Admission predictor modeling in pediatric interhospital transport

Objectives: The objective of this investigation was to determine if an existing general severity of illness measure describing pediatric emergency patients, calculated at referring hospitals, predicts the need for hospital admission and intensive care unit (ICU) admission at receiving hospitals. Methods: A consecutive series of interhospital transports to an urban pediatric tertiary care hospital from other emergency departments (EDs) during a I-year period were studied. The pediatric risk of admission score, a validated emergency department measure of severity of illness, was calculated by the transport team leader on arrival at the referring hospital using data available at that time. Outcomes examined in a logistic regression model and receiver operating characteristic curves included the need for hospital admission and ICU admission. Results: From 52 referring emergency departments, 1920 consecutive interhospital transport records were analyzed. Of these, 1557 (81.1%) patients were ultimately admitted to the receiving hospital, including 131 (6.8%) to the ICU. Logistic regression for hospital admission demonstrated a significant independent association with higher age, higher risk of admission, trauma diagnosis, and the lack of a pediatric inpatient service. The receiver operating characteristic curve for hospital admission [area under the curve = 0.612 (0.576, 0.647)] was not useful to determine a suitable cut point below which hospital admission was unlikely to occur. Pediatric risk of admission score performance as a predictor of ICU admission by receiver operating characteristic curve was only slightly better (area under the curve = 0.721 [0.653, 0.788]). Conclusions: This form of the pediatric risk of admission score is not practical as a predictor of hospital and ICU admission among pediatric interhospital transport. Specific calibration could increase its utility for the transport population. This in turn may contribute to more effective interhospital transport triage and more efficient allocation of transport resources.