Comparison of Three Different Timeframes for Pediatric Index of Mortality Data Collection in Transported Intensive Care Admissions*

Abstract

To identify the most appropriate timeframe for Pediatric Index of Mortality-2 data collection in patients transported to PICUs by specialist teams. Retrospective cohort study. A regional PICU transport team in London, United Kingdom. Children admitted for intensive care to a tertiary children's hospital PICU following transport by a PICU transport team between January 1, 2007, and December 31, 2008. None. Data on case mix and outcome from children transferred to the tertiary PICU during the study period were analyzed. The "standard" timeframe used in calculating Pediatric Index of Mortality-2 was compared with Pediatric Index of Mortality-2 calculated using data from two other 1-hour timeframes (during "retrieval" and during "admission"). A total of 759 transported admissions were studied. Eighty-three children died (mortality rate, 10.9%). Data were missing in up to 42.7% of admissions for some Pediatric Index of Mortality-2 variables from transport. However, missing data persisted even after the first hour of PICU admission in most cases. There was significant improvement in some physiological variables following transport (p < 0.01), but Pediatric Index of Mortality-2 did not change significantly. Pediatric Index of Mortality-2 from all three timeframes exhibited good discrimination (area under the receiver-operating characteristic curve ≥ 0.77). Calibration across deciles of mortality risk was poor for the "admission" Pediatric Index of Mortality-2 (Hosmer-Lemeshow goodness-of-fit test p = 0.04) but good for the other two calculated Pediatric Index of Mortality-2 models (p > 0.20). The findings of our single-center study do not support the need for different timeframes for Pediatric Index of Mortality-2 data collection in transported and direct PICU admissions. Uniformity in scoring procedure may simplify data collection and improve data quality.