Abstract
Objectives
To examine the characteristics of safety net (sn) and non-sn neonatal intensive care units (NICUs) in California and evaluate whether the site of care is associated with clinical outcomes.
Study design
This population-based retrospective cohort study of 34 snNICUs and 104 non-snNICUs included 22 081 infants born between 2014 and 2018 with a birth weight of 401-1500 g or gestational age of 22-29 weeks. Quality of care as measured by the Baby-MONITOR score and rates of survival without major morbidity were compared between snNICUs and non-snNICUs.
Results
Black and Hispanic infants were cared for disproportionately in snNICUs, where care and outcomes varied widely. We found no significant differences in Baby-Measure Of Neonatal InTensive care Outcomes Research (MONITOR) scores (z-score [SD]: snNICUs, −0.31 [1.3]; non-snNICUs, 0.03 [1.1]; P = .1). Among individual components, infants in snNICUs exhibited lower rates of human milk nutrition at discharge (−0.64 [1.0] vs 0.27 [0.9]), lower rates of no health care-associated infection (−0.27 [1.1] vs 0.14 [0.9]), and higher rates of no hypothermia on admission (0.39 [0.7] vs −0.25 [1.1]). We found small but significant differences in survival without major morbidity (adjusted rate, 65.9% [95% CI, 63.9%-67.9%] for snNICUs vs 68.3% [95% CI, 67.0%-69.6%] for non-snNICUs; P = .02) and in some of its components; snNICUs had higher rates of necrotizing enterocolitis (3.8% [3.4%-4.3%] vs 3.1% [95% CI, 2.8%-3.4%]) and mortality (95% CI, 7.1% [6.5%-7.7%] vs 6.6% [6.2%-7.0%]).
Conclusions
snNICUs achieved similar performance as non-snNICUs in quality of care except for small but significant differences in any human milk at discharge, infection, hypothermia, necrotizing enterocolitis, and mortality.
Each year in the US, ~ 45 000 infants with very low birth weight (VLBW; <1500 g) require care in a neonatal intensive care unit (NICU) and incur much of the $35 billion annual total cost of preterm birth.1,2 Stark variations in the quality of care and outcomes exist that are unexplained by underlying clinical risks. Racial and ethnic disparities could be due to disparities in care within NICUs, with Black and Hispanic infants receiving lower-quality care,3 or disparities between NICUs, with infants segregated by race and ethnicity into lower-quality NICUs.4-8
Safety net (sn) NICUs treat predominantly low-income and underserved populations and are at the forefront of providing care for Black and Hispanic populations, those insured by Medicaid, and other vulnerable patients,9,10 who shoulder many social risk factors that impact maternal health and prenatal care11-13 and disproportionately experience poverty, and housing and food insecurity, which affect health insurance, access to care, and health capital.14-16 sn hospitals may face resource limitations in physical equipment, staffing, ancillary services, and subspecialty support.9,10 Although not specific to snNICUs, studies in New York City found that birth hospitals contributed significantly to disparities in outcomes, and that lower-performing NICUs had greater proportions of Black and Hispanic patients.6,17,18 Although studies in California also found significant racial disparities in outcomes, site of care contributed less to these disparities than in New York City.3,8,19
Few studies have specifically examined snNICUs,20 and it remains unclear how snNICUs vary in care and outcomes and contribute to these patterns of neonatal care. The goal of this population-based study was to examine whether infants born in snNICUs face disparities in quality of care and health outcomes. To address this question, we examined the characteristics of snNICUs and non-snNICUs in California and evaluated the extent to which site of care is associated with quality of care, clinical outcomes, and racial/ethnic disparities.
Methods
We analyzed data from 138 NICUs in the California Perinatal Quality Care Collaborative (CPQCC), representing >90% of California NICUs.21 All infants with VLBW (401-1500 g) and gestational age of 22-29 weeks born between January 1, 2014, and December 31, 2018, and cared for in the CPQCC were included (Figure 1; available at www.jpeds.com). We excluded infants who died in the delivery room or within 12 hours after delivery, were previously discharged home, or had congenital anomalies. Using an algorithm developed with the Vermont Oxford Network to ensure correct attribution of outcomes to NICUs for the Baby-MONITOR score, we also excluded infants with >3 admissions from 2 separate hospitals, for a final cohort of 23 551 admissions from 22 081 infants (Figure 1). Details of the algorithm are available elsewhere.3,8,22,23 This study was approved by the Stanford University Institutional Review Board.
Figure 1.
Study flow chart.
