Analysis of different decision aids for clinical use in pediatric head injury in an emergency department of a general hospital.

INTRODUCTION: The diagnostic algorithm in children with head injury remains uncertain. The National Emergency X-Radiography Utilization Study II (NEXUS II) recently proposed a new decision aid. We analyzed the data prospectively recorded in a local database to evaluate the sensitivity and specificity of the variables proposed by NEXUS II, by comparing with an Italian proposal. METHODS: The clinical data of 2,391 children with head injury (0-10 years old) were prospectively collected, for an 8-year period. Any posttraumatic intracranial lesion was selected as the main outcome. The predictive ability of the selected variables was tested by the area under the receiver operating characteristic (ROC) curve. RESULTS: Eighteen of the 2,391 children (0.7%) showed the presence of intracranial lesions on computed tomography scan; neurosurgical intervention was needed in only one subject and an unfavorable outcome at 6-month follow-up occurred in only two subjects. The ability to correctly predict intracranial lesions was higher for the Italian proposal (ROC area, 0.896; 95% confidence interval, 0.887-0.904) than the NEXUS II (ROC area, 0.741; 95% confidence interval, 0.666-0.817; Fisher's exact test; p<0.001). The individual variables of the Italian proposal were both more sensitive (100% vs. 89%) and more specific (79% vs. 59%). No lesions remained undiagnosed by the variables considered in the Italian proposal, whereas two cases with lesions would have been missed by the NEXUS II rule. CONCLUSIONS: In our setting, the variables selected by the Italian proposal had higher discriminating capacity for intracranial lesions than those proposed by the NEXUS II rule, in children with head injury. These results should be considered in children with head injury attending an emergency department of a general hospital.