by Lisa Kearns, MS MA and Arthur Caplan, Ph.D.
There is a little discussed problem in academic publishing: the scant amount of information provided by disclosures of conflict that accompany journal articles. These brief lists of organizations with which authors have financial relationships convey frustratingly little detail about the nature of the relationships. Current disclosure practices fail to provide the transparency about an author’s relationships that they are intended to.
In the belief that an upgrade to this primitive state of COI disclosure is not only necessary but long overdue, the NYU School of Medicine Division of Medical Ethics recently issued an internal Conflict of Interest Disclosure Policy.It argues that “transparency in all facets of research and scholarship” cannot be achieved by “listing only an entity that one works with; without substantive information about the nature of that work… simple disclosure seems inadequate.” The policy asks faculty and relevant staff to go beyond filling out the standard COI statements that most publications require by also creating and maintaining their own detailed statements and making them publicly available online. The policy is not an indictment of journals’ and organizations’ disclosure requirements, many of which ask for extensive information and allow for “free-form” responses, but rather an enhancement of them. It is an initial step toward increasing transparency and reducing publication bias by making authors active drivers of, not just respondents to, disclosure policy.
In practical terms, the Division’s policy encourages faculty (voluntarily for now) to craft detailed personal COI disclosure statements that can be linked to publications, presentations, and elsewhere: an electronic long-form COI disclosure statement (ELFCOI). These will be housed on the Division website, as a part of faculty professional biographies. Authors would list all financial/funding relationships, including honoraria and travel reimbursements or other compensation. And because not all bias results from financial relationships, the statements will afford the opportunity to declare political, religious, advocacy, or other sources of possible influence. A recent JAMA Psychiatry article by Cristea and Ioannidis discussed some of these non-standard but potentially biasing relationships in research on psychosocial interventions. They posit that the dearth of solid COI guidelines in this area might be due to the “limited knowledge of possible biasing effects” in a field in which a wide range of heterogenous practices are employed by adherents to a variety of schools of thoughts with preferences for specific interventions. They emphasize the potential for difficult to detect financial conflict in this field, yet also point to areas of other types of conflict as well.
Although financial relationships are the more worrisome sources of COI, as it is through these that industry has the potential to wield the strongest influence, these other types of associations might also be of interest to readers in some instances — when an author is advocating regulatory changes to abortion laws, for example, or endorsing a specific type of parenting intervention. The space to provide detail will let readers decide for themselves if author relationships have created bias. And moving disclosure to an electronic, publicly available website increases visibility while reducing the need for journals to devote space to the listing of commercial ties in agate type. (NYU is undergoing a system-wide website overhaul that is scheduled to be completed this summer; a beta version of Caplan’s ELFCOI can be found on the disclosure policy page through the link to the policy above.)
While published disclosures can be tailored to specific audiences, the ELFCOI can be made readily available to anyone who may wish to see it by providing a link to it. In submitted and published journal articles, a link to the statement could be included along with author affiliation; the ELFCOI link could be added to a disclosure slide for presentations; and it could be part of the required submission information when registering trials on ClinicalTrials.gov or seeking membership on state or federal advisory panels.
The idea for the policy springs from several sources. One is the frequent albeit brief revelations of COI in research and publishing. Another is the confusion about whether actual bias exists simply because a relationship exists. Two recent examples illustrate this particularly well: In April, Kaiser Health News (KHN) debuted the Pre$cription for Powerdatabase, which lists pharmaceutical companies’ contributions to patient advocacy groups. A month earlier, American Journal of Bioethics (AJOB) Empirical Bioethics published “Undisclosed conflicts of interest among biomedical textbook authors,” which revealed that an “appreciable subset” of authors over the study period had received contributions totaling in the millions from drug and device manufacturers. In each of these reports only the names of the parties involved and the dollar amounts received were listed, yet the suggestion of bias was straightforward, even though the researchers in the textbook report acknowledged that “their findings do not necessarily suggest that payments by pharmaceutical manufacturers…adversely impacted the textbook content.”
In academia especially, associations with the pharmaceutical industry are automatically suspect, and readers may wonder whether a relationship has influenced the content of an article, even though not all work done in concert with companies is promotional. One example is Bioethics International (BI), a nonprofit founded by a former member of the Division. The organization houses the Good Pharma Scorecard, an index that evaluates and ranks new drugs and drug companies on ethics and patient-centricity performance standards. While it does not accept any industry funding, it does engage in multi-stakeholder deliberations, including an annual forum with pharmaceutical companies, academic experts, patients and patient groups, investors, payers, and the like. Another example is the Compassionate Use Advisory Committees (CompAC), a collaboration between the Division of Medical Ethics at NYUSOM and Janssen Pharmaceuticals to develop an ethical framework by which to allocate investigational agents on a compassionate use basis, and with which both authors are involved. In both of these examples, current common disclosure practices would list just the authors’ associations with the pharmaceutical companies, without revealing the nature of their work.
The little that has been written about author disclosure in publishing has generally focused on transparency between healthcare providers and the pharmaceutical industry. Bero, Krughoff, and Loewenstein’s “Model for Broader Disclosure” called for an expanded system through which disclosure statements would reside in a secure national database. Its aim was similar to that of the NYU policy, but its scope was more limited by requiring chiefly financial relationships and COI relating to drug, device, or biotech companies. As noted, though, not all conflict is financial — conflicts can arise due to relationships with gift givers, advocacy groups, and foundations, or adherence to various schools of professional thought — and not all financial ties generate conflicts. Also, as with virtually all publication disclosure requirements, even those that request explanations of potential conflicts, Bero and colleagues’ model asks participants to provide information in response to a standardized form. Similarly, Cristea and Ioannidis propose disclosure guidelines tailored to accommodate the wide range of psychosocial interventions that researchers may use, along with increased editorial oversight. However, the Division policy places the burden for full, detailed disclosure on researchers and authors, making them the tailors and the ultimate overseers of transparency and accuracy.
Of note, the committee to which Bero’s model was proposed opposed it. Members cited potential “mischaracterization by the mass media” and cost, among other reasons. The details that faculty would be expected to provide in an ELFCOI policy would hopefully prevent this kind of mischaracterization, and the only cost of implementing the policy is the time it takes faculty to draft it and the tech department to post it.
The new policy and ELFCOI have limitations. One is that an abundance of transparency in the form of an exhaustive list of all financial, ideological, and other relationship raises concerns about violations of personal privacy. Indeed, the committee that opposed the Bero model voiced this issue. Along the same lines, a Division colleague worried that so much personal information readily available online, especially about financial holdings, could be put to nefarious use by third parties. These are significant concerns, and how to address them while still ensuring that the spirit and the goals of full disclosure are met needs further exploration. Nonetheless, the ELFCOI is an important first step toward encouraging authors and researchers to take ownership of disclosure, and to bring COI disclosure into the electronic age.
ELFCOI still may not change the minds of those critics who believe that any relationship between industry and bioethics ought to be avoided, but that is not the intention of the policy. The goal of ELFCOI is to provide easily accessed space in any setting for detailed enough information so that readers can assess for themselves author bias or conflict, and, more important, to make authors and researchers creators of, not reactors to, improved disclosure policy. More is better than less in managing COI.