Outcomes of Interest
The outcomes were NICU quality of care as measured by the Baby-MONITOR score22 and survival without major morbidity. The risk-adjusted Baby-MONITOR composite indicator assesses NICU quality of care for infants with VLBW and includes 9 individually risk-adjusted measures, including antenatal steroid administration, hypothermia on admission, non–surgically induced pneumothorax, health care–associated infection, chronic lung disease, timely retinal exam, discharge on any human breastmilk, death during the birth hospitalization, and high growth velocity.22,24 Each metric uses clinical data from each CPQCC NICU and is individually risk-adjusted, standardized using the Draper–Gittoes method, and expressed such that a higher score represents a more positive outcome.25 A standardized observed minus expected z-score is calculated, equally weighted, and averaged to derive a Baby-MONITOR score for each NICU, expressed in standard units.3,22,25
Survival without major morbidity is a risk-adjusted measure that includes the absence of the following: infant death during birth hospitalization (counting transfers among NICUs) up to age 1 year, chronic lung disease, severe peri-intraventricular hemorrhage, nosocomial infection, necrotizing enterocolitis (NEC), severe retinopathy of prematurity or surgery for retinopathy of prematurity, and cystic periventricular leukomalacia.26 Each component was individually risk-adjusted for factors including sex, gestational age, 5-minute Apgar score, small for gestational age status, multiple birth status, race and/or ethnicity, inborn or outborn status, and receipt of prenatal care. The CPQCC race/ethnicity classification scheme is based on maternal race and includes non-Hispanic White, non-Hispanic Black, and Hispanic (of any race) and Asian/Pacific Islanders, and for this analysis, collapses the American Indian or Alaskan Native and other race groups into an “other” category because of small sample sizes. We attributed survival without major morbidity to the birth hospital. Standard CPQCC definitions were applied for each component, as described in detail elsewhere.26
Definition of Safety Net Hospitals
snNICUs treat predominantly low-income and underserved populations and are at the forefront of providing care for Black and Hispanic populations, those insured by Medicaid, and other vulnerable patients.9 The literature on sn hospitals has not established a standard definition.20 Previously proposed definitions vary, characterizing medically vulnerable populations by poverty level, lower educational attainment, unemployment, race, or shares of uninsured or Medicaid patients as a proxy for financial strain or health care access.9,20,27,28 Although definitions have varied, in the present study we defined sn hospitals by including county hospitals, as well as those serving >66% Medi-Cal (California’s Medicaid) patients and >15 infants with VLBW annually. Specifically, we began with lists from the California Association of Public Hospitals29 and America’s Essential Hospitals, which are hospitals dedicated to caring for the medically underserved.30 All county hospitals were included on the foregoing lists and automatically designated as snNICUs. For all others, we omitted level 1 or non-CPQCC NI-CUs. For the remaining member NICUs, we linked to 2017 birth certificate data from the California Department of Public Health. After examining distributions of Medi-Cal, NICUs with Medi-Cal shares >66% were designated as potential snNICUs. Among these candidates, noncounty hospitals treating <15 infants with VLBW annually were not designated as snNICUs, to ensure adequate sample size and verify that each snNICU treats a substantial number of vulnerable patients, for a final sample of 34 snNICUs and 104 non-snNICUs. All hospitals and their numbers of infants with VLBW and Medi-Cal shares are listed in Table I (available at www.jpeds.com). The geographic location of the snNICUs is shown in Figure 2. In a separate sensitivity analysis, we raised the Medi-Cal cutoff to >75%, resulting in 19 snNICUs, and also assessed differences in performance between snNICUs and non-snNICUs with this definition.
Table I.
Percentage of infants with VLBW with Medi-Cal as the payer source
| Non-sn hospitals |
sn hospitals* |
||||
|---|---|---|---|---|---|
| Hospital | Medi-Cal patients, % | Infants with VLBW annually | Hospital | Medi-Cal patients, % | Infants with VLBW annually |
| 1 | 0.0 | 17.8 | 105 | 60.3 | 19.4 |
| 2 | 0.0 | 39.8 | 106 | 65.2 | 53.4 |
| 3 | 0.0 | 1.8 | 107 | 66.2 | 165.8 |
| 4 | 0.0 | 9 | 108 | 67.0 | 64.4 |
| 5 | 0.0 | 9.8 | 109 | 67.1 | 45 |
| 6 | 0.0 | 20.2 | 110 | 68.9 | 88.6 |
| 7 | 0.0 | 0.6 | 111 | 69.0 | 30.2 |
| 8 | 0.6 | 69.8 | 112 | 69.4 | 116.6 |
| 9 | 1.1 | 39 | 113 | 69.4 | 15.4 |
| 10 | 1.3 | 41.4 | 114 | 69.7 | 68.8 |
| 11 | 1.7 | 43 | 115 | 70.8 | 19.2 |
| 12 | 2.2 | 40.6 | 116 | 72.0 | 28.6 |
| 13 | 2.3 | 36 | 117 | 74.0 | 26.2 |
| 14 | 2.6 | 21.4 | 118 | 75.4 | 30.6 |
| 15 | 3.0 | 50.6 | 119 | 76.0 | 96.4 |
| 16 | 3.6 | 3.8 | 120 | 76.1 | 171.6 |
| 17 | 4.3 | 10.4 | 121 | 80.6 | 24.8 |
| 18 | 6.0 | 47.2 | 122 | 83.7 | 23.6 |
| 19 | 6.2 | 54.6 | 123 | 83.7 | 46.8 |
