Aberrant diagnoses by individual surgical pathologists.

Methods of auditing the performance of histopathologists, such as external and internal quality assurance, clinicopathological conferences, and "double-reporting" of microscopic slides, show significant diagnostic errors in at least 1.2% of reports. Although some of these are in well-recognized areas of difficulty, such as melanoma or lymphoma, most errors are in common biopsy specimens. We have developed a method that compares diagnostic patterns of individual histopathologists. This aims to identify specific diagnoses that a pathologist makes more or less frequently than other colleagues and enables the individual to reflect on his or her own histologic expertise in reporting on specific biopsy results. The bottom line diagnoses of transurethral resection of prostate specimens; rectal, gastric, and bladder biopsy samples; and endometrial curettages were analyzed retrospectively. Analyses were performed on diagnoses made by at least 15 pathologists on each specimen type and expressed as a standardized ratio (SR) with 95% confidence intervals (CI). An SR of 1.0 indicated a pattern of diagnosis matching the combined pattern of other colleagues. An SR <1.0 indicated relative "underdiagnosis" and an SR >1.0 indicated relative "overdiagnosis." Diagnostic rates of individual pathologists whose CIs did not straddle the value of 1.0 were considered aberrant, although not necessarily incorrect. The 47 of 226 (20.8%) aberrant SRs included four pathologists' diagnoses of prostatic carcinoma, three each of endometrial, rectal, and bladder carcinoma, and one of gastric malignancy. This method, which could easily be automated and used regionally or nationally, should provide pathologists with a profile of their diagnostic patterns in comparison with their peers.