Interobserver Variability in Interpretation of Mammogram

ABSTRACT

PURPOSE: The purpose of this study was to evaluate the performance of radiologists for mammographic screening, and to analyze interobserver agreement in the interpretation of mammograms. MATERIALS AND METHODS: 50 women were selected as subjects from the patients who were screened with mammograms at two university hospitals. The images were analyzed by five radiologists working independently and without their having any knowledge of the final diagnosis. The interobserver variation was analyzed by using the kappa statistic. RESULTS: There were moderate agreements for the findings of the parenchymal pattern (k=0.44; 95% CI 0.39-0.49), calcification type (k=0.66; 95% CI 0.60-0.72) and calcification distribution (k=0.43; 95% CI 0.38-0.48). The mean kappa values ranged from 0.66 to 0.42 for the mass findings. The mean kappa value for the final conclusion was 0.44 (95% CI 0.38-0.51). In general, moderate agreement was evident for all the categories that were evaluated. CONCLUSION: The general agreement was moderate, but there was wide variability in some findings. To improve the accuracy and reduce variability among physicians in interpretation, proper training of radiologists and standardization of criteria are essential for breast screening.