| 20 | 8.2 | 37.2 | 124 | 84.4 | 25.2 |
| 21 | 8.6 | 25.8 | 125 | 85.1 | 32.4 |
| 22 | 8.6 | 24.8 | 126 | 86.7 | 13.8 |
| 23 | 8.9 | 10.2 | 127 | 88.4 | 29.4 |
| 24 | 9.9 | 86.8 | 128 | 89.3 | 44.4 |
| 25 | 10.9 | 80.6 | 129 | 90.6 | 15.8 |
| 26 | 11.6 | 64.2 | 130 | 90.6 | 27.6 |
| 27 | 12.7 | 55.6 | 131 | 90.8 | 50.6 |
| 28 | 14.3 | 5.8 | 132 | 92.7 | 43 |
| 29 | 15.1 | 48.8 | 133 | 93.8 | 35.8 |
| 30 | 15.2 | 8.2 | 134 | 94.6 | 29.2 |
| 31 | 16.0 | 22 | 135 | 95.8 | 11.4 |
| 32 | 16.7 | 13.4 | 136 | 98.1 | 41.2 |
| 33 | 17.9 | 20.4 | 137 | 100.0 | 27.4 |
| 34 | 19.8 | 99 | 138 | 100.0 | 43.2 |
| 35 | 21.4 | 21.4 | |||
| 36 | 22.0 | 13.4 | |||
| 37 | 23.1 | 18.6 | |||
| 38 | 25.0 | 5 | |||
| 39 | 25.0 | 17.8 | |||
| 40 | 27.8 | 51.8 | |||
| 41 | 29.4 | 14.8 | |||
| 42 | 30.4 | 54.2 | |||
| 43 | 31.2 | 43 | |||
| 44 | 32.1 | 22.4 | |||
| 45 | 32.5 | 77.6 | |||
| 46 | 33.3 | 1.4 | |||
| 47 | 34.0 | 20.2 | |||
| 48 | 34.2 | 6.8 | |||
| 49 | 36.4 | 26.6 | |||
| 50 | 37.5 | 13.4 | |||
| 51 | 38.3 | 109.8 | |||
| 52 | 40.1 | 87 | |||
| 53 | 40.7 | 22.8 | |||
| 54 | 40.7 | 14.6 | |||
| 55 | 41.3 | 27.6 | |||
| 56 | 42.7 | 27.2 | |||
| 57 | 44.4 | 21.2 | |||
| 58 | 44.8 | 35.2 | |||
| 59 | 45.9 | 20.2 | |||
| 60 | 46.2 | 1.6 | |||
| 61 | 46.6 | 87.6 | |||
| 62 | 47.2 | 76.8 | |||
| 63 | 47.5 | 16.8 | |||
| 64 | 48.9 | 39.2 | |||
| 65 | 49.3 | 29 | |||
| 66 | 50.0 | 9.8 | |||
| 67 | 52.0 | 53.8 | |||
| 68 | 52.7 | 12 | |||
| 69 | 53.8 | 15 | |||
| 70 | 54.1 | 18.6 | |||
| 71 | 54.3 | 7.6 | |||
| 72 | 57.9 | 7.8 | |||
| 73 | 58.2 | 24.2 | |||
| 74 | 58.3 | 68 | |||
| 75 | 58.3 | 9.6 | |||
| 76 | 60.0 | 4.8 | |||
| 77 | 60.8 | 140.8 | |||
| 78 | 61.0 | 117.6 | |||
| 79 | 61.5 | 32.4 | |||
| 80 | 62.1 | 17.8 | |||
| 81 | 62.5 | 9 | |||
| 82 | 62.9 | 12.6 | |||
| 83 | 63.1 | 12.6 | |||
| 84 | 63.5 | 71.8 | |||
| 85 | 63.6 | 16.6 | |||
| 86 | 65.4 | 15.4 | |||
| 87 | 66.7 | 0.8 | |||
| 88 | 69.7 | 13.8 | |||
| 89 | 70.8 | 14.6 | |||
| 90 | 73.5 | 13.2 | |||
| 91 | 73.9 | 19.8 | |||
| 92 | 77.4 | 14.2 | |||
| 93 | 78.7 | 6.2 | |||
| 94 | 87.5 | 13 | |||
| 95 | 100.0 | 9 | |||
| 96 | 100.0 | 2.4 | |||
| 97 | 100.0 | 2.6 | |||
| 98 | 100.0 | 12.8 | |||
| 99 | No data | ||||
| 100 | No data | ||||
| 101 | No data | ||||
| 102 | No data | ||||
| 103 | No data | ||||
| 104 | No data | ||||
County hospitals might not meet the threshold of 66% Medi-Cal or 15 VLBWs per year, as they were automatically designated sn hospitals.
Figure 2.
Locations of snNICUs in California, with the average number of infants with VLBW treated annually.
Statistical Analyses
We examined distributions of infant, maternal, and hospital characteristics between snNICUs and non-snNICUs using mean (SD) and the chi-square test. We obtained hospital-level data from California’s Office of Statewide Health Planning and Development and CPQCC databases, including American Academy of Pediatrics31 and California Children’s Services NICU levels32; maternal socioeconomic data from the linked birth certificates; and clinical data from the CPQCC. We used the Kolmogorov–Smirnov test to examine differences in racial/ethnicity distributions among snNICUs and non-snNICUs. We calculated Baby-MONITOR z-scores for each NICU,22,25 comparing mean scores for snNICUs and non-snNICUs with t tests.
In sensitivity analyses, we stratified Baby-MONITOR scores into process and outcome measures. To calculate adjusted rates of survival without major morbidity and its individual components for each NICU, we fitted multivariable logistic regression models, adjusting for the aforementioned covariates.26 We calculated the expected number of cases per year for each outcome and calculated a standardized incidence ratio, which is the ratio of observed cases to expected cases. Multiplying the standardized incidence ratio by the population rate results in the adjusted rate, with confidence intervals calculated using the Poisson exact method. We compared adjusted rates between snNICUs and non-snNICUs using the t test. We ranked NICUs by their quality scores or adjusted rates and constructed caterpillar plots to visualize where snNICUs stand across the network. We further stratified snNICUs and non-snNICUs by teaching and nonteaching hospitals and compared quality and outcomes. All analyses were performed with SAS version 9.4 (SAS Institute), and R Studio version 1.1.442 (R Foundation for Statistical Computing).
Results
Cohort and Hospital Characteristics
The distributions of infant, maternal, and hospital characteristics between snNICUs and non-snNICUs are shown in Table II. In the CPQCC network, 31% of infants were cared for in snNICUs. The snNICUs had higher American Academy of Pediatrics levels of care, representing higher proportions of hospitals capable of subspecialty intensive care,31 as well as higher numbers of licensed beds. There were no significant differences in nursing hours per patient-day and numbers of infants with VLBW cared for per year. However, 38% of mothers cared for in snNICUs had a high school degree (or General Educational Development) or less, compared with 22% of those in non-snNICUs (P < .001). We found no differences in the distribution of foreign-born mothers vs US-born mothers.
Table II.
