Neighborhood Factors and the Association With Ward-to-Pediatric Intensive Care Unit Transfers

Introduction: Neighborhood factors are associated with pediatric critical illness but remain understudied. Children from very low COI neighborhoods are overrepresented in PICU admissions. We assessed whether children from disparate neighborhoods are more likely to be transferred from the ward to the PICU in the first 72-hours of admission, hypothesizing that children from disparate neighborhoods are overrepresented and may have different illness trajectories.


Methods: We conducted a retrospective ecological cohort study of hospitalized children < 18 years of age admitted from the emergency department at Children’s Hospital of Philadelphia from 2014 to 2024. Mixed effects logistic regression clustered by census tract estimated adjusted odds ratios (aOR) for PICU transfer in the first 72-hours across COI quintiles (very high, high, moderate, low, and very low) adjusting for race/ethnicity, payor, age, season, and complex chronic conditions. Logistic regression estimated aOR across payor as a secondary analysis.


Results: Of 176,223 patients admitted to the ward, 3492 (~2%) required PICU transfer in the first 72-hours. For patients transferred versus not transferred, most identified as male (55% vs. 54%), Non-Hispanic Black (36% vs. mostly Non-Hispanic White (38%)), had government insurance (59% vs. 54%), and were from very low COI neighborhoods (37% vs 35%). In unadjusted analysis, high (OR 1.22, 1.05-1.40), moderate (OR 1.22, 1.05-1.41), low (OR 1.18, 1.00-1.38), and very low COI (OR 1.36, 1.18-1.56) had an increased risk of transfer when compared to very high COI (reference). After adjusting, there was no difference in PICU transfer across COI quintiles, including high (aOR 1.11, 0.97-1.27), moderate (aOR 1.05, 0.91-1.23), low (aOR 0.99, 0.85-1.17), or very low COI (aOR 1.12, 0.97-1.30). Charity care (aOR 2.44, 1.32-4.51) and government insurance (aOR 1.09, 1.01-1.18) had an increased risk of transfer when compared with commercial insurance (reference).


Conclusions: There was no association between COI and PICU transfer when adjusting for confounders. There was a higher risk of PICU transfer for patients with charity care and government insurance. Further studies should explore time to transfer and longitudinal modeling to assess risk of transfer across COI quintiles at different transfer time intervals.