How much reproducibility do we need in human and veterinary pathology?

Abstract

In diagnostic and research reports as well as text-books of human and veterinary pathology repeatability, reproducibility, inter- and intra-observer variation are mentioned rarely as a problem in preparing diagnosis from macroscopic and/or microscopic samples and discussed inconsistently. However, optimal care and restoration of health for a patient are dependent on reliability of diagnosis, therapy, prognosis and prophylaxis. This requires for all tests and procedures a maximal repeatability and reproducibility, a sensitivity and specificity of 85-95% for procedures and methodologies and a comparison of results procedures and methodologies to a gold standard. Looking at the various steps on the road to diagnosis in pathology this is influenced by a series of laboratory steps preparing tissue samples but most importantly reproducibility depends on the handling of visual information in the central nervous system of the individual diagnostician. Thus reproducibility in this context has to be divided into at least three levels: individual (epistemological, organoleptic, inter- and intra-observer variation, and formal/technological- and normative reproducibility). The aim of the present manuscript is to stimulate the reflection among the pathology experts on this most important topic.

Keywords: Inter- and intraobserver variation; Repeatability of diagnosis; Reproducibility.