How to assess the credibility of clinical trial findings?
KEY TAKEAWAYS
- Cochrane provides recommendations on how to assess the trustworthiness of clinical trial data and tackle suspected misconduct.
- Prof. Lisa Bero calls on journals, publishers, and research institutions to introduce routine data checks on manuscripts to help detect fraud.
Accurate reporting of clinical trials is essential to further our understanding of diseases and evaluate the efficacy and safety of treatments. In a recent World View article, published in Nature, Professor Lisa Bero discusses the approaches used by Cochrane reviewers to assess the trustworthiness of clinical trial findings and tackle suspected data accuracy issues.
Prof. Bero believes that fraudulent studies are widespread in the scientific literature. The fraud may not be intentional and can be linked to:
- inaccurate description of how interventions were administered
- use of inappropriate statistical analyses
- reporting fabricated data
- false representation of real data.
Cochrane provides tools to help their reviewers detect potential fraud, and templates for asking journals for investigations and retractions. The checks that the reviewers are encouraged to undertake to identify problematic studies include:
- looking for evidence of prospective clinical trial registration and ethical approval
- considering the plausibility of baseline and outcome data
- watching out for overlapping text and other inconsistencies across the article
- consulting platforms for post-publication peer review, such as PubPeer.
When reviewers find a problem, they are advised to request additional information from the authors and if the response is not satisfactorily reassuring, contact the journal editor. The journal can then launch its own investigation to decide whether the article should be retracted.
Detecting and removing fake studies from the literature requires coordinated efforts from all parties involved in the publication pipeline.
Prof. Bero does not agree with calls from some reviewers to exclude studies from certain countries or those that have not been prospectively registered. She points out that such measures would reduce global patient representation, and that trial registration does not ensure proper study conduct. Furthermore, prospective registration is uncommon for observational studies.
Prof. Bero emphasises that while the risks of mislabelling legitimate research as fraudulent cannot be ignored, detecting and removing fake studies from the literature is important and requires coordinated efforts from all parties involved in the publication pipeline. The article concludes with a call for research institutions, journals, and publishers to implement routine data checks on manuscripts and share information and technical resources to help identify anomalies.
—————————————————–
Categories