Do not make clinical decisions based on abstracts of healthcare research: A systematic review

Abstract Objective To summarize the reporting quality of healthcare abstracts and inconsistencies between abstracts and full texts. Study design and Setting This systematic review included overviews of randomized controlled trials (RCTs) and systematic reviews (SRs) that summarized data of healthcare abstracts on reporting of abstracts and consistency of abstracts with the full text. Searches were performed in PubMed, CENTRAL, Cochrane Library and EMBASE databases from 1900 to February 2019. Two authors screened the overviews and extracted the data. All analyses were descriptive and divided into two main groups: abstracts’ reporting quality and abstracts’ consistency with the full text. Abstracts were considered poorly reported and inconsistent with the full text if more than 5% of abstracts’ information was not fully reported or not consistent with the full text. Results 27 overviews analyzing 5,194 RCTs and 866 SRs were retrieved for reporting quality of abstracts. A total of 22 overviews analyzing 2,025 RCTs and 551 SRs were included for consistency of abstracts with the full text. Abstracts across all healthcare areas presented poor reporting quality and were inconsistent with the full texts, with results and conclusions as the most inconsistent sections. Conclusion Abstracts of healthcare RCTs and SRs have shown a large room for improvement in reporting quality and consistency with the full text. Authors, journal editors and reviewers need to give the highest priority to this matter.