Emergency Computed Tomography: What is Missed at First Reading?

Abstract

The aim of this study was to evaluate the accuracy of preliminary computed tomography (CT) interpretations made by radiology residents in the emergency department. For 4 periods of 20 consecutive days, attending radiologists prospectively recorded any disparities between their own CT interpretations (reference standard) and the preliminary interpretations of emergency radiology residents. Misinterpretations were defined as major when related to a potentially life-threatening pathology if not immediately managed after CT. The rate of disparities was calculated for all CT examinations, separately for body and neuroradiological cases, and for working and on-call hours. A total of 3044 emergency CT examinations were performed during the survey: 1568 (51%) body scans and 1476 (49%) neurological scans. Disparities were reported in 145 (4.8%) of all CT examinations, with 0.8% (24/3044) defined as major misinterpretations. There were 100 (6.3%) of 1568 disparities in body CT versus 45 (3.0%) of 1476 in neurological CT examinations (P = 0.001). No significant differences were found between disparities recorded during regular working hours (47/1083 [4.3%]) versus on-call hours (98/1961 [5.0%], P = 0.47). There were typically more disparities between resident and attending interpretations in emergency body CT. Major disparities represented only a small proportion of all misinterpretations. The detailed analysis of our data enables benchmarking over time and offers a reference for optimizing the training of residents in emergency radiology.