Characteristics of sn and non-sn hospitals, 2014-2018
| Characteristics | sn hospitals |
Non-sn hospitals |
P value* |
|---|---|---|---|
| Total admissions, n (%) | 7183(30) | 16 368(70) | — |
| AAP level of care, n (%)† | |||
| 2 | 0 (0) | 24 (23) | .02 |
| 3 | 26 (76) | 61 (59) | |
| 4 | 7 (21) | 17 (16) | |
| Unknown | 1 (3) | 2 (2) | |
| CCS level of care, n (%)† | |||
| Non-CCS | 0 (0) | 14 (13) | .004 |
| Community | 26 (76) | 58 (56) | |
| Intermediate | 0 (0) | 17 (16) | |
| Regional | 8 (24) | 15 (14) | |
| Hospital ownership, n (%) | |||
| Government | 11 (32) | 4 (4) | <.001 |
| Nonprofit | 15 (44) | 85 (82) | |
| Investor | 4 (15) | 13 (13) | |
| Other | 3 (9) | 2 (2) | |
| Teaching hospital, n (%) | |||
| No | 21 (62) | 87 (92) | <.001 |
| Yes | 8 (38) | 8 (8) | |
| Urban or rural, n (%)‡ | |||
| Rural | 1 (3) | 1 (1) | .45 |
| Urban | 33 (97) | 93 (99) | |
| Licensed NICU beds | |||
| Mean (SD) | 29 (19) | 22 (17) | .04 |
| 0-11 (Q1), n (%) | 3 (9) | 28 (27) | .03 |
| 12-19 (Q2), n(%) | 6 (18) | 29 (28) | |
| 20-29 (Q3), n (%) | 13 (38) | 23 (22) | |
| ≥30 (Q4), n (%) | 12 (35) | 23 (22) | |
| Participation in any QI collaboration, n (%) | |||
| No | 17 (50) | 58 (56) | .56 |
| Yes | 17 (50) | 46 (44) | |
| Participation in >1 QI collaboration, n (%) | |||
| No | 22 (65) | 79 (76) | .2 |
| Yes | 12 (35) | 25 (24) | |
| Nursing hours per patient day, mean (SD) | 16.5 (6.8) | 15.8 (5.3) | .58 |
| Infants with VLBW per year, mean (SD) | 42 (34) | 31 (32) | .1 |
| Medi-Cal, %, mean (SD) | 80.8 (11.7) | 36.8 (27.4) | <.001 |
| Infant characteristic | |||
| Racial/ethnicity distribution, n (%)§ | |||
| Asian/Pacific Islander | 472 (7) | 2650 (17) | <.001 |
| Hispanic | 4221 (59) | 6032 (38) | |
| Non-Hispanic Black | 1029 (14) | 1957 (12) | |
| Non-Hispanic White | 1218 (17) | 4705 (30) | |
| Other | 196 (3) | 522 (3) | |
| Any nonWhite | 5918 (83) | 11 161 (70) | <.001 |
| Maternal characteristics | |||
| Age, y, mean (SD) | 29.0 (6.5) | 30.9 (6.3) | <001 |
| Hypertension, n (%) | |||
| No | 4952 (69) | 10 850 (66) | <.001 |
| Yes | 2214 (31) | 5497 (34) | |
| Chorioamnionitis, n (%) | |||
| No | 6687 (93) | 14 922 (91) | <.001 |
| Yes | 481 (7) | 1427 (9) | |
| Diabetes, n (%) | |||
| No | 6320 (88) | 14 065 (86) | <.001 |
| Yes | 847 (12) | 2282 (14) | |
| Maternal education, n (%)¶ | |||
| <High school | 842 (12) | 871 (5) | <.001 |
| High school graduate/General Educational Development | 1866 (26) | 2825 (17) | |
| Some college/Associate’s degree | 1665 (23) | 3700 (23) | |
| Bachelor’s degree or above | 934 (13) | 4378 (27) | |
| Unknown/missing/withheld | 1876 (26) | 4594 (28) | |
| Country of birth, n (%)¶ | |||
| Foreign born | 1851 (26) | 4271 (26) | .82 |
| US born | 3844 (54) | 8751 (53) | |
| Unknown | 1488 (21) | 3346 (20) |
AAP, American Academy of Pediatrics; CCS, California Children’s Services.
Data are shown as the number of admissions.
Chi-square test or t test between sn and non-sn hospitals.
Most recent level of care recorded.
Rural hospitals are defined according to California Health and Safety Code Section 124840.
Race missing for 549 infants.
Data on maternal education and country of birth not yet available for 2018.
Distribution of Infants by Race and Ethnicity
NonWhite infants with VLBW, particularly Black and Hispanic infants, were cared for disproportionately in snNICUs. Specifically, 34% of Black infants and 41% of Hispanic infants received care in an snNICU, compared with 21% of White infants (P < .001). Overall, 35% of nonWhite infants and 21% of White infants were cared for in an snNICU (P < .001).
NICU Quality and Infant Outcomes
Figure 3, A and B shows the NICU level distributions of Baby-MONITOR scores and adjusted rates of survival without major morbidity for snNICUs and non-snNICUs. Among snNICUs, quality scores ranged from −2.0 to 5.7 standard units, and adjusted rates of survival without major morbidity ranged from 15.0 to 88.4 per 100 infants. SnNICUs were distributed similarly to non-snNICUs across the network (P = .36) and skewed toward lower performance, with 5 (15%) and 6 (18%) snNICUs achieving top quartile performance on Baby-MONITOR and survival without a major morbidity, and with 11 (32%) and 14 (41%) snNICUs in the bottom quartile of Baby-MONITOR and survival without major morbidity (Table III; available at www.jpeds.com).
Figure 3.
A, Baby-MONITOR composite scores by hospital type. NICUs are ranked according to their Baby-MONITOR scores on the x-axis, and their actual scores are plotted on the y-axis. B, Adjusted rates of survival without major morbidity by hospital type. NICUs are ranked according to their adjusted rates on the x-axis, and their actual adjusted rates are plotted on the y-axis.
Table III.
