Comparing apples and oranges and pears and kumquats: The misuse of index systems for processing histopathology data in fish toxicological bioassays

Histopathology index systems involve the application of weighted scores to various diagnostic findings for the purpose of calculating overall organ and/or whole‐animal health measurements. Such systems were originally developed as tools for monitoring the general health of fish populations. More recently, index systems have been applied to hypothesis‐based toxicological studies, the goal of which is to investigate whether a cause‐and‐effect relationship exists between exposure to a particular test substance or environmental contaminant and morphologic effects in fish tissues. However, the application of index systems in that context is problematic for various reasons: a dependency on untested assumptions of toxicological importance for different types of histopathologic findings; organ scores that combine mechanistically unrelated and potentially contradictory diagnoses; calculations that include excessive numbers of findings, some of which may be incidental to the study outcome; failure to incorporate additional relevant results into the data interpretation, such as clinical observations, macroscopic findings, organ/body weights, clinical pathology data, and the results of hormonal or other biochemical assays; the inappropriate mathematical manipulation of ordinal categorical data (e.g., severity scores and “importance factors”); and a tendency of these systems to amplify, mask, and divert attention from methodological weaknesses and inaccurate diagnoses. The purpose of the present article is to demonstrate why the use of index systems is a misguided approach for handling histopathology data in studies of potential fish toxicants. Environ Toxicol Chem 2018;37:1688–1695. © 2018 SETAC