Reporting quality of diagnostic accuracy studies: a systematic review and meta-analysis of investigations on adherence to STARD

Background Poor reporting of diagnostic accuracy studies impedes an objective appraisal of the clinical performance of diagnostic tests. The Standards for Reporting of Diagnostic Accuracy Studies (STARD) statement, first published in 2003, aims to improve the reporting quality of such studies. Objective To investigate to which extent published diagnostic accuracy studies adhere to the 25-item STARD checklist, whether the reporting quality has improved after STARD's launch and whether there are any factors associated with adherence. Study selection We performed a systematic review and searched MEDLINE, EMBASE and the Methodology Register of the Cochrane Library for studies that primarily aimed to examine the reporting quality of articles on diagnostic accuracy studies in humans by evaluating adherence to STARD. Study selection was performed in duplicate; data were extracted by one author and verified by the second author. Findings We included 16 studies, analysing 1496 articles in total. Three studies investigated adherence in a general sample of diagnostic accuracy studies; the others did so in a specific field of research. The overall mean number of items reported varied from 9.1 to 14.3 between 13 evaluations that evaluated all 25 STARD items. Six studies quantitatively compared post-STARD with pre-STARD articles. Combining these results in a random-effects meta-analysis revealed a modest but significant increase in adherence after STARD's introduction (mean difference 1.41 items (95% CI 0.65 to 2.18)). Conclusions The reporting quality of diagnostic accuracy studies was consistently moderate, at least through halfway the 2000s. Our results suggest a small improvement in the years after the introduction of STARD. Adherence to STARD should be further promoted among researchers, editors and peer reviewers.