Background

An Institute of Medicine Report defined conflict of interest (COI) as “a set of circumstances that creates a risk that professional judgment or actions regarding a primary interest will be unduly influenced by a secondary interest” [1]. That report focused on financial COIs, “because they are at the heart of concerns and debates about conflicts of interest”. Maharaj has recently proposed a method for operationalizing financial COI, through a scoring that takes into account the nature and extent of financial conflicts. [2]

A recent study surveyed 100 trial reported in medical journals, with at least one Danish author not employed by the pharmaceutical industry. [3] The investigators found that 27% of those authors reported one or more financial COIs, while 13% had undisclosed COIs related to the trial sponsor or manufacturer of trial drugs. [3]

While the focus has been on "financial" COI, about half of the conflicts of interest cases brought to Committee on Publication Ethics (COPE) forum for advice involved non-financial COIs [4]. As an example of non-financial COI, a study by Luborsky et al found that the researcher's commitment to a particular school of thought biased the study design and outcome of psychotherapy research [5].

One type of non-financial COI that has received increased attention is intellectual COI [6–9]. We defined intellectual COI in the context of clinical practice guidelines as “academic activities that create the potential for an attachment to a specific point of view that could unduly affect an individual’s judgment about a specific recommendation” [7]. The operational definition of ‘important intellectual COI’ consisted of authorship of original studies, and peer-reviewed grant funding directly bearing on a recommendation. [7]

The International Committee for Medical Journals Editors (ICMJE) started to address conflicts of interest of authors in the mid-1980's [10]. Although criticized as being weak, inconsistent, and inadequately enforced [11], COI disclosure policies succeeded in making disclosure an integral part of medical articles [12]. In 2008, 98% of journals that officially endorsed ICMJE guidelines had COI policies as compared to 84% of journals not endorsing the guidelines [13].

One of the reasons for deficiencies in COI disclosures has been the absence of standardized COI forms [12]. A satisfactory form requires information addressing: the work under consideration for publication, relevant financial activities outside the submitted work, intellectual property, and other relationships not covered in previous sections. The most recent ICMJE form assesses these aspects more appropriately than previous versions [14].

Three recent cross sectional studies assessed the characteristics of the COI policies in high-impact biomedical journals. Blum et al and Cooper et al found that 89% of peer-reviewed biomedical journals and 93% of peer-reviewed medical journals had author COI policies [13, 15]. Both Blum et al and Bosch et al found that the definition and the requirements for COI disclosure referred to financial rather than non-financial COI [13, 16]. None of these studies specifically addressed intellectual COI, and none evaluated the implementation of the COI disclosure policy during the submission process.

The primary objective of the study was to assess the policies of clinical journals for authors to disclose their financial and non-financial conflicts of interest as implemented during the submission process. This study was conducted in the context of a wider research project to better understand the reporting and handling of financial and non-financial COIs in health research.

Methods

The Institutional Review Board of the American University of Beirut deemed the study as not human subject research. As detailed below, the research addressed the administrative process through review of public documents and a simulation of a manuscript submission, to assess the actual implementation of the COI disclosure policy.

Results

Of the 119 journals indexed under Abridged Index Medicus, we included 117 as two journals proved to be non-peer reviewed (Medical Letter and Hospitals and Health Networks). A total of eight journals published papers only by invitation; therefore, we were not able to simulate a submission of a manuscript to check for email replies. For eight journals, we limited our investigation to the review of public documents but did not simulate a manuscript submission due to associated submission fees (i.e., that did not lead to the exclusion of any journal).

General characteristics of the journals

Table 1 shows the general characteristics of the journals. Half of the journals were classified under the medical category (50%). Few journals were ICMJE members (n = 5; 4%), and most journals (n = 94; 80%) were affiliated with a professional organization.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. General characteristics of the journals (N = 117).

https://doi.org/10.1371/journal.pone.0152301.t001

COI policy

Out of the 117 included journals, 116 had a COI policy. Table 2 provides the detailed results pertaining to journals’ COI policies. Most journals had their own policies (n = 81; 70%), whereas some based their policies on that of their publishers' (n = 37; 32%). Indeed, some journals relied on both the publisher's policy and their own policy. For example, nine journals reported their policy to be based on both ‘journal’s own policy’ and ‘publisher’s policy’. The journals required COI to be disclosed in one or more of the following forms: narrative statement (n = 72; 62%), ICMJE Uniform Disclosure Form (n = 26; 22%) or a COI disclosure form different from the ICMJE Uniform Disclosure Form (n = 38; 33%). While all journals required COI to be submitted online before the submission is finalized, we learned through mock submission that 17 journals (15%) contact the authors individually (e.g., by email) to ask for COI disclosure. All journals required COI disclosure in relation to the submitted work (n = 116; 100%) and some to, work outside the submitted work (n = 43; 37%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of conflicts of interest (COI) policies (N = 116).

https://doi.org/10.1371/journal.pone.0152301.t002

Requirements for Financial COI

Table 3 provides the details of the journals’ requirements for disclosure of financial COI. Journals required disclosures of financial relationships pertaining to: the authors (n = 116; 100%), the authors' family members (n = 40; 35%), and the authors' institution (n = 34; 29%). Journals required specification of: source of payment (n = 108; 93%); amount of payment when above a specific cutoff (n = 2; 2%); and amount of payment irrespective of the amount (n = 8; 7%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Requirements for financial conflicts of interest (COI) disclosure; N = 116.

