Assessing diagnostic errors: when is suspension of a pathologist justified?

In 1995, of 120 doctors who had been suspended and investigated for incompetence by the study group of the Society of Clinical Psychiatrists, five were haematologists and 14 histopathologists. Only 16% of suspensions were proven justified at a tribunal. Two haematologists and three pathologists were reinstated, three haematologists and five pathologists accepted a settlement, four pathologists were dismissed, and two cases were sub judice. Furthermore, this study showed that “women doctors are much less likely to get their jobs back than men whilst, at the same time, more often found not guilty of any professional misdoing”. In the past few years histopathologists in Great Britain have frequently been accused of diagnostic or professional incompetence, and their employers sought the names of assessors, reputable pathologists, to review a list of diagnostic mistakes. The results of the reviews have usually not been made known to the Royal College of Pathologists; the assessors, having negotiated their contracts directly with the employing NHS Trusts, produced the reports for the trust while the investigated pathologist was suspended or on the sick list. Reasons for an investigation are often hostility within the department or personality clashes, but frequently they remain obscure. Often the case is initiated by an article in a local newspaper, the pathologist being accused of a major mistake, the consequences of which may or may not be spelt out. Histopathologists are particularly vulnerable if they are involved in cancer screening programmes. Zero error is expected by the public and administrators, and the fear of litigation following a false negative diagnosis is beginning to threaten various screening programmes. In June 1996 the College of American Pathologists hosted an international conference on liability and quality issues in cervical vaginal cytology where the standard of practice and possible solutions were discussed. The Board of Directors of the California Society of Pathology issued guidelines for review of cervical smears in the context of potential litigation. Among many important points, the guidelines clearly state that zero error is an unreasonable expectation that cannot be achieved even with the use of automated re-screening devices, and that even skilled screening cytologists have an irreducible false negative rate of at least 5%. The California pathologists have accepted that all laboratories make mistakes and a false negative result is not necessarily evidence of substandard practice. Errors should not be judged in isolation but in the context of the overall performance of that laboratory. In their guidelines, the California Society of Pathology also warned against biased reviews performed with hindsight knowledge, and regarded as unethical reviewing by experts who did not practise routine diagnostic cytopathology. The society advised that the degree of diYculty should be established in a fashion simulating normal practice of “reasonable” cytologists without knowledge of the clinical outcome or litigation. It seems logical to extend this method of reviewing errors to histopathological diagnosis where the variation of opinion and subjectivity of assessment by pathologists may frequently lead to variation in diagnoses that may be seen as errors. The results of studies concerned with error rates in histopathology vary widely (no serious errors, 0.26%, up to 1.2% of histopathological reports) but these were performed in academic or teaching institutions. During the past decade various presentations to the Pathological Society of Great Britain and Ireland demonstrated true interobserver variation, and this should be considered in calculating the likelihood of diagnostic errors in histopathology. In establishing what is an acceptable or realistic error—what variation of opinion is unavoidable—we could use the experience of the Scottish Panel for the Consistency of Histopathological Reporting who, over the years, examined observer variability in the reporting of biopsies results of the cervix, bladder, rectum, bronchus and, recently, prostate. The panel established that pathologists are reliable in recognising the entity but inconsistent in quantifying the degree (grade of tumours, depth of invasion). Similarly, a review of 100 randomly selected bronchial biopsies by 11 histopathologists demonstrated generally good ability to distinguish between small cell and non-small cell carcinoma, but great variation in subclassifying the non-small cell tumours. Dermott and colleagues found considerable disagreement between six observers assessing radial and vertical growth of malignant melanoma in a large J Clin Pathol 1998;51:649–651 649