and December 31, 2014. | Baby-MONITOR score (a composite of 9 process and outcome measures of quality). For each NICU, a risk-adjusted composite and individual component quality score for each race and ethnicity. | Composite quality scores ranged by 5.26 standard units (range: −2.30 to 2.96). Non-Hispanic white infants higher on measures of process compared with non-Hispanic black infants and Hispanic infants. Compared with white infants, non-Hispanic black infants scored higher on measures of outcome; Hispanic infants scored lower on 7 of the 9 Baby-MONITOR subcomponents. | Difference between highest and lowest performing NICUs was large (5.2 standard units). | Some of the disparity created by inferior performance among modifiable measures of process rather than outcome, suggesting a critical role for QI efforts. |
| Howell et al20 | 7177 infants, GA 24–31 wk (very preterm) | Mortality or severe neonatal morbidity | Morbidity and mortality higher among black and Hispanic than white VPTBs. Risk-adjusted morbidity and mortality rate was twice as high for VPTBs in hospitals in the highest morbidity and mortality tertile than those born in the lowest tertile hospitals. Black and Hispanic VPTBs were more likely to occur in the highest morbidity and mortality hospitals than white. Most of the disparity can be attributed to differences in infant health risks, but birth hospital also plays a significant role. | Percentage of disparity explained by birth hospital: black-white = 40% (95% CI, 30%–50%); Hispanic-white 30% (95% CI, 10%–49%). | 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 may all contribute to black and Hispanic women giving birth at hospitals with higher mortality and morbidity rates. |
“African American” and “black” are often used interchangeably in the literature reviewed. In this table, we use the same language as the articles cited. aOR, adjusted odds ratio; BM, breast milk; CI, confidence interval; OR, odds ratio; VON, Vermont Oxford Network; VPTB, very preterm birth; —, not applicable.