TABLE 2

Systematic Review Process

Variable, Author, and ReferenceDemographics of Study (If Available)Outcome(s) of InterestSummary of Main FindingsRelative Risk or Measures of Effect (If Applicable)Suggested Mechanism
Breast milk
 Lee et al2190% of infants admitted into Californian NICUs with birth weight of 500–1500 g and born at or transferred to a CPQCC center within 2 d of birth from 2006 to 2008.Any amount of BMF at time of discharge. Hospitals were divided into quartiles by hospital percentage of each race, and BMF rates by race were compared.BMF found to be higher for all racial and ethnic groups in hospitals with more white mothers. Differences were less pronounced for Hispanic mothers than for black mothers. African Americans had the lowest rates of BMF when cared for in hospitals with more African Americans (43%) versus hospitals with fewer African Americans (59%, P = .003). On the other hand, white mothers had the lowest BMF rates (49%) in hospitals with the fewest white mothers.With risk adjustment, white mothers were more likely to engage in BMF when there were more white mothers at that hospital (odds ratio 1.15 for 10% increase in white mothers, 95% confidence interval 1.04–1.28). Black mothers were less likely to engage in BMF when more black mothers were at the hospital (odds ratio 0.80 for each 10% increase in black mothers, 95% confidence interval 0.67–0.97).Hospitals serving more patients of color were less likely to have premature infants fed breast milk for all races. Targeting such hospitals for QI may help to improve BMF rates overall and reduce disparities.
 Profit et al918 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 was among modifiable measures of process rather than outcome, suggesting a critical role for QI efforts.
 Riley et al38410 VLBW infants born between February 2008 and 2012 and admitted to the Level III NICU at Rush University Medical Center in Chicago, Illinois.HM feeding at discharge as mitigated by neighborhood structural factors.In bivariate analysis HM feeding at discharge was negatively correlated with neighborhood structural factors including neighborhood concentrated disadvantage (OR: 0.72; CI: 0.60–0.86), neighborhood violent crime rate (OR: 0.93; CI: 0.89–0.98), and negatively correlated with patient factors including black race and/or ethnicity (OR: 0.41; CI: 0.24–0.70). In multivariate analysis, only maternal race and/or ethnicity (OR: 1.04; CI: 1.00–1.09), WIC eligibility, and length of NICU hospitalization predicted HM feeding at discharge for the entire cohort. The interaction between access to a car and race and/or ethnicity significantly differed between black and white or Asian mothers, although the predicted probability of HM feeding at discharge was not significantly affected by access to a car for any racial and/or ethnic subgroup.In multivariate analysis, maternal race and/or ethnicity (OR: 1.04; CI: 1.00-1.09) predicted HM feeding at discharge for entire cohort.Socioeconomic status affected breastfeeding rates in all racial/ethnic groups, but disproportionately affected black mothers to a greater degree than white or Hispanic mothers.
 Cricco-Lizza22130 black non-Hispanic mothers enrolled in the New York Special Supplemental Nutrition Program for WIC were general informants. From this group, 11 key informants had close follow-up during pregnancy and the first postpartum year.Audiotaped interviews and field notes were analyzed for mothers’ descriptions of infant-feeding education and support from nurses and physicians.Found limited breastfeeding education and support during pregnancy, childbirth stay in NICU, postpartum, and recovery in the community. They also expressed trust or distrust concerns and varying degrees of anxiety about how they were treated by nurses and physicians.N/ALack of trusting relationships (neglecting the affective elements of health care).
