The reporting quality of studies of diagnostic accuracy in the urologic literature.

Abstract

High-quality evidence regarding questions of diagnostic accuracy relies on transparent reporting of study results. The quality of reporting for such studies in the urologic literature is unknown. In accordance with an a priori protocol, we systematically searched for all articles on diagnostic accuracy studies published in four major urologic journals in 2015. Using the 2015 STAndards for Reporting Diagnostic accuracy studies (STARD) checklist, two of us independently abstracted data. For each article, we calculated STARD summary scores (scale of 0-30, with higher scores reflecting higher-quality reporting). We compared scores by journal, topic, and sample size. We screened 819 references of which 61 met inclusion criteria. Nearly two-thirds of studies (39/61%; 63.9%) addressed prostate cancer diagnosis or staging; less than one in ten (6/61%; 9.8%) was conducted in non-oncological disease settings. The major focus for the investigation of new index tests lay in imaging modalities (33/61%; 54.1%); over half of these imaging studies addressed magnetic resonance imaging (18/61%; 29.5%). The average STARD score was 18.9 ± 2.4 (range 12-24). Six criteria had poor reporting compliance and were met by less than 20% of studies. We found no association between reporting quality and topic, journal or study size. The reporting quality of studies of diagnostic accuracy appears modest and independent of topic, journal or study size. There is an urgent need for greater awareness for the reporting quality of these studies among readers, editors, and investigators to raise evidentiary standards on issues of diagnosis.