Distribution of hospitals within quartiles of Baby-MONITOR and survival without major morbidity
| Outcomes | Non-sn hospitals (N = 104) |
sn hospitals (N = 34) |
|---|---|---|
| Baby-MONITOR, n (%) | ||
| Lowest quartile | 23 (22) | 11 (32) |
| 2nd quartile | 26 (25) | 9 (26) |
| 3rd quartile | 25 (24) | 9 (26) |
| Highest quartile | 30 (29) | 5 (15) |
| P = .36 | ||
| Survival without major morbidity | ||
| Lowest quartile | 21 (20) | 14 (41) |
| 2nd quartile | 26 (25) | 8 (24) |
| 3rd quartile | 28 (27) | 6 (18) |
| Highest quartile | 29 (28) | 6 (18) |
| P = .09 | ||
Tables IV and V show patient-level Baby-MONITOR scores and adjusted rates of survival without a major morbidity. Overall, we found no significant differences in NICU quality scores between snNICUs and non-snNICUs, with a mean Baby-MONITOR score of −0.31 (1.3) for snNICUs and 0.03 (1.1) for all other NICUs (P = .13). When quality scores were aggregated into process and outcome measures, there were no significant differences between snNICUs and non-snNICUs for process measures (P = .90), but there was a nonsignificant trend toward lower performance in snNICUs for outcome measures (P = .06). However, we did find some significant differences across individual components of the Baby-MONITOR. In particular, snNICUs had significantly lower rates of any human milk at discharge and higher rates of health care–associated infection; however, they achieved significantly lower rates of moderate hypothermia on admission (Tables IV).
Table IV.
Quality scores of various outcomes across hospital types, 2014-2018
| Outcomes | sn hospitals, z-score, mean (SD)* | Non-sn hospitals, z-score, mean (SD) | P value† |
|---|---|---|---|
| Baby-MONITOR quality score‡ | −0.31 (1.3) | 0.03 (1.1) | .13 |
| No hypothermia | 0.39 (0.7) | −0.25 (1.1) | <.001 |
| No pneumothorax | 0.29 (1.0) | 0.09 (1.0) | .3 |
| Timely eye exam | 0.10 (0.7) | −0.28 (1.1) | .06 |
| Survival | −0.08 (1.2) | 0.14 (0.9) | .34 |
| High growth velocity | −0.16 (1.0) | 0.02 (1.0) | .35 |
| No chronic lung disease | −0.23 (1.2) | 0.12 (0.9) | .13 |
| No health care–associated infection | −0.27 (1.1) | 0.14 (0.9) | .04 |
| Antenatal steroids | −0.37 (1.1) | −0.18 (1.0) | .36 |
| Breastmilk at discharge | −0.64 (1.0) | 0.27 (0.9) | <.001 |
| Outcome measures | −0.20 (1.3) | 0.23 (1.1) | .06 |
| Process measures | −0.24 (1.2) | −0.21 (1.1) | .90 |
All components of Baby-MONITOR are risk-adjusted. Baby-MONITOR components are adjusted for sex, gestational age, 5-minute Apgar score, small for gestational age status, multiple birth status, inborn or outborn status, receipt of prenatal care, and cesarean birth.
t test for equal means across hospital type.
Admission-level data.
Table V.
Adjusted risk of various outcomes across hospital types, 2014-2018
| Outcomes | sn hospitals, adjusted rate (95% CI)*,† | Non-sn hospitals, adjusted rate (95% CI) | P value‡ |
|---|---|---|---|
| Survival without major morbidity§ | 65.9 (63.9-67.9) | 68.3 (67.0-69.6) | .02 |
| Severe peri-intraventricular hemorrhage | 6.0 (5.5-6.6) | 6.7 (6.3-7.1) | .27 |
| Severe ROP or ROP surgery | 5.9 (5.3-6.6) | 5.9 (5.4-6.3) | .07 |
| Cystic PVL | 2.3 (1.9-2.7) | 2.1 (1.9-2.4) | .21 |
| Necrotizing enterocolitis | 3.8 (3.4-4.3) | 3.1 (2.8-3.4) | .04 |
| Mortality | 7.1 (6.5-7.7) | 6.6 (6.2-7.0) | .01 |
| Chronic lung disease | 21.4(20.3-22.6) | 20.6 (19.8-21.4) | .21 |
| Health care–associated infection | 8.8 (8.1-9.5) | 7.5 (7.1-8.0) | .07 |
PVL, periventricular leukomalacia; ROP, retinopathy of prematurity.
Rate per 100 infants.
All components of survival without major morbidity are risk-adjusted. Survival without major morbidity components is adjusted for sex, gestational age, 5-minute Apgar score, small for gestational age status, multiple birth status, race/ethnicity, inborn or outborn status, and receipt of prenatal care
t test for equal means across hospital type.
Attribution to birth hospital.
We found clinically small but statistically significant differences in adjusted rates of survival without major morbidity between snNICUs and non-snNICUs (Table V). snNICUs showed an adjusted rate of 65.9 (95% CI, 63.9-67.9) per 100 infants, and non-snNICUs showed an adjusted rate of 68.3 (95% CI, 67.0-69.6) per 100 infants (P = .02). Care in an snNICU was associated with higher infant mortality (7.1 [95% CI, 6.5-7.7] vs 6.6 [95% CI, 6.2-7.0] per 100 infants; P = .01) and higher rates of NEC (3.8% [95% CI, 3.4%-4.3%] vs 3.1% [95% CI, 2.8%-3.4%] per 100 infants; P = .04). In a separate sensitivity analysis, we raised the Medi-Cal cutoff for defining safety net hospitals to >75%, which resulted in 19 snNICUs. Although using this cutoff resulted in fewer snNICUs, our findings remained robust, showing no significant difference in performance between snNICUs and non-snNICUs.
Further stratification showed that teaching snNICUs performed as well as nonteaching snNICUs. Among non-snNICUs, however, teaching NICUs showed a nonsignificant trend toward lower quality of care compared with nonteaching NICUs. The only exception was survival without major morbidity, for which teaching NICUs had significantly lower rates than nonteaching NICUs (59.0 [95% CI, 54.6-63.6] vs 69.1 [95% CI, 67.7-70.5] per 100 infants). Similarly, outcomes in infants did not differ by gestational age; outcomes in infants of <28 weeks were not significantly different from those of infants of >28 weeks (data not shown). Finally, we found that designation as an snNICU was not significantly associated with performance in the bottom quartile of either Baby-MONITOR or survival without major morbidity.
Discussion
In this study, we characterized differences in quality of care and survival without major morbidity among a population-based cohort of infants with VLBW born in snNICUs and non-snNICUs in California and examined the distribution of infant race and ethnicity cared for in these NICUs. Our primary aim was to understand how snNICUs may differ on quality and outcomes, and whether disparities or variations among snNICUs in California may serve as a learning point for safety net hospitals that are similarly positioned to serve vulnerable and minority populations.