 Fleurant et al23362 racially diverse mothers of infants with VLBW.HM feeding at discharge.For all 362 mothers, WIC negatively predicted HM feeding at discharge and maternal goal near time of discharge positively predicted HM feeding at discharge. Perceived breastfeeding support from infant’s maternal grandmother negatively predicted HM feeding at discharge.For all mothers: WIC eligibility (OR: 0.34; 0.15–0.75; P = .008), breastfeeding support from mothers’ mother (OR: 0.45; 0.26–0.79; P = .005), goal of any HM near discharge (OR: 8.38; 3.42–20.53; P ≤ .001). Stratified by race and/or ethnicity: “For black mothers, support from the mother’s mother (OR 0.27 [95% CI 0.11–0.68], P = .006).Goal of HM at discharge predicting HM feeding consistent with other research. Surprised by finding that higher maternal grandmother support predicts lower BM. Suggestion that this relates to maybe the perception of support (rather than actual support) and that mothers coparenting with their own mothers are more likely to experience parenting stress.
 Brownell et al24Mothers of infants eligible to receive donor milk (≤32 weeks’ gestation or ≤1800 g) born between August 2010 and 2015.Odds of nonconsent.Of the 486 mother-infant dyads from the first 5 y of the donor milk program, nonwhite race (aOR: 1.69; 95% CI: 1.04–2.76) and increasing GA (aOR: 1.11; 95% CI: 1.03–1.21) independently predicted nonconsent. Each year the program existed, there was a 48% reduction in odds of nonconsent (aOR: 0.52; 95% CI: 0.43–0.62). The most common reason given for nonconsent was ‘‘it’s someone else’s milk.’’Nonwhite race (aOR: 1.69; 95% CI: 1.04–2.76); increasing GA (aOR: 1.11; 95% CI: 1.03–1.21) independently predicted nonconsent.Program duration was associated with reduced nonconsent rates and may reflect increased exposure to information and acceptance of donor milk use among NICU staff and parents. Despite overall improvements in consent rates, race-specific disparities in rates of nonconsent for donor milk persisted after 5 y of this donor milk program.
 Boundy et al25Data from CDC’s 2015 Maternity Practices in Infant Nutrition and Care survey, linked to the 2011–2015 US Census Bureau’s American Community Survey.The use of mother’s own milk and donor milk in hospitals with NICUs by the percentage of non-Hispanic black residents in the hospital postal code area, categorized as being above or below the national average (12.3%).In postal codes with >12.3% black residents, 48.9% of hospitals reported using mothers’ own milk in ≥75% of infants in the NICU, and 38.0% reported not using donor milk, compared with 63.8% and 29.6% of hospitals, respectively, in postal codes with ≤12.3% black residents.N/ADifferences found may be related to variations in health care personnel support, hospital policies and practices, mothers’ knowledge and access to information, and community-level support for breastfeeding. Donor milk use might also be affected by hospital proximity to milk banks, state regulations, and hospital policies related to the provision of donor milk and insurance reimbursement.
 Vohr et al26Infants weighing ≤1250 g cared for in an open-bay NICU (January 2007–August 2009) (n = 394) versus those in a SFR NICU (January 2010–December 2011) (n = 297).Human milk provision at 1, 4 wk and discharge, and 4 wk volume (mL/kg/d). Also, 18–24 mo Bayley III.Infants cared for in the SFR NICU had higher Bayley III cognitive and language scores, higher rates of human milk provision at 1 and 4 wk, and higher human milk volume at 4 wk. The SFR NICU was associated with a 2.55-point increase in Bayley cognitive scores and 3.70-point increase in language scores. Every 10 mL/kg per d increase of human milk at 4 wk was associated with increases in Bayley cognitive, language, and motor scores (0.29, 0.34, and 0.24, respectively). Medicaid was associated with decreased cognitive and language scores, and low maternal education and nonwhite race with decreased language scores.Nonwhite race with decreased language scores (−5.8).Authors do not comment on race or ethnicity but note that low maternal education, poverty, and insurance status likely mean less ability to spend time in the NICU and provide breast milk, thus leading to lower Bayley III cognitive scores.
