Posted: March 3, 2015

The March 17, 2015 issue of the Journal of the American Medical Association has a disturbing article from The University of Washington and a large number of distinguished collaborating medical schools in which pathologists compared the accuracy of the average community hospital pathologist to compared to a panel of experts in the reading of breast biopsies.  They found that the doctor agreed with the panel only about 75% of the time, and that most of the uncertainty occurred in lesions of borderline malignant potential such as ductal carcinoma in situ or atypical hyperplasia.  Doctors who read a higher number of breast biopsies per unit time were more likely to be accurate.  The miss rate for frank cancer was fortunately quite low, with only 4% of cancers missed.  In cases where the biopsy was truly benign, about 13% were overinterpreted as showing something abnormal.   In women whose mammograms were difficult to interpret because they have dense breasts, the likelihood of a mistake occurring was also greater.  The authors do not have a solution for this problem other than to say that higher volumes make for greater accuracy.  Dr. Stark weighs in: “For those women whose biopsies show premalignant changes — either DCIS or atypia — given these results it would not be out of line for the woman or her surgeon to request that the slides from the biopsy be sent to a high volume breast center for a second opinion.  Despite the improvements in the disseminiation of knowledge, doctors are still human and errors occur.”