A recent article in the nursing literature (Gray, Al-Ghareeb, Davis, Mckenna, & Hillel, 2017) has generated concern in the broader academic community about author, reviewer, and editor responsibilities related to inclusion of retracted literature in systematic reviews. In an interview with the first author of that paper, Retraction Watch raised the issue of the difficulty of identifying retracted studies that might be included in systematic reviews (http://retractionwatch.com/2018/01/08/concerning-largely-unrecognised-threat-pat). For many authors who follow a specific field of research, it is common to retrieve citations and download a copy of new research as soon as it is published, with the intention that this new information would inform their own work at some time in the future. When the time arrives to write a review, a grant application, or an article for submission, the files on the subject are neatly arranged in a folder, and the accompanying citations are organized in some version of a citation management program. This is work efficiency for the busy academic researcher. If the author does an additional search prior to finalizing the paper, a retraction notice might pop up, but in some cases, the retractions are not linked to the article, they are not identified as retractions, or they are never even published. As a result, retracted articles may not be identified and erroneously included in new studies.

Scope of the problem

In the study by Gray et al. (2017) of nursing literature, a total of 37 systematic reviews contained citations to retracted trials; 23 of those retracted studies were included in the analysis (i.e., not just part of background information). What is not known is if the inclusion of retracted studies influenced the conclusions of the review. That is the problem, because clinical practice is based on evidence and evidence is derived from clinical trials and systematic reviews. A check of retracted publications in PubMed on January 15, 2018 (https://www.ncbi.nlm.nih.gov/pubmed?term=retracted+publication+%5Bpt%5D) reveals a total of 5,631 articles. Of those, 60 are found by adding the term nursing, including a few in the nurse practitioner (NP) literature. But, NPs do not just cite nursing or NP literature in their articles. For example, adding the term "statins" to the search results for retractions results in 15 retractions; several of those articles could have been included in systematic reviews produced nurses, NPs, pharmacists, medical doctors, or basic science researchers.

The reality is that the PubMed database is not a comprehensive listing of retracted literature. Retraction Watch has been developing a searchable database for retractions that will be more current and user friendly for searching by authors, reviewers, and editors (http://retractiondatabase.org/RetractionSearch.aspx?). Searching the Retraction Watch database on the subject field of nursing, 43 entries are available, along with specific information about the reasons for the retractions. There is a discrepancy between the numbers on the PubMed and Retraction Watch lists: 60 versus 43. The difference seems to be that nurses published some of their work in journals from other disciplines, and the Retraction Watch list specifically sorts on nursing journals not on whether an author is a nurse. Information about the reasons for each retraction is also available on the Retraction Watch database, which is very useful for anyone conducting reviews. Duplicate publications mean there is another article published somewhere. Including more than one study using the same population could skew the results of the review or meta-analysis because the same subjects are counted as two different cohorts.

Responsibilities of authors, reviewers, and editors

First, authors are responsible for the data they use to formulate their conclusions. Thus, authors of systematic reviews must be able to verify they have not based their findings on fraudulent data and retracted literature. They can do this by repeating the search just prior to submission and checking for any links to retractions for their references. Reviewers, who are usually experts in their respective fields, should check the references and citations for any obvious errors, including any known retractions. Maintaining currency in a field of expertise requires researchers be aware of serious misconduct and subsequent retractions within their domain of expertise. This is one reason journal editors request expert peer review. Finally, editors have an obligation to follow through on accusations of misconduct, retract literature when appropriate, and carefully review submissions and publications in their journals. Journals that are members of COPE (Committee on Publication Ethics) have guidelines to follow and are obligated to respond to all issues that question the integrity of the science they publish. Gray et al. (2017) contacted the editors of the 23 articles they discovered in the nursing literature to transmit their findings. Conscientious editors will respond appropriately by contacting the authors, assessing the impact of the retracted literature on the evidence base, and taking appropriate action based on that assessment.

Reference