High-risk infant follow-up referrals
 Barfield et al271233 infants with VLBW (<1200 g) in Massachusetts born January 1998–June 2003Rates of referral and time to referral by race and/or ethnicity.Black infants had lowest referral rates, especially at 0–12 mo. 12% black infants not referred at all, compared with 6.8% overall.aHR of referral of black non-Hispanic infants compared with white non-Hispanic infants was 0.85.Possible that minority families are less confident in their ability to access specialty health care services, such as EI. Also mentions barriers to access, physician mistrust, concerns and/or misunderstandings regarding treatment, and disparities in insurance status (which was included and adjusted for when calculated aHR, but mentioned in discussion).
 Hintz et al28CPQCC infants <1500 g, 2010–2011High-risk infant follow-up referral rates.Lower odds of referral for infants with maternal race Hispanic or African American versus white.aOR compared with white: African American: 0.58, Hispanic: 0.65.Previous research has pointed to issues of noncompliance and lacking support systems; in unadjusted analyses, hospitals serving highest proportion of African American infants have lower referral rates overall.
Parental satisfaction and family experience
 Martin et al29Self-reports from non-Hispanic black and non-Hispanic white participants who had an infant born at a GA ≤35 wk or birth wt <2000 g, presenting within 2 mo after NICU discharge to any 1 of 30 Children’s Hospital of Philadelphia primary care centers between January 1, 2010, and January 1, 2013.Parental reported trust, communication style, expectations of the health care system, and satisfaction with their physician and the NICU course. Collected parental data included sex, age, employment, education, home ownership, ethnicity, race, marital status, household resident information, and income.Although more commenting parents were white (62%) than black (38%), black parent comments were more negative (58% negative vs 42% positive) compared with white parent comments (33% negative vs 67% positive). The nurse-parent relationship and distinct positive and negative nursing behaviors are important factors affecting parental satisfaction with NICU care. Black parents were most dissatisfied with nursing support, wanting compassionate and respectful communication and nurses that were attentive to their children. White parents were most dissatisfied with inconsistent nursing care and lack of informative exchanges, wanting education about their child’s short- and long-term needs. Both groups described a chaotic NICU environment with high nursing turnover.N/ABlack patients are less likely to report satisfaction with health care, possibly stemming from differences in provider communication styles, clinician’s attitudes, medical mistrust, and perceived racism. Provider uncertainty, biases, and stereotyping may also contribute to unequal treatment.
 Sigurdson et al30Attendees at the 2016 Vermont Oxford Network Quality Congress composed of providers (nurses, physicians, and other clinical specialists) and NICU family advocates.Open-ended survey of accounts of disparate care.Study gathered 324 accounts of disparate care, majority described perceived worse care of families, not strictly infants. Accounts described worse care based on intersecting factors of, eg, language, culture or ethnicity, race, etc. Respondents described families receiving neglectful care, judgmental care, and systemic barriers to care.N/AAdverse consequences of disparate care for infants mediated through adverse interactions with families more so than through differences in care to infant. Social inequalities shape the content and tone of health care encounters such that families receive worse care based on intersecting factors.
Shared parent-provider decision-making
 Van McCrary et al31Two cases in a New York City tertiary care unit.Nonmedical barriers that impede decision-making in the NICU.Many NICUs are not equipped to deal with complicated cultural and linguistic barriers that prevent them from communicating effectively with their patients.N/AFor both families of color in these case studies, a reinforcing cycle of disapproval had formed. Each family made, or had previously made, decisions that the NICU staff did not agree with, so the staff implicitly resented the family. In both cases, care decisions were influenced by social and cultural factors that were misunderstood by NICU medical staff.
 Tucker Edmonds et al32Periviable infants (23.0–24.6 wk) in California, Missouri, and Pennsylvania, 1995–2005.Resuscitation of periviable infants.Black race a predictor for neonatal intubation, even when controlling for clustering at the delivery hospital level.aOR for intubation of black infants compared with white: 1.25.Likely due to differences in patient preferences (eg, religious beliefs, cultural preferences); hospital-level practices or resources could also be a contributing factor.
