Toward a checklist for reporting of studies of diagnostic accuracy of medical tests.

BACKGROUND : "Diagnostic accuracy" refers to the ability of medical tests to provide accurate information about diagnosis, prognosis, risk of disease, and other clinical issues. Published reports on diagnostic accuracy of medical tests frequently fail to adhere to minimal clinical epidemiological standards, and such failures lead to overly optimistic assessments of evaluated tests. Our aim was to enumerate key items for inclusion in published reports on diagnostic accuracy, with a related aim of making the reports more useful for systematic reviews. METHODS : We examined published reports on shortcomings of studies of diagnostic accuracy. We prepared an initial draft of a checklist to address common errors and presented it at a meeting of editors. After incorporation of comments from editors, we published a revised version in Clinical Chemistry in 1997 for comment from readers. One of us (E.M.) additionally circulated copies of the draft to methodologists and others interested in Evidence-Based Medicine. We updated the checklist with input from these sources. RESULTS : The updated document lists items for inclusion in the title, abstract, methods, results, and discussion sections of published papers. Depending on the nature of the study, the total number of items for a single paper is approximately 40. We invite comments on this document, which is freely available at Clinical Chemistry Online, where it can accessed readily from the Table of Contents for the July 2000 issue at www. clinchem.org/content/vol46/issue7/. Comments (eLetters) can be posted there for general reading. CONCLUSIONS : The suggested revisions incorporated in this report appear useful to ensure inclusion of additional information that can allow assessment of the validity of the conclusions and the applicability of the study in other settings. The list can be useful in formulating guidelines and a checklist, which will require testing by authors and study of their effect on published studies of diagnostic accuracy.