The main findings are that Black and Hispanic infants with VLBW are disproportionately cared for in snNICUs in California, and that variation in the quality of care and outcomes exists in both snNICUs and non-snNICUs. snNICUs were not associated with worse measures of overall NICU quality but exhibited slightly worse rates of survival without a major morbidity. Although previous research has shown wide variations in quality and outcomes in the CPQCC network,3,23,26 this study specifically highlights snNICUs, which were distributed evenly across the network in terms of performance, with 5 (15%) and 6 (18%) snNICUs achieving top quartile performance on Baby-MONITOR and survival without major morbidity, respectively (Tables IV and V). Although we noted a shift toward lower performance among snNICUs, with 11 (32%) and 14 (41%) snNICUs in the bottom quartile of Baby-MONITOR and survival without major morbidity, respectively, snNICUs were distributed similarly to non-snNICUs overall (P = .36), and the observed variation across snNICUs was similar to that seen across NICUs as a whole in both quality of care and survival without major morbidity.3,23,26 However, both variation and rates of outcomes were not as pronounced as those seen in New York City hospitals6,33 and across the Vermont Oxford Network,34 revealing possible learning opportunities from snNICUs in California to reduce variation and improve outcomes.
The reasons for the variations in snNICU performance are unknown. Potential contributors to overall similarities in the quality of care and outcomes between snNICUs and non-snNICUs may include California’s stringent regulatory oversight and high standards for staffing, NICU designations, and quality control and improvement (ie, CPQCC). NICUs can audit and benchmark their data and are part of a statewide community of providers dedicated to optimizing the quality of care and sharing best practices. Beyond population differences, several underlying pathways could explain the observed variations. Separate geographic regions may see different manifestations of structural racism in their communities, such as residential segregation, health care system policies, and organization, or structures in place for prenatal or perinatal care access and neonatal or maternal transport. The role of quality infrastructure also may explain the different patterns of site-specific racial and ethnic disparities in quality of care and outcomes. California’s history of quality improvement (QI) collaboration, statewide QI initiatives, and efforts to address inequities in health care delivery could contribute to reduced gaps in performance between snNICUs and non-snNICUs. However, high quality of care does not always translate to equitable care, and QI initiatives may widen the quality gap for vulnerable populations.35
Our findings are consistent with the literature on care provision in resource-constrained settings.27 In adult safety net health systems in California, wide variations in performance have been found among measures for ambulatory patient safety.36 Importantly, even in a unique state with large private health care systems, such as Kaiser Permanente, providing a high quality of care, some snNICUs manage to achieve equally high performance and quality of care despite serving vulnerable populations and possible resource limitations. These high-performing snNICUs may hold important lessons and serve as models for other snNICUs.
Our characterization of snNICUs and non-snNICUs by distribution of race and ethnicity provides a lens through which to examine and address health disparities and inequity, as snNICUs disproportionately care for Black and Hispanic infants. Given this uneven distribution, our finding of no difference in NICU quality between snNICUs and non-snNICUs is surprising as it contrasts with previous literature that found disparities in NICU quality and outcomes by race and ethnicity that were attributable to the site of care. In studies from New York City, Black and Hispanic infants were more likely to be born in hospitals with higher morbidity and mortality rates, and birth hospital explained 40% of the Black–White disparity and 30% of the Hispanic–White disparity in NICU outcomes and maternal morbidity.6,17,18 The Vermont Oxford Network also found lower quality of care and worse neonatal outcomes in hospitals that care for a larger population of minority infants.7 In our study, several snNICUs, which care for predominantly Black and Hispanic infants, achieved high performance, and in California, the birth hospital accounted for only 8% of the Black–White disparity in maternal morbidity.19 Closer investigations of high- and low-performing safety net hospitals are needed to improve care for vulnerable communities as well as to elucidate factors contributing to different patterns of care.
Composite measures provide a summary score of quality or outcomes, but they may obscure individual metric results. Although we found no significant differences in overall Baby- MONITOR scores, the difference in effect size between snNI-CUs and non-snNICUs still may be clinically relevant. The difference was driven primarily by significantly lower rates of any human milk at discharge (P < .001) and higher rates of health care–associated infection (P = .04) among snNICUs. Previous studies have found significantly poorer outcomes for health care–associated infection and breastmilk feeding, as well as lower rate of antenatal steroid use, among Black and Hispanic infants, who were cared for mostly in snNICUs, thus explaining the difference in effect size seen between snNICUs and non-snNICUs.3,6,8,37-40
The findings of a lower rate of breastmilk use and a higher rate of NEC among snNICUs are notable, as we recently reported that human milk use was a significant mediator of racial/ethnic disparities in NEC and, when adjusted for, attenuated the disparity in NEC incidence between Black or Hispanic and White infants.41 Previous studies also have demonstrated a lower risk for NEC in VLBW infants who received human milk.42,43 These results further highlight the disparities within snNICUs. Our findings of higher NEC rates in snNICUs fit into a larger narrative that we have shown in previous research, as Black and Hispanic infants, whose mothers historically have lower rates of providing milk in the NICU, were cared for predominantly in snNICUs.3,43-46 Future interventions and policy changes should target barriers to human milk provision in NICUs, including family leave, access to donor milk, and maternal factors such as work, transportation, and comorbidities.44
Our findings also show a trend toward poorer performance among teaching NICUs. However, these findings on the effect of teaching hospitals need further investigation, as California’s Office of Statewide Health Planning and Development defines a teaching hospital as a hospital with any residency program, and many hospitals classified as teaching may have small residency programs and may perform very differently from large academic medical centers.
