Reporting Completeness of Diagnostic Accuracy Studies: A Meta-Review of Investigations on Adherence to STARD 2015

Introduction: This meta-review aims to evaluate meta-analyses on adherence of diagnostic test accuracy (DTA) studies to the STAndards for Reporting of Diagnostic Accuracy Studies (STARD) 2015 checklist. Methods: We searched MEDLINE and EMBASE and included reviews using STARD 2015 to evaluate completeness of reporting of primary DTA studies in humans in any field of research and presented their results quantitatively. We extracted data independently, in duplicate. Random-effects meta-analysis models using restricted maximum likelihood estimation were used to determine overall mean and 95% confidence intervals (CI) of adherence to STARD 2015, compare fields of research, and evaluate adherence over time. Subgroup analyses to compare adherence across fields of research (diagnostic imaging, lab/biomarker, and other), were performed with pairwise differences between subgroups tested using meta-regression models with field of research as a categorical moderator. Results: A total of 14 reviews evaluating 1115 primary DTA studies were included. The range of primary studies evaluated in each review was 6 to 158. Included reviews were from Canada (n = 6), Germany (n = 3), Australia (n = 1), USA (n = 2), Spain (n = 1), and South Korea (n = 1). Field of research of included reviews were diagnostic imaging (n = 7), lab/biomarker (n = 4), and other (n = 3). The overall mean STARD 2015 adherence was 53.2% (95% CI: 45.9-60.5). Mean adherence was higher among diagnostic imaging studies (61.7% [95% CI: 56.8-66.6]) compared to lab/biomarker studies (48.5% [95% CI: 33.5-63.4]; P = .02) and other fields of research (39.4% [95% CI: 23.9-54.9]; P < .001). Time trends analysis found that STARD adherence did not change over time ( P = .28). Conclusion: Adherence of primary DTA studies to STARD 2015 was evaluated to be incomplete.