https://doi.org/10.1371/journal.pone.0152301.t003

Requirements for Non-financial COI disclosure

Sixty-six journals (57%) required disclosure of non-financial COI. Table 4 shows the details of the journals’ requirements for disclosure of non-financial COI. Some of the terms used by the journals to instruct authors to disclose non-financial COI included: "non-financial COI" (n = 4; 6%), "non-financial affiliations"(n = 3; 5%). Journals also used other terms to express requirements of disclosure of non-financial COIs: "other" (n = 44;67%), "academic association" (n = 9; 14%), "professional" (n = 5; 8%), and "intellectual" (n = 2; 3%). S1 Table describes the non-financial COIs as defined by individual journals.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Types for non-financial COI for which disclosure is required; N = 66.

https://doi.org/10.1371/journal.pone.0152301.t004

Discussion

All but one of 117 core clinical journals has a COI policy, and all these journals require disclosure of financial COI pertaining to the authors; a minority asks for financial COI disclosure pertaining to the family members or institutions of the authors (Table 3). About half of the journals required the disclosure of at least one form of non-financial COI, out of which only a few specifically referred to intellectual COI (Table 4).

The results of this study are to a certain extent consistent with those of existing research, and may point to some positive trends. Our finding that all but one of 117 core clinical journal had a COI policy represents a proportion slightly higher than those found by previous similar studies. Blum et al found that, in 2008, 89% of “high-impact medical journals across 35 subject categories” had author COI policies [13]. Bosch et al found that, in 2011, 91% of high impact biomedical journals had language requiring COI disclosure [16]. While this might suggest improvement over the last few years, it might also be due to the different mix of journals included. A recently published study of 64 general and abdominal surgery journals found that 88% of those journals mandated authors to disclose their COIs. [17]

In our study 57% of journals required disclosure of at least one form of non-financial COI (Table 4). Blum et al found that, in 2008, respectively 42% and 26% of high-impact medical journals required disclosure of “personal non financial relationships” and of “non financial relationships with organizations” [13]. Kesselheim et al found that, in 2009, 42% of major oncology journals sought non-financial disclosures [18]. Bosch et al found that, in 2011, 70% of high impact biomedical journals required disclosure of non-financial COI [16]. Those numbers remain relatively low considering the increasing concern about non-financial COI [4]. There is similarly concern that only two journals required disclosure of intellectual COI, though this could be a labeling issue.

A small minority of journals (17%) described a potential impact of disclosed COI on the editorial process (S2 Table). Even amongst those, very few had clear-cut criteria for rejection based on the content of the disclosures.

There is evidence that experts may fail to disclose all of their COI. One study found that 8% of experts failed to disclose their financial COI in clinical practice guidelines for hyperlipidemia or diabetes [19]. A minority of journals (17%) reported having procedures to verify authors' COI disclosures (S3 Table). Only 24% of journals reported potential impact of non-disclosure on the editorial process (S4 Table).

One study investigated perceptions of COIs of reviewers of and applicants to French academic grants. [20] This qualitative study found that participants believed non-financial COIs have a greater influence than financial COIs. Another study investigated the relationship between screening mammography recommendations and the COI of the guideline panel members. [21] It found that the odds of recommending routine screening was associated with the number of recent related publications by the lead guideline author. Also, investigators have assessed the interpretation of a meta-analysis with highly heterogeneous results by authors of primary studies and by methodologists. [22] They specifically asked authors of primary studies and methodologists about the effect size and potential causality of the association. Authors were more likely than methodologists to interpret the results of a meta-analysis as indicating a strong association. Moreover, the higher the number of papers related to the review topic published by the authors, the higher their estimates of the “true odds ratio” were. While authors’ estimates of the true odds ratio were correlated with the size of odds ratio in their primary studies, the association was not statistically significant.

This study has a number of strengths. This is the first study to specifically address disclosure of intellectual COI, whether the journal makes the disclosure available, and the effect of disclosures on the editorial process. We have used systematic methods for data abstraction (duplicate process). This is the first study to evaluate the actual implementation of the COI disclosure policy during the submission process itself.

The study also has limitations. Our design would not have captured journal requests for COI disclosure that might occur at later steps of the editorial process (e.g., after acceptance of the manuscript). We were not able to simulate a manuscript submission for seven journals that did not have a submission site and for eight journals that imposed submission fees.

Our findings have a number of implications for editorial policies. Those policies should provide clear and explicit guidance to authors on disclosing non-financial COI. Based on the findings reported in S1 Table, the policies could request the disclosure of non-financial COIs such as intellectual COI (related to authorship of studies and opinion pieces), academic affiliations, professional COIs, political and religious affiliations. While there is a need for improving policies in terms of the management of disclosed COI beyond their publication, this is a more challenging and controversial objective. Other areas to improve include the verification of financial COI of family members and institutions of authors.

The findings have a number of implications for future research. It would be valuable to explore how the standardized COI form (e.g., the ICMJE form) impacts the completeness and veracity of COI disclosures. It would also be helpful to address how frequently COI disclosures affect editorial decisions (e.g., rejection), and how frequently COI disclosures affect the perceptions of the readers regarding the trustworthiness of study findings. On a more general level, future research should assess whether all non-financial relationships discussed above should be deemed COI, whether they could be avoided, and whether they have as much influence as financial COI.

Supporting Information