KEY TAKEAWAYS

• Although the number of retracted scientific articles is rising, their impact on clinical practice is unknown.
• Including retracted articles in reviews may even be acceptable, depending on the reason for retraction.

With existing concerns about the potential impact of retracted scientific articles making their way into systematic reviews and clinical practice guidelines, Dr Jonathan Livingstone-Banks questioned in an article published on Gavi whether this makes a material difference to the interpretation of the affected reviews or guidelines.

Retractions are a hot topic in academic publishing, with an ever-increasing number of papers appearing in the Retraction Watch database. Although progress has been made in the retraction process since its inception, the founders of Retraction Watch have called for more action, such as compensating those who detect fraudulent papers, and perhaps even implementing the ‘bug bounty’ model used in the computer security industry.

Although Dr Livingstone-Banks acknowledges that a striking number of retracted papers are included in reviews and clinical practice guidelines, he expresses doubts over whether they actually impact clinical practice. With over 160,000 reviews published between 2000 and 2019, he suggests that the proportion citing retracted studies is likely to be small, and questions whether they are the high-quality articles that influence decision making. Further, even when they are included in high-quality articles, the meta-analysis process is designed to reduce their impact (eg, by assigning lower weighting to lower quality studies, or by conducting sensitivity analyses that remove questionable studies to test the reliability of the meta-analysis conclusions).

With over 160,000 reviews published between 2000 and 2019, the proportion that cite retracted studies is likely to be small.

Dr Livingstone-Banks notes that Cochrane, a global non-profit group that publishes many meta-analyses across healthcare, conducted a study on whether retracted studies affect their reviews. Cochrane does not always exclude retracted papers and instead considers the reason for retraction. Reasons for exclusion may include clear evidence that the findings are unreliable, either due to falsified data, plagiarism, faked peer review, or ‘major error’. However, a blanket policy of excluding retracted studies may omit relevant data (eg, in the case of lack of ethical approval rather than data error) or introduce publication bias.

We look forward to following how other publishers choose to deal with retracted articles in the future.

————————————————–