The Accuracy of Computed Tomography in the Diagnosis of Chronic Rhinosinusitis

Abstract

To determine the sensitivity, specificity, and diagnostic accuracy of paranasal sinus computed tomography (CT) in the diagnosis of chronic rhinosinusitis (CRS). Prospective dual cohort study. One hundred seventy-one consecutive patients undergoing endoscopic sinus surgery for CRS were evaluated with CT and staged according to the Lund system. Histopathological findings from sinus specimens were reviewed and graded. A second contemporaneous control group of 130 patients undergoing CT of the sinus regions for other reasons but without a diagnosis of CRS was also staged. Sensitivity, specificity, and the receiver-operator characteristic were determined for the sinus CT in the diagnosis of CRS. Positive and negative predictive values were also computed. In the disease-positive group of patients with CRS, the mean Lund score was 9.8 (95% confidence interval, 9.0-10.6). The mean inflammatory grade on histopathological study was 2.3 (range, 0-4). For the control group (without disease), the mean Lund score was 4.3 (95% confidence interval, 3.5-5.0). The AUC for the receiver-operator characteristic was 0.802 (P <.001). Selecting a Lund score cut-off value of greater than 2 as abnormal, the sinus CT exhibited sensitivity and specificity of 94% and 41%, respectively. Increasing the cut-off value to 4 changed the sensitivity and specificity to 85% and 59%, respectively. The paranasal sinus CT scan exhibits good sensitivity and above-average specificity for the diagnosis of CRS. When added to the history and physical findings, CT may add to the diagnostic accuracy of CRS.