Author reply

Abstract

Dr. Sirota's responses to the article by Kronz et al. and our accompanying editorial on the subject are informative and well taken. Medical error can be reduced by a number of strategies, including standardization, task simplification, the use of computer-assisted technologies, and redundant system buffers. In our editorial we focused our discussions on false-positive errors; however, as Dr. Sirota correctly notes, false-negative errors are equally serious and their detection requires additional sampling methods. A logical extension of our proposals for detecting false-negative errors would be to focus second reviews of “negative reads” in those circumstances that historically have high frequencies of “missed diagnoses.” The particular clinical, histologic, immunohistochemical, and molecular characteristics of these blind pockets for the practicing pathologist will need much investigation. The devil will be in the details of defining the landscapes on which diagnostic errors in pathology occur with significant frequency. We disagree somewhat with Dr. Sirota's comment that surgical pathology diagnoses are opinions rather than facts. Image-based diagnoses do not have the numeric precision that is characteristic of some other laboratory data; however, they represent powerful integrated statements concerning complex biologic systems. Our clinical colleagues consider these diagnoses to be our closest approximation to the reality of disease, and they act on these data. We prefer to consider these diagnoses probability statements based on data that are more or less developed, depending on the context. We appreciate Dr. Sirota's thoughts and believe that continued outcomes analysis regarding morphologic data should be developed. John E. Tomaszewski M.D.*, Virginia A. LiVolsi M.D.*, * Department of Pathology and Laboratory Medicine, University of Pennsylvania Medical Center, Philadelphia, Pennsylvania