Clinical trial data is often utilised for the purposes of clinical and health policy decision-making, with journal publications and registry reports the main sources for this information. Therefore, it is vital that the data provided in these sources are comprehensive and accurate. The clinical study report (CSR), which is provided by the sponsor as part of regulatory requirements includes extensive detail regarding trial methodology and the efficacy and safety of an intervention.
In a recent study published in Trials, Hodkinson et al. compared the reporting of harms in publications related to clinical trials of orlistat, a treatment for obesity, with the corresponding CSRs obtained from the manufacturer. Of the 31 articles reporting orlistat clinical trials, CSRs were unavailable for 26 publications because either the manufacturer was not the trial sponsor or the trial predated their policy for sharing CSRs.
For the remaining five trials, the study revealed that journal publications often did not fully report all adverse events and serious adverse events. In addition, while 70–90% of the CONSORT harms reporting criteria for methods were satisfied in CSRs, journal publications satisfied only 10–50%. Both documents satisfied a high proportion of the results section reporting criteria.
With such discrepancies in completeness and quality of harms reporting in publically available journal articles and unpublished CSRs, the authors suggest that researchers’ carrying out evidence synthesis for decision-making should not solely rely on journal publications, emphasising that accessibility to CSRs and anonymised participant-level clinical data is a necessity. Such a move has widespread support with some journals, regulatory authorities and healthcare companies now making the necessary steps towards greater trial data transparency.