Background: the Consolidated Standards of Reporting Trials (CONSORT) Statement is a guideline to improve the quality of the reporting of randomized controlled trials (RCTs). It was developed in 1996 and was most recently revised in 2010. Specific guidance for the reporting of harms in RCTs can be found in CONSORT Harms, an extension to CONSORT released in 2004 and not yet updated.
Objectives: to find out whether an update to CONSORT Harms was needed, we reviewed the reporting of harms in RCTs and assessed the impact of CONSORT Harms.
Methods: we performed a scoping overview comprised of reviews evaluating the quality of the reporting of harms in RCTs against the 10 items in the CONSORT Harms checklist. We used comprehensive searches conducted in nine electronic databases, including Embase, MEDLINE and the Cochrane Methodology Register (CMR). We did not apply any language restrictions; however, we applied a date limit of 2004 onwards. We extracted general and demographic data from included studies and information on the number of RCTs assessed as complying with the reporting items of CONSORT Harms. We summarized results through descriptive analysis.
Results: we screened 3436 hits and included a total of 46 reviews. Five reviews evaluated RCTs of non-pharmacological interventions (11%). Thirteen (28%) reviews explicitly assessed adherence to the 10 items of CONSORT Harms; among these, eight documented the absolute number of trials reporting the CONSORT Harms items (n = 763 RCTs). Most of the CONSORT Harms items were reported by less than half of the trials (range 2.4% to 49%). For instance, less than 3% of the assessed RCTs reported item 9 (subgroup analyses and exploratory analyses for harms). Only two items of CONSORT Harms were reported by more than half of the RCTs assessed (52%). Twelve reviews (12/46; 26%) developed checklists by modifying CONSORT Harms items comprising several components. Six reviews (n = 872 RCTs) suggested a limited impact of CONSORT Harms when comparing the reporting of the guideline items in RCTs published before and after 2004 (median of 47%, interquartile range (IQR) 36% to 54%; and 53%, IQR 44% to 57%, respectively). Only 42% of the trials published after 2004 reported on the specific adverse effects assessed.
Conclusions: the reporting of harms remains inadequate in RCTs of pharmacological and non-pharmacological interventions. The publication of CONSORT Harms in 2004 appears of limited impact and items relevant to the interpretation of the results on harms outcomes remain inadequately reported. Based on these results, we are collaborating to update CONSORT Harms 2004, including a discussion of whether harms reporting should be included in the main CONSORT checklist, rather than only as an extension.
Patient or healthcare consumer involvement: the initiative to update CONSORT Harms 2004 is inviting the involvement of patient’s representatives.