This study must be viewed in light of its design. There is currently no standard definition of a safety net hospital, and studies have shown that different definitions may affect observed results.20,47 In a separate sensitivity analysis, we raised the Medi-Cal cutoff to >75%, resulting in 19 snNICUs, but our findings remained robust. Moreover, decisions regarding metric calculation may introduce bias from misclassification of outcomes. For example, attribution of outcomes for the approximately 20% of transferred infants is complex. For survival without major morbidity, the outcome was attributed to the birth hospital. This attribution choice also results in bias and might not be fair to the birth hospital. Fortunately, previous sensitivity analyses have shown that the calculation of outcomes is robust to different transfer scenarios.3 Furthermore, there may be unmeasured confounding from factors contributing to hospital performance that are not captured in our data. Finally, this study encompasses only a single state and may not generalize to other regions of the US. However, our study is population-based from the nation’s most populous state that is racially, ethnically, and geographically diverse, and it was an intention of this study to draw a regional contrast with previous work done in New York City.
In conclusion, California snNICUs had similar quality of care as non-snNICUs but lower survival without major morbidity. Infants in snNICUs had significantly lower rates of human milk nutrition at discharge, and higher rates of health care–associated infection and infant mortality. Elucidating factors contributing to these differing patterns in quality and outcomes is important for further optimizing care and outcomes for sick newborns requiring intensive care in California and beyond. ■
Supplementary Material
Funding sources:
Supported by grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01 HD083368-01 and R01 HD08467-01, PI J.P.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the Eunice Kennedy Shriver National Institute of Child Health and Human Development or the National Institutes of Health. The funders/sponsors did not participate in any part of the work. The authors declare no conflicts of interest.
Glossary
- CPQCC
California Perinatal Quality Care Collaborative
- MONITOR
Measure Of Neonatal InTensive care Outcomes Research
- NEC
Necrotizing enterocolitis
- NICU
Neonatal intensive care unit
- QI
Quality improvement
- Sn
Safety net
- VLBW
Very low birth weight
References
- 1.Institute of Medicine. A systematic review of costs associated with preterm birth. In: Behrman RE, Butler AS, eds. Preterm birth: causes, consequences, and prevention. Washington: National Academies Press; 2007. p. 688–724. [PubMed] [Google Scholar]
- 2.Centers for Disease Control and Prevention. Natality information: live births. Natality 2016-2020 (expanded). Accessed October 19, 2021. https://wonder.cdc.gov/natality.html [Google Scholar]
- 3.Profit J, Gould JB, Bennett M, Goldstein BA, Draper D, Phibbs CS, et al. Racial/ethnic disparity in NICU quality of care delivery. Pediatrics 2017;140:e20170918. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Horbar JD, Edwards EM, Greenberg LT, Profit J, Draper D, Helkey D, et al. Racial segregation and inequality in the neonatal intensive care unit for very low-birth-weight and very preterm infants. JAMA Pediatr 2019;173:455–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lake ET, Staiger D, Horbar J, Kenny MJ, Patrick T, Rogowski JA. Disparities in perinatal quality outcomes for very low birth weight infants in neonatal intensive care. Health Serv Res 2015;50:374–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Howell EA, Janevic T, Hebert PL, Egorova NN, Balbierz A, Zeitlin J. Differences in morbidity and mortality rates in black, white, and Hispanic very preterm infants among NewYork City hospitals. JAMA Pediatr 2018;172:269–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Morales LS, Staiger D, Horbar JD, Carpenter J, Kenny M, Geppert J, et al. Mortality among very low-birthweight infants in hospitals serving minority populations. Am J Public Health 2005;95:2206–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Liu J, Sakarovitch C, Sigurdson K, Lee HC, Profit J. Disparities in health care-associated infections in the NICU. Am J Perinatol 2020;37:166–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Institute of Medicine (US), Committee on the Changing Market, Managed Care, Future Viability of Safety Net Providers. In: Lewin ME, Altman S, eds. Americas’s health care safety net: intact but endangered. Washington: National Academies Press; 2000. [PubMed] [Google Scholar]
- 10.Werner RM, Goldman LE, Dudley RA. Comparison of change in quality of care between safety-net and non-safety-net hospitals. JAMA 2008;299:2180–7. [DOI] [PubMed] [Google Scholar]
- 11.Becker G. Deadly inequality in the health care “safety net”: uninsured ethnic minorities’ struggle to live with life-threatening illnesses. Med Anthropol Q 2004;18:258–75. [DOI] [PubMed] [Google Scholar]
- 12.Advisory Committee on Minority Health. Sustaining and advancing the healthcare safety net: ensuring that the healthcare safety net is not left behind. 2013. Accessed October 19, 2021 https://minorityhealth.hhs.gov/Assets/pdf/ACMH_Healthcare_Safety_Net_Report_Sep_2013_final_508.pdf
- 13.Beck AF, Edwards EM, Horbar JD, Howell EA, McCormick MC, Pursley DM. The color of health: how racism, segregation, and inequality affect the health and well-being of preterm infants and their families. Pediatr Res 2020;87:227–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Crear-Perry J, Correa-de-Araujo R, Lewis Johnson T, McLemore MR, Neilson E, Wallace M. Social and structural determinants of health inequities in maternal health. J Womens Health (Larchmt) 2021;30:230–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Howell EA. Reducing disparities in severe maternal morbidity and mortality. Clin Obstet Gynecol 2018;61:387–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Howell EA, Ahmed ZN. Eight steps for narrowing the maternal health disparity gap: step-by-step plan to reduce racial and ethnic disparities in care. Contemp Ob Gyn 2019;64:30–6. [PMC free article] [PubMed] [Google Scholar]
