Development of a Screening Tool for the Identification of Sacroiliitis in Computed Tomography Scans of the Abdomen

Objective. To develop a screening tool for the identification of sacroiliitis on abdominal computed tomography (CT) scan. Methods. Variables including erosions (number and size), sclerosis (depths of > 0.3 cm or > 0.5 cm), and ankylosis were identified through a training exercise involving 12 CT scans containing the sacroiliac joints. Two blinded readers read 24 CT scans from a derivation cohort to propose a screening tool for identifying discriminating features of sacroiliitis. A test cohort of 68 patients was used to confirm the utility of this tool. Inter-and intraobserver values, sensitivity, specificity, and positive/negative likelihood ratios were calculated for individual as well as combinations of variables. Erosions were evaluated using receiver-operating characteristic curves. Results. Analysis of the derivation cohort determined that counting the number of erosions on the worst coronal slice in each of 4 articular surfaces was not inferior to analyzing each individual slice in either transverse or coronal view. In the test cohort, interreader reliability for ankylosis and iliac and sacral erosions was very good (kappa = 1, ICC = 0.989 and 0.995, respectively) whereas for sclerosis, it was moderate (kappa = 0.39-0.96). A total erosion score of = 3 was found to have the highest sensitivity and specificity for sacroiliitis (91% for each). The addition of a > 0.5 cm of iliac sclerosis or a > 0.3 cm of sacral sclerosis marginally increased the sensitivity (94%) but decreased specificity (85%). Conclusion. The presence of ankylosis or a total erosion score of >= 3 on CT is sufficient for identifying patients at high risk of sacroiliitis and may prompt more timely referrals to a rheumatologist.