Initial predictive factors of outcome in severe non-accidental head trauma in children

Object The aim of this study is to evaluate the outcome of young children hospitalized for non-accidental head trauma in our PICU, to evaluate PRISM II score in this sub-population of pediatric trauma and to identify factors that might influence the short-term outcome. Materials and methods Files of all children less than 2 years old with the diagnosis of non-accidental head trauma over a 10-years period were systematically reviewed. We collected data on demographic information, medical history, clinical status, and management in the PICU. Three severity scores were then calculated: PRISM II, Glasgow Coma Scale (GCS), and Pediatric Trauma Score (PTS). Prognosis value of qualitative variables was tested with a univariate procedure analysis (anemia, diabetes insipidus...). Then, quantitative variables were tested with univariate procedure too (age, weight, PRISM II, GCS, Platelet count, fibrin, prothrombine time (PT)...). Potential association between variables and death was tested using univariate procedure. Variables identified by univariate analysis were then analyzed with multivariate analysis through a forward-stepping logistic regression. Results Thirty-six children were included. Mean age was 5.5 months (8 days-21.5 months). Mortality rate was 27.8%. At admission, PTS, PRISM II, GCS, PT, PTT, and diabetes insipidus were significantly altered or more frequent in non survivors. Cutoff value for PRISM II at which risk of mortality increased was 17.5 (sensitivity = 0.8; specificity = 0.88). Conclusion PRISM II is a reliable and easy performing tool for assessing the prognosis of non-accidental cranial traumatism in young children. GCS and PTS, scores even simpler than PRISM II, showed good accuracy regarding survival prediction.