- 17.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;215:143–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Howell EA, Egorova NN, Janevic T, Balbierz A, Zeitlin J, Hebert PL.Severe maternal morbidity among Hispanic women in New York City: investigation of health disparities. Obstet Gynecol 2017;129:285–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mujahid MS, Kan P, Leonard SA, Hailu EM, Wall-Wieler E, Abrams B, et al. Birth hospital and racial/ethnic differences in severe maternal morbidity in the state of California. Am J Obstet Gynecol 2021;224:219. e1–e15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Matty R, Heckmann R, George E, Barbo Barthel A, Suter LG, Ross JS, et al. Identification of hospitals that care for a high proportion of patients with social risk factors. JAMA Health Forum 2021;2:e211323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Gould JB. The role of regional collaboratives: the California Perinatal Quality Care Collaborative model. Clin Perinatol 2010;37:71–86. [DOI] [PubMed] [Google Scholar]
- 22.Profit J, Kowalkowski MA, Zupancic JA, Pietz K, Richardson P, Draper D, et al. Baby-MONITOR: a composite indicator of NICU quality. Pediatrics 2014;134:74–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Profit J, Gould JB, Bennett M, Goldstein BA, Draper D, Phibbs CS, et al. The association of level of care with NICU quality. Pediatrics 2016;137:e20144210. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Profit J, Gould JB, Zupancic JA, Stark AR, Wall KM, Kowalkowski MA, et al. Formal selection of measures for a composite index of NICU quality of care: Baby-MONITOR. J Perinatol 2011;31:702–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Draper D, Gittoes M. Statistical analysis of performance indicators in UK higher education. J R Stat Soc Ser A Stat Soc 2005;167:449–74. [Google Scholar]
- 26.Lee HC, Liu J, Profit J, Hintz SR, Gould JB. Survival without major morbidity among very low birth weight infants in California. Pediatrics 2020;146:e20193865. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Dynan L, Goudie A, Smith RB, Fairbrother G, Simpson LA. Differences in quality of care among non-safety-net, safety-net, and children’s hospitals. Pediatrics 2013;131:304–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Popescu I, Fingar KR, Cutler E, Guo J, Jiang HJ. Comparison of 3 safety-net hospital definitions and association with hospital characteristics. JAMA Netw Open 2019;2:e198577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.California Association of Public Hospitals and Health Systems. CAPH member systems. 2020. Accessed February 10, 2021 https://caph.org/
- 30.America’s Essential Hospitals. Our members by state. 2020. Accessed February 10,2021 https://essentialhospitals.org/about/listing-of-americas-essential-hospitals-members/
- 31.American Academy of Pediatrics, Committee on Fetus and Newborn, Barfield WD, Papile LA, Baley JE, Benitz W, Cummings J, et al. Levels of neonatal care. Pediatrics 2012;130:587–97. [DOI] [PubMed] [Google Scholar]
- 32.Department of Health Care Services. Standards for neonatal intensive care units (NICUs). 1999. Accessed October 19, 2021 https://www.dhcs.ca.gov/services/ccs/Pages/ProviderStandards.aspx
- 33.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:e407–15. [DOI] [PubMed] [Google Scholar]
- 34.Horbar JD, Edwards EM, Greenberg LT, Morrow KA, Soll RF, Buus-Frank ME, et al. Variation in performance of neonatal intensive care units in the United States. JAMA Pediatr 2017;171:e164396. [DOI] [PubMed] [Google Scholar]
- 35.Stevenson DK, Wong RJ, Profit J, Shaw GM, Wang CJ, Lee HC. “Following through”: addressing the racial inequality for preterm infants and their families. Pediatr Res 2020;87:192–3. [DOI] [PubMed] [Google Scholar]
- 36.Khoong EC, Cherian R, Rivadeneira NA, Gourley G, Yazdany J, Amarnath A, et al. Accurate measurement in California’s safety-net health systems has gaps and barriers. Health Aff (Millwood) 2018;37:1760–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Liu J, Parker MG, Lu T, Conroy SM, Oehlert J, Lee HC, et al. Racial and ethnic disparities in human milk intake at neonatal intensive care unit discharge among very low birth weight infants in California. J Pediatr 2020;218:49–56.e3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Parker MG, Greenberg LT, Edwards EM, Ehret D, Belfort MB, Horbar JD. National trends in the provision of human milk at hospital discharge among very low-birth-weight infants. JAMA Pediatr 2019;173:961–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Rogowski JA, Staiger DO, Horbar JD. Variations in the quality of care for very-low-birthweight infants: implications for policy. Health Aff (Millwood) 2004;23:88–97. [DOI] [PubMed] [Google Scholar]
- 40.Boghossian NS, Geraci M, Lorch SA, Phibbs CS, Edwards EM, Horbar JD. Racial and ethnic differences over time in outcomes of infants born less than 30 weeks’ gestation. Pediatrics 2019;144:e20191106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Goldstein GP, Pai VV, Liu J, Sigurdson K, Vernon LB, Lee HC, et al. Racial/ethnic disparities and human milk use in necrotizing enterocolitis. Pediatric Res 2020;88(Suppl 1):3–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Sisk PM, Lovelady CA, Dillard RG, Gruber KJ, O’Shea TM. Early human milk feeding is associated with a lower risk of necrotizing enterocolitis in very low birth weight infants. J Perinatol 2007;27:428–33. [DOI] [PubMed] [Google Scholar]
- 43.Pineda RG. Predictors of breastfeeding and breastmilk feeding among very low birth weight infants. Breastfeed Med 2011;6:15–9. [DOI] [PubMed] [Google Scholar]
- 44.Fleurant E, Schoeny M, Hoban R, Asiodu IV, Riley B, Meier PP, et al. Barriers to human milk feeding at discharge of very-low-birth-weight infants: maternal goal setting as a key social factor. Breastfeed Med 2017;12:20–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Riley B Schoeny M, Rogers L Asiodu IV, Bigger HR, Meier PP, et al. Barriers to human milk feeding at discharge of very low-birthweight infants: evaluation of neighborhood structural factors. Breastfeed Med 2016;11:335–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Merewood A, Brooks D, Bauchner H, MacAuley L, Mehta SD. Maternal birthplace and breastfeeding initiation among term and preterm infants: a statewide assessment for Massachusetts. Pediatrics 2006;118:e1048–54. [DOI] [PubMed] [Google Scholar]
- 47.Ibrahim S. Identifying safety net hospitals consistently. JAMA Health Forum 2021;2:e211240. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.