Clinicians and lay people tend to overestimate the effectiveness of a treatment when only the relative effect is presented, particularly if the relative effect is large, but the absolute effect is small. In recognition of this problem, item 17b of The Consolidated Standards of Reporting Trials (CONSORT) 2010 statement stipulates authors present both absolute and relative effects for binary outcomes in randomised controlled trials (RCTs). Adherence to item 17b and the effect of differing levels of CONSORT endorsement by journals on adherence is not well known. We assessed the extent to which item 17b is adhered to in 258 RCTs published in five leading medical journals (Annals of Internal Medicine, BMJ, JAMA, The Lancet and The New England Journal of Medicine) between January and December 2019 that all endorsed the CONSORT statement to varying degrees. Only 53 of 258 (20.5%; 95% CI 15.8% to 26.0%) included studies adhered fully to item 17b. Proportional adherence was higher in journals that endorsed the statement more strictly (BMJ and JAMA, 47.4% [34.0% to 61.0%]) compared with journals less strict in their endorsement (NEJM and Ann Intern Med, 12.2% [7.0% to 19.3%]; The Lancet, 14.1% [7.3% to 23.8%]). Journals that only recommend author adherence to CONSORT had a greater proportion of studies reporting only relative effects in the main results section (62.6%) and abstract (64.2%) compared with journals that require authors to submit a completed checklist (24.6% and 29.8%, respectively). The majority of RCTs (79.5%) with binary primary outcomes published in five leading medical journals during 2019 do not report both absolute and relative effect estimates as per item 17b of the CONSORT guideline despite its universal endorsement. Differences in adherence were observed between journals that endorsed the CONSORT statement to differing extents.
BMJ Evid Based Med