Multilevel social factors and NICU quality of care in California

The study addresses persistent racial/ethnic inequities in adverse birth outcomes, particularly among very low birth weight (VLBW) infants, who are at heightened risk of mortality and morbidity. Prior work indicates that inequities persist both between and within NICUs, where vulnerable populations often receive lower quality of care and less family-centered care. The California Perinatal Quality Care Collaborative (CPQCC) has introduced a health equity dashboard to support quality improvement (QI) by auditing and providing feedback on equity-related measures. A key missing component for NICU QI efforts has been granular data on multilevel social determinants, including maternal social status, neighborhood conditions, and NICU catchment area characteristics. This study’s purpose is to incorporate social and built environment factors into an equity framework for NICU quality assessment and to evaluate whether conditions in NICU catchment areas are associated with Baby-MONITOR quality scores. The hypothesis is that social and built environment conditions in NICU catchment areas are associated with NICU quality of care as measured by the Baby-MONITOR composite.

The paper highlights prior research linking neighborhood and environmental contexts to birth outcomes, including racial residential segregation, income inequality, greenspace, socioeconomic status, and built environment characteristics. Social and built environments, such as neighborhood deprivation and walkability, influence pregnancy outcomes and mortality and may affect NICU care delivery through pathways including access to high-quality NICUs, maternal advocacy capacity, language and transportation barriers, and the need for coordinated post-discharge support. These findings suggest that contextual factors may shape care processes and outcomes in NICUs, motivating the integration of such measures into equity-focused QI initiatives.

Study population: The cohort derived from the CPQCC registry included VLBW infants admitted to 119 California NICUs between 2008 and 2011. Inclusion criteria were admissions with birthweight 401–1500 g or gestational age 22–29 weeks, excluding deaths in the delivery room or before 12 hours of life and severe congenital abnormalities. Clinical infant variables (sex, GA, Apgar, outborn status) and maternal variables (race/ethnicity, prenatal care, parity, mode of delivery) were included. Linkage with state data sources provided maternal residence at birth, education, payer source, and nativity. Geocoding: Maternal addresses were geocoded (SAS 9.4 PROC GEOCODE) to latitude/longitude and linked to 2010 census tracts (0.5% not geocoded). Neighborhood and built environment measures: From the California Neighborhoods Data System, U.S. Census and ACS (2007–2011), the study derived tract-level measures: racial/ethnic composition, neighborhood socioeconomic status (nSES; PCA-based composite of education, housing cost, employment, occupation, income, poverty), population density (persons/km²), commute by car/motorcycle, and household crowding. Built environment metrics included street connectivity (gamma index), parks per 1000 residents (NAVTEQ NAVSTREETS), business and recreational facilities per 1000 residents (NETS, 2006–2008 window), retail food environment index (ratio of unhealthy to healthy outlets), restaurant environment index (unhealthy/healthy), and traffic density (California DOT). Measures were categorized into statewide quintiles. NICU catchment areas: For each NICU, census tracts were ranked by number of resident mothers/infants served; tracts were added until at least 80% of the NICU’s infants were included, prioritizing geographically nearer tracts in case of ties. Tracts without infants but surrounded by included tracts were added to ensure contiguity. The resulting catchment areas covered 7,501 of 8,057 California tracts (1,176 without infants), with a median of 228 tracts per NICU catchment (IQR 318); catchments frequently overlapped (median 4, range 1–17). Catchment-level attributes were calculated as averages across included tracts. NICU quality measure: The Baby-MONITOR composite included nine risk-adjusted, standardized components: (1) any antenatal steroids; (2) moderate hypothermia on admission (<36°C); (3) nonsurgical pneumothorax; (4) healthcare-associated infection; (5) oxygen requirement at 36 weeks; (6) timely retinopathy of prematurity screening; (7) discharge on any human milk; (8) in-hospital mortality; (9) growth velocity. Each component’s observed-minus-expected value was standardized to the California VLBW reference population, aggregated, and averaged; scores <0 indicate worse-than-expected quality. Statistical analysis: NICU-level social and built environment factors (both patient composition and catchment attributes) were compared across tertiles of Baby-MONITOR scores using chi-square tests for categorical variables and ANOVA for continuous variables. Descriptive statistics characterized the cohort and NICU/catchment characteristics.

• Cohort: 17,781 admissions from 15,901 unique VLBW infants across 119 California NICUs (2008–2011). Mean birthweight 1,070.6 g (SD 284.4), mean gestational age 28.2 weeks (SD 2.9). Maternal race/ethnicity: 47.8% Hispanic, 25.9% non-Hispanic White, 13.6% non-Hispanic Black, 10.3% Asian/Pacific Islander. Education: 50.0% high school or less. Payer: 49.1% Medi-Cal, 44.0% private insurance. - Baby-MONITOR component rates (unadjusted counts): antenatal corticosteroids 86.3%; moderate hypothermia 15.0%; pneumothorax 3.7%; healthcare-associated infection 12.9%; chronic lung disease 21.5%; timely eye exam 93.7%; discharge on any human milk 64.3%; mortality 6.8%; high growth velocity 53.8%. - Main associations: NICUs serving higher proportions of minority racial/ethnic patients and lower SES patients had lower Baby-MONITOR quality scores. NICU catchment areas characterized by lower SES, higher proportions of minority residents, and greater household crowding were associated with lower Baby-MONITOR scores.

Findings support the hypothesis that multilevel social determinants, including both the composition of NICU patient populations and the socioeconomic and built environment characteristics of NICU catchment areas, are associated with NICU quality of care as measured by Baby-MONITOR. These results underscore the importance of integrating contextual social and environmental information into QI efforts and equity monitoring. Incorporating catchment area metrics into the CPQCC health equity dashboard can help providers identify potential barriers (e.g., deprivation, crowding) influencing care processes and outcomes, inform targeted resource allocation (e.g., lactation support, transportation/language services), and promote strategies to reduce disparities in care delivery and outcomes for VLBW infants.

Multilevel social and built environment factors are linked to NICU quality of care in California. Incorporating these contextual measures into a health equity dashboard offers actionable insights for providers and NICUs to identify needs and tailor interventions aimed at improving equity in care delivery and outcomes for VLBW infants. Future research should expand beyond bivariate analyses to multivariable and longitudinal designs, explore mechanisms linking contextual factors to specific care processes, and evaluate the impact of dashboard-informed QI interventions on reducing disparities.