• Julie L Holihan, Deepa Cherla, Katherine J Blair, Steven S Chua, Joseph P Hasapes, Eduardo J Matta, Kaustubh G Shiralkar, Venkateswar R Surabhi, Varaha S Tammisetti, Tien C Ko, and Mike K Liang.
• Department of Surgery, University of Texas Health Science Center at Houston, Houston, Texas. Electronic address: Julie.L.Holihan@uth.tmc.edu.
• J. Surg. Res. 2018 Apr 1; 224: 97-101.

BackgroundPrevious studies suggest that agreement between readers of computed tomography (CT) scans for the diagnosis of a ventral hernia (VH) is poor (32% agreement, κ = 0.21). Recommendations were developed by surgeons and radiologists after determining common reasons for disagreement among CT reviewers; however, the long-term effect of adoption of these recommendations has not been assessed. The aim of this quality improvement (QI) project was to determine whether the incorporation of recommendations developed by surgeons and radiologists improves agreement among reviewers of CT scans in diagnosing a VH.MethodsA prospective cohort of patients, with a CT scan of the abdomen and pelvis in the past 1 y, attending a surgery clinic at a single institution was enrolled. Enrolled subjects underwent a standardized physical examination by a trained hernia surgeon to determine the likelihood of a clinical VH (no, indeterminate, or yes). The QI intervention was the distribution and implementation of previously described recommendations. After a year of intervention, independent radiologists assessed patients' CT scans for the presence or absence of a VH. Percent agreement and kappa were calculated to determine interobserver reliability. In-person discussion on scans with disagreement was held, and the results were used as a "gold standard" to calculate sensitivity, specificity, positive, and negative predictive values for CT scan diagnosis of a VH.ResultsA total of 79 patients were included in the study. After QI intervention, seven radiologists agreed on 43% of the scans, and κ was 0.50 (P < 0.001). Agreement was highest among patients with a high clinical likelihood of a VH and lowest among patients with an indeterminate clinical likelihood. Sensitivity and specificity were 0.369 and 0.833, respectively.ConclusionsAfter the implementation of recommendations, there is improved agreement among radiologists reading CT scans for the diagnosis of a VH. However, there is substantial room for improvement, and CT scans for the diagnosis of VH is not ready for widespread use.Copyright © 2017 Elsevier Inc. All rights reserved.

17 keywords...

hide…