285 - Associations of residential greenspace with emergency department visits and readmissions among infants with bronchopulmonary dysplasia (BPD)

Publication Number: 2275.285

Rose M. Albert, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Allan C. Just, Brown University, Providence, RI, United States; John Feister, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States; Kristan Scott, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Sara DeMauro, Children's Hospital of Philadelphia, Philadelphia, PA, United States; Aimin Chen, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, United States; Heather Burris, Children's Hospital of Philadelphia, Philadelphia, PA, United States; Timothy Nelin, Childrens Hospital of Philadelphia, Philadelphia, PA, United States

Background: Emergency department (ED) visits and inpatient (IP) readmissions for acute respiratory illness are common after discharge from the neonatal intensive care unit (NICU) among infants with BPD. Whether neighborhood greenspace is associated with acute healthcare utilization in this population is unknown.

Objective: To quantify the association of neighborhood greenspace exposure, as measured by the Normalized Difference Vegetation Index (NDVI), with medically attended acute respiratory illness in the first year after NICU discharge among preterm infants with BPD.

Design/Methods: We analyzed associations of NDVI with medically attended acute respiratory illness (ED visits and IP readmissions) among a cohort of 378 infants with BPD born between 2010-2020 and discharged to a residential address in metropolitan Philadelphia. NDVI was calculated using Landsat 7 (2010-2012) and Landsat 8 (2013-2021) imagery queried using Google Earth Engine. Satellite imagery with < 25% cloud cover was used to calculate mean NDVI for each year. NDVI buffer zones (200m, 400m, 800m) were calculated using each infant's geocoded address. Sequential logistic regression models quantified associations of NDVI with ED visit or IP readmission in the first year after NICU discharge, adjusting for year, neighborhood deprivation, annual mean PM2.5 exposure, race/ethnicity, insurance status, gestational age, birth weight, BPD grade, discharge age, discharge respiratory support, and infant sex. Secondary analyses evaluated ED visits and IP readmissions separately.

Results: Among 378 infants with BPD, 175 (46.3%) had medically attended respiratory illness. These infants were more likely to have higher grade BPD, require respiratory support at discharge, be identified as non-Hispanic Black, reside in areas with greater neighborhood deprivation, and have lower residential NDVI (Table 1). Substantial regional variation in NDVI existed (Figure 1). In fully adjusted models, each standard deviation increment of NDVI was associated with lower adjusted odds of composite medically attended respiratory illness (200m: aOR 0.74, 95% CI 0.56-0.98; 400m: 0.71, 95% CI 0.53-0.95; 800m: 0.73, 95% CI 0.54-0.97, Table 2). Adjusted estimates of effect were consistent in direction for both ED visits and IP readmissions analyzed separately.

Conclusion(s): Infants with BPD discharged to neighborhoods with more greenspace had lower odds of acute respiratory illness in the year after discharge. The neighborhood environment may represent a modifiable factor influencing post-discharge outcomes in this high-risk population.

Table 1. Bivariable associations of cohort characteristics with medically attended acute respiratory illness (ED visit or IP readmission) in the first year after NICU discharge among infants <32 weeks' gestation with BPD
Abbreviations: ED, emergency department; BPD, bronchopulmonary dysplasia; GA, gestational age; IQR, interquartile range; NICU, neonatal intensive care unit; SD, standard deviation; PM2.5, particulate matter <2.5µm in diameter; NDVI, Normalized Difference Vegetation Index
Two-way t-test used for continuous variables. Chi-square used for categorical variables.
*BPD severity stratified at 36 weeks or hospital admission if hospital admission to CHOP occurred after 36 weeks post-menstrual age using 2019 Neonatal Research Network criteria (Jensen et al. 2019).
Missing data: birth weight (n = 7).
#Mean PM2.5 exposure in the year after NICU discharge at the census block group level estimated using XIS-PM2.5, a spatiotemporal machine learning model (Just et al. 2025).
^Census tract-level deprivation index from 2018 (Brokamp et al. 2019).


Figure 1.
Mean NDVI in 2015 by census block group in metropolitan Philadelphia comprising 11 counties within Pennsylvania, New Jersey, and Delaware. Landsat 7 (2010-2012) and Landsat 8 (2013-2021) imagery were queried for each calendar year using Google Earth Engine, filtering for data with less than 25% cloud cover, to generate an annual mean. NDVI was calculated by subtracting the red band from near infrared band, then dividing by the sum of the near infrared and red band. NDVI values calculated from Landsat 7 were adjusted as described by Robinson et al., 2017.

Table 2. Adjusted associations of NDVI with medically attended acute respiratory illness in the first year after NICU discharge among infants with BPD (n = 378)
Abbreviations: NDVI, normalized difference vegetation index; NICU, neonatal intensive care unit; BPD, bronchopulmonary dysplasia; aOR, adjusted odds ratio; PM2.5, particulate matter <2.5µm in diameter
#Mean PM2.5 exposure in the year after NICU discharge at the census block group level estimated using XIS-PM2.5, a spatiotemporal machine learning model (Just et al. 2025).
^Census tract-level deprivation index from 2018 (Brokamp et al. 2019).