Conflict of Interest in Journal Peer Review

Although there is no unified definition of conflict of interest (COI) within the medical and scientific community, it is mostly described as a set of circumstances in which a primary professional interest is excessively influenced by an individual’s secondary interest(s) (Lo and Field 2009; Thompson 1993). Such COIs may apply to authors, editors, and reviewers. Most of the discussion of COI in scientific publications has focused on the impact that authors’ financial and nonfinancial (i.e., personal or professional) interests can have on research integrity (Bekelman, Li, and Gross 2003; Shamoo and Resnik 2015), but there has been less focus on editor and reviewer bias. Awareness of the ethical concerns created by COIs in research has grown steadily since the 1990s, and the vast majority of journals have adopted policies that require authors to disclose financial interests (Resnik, Konecny, and Kissling 2017). Fewer journals have adopted policies that require reviewers or editors to disclose COIs or potential COIs. A survey of 91 biomedical journals conducted in 2006 found that while 93% require authors to disclose COIs, only 46% require reviewers to disclose COIs and 40% require editors to make such disclosures (Cooper et al. 2006). Other studies have reported similar findings. For example, a study of 224 environmental, public, and occupational health journals found that while 96% require authors to disclose COIs, only 33.9% apply this policy to reviewers or editors (Resnik, Konecny, and Kissling 2017). A survey of 256 medical journals for author manuscript submission documents found that 89% had COI policies (Blum et al. 2009). Another survey of 399 high-impact biomedical journals found that 89.7% require authors to disclose COIs but only 38.8% apply this policy to editors (Bosch et al. 2013), and a survey of 37 ophthalmology journals found that 100% require authors to disclose COIs but that percentage was much lower for reviewers (60%) and editors (30%; Anraku et al. 2009).

COI in journal peer review is an important problem that merits more attention than it currently receives. COIs can bias judgments made by reviewers or editors in favor of or against a manuscript and undermine authors’ trust in the peer-review process (Resnik and Elmore 2016). A recent study of 52 top influential medical journals found that 50.6% of editors received payments from industry in 2014 (Liu et al. 2017). But it’s not just financial profit that may bias a review process. COIs may also involve the relationship that an editor or reviewer has with the author. Some types of COIs that could lead to bias include:

a researcher with stock in company X is asked to review a study showing that a drug manufactured by the company is superior to competing products;

a researcher with stock in company Y that manufactures a competing product is asked to review the same study;

an associate editor with stock in company X handles the review process;

a researcher is asked to review a paper submitted by someone he or she has a professional relationship with, such as a colleague at his or her institute, a collaborator, or a former student or advisor;

a researcher is asked to review a paper submitted by a competing research group; and

an associate editor at a journal submits a paper to the journal.

Many journals allow authors to suggest preferred or nonpreferred reviewers. Authors may take advantage of this option by suggesting colleagues or collaborators who are likely to be receptive to their manuscripts. Several studies have shown that reviewers suggested by authors tend to rate manuscripts more favorably than those suggested by editors. For example, Schroter and colleagues examined 788 reviewer reports from 10 top biomedical journals and found that author-suggested reviewers made significantly more favorable publication recommendations than editor-suggested reviewers, although there were no significant differences in the quality of reviews based on an independent assessment (Schroter et al. 2006; Wager, Parkin, and Tamber 2006). Wager and colleagues examined 100 reviewer reports submitted to BioMed Central journals and also found that author-suggested reviewers made significantly more favorable publication recommendations than editor-suggested reviewers, and Bornmann and Daniel made a similar finding based on an examination 1,145 reviews submitted to the journal Atmospheric Chemistry and Physics (Bornmann and Daniel 2010; Wager, Parkin, and Tamber 2006).

Most granting agencies prohibit some types of conflicts and require scientists who review research proposals to disclose COIs (Shamoo and Resnik 2015). The National Institutes of Health, for example, does not allow an individual to review proposals he or she could receive a direct financial benefit from. Reviewers are also restricted if they have had a major role on the proposal or have had a professional relationship (e.g., collaborator, supervisor, or student) with someone listed as having a major role on the proposal within the last 3 years (National Institutes of Health 2014). The National Science Foundation (2004) has a similar policy.

What can journals do to address COI in peer review? The most important step that journals can take is to develop clear and effective COI policies for reviewers and editors. A policy should define COIs and provide individuals with examples of the types of situations covered by the policy and include a process for reviewers and editors to report the COI and recuse themselves from the review process, if appropriate. If a reviewer discloses a conflict that requires recusal, an editor could invite a different scientist to review the manuscript. If an editor discloses a conflict, a nonconflicted editor could handle the review process, assuming that the journal has more than 1 editor. The Committee on Publication Ethics (2016, 2017) has developed guidelines that journals could adopt or use for addressing reviewer and editor COIs.

Prohibiting all COIs in journal peer review could compromise the quality of the review, since many of the individuals with the required expertise may have personal or professional relationships with the authors (Resnik and Elmore 2016). Finding nonconflicted reviewers can be especially difficult in smaller research fields where only a handful of people may be qualified to review a manuscript. If an editor has difficulty finding nonconflicted reviewers, he or she could try to manage this situation by including some nonconflicted reviewers in the review process to compensate for potential bias (Shamoo and Resnik 2015).

Blinding reviewers to the identities of authors may help to minimize some COIs in peer review, but it is no panacea, since studies have shown that blinded reviewers can correctly identify first authors more than 40% of the time (Baggs et al. 2008; Cho et al. 1998; van Rooyen et al. 1998). Also, while blinding may minimize biases due to personal or professional relationships, it is not likely to impact those due to financial relationships since blinding does not involve removal of the names of products or research sponsors.

It would be useful for journals to collect information on the prevalence of reported COIs among reviewers and editors and the types of conflicts that occur (e.g., financial, professional). A declaration of all real or potential competing interests for editors and reviewers should be received by the journal before an article is reviewed and accepted for publication. Many journals already collect information on reviewer performance, so it would not be too difficult to track COI data. This information may assist journal editors in developing and refining their COI policies and in assuring the integrity of published research.