학회지검색
대한유방검진의학회 학회지
Original Article
1National Cancer Control Institute, National Cancer Center 2Department of Radiology, Soonchunhyang University College of Medicine, Bucheon Hospital 3Department of Radiology, Center for Breast Cancer, National Cancer Center Hospital 4Department of Radiology, Seoul St. Mary’s Hospital, College of Medicine, The Catholic University of Korea 5Department of Radiology and Research Institute of Radiology, Asan Medical Center, University of Ulsan, College of Medicine 6Department of Radiology, Seoul National University Hospital 7Department of Radiology, Bundang CHA Medical Hospital, CHA University
Purpose: To quantify inter-radiologist variability in interpreting screening mammograms and analyze the experience-related factors affecting inter-radiologist agreement.
Materials and Methods: A random sample of 300 mammograms was selected from women aged ≥ 40years who were examined under the National Cancer Screening Program (NCSP) in Korea from 2007 to 2009. Inter-radiologist agreement was assessed with a weighted kappa statistic (Kw). The factors affecting the radiologists’agreement were assessed using logistic regression.
Results: The inter-radiologist agreement was moderate (overall K = 0.60, 95% confidence interval [CI]: 0.44-0.78) regarding the decision to recall vs. no recall and substantial (overall Kw = 0.64, 95% CI: 0.57-0.71) using the four-category scales for final assessment in the NCSP. Radiologists who read more than 5000 screening mammograms annually had significantly higher agreement (OR, 1.36; 95% CI, 0.95-
1.95), compared to those who did not. Similarly, those who spent ≥ 6 hours weekly reading screening mammograms had significantly higher agreement (OR, 1.76; 95% CI, 1.34-2.31), compared to those who spent less time.
Conclusion: Inter-radiologist agreement varied widely. However, radiologists who read a higher annual volume of screening mammography and spent more time per week doing so demonstrated higher agreement in the final assessment.
Key Words: Mammography; Breast cancer; Observer variation; Reading volume