Kangaroo care
 Hendricks-Muñoz et al3342 nurses and 143 mothers at 2 New York City hospitals.Maternal and provider perspectives on KMC and MCP.Mothers of color perceived that they received less education and access to KMC; nurses of color were more supportive of MCP and KMC.61% of mothers of color strongly identified that access to KMC had been limited to them compared with 39% of white mothers; 77% of white mothers compared with only 50% of mothers of color strongly perceived that nurses were supportive in helping them provide KMC for their infant; OR of nurses of color encouraging parent presence compared with white: 2.8.Cultural and linguistic differences leading to communication difficulties.
Surfactant and RDS
 Hamvas et al34Infants with VLBW born 1987–1989 and 1991–1992 in Saint Louis area; 1563 infants in total, 315 deaths.Effects of surfactant approval on neonatal mortality among black and white infants.Neonatal mortality decreased more for white than black infants after surfactant was approved; after approval, white VLBW mortality dropped 41% and VLBW mortality for black infants did not change.Relative risk of death among black newborns with VLBW compared with white, 1987–1989: 0.7. Increased to 1.3 in 1991–1992.Larger portion of neonatal mortality in white neonates was attributed to RDS, so introduction of surfactant had a larger effect; RDS more prevalent among white infants because fetal pulmonary surfactant matures more slowly in white infants.
 Ranganathan et al3544 712 African American infants and 73 942 non-Hispanic white infants.Effects of surfactant approval on neonatal mortality among black and non-Hispanic white infants with VLBW.Introduction of surfactant therapy in 1990 improved the outcomes for white infants with VLBW more than black. No difference between African American and non-Hispanic white in rate of decline for all categories of mortality between 1985 and 1988; Difference between African American and non-Hispanic white in rate of decline for all except nonrespiratory neonatal between 1988 and 1991.1988–1991: Odds of neonatal death from RDS declined 34% in non-Hispanic white infants and only 16% in African American infants (P < .01); odds of death from all respiratory causes declined 41% in non-Hispanic white infants and 22% in African American infants (P < .01).Differential efficacy of surfactant on non-Hispanic white and African American infants with VLBW. Lungs of African American fetuses mature more rapidly and begin to produce pulmonary surfactant earlier; therefore, exogenous surfactant may have fewer additional benefits; African American infants with VLBW may also respond less favorably than non-Hispanic white infants of same severity.
 Frisbie et al36All US infants born 1989–1990 (5 407 166) and 1995–1998 (10 809 746).Black-white disparity in infant mortality due to RDS and other causes.Disparity in RDS mortality between black and white infants increased after introduction of surfactant. Absolute declines in mortality greater for white infants than black infants among LBW infants after the introduction of surfactant. Black infants had a relative survival advantage in presurfactant era; this became a survival disadvantage postsurfactant.Infant mortality due to RDS in black infants (versus white infants): 1989–1990, OR = 0.832. 1995–1998, OR = 1.114.Social inequality and differential access to intervention between black and white infants.
 Howell et al37All US infants born 1989–1990 (5 407 166) and 1995–1998 (10 809 746).Black-white disparity in infant mortality due to RDS and other causes.Disparity in RDS mortality between black and white infants increased after introduction of surfactant. Absolute declines in mortality greater for whites than blacks among LBW infants after the introduction of surfactant. Black infants had a relative survival advantage in presurfactant era; this became a survival disadvantage postsurfactant.Infant mortality due to RDS in black infants (versus white infants): 1989–1990, OR = 0.832. 1995–1998, OR = 1.114.Social inequality and differential access to intervention between black and white infants.
  • “African American” and “black” are often used interchangeably in the literature reviewed. In this table, we use the same language as the articles cited. aHR, adjusted hazard ratio; aOR, adjusted odds ratio; BM, breast milk; BMF, breast milk feeding; CDC, Centers for Disease Control and Prevention; CI, confidence interval; CPQCC, California Perinatal Quality Care Collaborative; EI, early intervention; HM, human milk; KMC, kangaroo mother care; LBW, low birth weight; MCP, maternal care partnerships; N/A, not available; OR, odds ratio; SFR, single family room.