Interobserver variability in CT scan interpretation for suspected acute appendicitis

Abstract

Objective: A prospective study was undertaken to assess the interobserver variability in CT scan interpretation in patients with suspected acute appendicitis.

Methods: 103 adult patients with suspected acute appendicitis underwent unenhanced helical multislice CT scanning of the abdomen. All patients subsequently underwent laparoscopy by a surgeon who was blind to the diagnosis suggested by the CT scan. All CT scans were interpreted by group A, B and C radiologists with different levels of expertise.

Results: Acute appendicitis was diagnosed on multislice CT scans in 69%, 74% and 80% by group A, B, and C radiologists, respectively. At laparoscopy, 83% of patients were diagnosed with acute appendicitis. For group A, B and C radiologists, the specificity of CT scanning for the diagnosis of acute appendicitis was 94%, 94% and 100%, respectively, the sensitivity was 81%, 88% and 95%, the positive predictive value was 98.6%, 98.7% and 100%, and the negative predictive value was 50%, 68% and 81%. The difference in the proportion of positive outcomes between observer groups A and C was significant. During laparoscopy, 12 patients were diagnosed with other diseases. These were all correctly diagnosed by group C radiologists; groups A and B radiologists missed the diagnosis of colitis in one patient. All other disorders were diagnosed correctly.

Conclusions: The sensitivity of CT interpretations for the diagnosis of acute appendicitis differs considerably between radiologists. This interobserver variability has to be taken into account during implementation of routine CT scanning in patients with suspected acute appendicitis. Only in centres with expert CT radiologists is the implementation of routine CT scanning in patients with suspected acute appendicitis justified.