The concept of conflict of interest (COI) has garnered increasing attention in medical journals and the broader media (1–8). There has been particular concern about how inappropriate relationships between physicians and the pharmaceutical industry may impair the integrity, independence, scientific rigor, and/or clinical judgment of physicians and researchers (8, 9). COI pertaining to psychiatry has come under heightened scrutiny because of the pervasive media focus on psychiatry, allegations of nondisclosure among prominent psychiatrists, the fact that psychiatrists often top lists of those receiving financial support from drug-makers, and because several of the most profitable medications are produced for psychiatric indications (3–5). Recently, Congressional investigations and high-profile retractions and letters of apology in journals have galvanized editors and institutions to reevaluate their policies to ensure that they have comprehensive and clear standards with respect to COI (10–12).

COI is defined as a set of conditions in which professional judgment concerning a primary interest such as a patient's welfare or the accuracy of research findings may be unduly influenced by a secondary interest (7, 13). The International Committee of Medical Journal Editors (ICMJE) specifies that “COI exists when an author (or the author's institution), reviewer, or editor has financial or personal relationships that inappropriately influence (bias) his or her actions (such relationships are also known as dual commitments, competing interests, or competing loyalties).”

Disclosure of potential COI is a cornerstone of journal policies regarding COI. Understanding research findings includes taking into account how study design and authors' interpretation may be affected when a sponsor stands to gain from the findings (13–16). In the last 25 years, drug- and device-makers have become the primary source of research funding, even ahead of government and academic institutions (17, 18). Disclosure at the reviewer and editor levels is also recognized as an area where COI may affect the integrity of the science (15, 19). For journals, one way to ensure transparency around COI is for authors, reviewers, and editors to disclose their COIs. To this end, journals have developed policies to address how and what authors, reviewers, and editors disclose.

Despite COIs' being “ubiquitous and inevitable in academic life,” limited data are available regarding COI policies (2). Initial research indicates that there is wide variation among journals in requirements for publishing author, reviewer, and editor disclosures (20, 21). In this study, we aimed to characterize COI and disclosure policies published in peer-reviewed journals and, based on heightened awareness of COI in psychiatry, to assess whether there is a qualitative and quantitative difference between psychiatric and nonpsychiatric publications.

Sample

Our sample consisted of 40 journals, 20 selected from psychiatry and 20 from the general medical index. We selected 20 journals from each category with the highest Impact Factor (IF), on the basis of 2007 IF. We excluded the 14^th-highest rated nonpsychiatric journal, Cochrane Database Systematic Review, because its articles are meta-analyses and do not include individual disclosures; we instead added a child- and adolescent-specific journal (in addition to the Journal of the American Academy of Child and Adolescent Psychiatry) to provide a broader sampling of policies beyond adult domains. Of the nonpsychiatry journals, we selected the 18 highest-ranking, peer-reviewed, nonpsychiatry journals. Because two of the psychiatry journals had a pediatric orientation, we added Pediatrics and the Journal of Pediatrics, which were closely matched to the IF of the two pediatric-psychiatry journals. The 20 psychiatric journals consisted of Archives of General Psychiatry, Molecular Psychiatry, American Journal of Psychiatry, Biological Psychiatry, Neuropsychopharmacology, Schizophrenia Bulletin, British Journal of Psychiatry, Journal of Clinical Psychiatry, Psychotherapy and Psychosomatics, International Journal of Neuropsychopharmacology, Journal of the American Academy of Child and Adolescent Psychiatry, CNS Drugs, Bipolar Disorders, Journal of Child Psychology and Psychiatry, European Neuropsychopharmacology, Schizophrenia Research, American Journal of Medical Genetics Part B: Neuropsychiatric Genetics, Psychological Medicine, Addiction and Journal of Child and Adolescent Psychopharmacology, with a mean IF of 5.959, median IF of 5.041, and a range of 3.139 to 15.976. The 20 nonpsychiatric journals included New England Journal of Medicine, Lancet, Journal of the American Medical Association, Annals of Internal Medicine, Annual Review of Medicine, PLoS Medicine, British Medical Journal, Archives of Internal Medicine, Canadian Medical Association Journal, Annals of Medicine, American Journal of Medicine, Journal of Internal Medicine, Medicine, Annals of Family Medicine, Mayo Clinic Proceedings, Pediatrics, Journal of Pediatrics, American Journal of Preventative Medicine, Current Medical Research and Opinion, and Journal of General Internal Medicine, with a mean IF of 11.026, a median IF of 7.729, and a range of 2.876 to 52.589.

