Measurement of pediatric illness severity using simple pretransport variables

To test the hypothesis that pretransport variables can predict in-hospital mortality that will correlate with major interventions and unplanned events during interfacility transport.

A cohort of children (n = 2,253) transported by a specialized pediatric team to a children's hospital were studied. At the time of referral, data collected included age (months), heart rate, systolic blood pressure, respiratory rate, retractions, stridor or wheezing, seizures, skin perfusion, oxygen requirement, and mental status. Using univariate and stepwise logistic regression, variables predictive of in-hospital mortality were selected from a training set (n = 1,111) and assigned integers based on their computed coefficients. Probability of in-hospital mortality was calculated using the total integer score and age. The risk of mortality derived from the training set was validated in the remaining patients (n = 1,142) by comparing the observed and predicted mortalities. Major interventions performed and unplanned events were determined for each of five predetermined mortality risk groups.

Variables (integers) predicting in-hospital mortality included systolic blood pressure (11), respiratory rate (6), oxygen requirement (11), and altered mental status (11). Observed mortality was similar to predicted mortality in all risk categories for the validation sample. As risk of mortality increased, so did the performance of major interventions and the occurrence of unplanned events.

Four pretransport variables predicted in-hospital mortality. Risk of mortality correlated with the incidence of major patient interventions, and the occurrence of unplanned events increased as well. This model might be useful in comparing different transport systems using severity-adjusted assessment of children requiring interfacility transport.