Quality of reporting and risk of bias: A review of randomised trials in occupational health

Abstract

Objectives To assess the reporting quality of randomisation and allocation methods in occupational health and safety (OHS) trials in relation to Consolidated Standards of Reporting Trials (CONSORT) requirements of journals, risk of bias (RoB) and publication year. Methods We systematically searched for randomised controlled trials (RCTs) in PubMed between 2010 and May 2019 in 18 OHS journals. We measured reporting quality as percentage compliance with the CONSORT 2010 checklist (items 8-10) and RoB with the ROB V.2.0 tool (first domain). We tested the mean difference (MD) in % in reporting quality between CONSORT-requiring and non-requiring journals, trials with low, some concern and high RoB and publications before and after 2015. Results In 135 articles reporting on 129 RCTs, average reporting quality was at 37.4% compliance (95% CI 31.9% to 43.0%), with 10% of articles reaching 100% compliance. Reporting quality was significantly better in CONSORT-requiring journals than non-requiring journals (MD 31.0% (95% CI 21.4% to 40.7%)), for studies at low RoB than high RoB (MD 33.1% (95% CI 16.1% to 50.2%)) and with RoB of some concern (MD 39.8% (95% CI 30.0% to 49.7%)). Reporting quality did not improve over time (MD -5.7% (95% CI -16.8% to 5.4%). Conclusions Articles in CONSORT-requiring journals and of low RoB studies show better reporting quality. Low reporting quality is linked to unclear RoB judgements (some concern). Reporting quality did not improve over the last 10 years and CONSORT is insufficiently implemented. Concerted efforts by editors and authors are needed to improve CONSORT implementation.