Canadian Journal of Emergency Medicine
URL with Digital Object Identifier
Objective: To evaluate the accuracy of physician estimates of the probability of intracranial injury in children with minor head trauma. Methods: This is a subanalysis of a large prospective multicentre cohort study performed from July 2001 to November2005. During data collection for the derivation of a clinical prediction rule for children with minor head trauma, physicians indicated their estimate of the probability of brain injury visible on computed tomography (P-Injury) and the probability of injury requiring intervention (P-Intervention) by choosing one of the following options: 0%, 1%, 2%, 3%, 4%, 5%, 10%, 20%, 30%, 40%, 50%, 75%, 90%, and 100%. We compared observed frequencies to expected frequencies of injury using Pearson’s χ2-test in analyses stratified by the level of each type of predicted probability and by year of age. Results: In 3771 eligible subjects, the mean predicted risk was 4.6% (P-Injury) and 1.4% (P-Intervention). The observed frequency of injury was 4.1% (any injury) and 0.6% (intervention). For all levels of P-Injury from 1% to 40%, the observed frequency of injury was consistent with the expected frequency. The observed frequencies for the 50%, 75%, and 90% levels were lower than expected (p<0.05). For estimates of P-Intervention, the observed frequency was consistently higher than the expected frequency. Physicians underestimated risk for infants (mean P-Intervention 6.2%, actual risk 12.3%, p<0.001). Conclusions: Physician estimates of probability of any brain injury in children were collectively accurate for children with low and moderate degrees of predicted risk. Risk was underestimated in infants.