Reliability of diagnosis of traumatic brain injury by computed tomography in the acute phase.

The purpose of our study was to determine the accuracy and reliability of the computed tomographic (CT) diagnosis of acute traumatic brain injury (TBI) and to evaluate the inter-observer variation of CT reports of acute TBI between two experienced neuroradiologists and a neuroradiologist in training. One hundred cranial CT examinations of suspected TBI were chosen randomly from those taken during 1 year at a university central hospital, with institutional ethics committee approval. Two neuroradiologists and one neuroradiologist in training read the scans independently and were blinded to the clinical data. The original reports of radiologists on call, who were either radiology residents or general radiologists, were reviewed. Main outcome measures were false-positive and false-negative findings, and the congruency of different readers' reports. The reference standard was a group consensus formed by three neuroradiologists. Radiologists on call missed a significant number of brain contusions (67%), but with regard to intraventricular and subarachnoid hemorrhages and edema, their accuracy was moderate. They reported practically no false-positive brain contusion findings, and all readers found subdural hemorrhages reliably. The two experienced neuroradiologists had the smallest number of false-negative findings, but their reports differed significantly. Of the other neuroradiologists' mistakes, 75% were false-positive, nearly all of these concerning contusions, whereas the other made random mistakes. In conclusion, there was a marked variation between readers in the detection of brain contusion findings on acute brain CT. Experience increased accuracy, yet even between the reports of the most experienced readers, there were marked differences.