The Accuracy of Colorectal Cancer Detection by Computed Tomography in the Unprepared Large Bowel in a Community-Based Hospital

Abstract

Purpose This retrospective study examined the performance of general radiologists in a community-based hospital in detecting colorectal cancer (CRC) with computed tomography (CT) in the unprepared large bowel. Methods The pathology database at a community hospital over the past 7 years (2009–2015) was retrospectively analysed for pathologically proven CRC (924 cases). The provincial hospital information profile for these patients was reviewed to determine if they had an abdominal CT for any reason in the year prior to biopsy. Metrics such as age, sex, time between the CT and biopsy or surgery, whether CRC was initially detected by the radiologist, and if this was an emergency presentation was evaluated. In the cases where CRC was not identified, the CT scans were reanalysed to determine if the CRC was identifiable in retrospect. The sensitivity of detecting CRC by CT scan in the unprepared large bowel was calculated. Results Of the 924 biopsy proven CRC cases, 22% (207 of 924) of the patients had a CT prior to biopsy. Of these cases, 47% (97 of 207) presented on an emergency basis. Of the cases with imaging in the year prior, about 60% (125 of 207) had cancer prospectively detected by the radiologist. Upon re-examination of the cases in which CRC was not initially detected, 59% were visualized in retrospect. Conclusions Community general radiologists can successfully detect CRC with a high degree of accuracy. Reformatted images, bowel wall thickening when regional nodes are prominent, and minimizing oral contrast were helpful in improving detection.