P0591 - How generalizable is freestanding children's hospital data such as PHIS (Pediatric Health Information System)?

Background/Purpose: The Pediatric Health Information System (PHIS) is often used to study pediatric inpatient care; PHIS includes data from over 52 freestanding children's hospitals who are part of the Children's Hospital Association. As listed in PubMed, 125 publications have utilized the PHIS database in the last 5 years. While PHIS may be representative of most tertiary or quaternary pediatric care, the majority of hospitalized children receive care in non-children's hospitals, and it is unclear if PHIS accurately represents the care of common pediatric diagnoses often cared for in non-children's hospitals or community settings. The purpose of this study is therefore to compare inpatient populations and length of stay between freestanding children's hospitals and other pediatric inpatient sites

Methods: We utilized the 2016 Kid's Inpatient Database (KID) from the Agency for Healthcare Research and Quality, the largest all-payer pediatric care database. After excluding newborn diagnoses, psychiatric and surgical diagnoses, and medical diagnoses that reflected predominantly tertiary care such as oncologic issues, the 10 most frequent medical diagnoses were determined. For these 10 diagnoses, we then compared inpatient care characteristics between freestanding children's hospitals (FCH), and non-freestanding children's hospitals (NFCH, which includes children's hospitals in hospitals and community sites) using chi-square or Wilcoxon rank sum. FCH were used as a proxy for PHIS hospitals. Total number of diagnoses (#dx) was used as a proxy for medical complexity.

Results: Among the top 10 medical diagnoses in the KID, there were significant differences in the frequency of individual diagnoses between FCH and NFCH (table 1). Length of stay, patient age, and medical complexity also differed significantly between the groups, with children at FCH having longer LOS (median 2 days, 3rd quartile 4 days vs. 3, p<0.0001), younger age (median 2 years vs. 3, p<0.0001), and greater #dx (median 5 vs. 4, p<0.0001) compared with NFCH (table 2).

Conclusion: When looking at the top 10 medical diagnoses, the 2016 KID data shows significant differences between children cared for in FCH and NFCH. Even with NFCH including children's hospitals in hospitals where tertiary and quaternary care does take place, children in FCH are younger and more medically complex, with longer LOS. If community hospitals were compared to FCH, we suspect the differences would be greater. This demonstrates that while studies using freestanding children's hospital data such as PHIS may be helpful to understanding care in FCH, the results may not be generalizable to the population of hospitalized children in NFCH.