Assessing clinical performance of gynecology residents: sonographic evaluation of adnexal masses based on morphological scoring systems

Abstract

OBJECTIVE: To assess the performance of inexperienced gynecology residents in the evaluation of adnexal masses using sonographic scoring systems. METHODS: From April 2003 to October 2004, transvaginal sonography was performed preoperatively by junior gynecology residents on 1-month shifts on 123 patients suspected of having adnexal masses. A total of 137 adnexal masses were found and evaluated using two different morphological scoring systems (those of DePriest and Lerner). The diagnostic performance of the sonographic scoring systems was assessed using the McNemar test and receiver-operating characteristics (ROC) curve analysis was used for lesion characterization. The diagnostic accuracy when the results of both scoring systems satisfied malignancy cut-off values was also investigated. RESULTS: Histopathological analysis revealed that, of the 137 lesions, 109 were benign, 23 were malignant and five were borderline. Best clinical cut-off levels were > or = 5 on the DePriest and > or = 3 on the Lerner scores. Both systems achieved good performance for characterizing malignancy. No significant difference was found in terms of the accuracy of the two systems as determined by mean areas under the ROC curves (0.816 and 0.783, P = 0.562). The combined approach using both scoring systems resulted in higher specificity (77.1%, P < 0.05) and positive predictive value without a significant decrease in sensitivity (82.1%) compared with either system alone. Of the 43 histologically confirmed false-positive cases, mature cystic teratoma was most common, with 13/22 (59%) cases being misinterpreted as malignancies. CONCLUSION: Junior residents, inexperienced at sonography, performed fairly in terms of evaluating adnexal masses with the help of morphological scoring systems.