Racial and Ethnic Inequity in High-risk Infant Follow up Program (HRIF) Participation: A Multilevel Analysis of Individual- and Hospital-level Factors in a Nationally Representative Sample

Background: Extremely low birth weight (ELBW) infants face increased risk for chronic medical problems and developmental delays compared to term-born peers. The American Academy of Pediatrics recommends that all high-risk preterm infants discharged from NICUs receive coordinated health and developmental surveillance in high-risk infant follow-up programs (HRIF). HRIF participation is variable, with known racial and ethnic disparities. The hospital-level drivers of these disparities are largely ill-defined.

Objectives:
(1) To determine if HRIF participation in the United States varies by race/ethnicity.
(2) To identify hospital-level factors that contribute to HRIF program participation.

Methods: We used data from 58 U.S. hospitals between 2006-2017 participating in the Vermont Oxford Network (VON) ELBW Infant Follow-Up Project. Primary outcome was evaluation in HRIF at 18-24 months of age corrected for gestational age. The primary predictor was infant race/ethnicity, defined as maternal race (non-Hispanic white, non-Hispanic Black, Hispanic, Asian, Native American, Other). We excluded infants with missing outcome, race, or hospital discharge, and those who died prior to follow up. Individual-level covariables included gestational age, sex, NICU co-morbidities, and need for durable medical equipment. Hospital-level covariables included hospital type, academic hospital status, number of ELBW admissions annually, and proportion of white ELBW admissions annually. We utilized generalized linear mixed models to test within- and between-NICU variation in HRIF participation. We created a null random intercepts model to calculate the median odds ratio (MOR) for follow-up and the interclass correlation coefficient (ICC). We added race/ethnicity to the model to estimate HRIF participation inequity. Because the effect of race/ethnicity on HRIF participation may vary between hospitals, we included a random term to allow the effect to vary between hospitals. We added terms for individual- and hospital-level covariables meeting p<0.1 in bivariate analyses.

Results: Among the 19,502 infants included, 44.7% participated in HRIF. MOR of HRIF participation was 0.66 (95% CI [0.49,0.88]). The ICC was 26.3%, meaning 74.7% of the variance in HRIF participation was attributable to the individual level. In adjusted models, Black infants had significantly lower odds of HRIF participation compared to white infants (aOR=0.72, 95% CI [0.62,0.80]). The effect of race varied significantly between hospitals (covariance parameter estimate p=0.001). No hospital-level factors were found to be significantly associated with the outcome; however, including terms for hospital type, the number of ELBW admissions annually, and the proportion white ELBW admissions annually improved the model fit and decreased the variance at the hospital level.

Conclusion: Hospitals account for 26% of the variability in HRIF participation. Black infants are significantly less likely to participate in HRIF. The effect of race on participation varies across hospitals. We identified several hospital-level factors that explain a proportion of the variation in HRIF participation that may serve as potential targets for interventions to reduce neonatal health inequity.