129 - Outcomes of Transfer to Level I Pediatric Intensive Care units in Michigan

Poster Number: 129
Publication Number: 129.2898

Background: Current guidelines recommend transfer of the most severely ill children from Level II to Level I pediatric intensive care units (PICUs) to enhance their outcomes. There are, however, no defined clinical criteria to inform medical decision-making by Level II PICU physicians regarding which children to transfer and the timing of transfer.

Objective: To test the hypothesis that pre-transfer illness severity at the Level II PICU will be associated with patient mortality at the Level I PICUs.

Design/Methods: Retrospective cohort study of children 0-20 years old hospitalized with sepsis or respiratory failure at 6 Level II PICUs in Michigan who underwent transfer to two Level I PICUs between 1997 and 2007. The outcome variable was time-to-death since arrival at the Level I PICU. Independent variables included pre-transfer data at the Level II PICU including patient demographics (age, sex) and comorbidities, Pediatric Risk of Mortality version III score within the first 12 hours of PICU admission (PRISM III-12), and the daily Pediatric Logistic Organ Dysfunction (PELOD) score. Cox proportional hazard regression and backward selection with Akaike information criterion was used to select a parsimonious prediction model for the chance of survival at the Level I PICU after transfer.

Results: 212 children were transferred from Level II to Level I PICU care. Of these, 55% were male, with a median age of 0.7 years (Interquartile range [IQR]: 0.2 – 4.7 years). The median PRISM III-12 score was 6.6 (IQR: 0 – 10). Of the 212 children, 165 children had data on length of stay and independent covariates and were included in the analysis (Figure). 24 (14.5%) of these children died in the Level I PICU hospital. The resulted model had excellent discrimination (C-index=0.89), using four predictors: sex, PRISM III-12, change in PELOD score from admission to maximum PELOD score at the Level II PICU, and time from maximum PELOD score to transfer. As the most significant predictor, one unit of increase in the PRISM III-12 score was associated with 25% increase in hazard of death after transfer to the Level I PICU (Hazard ratio: 1.25, 95% Confidence Interval: 1.16 – 1.35, p-value < 0.0001).

Conclusion(s): Higher pre-transfer illness severity among critically ill children with sepsis or respiratory failure is associated with higher mortality at receiving Level I PICUs. This finding calls for in-depth research into decision-making underlying interhospital transfer, and determination of thresholds of illness severity that should prompt transfer.