PREDICTING OUTCOME IN PEDIATRIC SUBMERSION VICTIMS

Study objective: To predict outcome in children after near-drowning. D esign: Retrospective cohort study. Vegetative state and death were cla ssified as unfavorable outcomes, whereas all other outcomes were class ified as favorable. Demographic, episode-related, clinical, laboratory , and treatment variables available at the time of admission were eval uated for their usefulness in predicting outcome. Setting: Pediatric r eferral hospital. Participants: Children admitted after submersion inj ury in non-icy waters. Results: The study cohort comprised 194 childre n (median age, 2.6 years; range, 5 months to 18 years); 131 were neuro logically normal at the time of discharge, 10 had some degree of neuro logic impairment, 15 were in a vegetative state at the time of dischar ge, and 38 died. We used a combination of partitioning and logistic re gression to combine variables in a prediction rule that was always cor rect when unfavorable outcome was predicted. The final rule predicted favorable outcome for all children who were not comatose. Among comato se children, unfavorable outcome was predicted by a combination of abs ent pupillary light reflex, increased initial blood glucose concentrat ion, and male sex. This rule had a specificity of 100%-children with f avorable outcomes were always predicted to do well-and a sensitivity o f 65%. Therefore the rule was overly optimistic for 35% of patients wi th unfavorable outcomes. Conclusion: Pediatric submersion victims can be assigned to high or low likelihoods of unfavorable outcome with the use of four variables: comatose state, lack of pupillary light reflex , sex, and initial blood glucose concentration. This prediction rule m ay be useful if it can be validated in another cohort.