After a paper has been through peer review and has been published it is the obligation of the scientific community to scrutinise an author’s work. If a serious error or misconduct is spotted the paper should be retracted and the work is removed from the evidence base. Over the past ten years there has been an exponential growth in the number of retracted papers. Much of the increase may be explained by the use of technology that has made it easier to spot duplicate publications, or fabricated data, for example. Once a paper is retracted researchers should not cite this work in future publications; this is, however, not the case. Many papers continue to be cited long after they have been retracted. Retraction Watch has a list of the ten most highly cited retracted papers. The paper that currently holds the number one spot has been cited a total of 942 times, after retraction. It is plausible that researchers are using retracted work to justify further study. This may be the scientific equivalent of “fruit of the poisonous tree”. That is to say, if the research is based on tainted work then that work is itself tainted. Authors may also include retracted work in systematic reviews and meta-analyses. In clinical disciplines – such as nursing or medicine – this is particularly worrisome.
Clinical practice should be based on the best available evidence, i.e. from systematic reviews. If a review were to include a retracted paper then the resulting meta-analysis would be contaminated and recommendations for practice emerging from the study would be unsound; ipso facto putting patients at risk because a clinician is using evidence that is flawed. To date we have found five examples in the nursing literature where this has happened. We have written to the journal editors to advise then of the error that authors have made. In our minds this is a cut and dry issue. The author has clearly made an error, potentially a serious error and one that will need to be resolved. Either the editor will need to issue an erratum or potentially retract the review (and there are examples in the literature where this has happened).
There is a second way in which a systematic review may include research that is retracted. This is when the authors of the review cite a paper that is retracted after the review is published. A more nuanced debate is perhaps required given that the review author has not made a mistake. Would it not be punitive to the author – potentially damaging their career prospects – to retract a review when they have not made a mistake? However, the inclusion of a paper that has subsequently been retracted has the potential to impact effect sizes in meta-analysis and/or review conclusions. My group undertook a study to explore how often retracted clinical trials were included in systematic reviews. The answer; more common than you might think. We followed up the citations of eleven retracted nursing trials and determined that they were included in 23 systematic reviews. Currently there is no mechanism that will alert authors (or publishing editors) that their systematic review includes a study that has subsequently been retracted. We suspect, but don’t know for certain, that in medicine and the allied health professions there are many more systematic reviews that include retracted studies. Clinical practice guidelines, such as those produced by the National Institute of Health and Care Excellence (NICE) rely on evidence from systematic reviews. And this is where our observation flips from being an interesting intellectual exercise to one that may impact patient safety. Could it be that patients are being exposed to ineffective treatments because guidelines are based on flawed reviews?
Journal editors, reviewers and researchers need to be aware and mindful that systematic reviews may contain citations that have been retracted. There is a compelling argument that the editor who issues a retraction notice for a paper also has a duty to alert authors citing this work of the retraction decision. Part of the peer review process should be checking that included references (particularly those included in meta-analysis) are not retracted, it might also be argued. Finally, not only do review authors need to ensure that they have not cited retracted papers, but they also have a responsibility to periodically check (something the Cochrane collaboration encourage authors to do) the status of included studies.
The inclusion of retracted trials is a threat to the integrity of systematic reviews. Consideration needs to be given to how the scientific community responds to the issue with the ultimate goal of keeping patients safe.
Professor Richard Gray is the editor of the Journal of Psychiatric and Mental Health Nursing. No other conflict of interest declared.
This post may be cited as:
Gray R. (26 May 2018) The inclusion of retracted trials in systematic reviews: implications for patients’ safety. Research Ethics Monthly. Retrieved from: https://ahrecs.com/research-integrity/the-inclusion-of-retracted-trials-in-systematic-reviews-implications-for-patients-safety