Access to Policies

We used qualitative and quantitative approaches to evaluate the COI and disclosure policies that appeared in print or journal websites. If information was password-protected, we registered for accounts and gleaned as much information as was available on the journal website. If information was not available, our data indicate it as “NA.” The COI policies were analyzed by use of deductive content analysis, wherein the contents of the policies were analyzed according to predesignated domains (presented in Table 1 and Table 2) (22).

COI policies for authors, reviewers, and editors in the psychiatric and nonpsychiatric journals are shown in Table 1. Table 2 provides data about specific aspects of COI policies for authors. Differences across categories were compared by Fisher's exact test.

Publication of COI/Disclosure Policies

All journals published COI/disclosure policies that were accessible in print and online. Most journals provided comprehensive information online and referred readers to their websites. Both groups (75% of the psychiatric and 80% of nonpsychiatric) provided general definitions of COI.

Definitions of COI

For author-specific COI, journals across disciplines had similar rates of providing COI definitions and timelines for the declared COI. One apparent difference was whether the journal provided examples of COI for the authors: only 40% of the psychiatric journals gave examples, versus 85% of the nonpsychiatric journals (p<0.05). Eight of the psychiatric journals and one of the nonpsychiatric journals required “complete” (versus “relevant”) disclosure (p<0.001), and 19 of the nonpsychiatric journals required “relevant” disclosure, versus 12 of the psychiatric journals (p<0.05). However, nonpsychiatric journals tended to provide more detailed information about what could constitute potential conflict and asked for broader potentially relevant funding sources.

Publication of COI

Sixteen psychiatric and 19 nonpsychiatric journals asked authors to explicitly disclose if COI was not present. Every psychiatric journal published declared COI, whereas 70% of nonpsychiatric journals had such a policy (p<0.05), the remaining 30% publishing COI statements based upon the discretion of the editors. Only 25% of psychiatric and 20% of the nonpsychiatric journals described penalties for misstating COI. Some variability also existed regarding which submissions required COI declarations: 55% of psychiatric journals, versus 70% of nonpsychiatric journals, required “all” submissions to have COI statements. Forty percent of psychiatric journals and 15% of nonpsychiatric journals required statements for “manuscripts,” which was sometimes defined and sometimes vague. Nine psychiatric and 11 nonpsychiatric journals had COI forms for the authors (variably, a group form or form for each author); the remainder required authors to submit their own disclosures. Forty-five percent of psychiatric and 70% of nonpsychiatric journals did not provide a form for the authors. Interestingly, some journals provided both group forms at the time of initial submission and then individual forms at the time of acceptance/publication (which explains the percentages that total over 100%).

COI Policies for Editors and Reviewers

Eight psychiatric and eight nonpsychiatric journals provided definitions of COI pertaining to reviewers and required reviewers to declare their COI. The remainder of the journals either did not have such policies or the policies were not easily accessible. One psychiatric and two nonpsychiatric journals specified timelines for the COI disclosures for the reviewers. Three psychiatric and eight nonpsychiatric journals required disclosures for editors, and, in one psychiatric and four nonpsychiatric journals, editor COI was accessible.

Specific Disclosures for Authors

Table 2 delineates the range of disclosures required of authors by the journals. Overall, the nonpsychiatric journals specified more categories than psychiatric journals, such as personal relationships, academic competition, and “intellectual passion.” Seven psychiatric journals allowed authors to determine what to disclose, whereas five nonpsychiatric journals provided authors with that opportunity. Seven psychiatric and 12 nonpsychiatric journals reported honoraria. Eleven psychiatric and 16 nonpsychiatric journals published stock ownership, often if ownership was above a certain number of shares. Patents were evenly spread, in that 50% of each journal type requested this information. It is important to note that some journals provided both a range of categories and encouragement of authors to report conflicts not delineated by that journal.

