Prediction of intracranial computed tomography findings in patients with minor head injury by using logistic regression.

Object: The aim of this study was to develop a decision rule for physicians in developing countries to identify patients with minor head injury who will benefit from emergency brain CT scanning.
Methods: Three hundred eighteen patients with a history of blunt head trauma and a Glasgow Coma Scale (GCS) score >or= 13 who had presented within 12 hours of trauma underwent nonenhanced brain CT and were included in this prospective study. Computed tomography findings that necessitated neurosurgical care (either observation or intervention) were considered as positive findings. Logistic regression was used to develop the decision rule.
Results: Computed tomography scans were always normal in patients < 65 years old who did not have an obvious head wound, a raccoon sign, vomiting, memory deficit, or a decrease in their GCS score. Patients with 1 major criterion (GCS score < 14, raccoon sign, failure to remember the impact, age > 65 years, or vomiting) or 2 minor criteria (wound at the scalp or GCS score < 15) had an abnormal CT scan in 13% of the cases.
Conclusions: The decision rule developed by the authors appears to be 100% sensitive and 46% specific for positive findings on brain CT and will, in developing countries, help clarify the decision to obtain scans.