RSNA 2009 

Abstract Archives of the RSNA, 2009


SSE09-01

CT of Acute Appendicitis: Can Diagnostic Accuracy Be Used as a Practical Performance Metric for Readers Specialized in Abdominal Imaging?

Scientific Papers

Presented on November 30, 2009
Presented as part of SSE09: Gastrointestinal (Appendicitis)

 Research and Education Foundation Support

Participants

Lisa Lynn Chu BA, Presenter: Nothing to Disclose
Ying Lu PhD, Abstract Co-Author: Nothing to Disclose
Emily M. Webb MD, Abstract Co-Author: Nothing to Disclose
Joseph Stengel DO, Abstract Co-Author: Nothing to Disclose
Benjamin M. Yeh MD, Abstract Co-Author: Nothing to Disclose
Fergus V. Coakley MD, Abstract Co-Author: Nothing to Disclose

PURPOSE

To investigate if accuracy in the CT diagnosis of acute appendicitis can be used as a practical performance metric for readers specialized in abdominal imaging.

METHOD AND MATERIALS

We retrospectively identified all faculty radiologists specialized in abdominal imaging at our institution who independently interpreted more than 100 CT studies performed for suspected acute appendicitis between July 2002 and July 2007. The original reports were reviewed and classified as positive, negative, or equivocal for the diagnosis of acute appendicitis (reports of a non-visualized appendix and no secondary signs of inflammation were considered negative). Diagnostic accuracy for each reader was determined by comparison with the reference standard of surgical pathology or patient outcome. Differences in diagnostic accuracy between readers were analyzed using a trinomial model and Z-test for statistical significance (p<0.05). The minimum number of cases needed to significantly differentiate readers was determined by the normal approximation test and observed proportion of readers (alpha=0.05, 1-beta=0.80).

RESULTS

Six qualifying readers were identified and interpreted a total of 1947 CT studies for suspected appendicitis, with a mean of 324 cases per reader (range, 139 to 415). Based on the reference standard, 411 (21%) of the patients had acute appendicitis. The mean reader diagnostic accuracy was 93.6% (range, 90.6 to 95.4%). Only one reader had a diagnostic accuracy (90.6%) that was significantly lower than the other readers, but the absolute differences from the other radiologists were small (2.6 to 4.8%). The minimum number of cases needed to significantly differentiate the least accurate reader from any of the other readers was 1150.

CONCLUSION

Readers specialized in abdominal imaging appear to have uniformly high accuracy in the diagnosis of acute appendicitis and large numbers of studies accumulated over long time periods may be required to demonstrate only minor inter-observer variation, suggesting diagnostic accuracy for acute appendicitis at CT does not serve as a practical ongoing performance metric for such radiologists.

CLINICAL RELEVANCE/APPLICATION

In practice, accuracy in the diagnosis of acute appendicitis by CT is unlikely to be a useful ongoing performance metric for radiologists specializing in abdominal imaging.

Cite This Abstract

Chu, L, Lu, Y, Webb, E, Stengel, J, Yeh, B, Coakley, F, CT of Acute Appendicitis: Can Diagnostic Accuracy Be Used as a Practical Performance Metric for Readers Specialized in Abdominal Imaging?.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8016132.html