The debate over COI in academic medicine has evolved substantially since Relman addressed it in the 1980s (23, 24). Government officials, members of the media, and the broader public have become vocal participants in advocating for the reform of COI policies (20, 24–27). This growing interest highlights the investment of these groups in the scientific process and the widespread concerns about how COI in research can affect scientific knowledge. Despite these concerns, little has been published regarding COI policies (21). This preliminary study suggests that journals can be reviewed qualitatively and quantitatively to ascertain COI policies and to compare journals across various disciplines.

Of the medical specialties, psychiatry has been most closely scrutinized by government officials and the media (3–5). Overall, in our study, COI policies did not differ largely between psychiatric and nonpsychiatric journals. What is apparent is that the variability observed highlights the challenges in developing COI policies. For instance, some journals choose to use examples and some do not (there were significantly more examples in nonpsychiatric journals), which raises the question of whether examples assist authors in crafting COI declarations or, conversely, provide a rationale for withholding disclosures if the potential COI does not closely match the examples given.

There is lack of uniformity in how COI is described, what is considered COI, what types of COI need to be disclosed, the extent of the disclosure, and who must disclose COI. Written policies can guide authors, editors, and reviewers, but establishing formal policies on a journal-by-journal basis has resulted in substantial discrepancies. Should disclosure be “complete” or “relevant”? The psychiatric journals were significantly more likely to request “complete” and nonpsychiatric journals “relevant” disclosures. On the basis of our review, “complete” appears to include any funding that an author receives, whereas “relevant” would pertain to the paper being submitted (19). Nevertheless, these terms are ambiguous. Authors are put into a position of deciding how complete “complete” must be (a grant received to sponsor attendance at a meeting while a trainee? all investments, even those outside biotechnology?) and how relevant “relevant” is (relevant to the exact drugs studied, or relevant to the broader topic?).

Another example of lack of uniformity relates to timing of COI. Requirements ranged from disclosure of current COI to any potential COI within the past 2-to-5 years previously, but also 2, 5, and 10 years hence, or “for the foreseeable future.” One journal asked authors to declare potential COI from “at least the past 12 months” (putting authors in the unenviable position of deciding how much further than 12 months they ought to go). Approximately half the journals provided timelines for declaring COI, but only one psychiatric and two nonpsychiatric journals provided similar guidelines for reviewer COI, and none provided explicit information on the time-frame for editors.

This, in turn, raises a number of questions about different expectations for authors, reviewers, and editors. Overall, the policies were directed toward authors, rather than reviewers and editors. It may be that authors have historically been held to a higher standard of judgment, as they are the ones submitting the data. Nevertheless, recent articles have begun to highlight potential sources of bias in both reviewers' and editors' conflicts of interest (15, 19), and our study suggests that far more clarity is required.

Limitations of the current study include its descriptive nature and the small number of journals studied. Further research could increase the number of journals, expand the number of medical specialties studied, and look at differences between clinical and research-oriented journals. In this preliminary study, we divided the journals into two groups (psychiatric and nonpsychiatric), but there was a difference in the mean impact factor between the two groups; furthermore, many of the “nonpsychiatric” journals publish a large amount of psychiatric material, including original research, and therefore the distinction between “psychiatric” and “nonpsychiatric” is itself limited. Finally, this study did not examine whether authors, editors, or reviewers adhered to policies, or how they understood and interpreted the policies. Given the discrepancies between journals and the potential for ambiguity, this would be an important line of inquiry.

Although COI has been a topic of substantial debate in medical fields and in larger society, there are challenges to codifying COI policies and creating standardized approaches. How COI is presented by journals may affect its perceived importance by authors and reviewers (26, 28). Going forward, the role of standardized forms for disclosure is worthy of attention and prospective study, particularly as the ICJME recently published a uniform Author COI Disclosure form to be trialed by ICMJE member journals (29). Would harmonization or standardization policies and procedures across journals improve accountability and disclosure practices?

Challenges in adopting a standardized approach may reflect ongoing debates about what exactly constitutes COI, what needs to be disclosed, and who is responsible for disclosing COI. Universal standardized forms would require very clear guidelines and definitions, and, ideally, non-arbitrary timelines. It would also be worthwhile considering whether differences in policies between journals reflect the different concerns of specific disciplines or fields, or, taking a slightly different approach, whether journals need to adopt different standards based on discipline-specific concerns. Further study into these and other areas will help to improve journal COI policies and practices, with the ultimate goal of enhancing the quality and credibility of biomedical publications.

The authors report no competing interests.