There is nothing staid about STARD

Abstract

Incomplete reporting hampers the evaluation of results and bias in randomized trials, systematic reviews, observational studies, and diagnostic accuracy studies. Guidelines for reporting study design and methods have been developed to encourage authors and journals to include the required elements, e.g., the CONSORT guidelines for reporting randomized controlled trials data,1 the QUOROM guidelines for reporting systematic reviews and meta-analyses,2 and the STROBE guidelines for reporting observational studies in epidemiology3,4. Despite their importance in critical assessment of the literature, the impact of these guidelines has been disappointing. The original CONSORT guidelines came out in 1996.5 A direct comparison of pre-CONSORT reporting from 1994 to post-CONSORT reporting in 1998 demonstrated a small but significant improvement in reporting the CONSORT criteria6; however, by 2003, only 22% of high-impact medical journals referred to CONSORT in their instructions to authors and one-fourth of these referred to out-of-date criteria.7 The Standards for Reporting of Diagnostic Accuracy (STARD) criteria were first reported in 2003.8,9 These guidelines for reporting the accuracy of diagnostic studies …