Independent predictors of acute appendicitis on CT with pathologic correlation

Rationale and Objectives. To assess computed tomographic (CT) signs that have been described in published studies for the diagnosis of appendicitis to identify independent findings that predict appendicitis. Methods and Materials. A retrospective database search identified 67 patients with a CT scan of the abdomen/pelvis and pathologic evaluation of the appendix, including 41 with appendicitis and 26 with a normal appendix on pathologic examination. Each computed tomogram was re-evaluated by three independent, blinded observers who evaluated appendix diameter, enhancement of the appendix, thickening of the appendix, presence of an appendicolith, infiltration of peri-appendiceal fat, focal cecal thickening, local lymphadenopathy, fluid collections, non-appendiceal bowel thickening, non-periappendiceal infiltration of fat, and comparison of peri-appendiceal fat infiltration to thickening of adjacent bowel loops. Results. Mean diameter of the normal appendix (6.7 +/- 2.2 mm) was significantly lower than that of the inflamed appendix (12.1 +/- 4.3 nun; P <.001). Significant univariate predictors of appendicitis included appendix diameter > 8 nim (odds ratio [OR] 34.8), enhancement of the appendix (OR 4.4), thickening of the appendix (OR 4.3), infiltration of periappendiceal fat (OR 5.5), focal cecal thickening (OR 5.1), non-appendiceal bowel thickening (OR 0.4), and non-periappendiceal infiltration of fat (OR = 0.3). Of these variables, only appendix diameter and enhancement of the appendix were significant independent predictors of appendicitis on multivariate analysis. An overall diagnostic impression based on all secondary signs was less accurate than a diagnosis based on appendix diameter alone (receiver-operating characteristic analysis: Az = 0.80 vs. Az = 0.91, P =.02). Sensitivity/specificity of appendix diameter was 84%/87% using a cutoff between 8 and 9 mm and 97%/48% using a cutoff between 6 and 7 mm. Conclusion. Appendix diameter is the best single diagnostic criterion for appendicitis on CT scan. A cutoff between 8 and 9 mm provided the best balance of sensitivity/specificity in our study population, whereas a cutoff between 6 and 7 mm improved sensitivity at the expense of specificity. The presence of appendiceal enhancement provided additional diagnostic information, but other secondary signs of appendicitis did not improve diagnostic accuracy.