Prepublication abstract-only reports compared with full-text manuscripts for randomised controlled trials in inflammatory bowel disease: a systematic review.

ABSTRACT

INTRODUCTION: Randomised controlled trials (RCTs) of key therapies in inflammatory bowel disease (IBD) are often presented and available as abstracts for significant periods of time prior to full publication, often being employed to make strategic and clinical prescribing decisions. We compared the concordance of prepublication abstract-only reports and their respective full-text manuscripts. METHODS: Pairs of full-text manuscripts and their respective prepublication abstract-only reports for the same RCT outcomes, at the same time point of analysis were included. The RCTs were on treatments for IBD with full-text manuscripts published between 2010 and 2023. RESULTS: We found 77 pairs of full-text manuscripts and their prepublication abstract-only reports. There were significant mismatches in the reporting of stated planned outcomes (65/77 matched, p<0.001) and primary outcomes reported in their results sections (67/77, p<0.001); trial registrations (34/65, p<0.001); the number of randomised participants (49/77, p=0.18); participants reaching end of study (21/71, p<0.001) and primary outcome data (40/73, p<0.001). Authors conclusions matched (75/77, p=0.157). Authors did not provide explicit or implied justifications for the absence or non-concordance for any of the above items. CONCLUSIONS: Abstract-only reports have consistent issues with both limited reporting of key information and significant differences in data when compared with their later full-text publications. These are not related to further recruitment of patients or word count limitations and are never explained. As abstracts are often used in guidelines, reviews and stakeholder decision-making on prescribing, caution in their use is strongly suggested. Further work is needed to enhance minimum reporting standards in abstract-only works and ensure consistency with final published papers.