Disclosing interests and support
Authors declare their financial and non-financial interests and the support behind a piece of work so that readers can judge it: what counts as an interest, how journals and other bodies ask for it, how a declaration is gathered and worded, and what the evidence says a declaration actually changes.
What counts as an interest
A conflict of interest is any tie, relationship or activity, financial or otherwise, that could be perceived to influence the objectivity or veracity of what is written. JAMA's instructions for authors state that a conflict of interest may exist when an author, or the author's institution or employer, has financial or personal relationships or affiliations that could influence or bias the author's decisions, work or manuscript, and require all authors to report potential conflicts of interest, including specific financial interests relevant to the subject of the manuscript, in the Acknowledgment section.
A conflict of interest may be financial, such as funding support for the project or speaking honoraria from a company whose product the article recommends, and it may also be consultancy, expert testimony, stock ownership or employment by a company whose product the article mentions, or involve family members, academic competition and personal relationships. A note that a trial was supported by a grant from the maker of the product it tested can undermine the report's credibility for a reader. Peer reviewers and editors, as well as authors, can have conflicts of interest.
In 2019 JAMA's instructions to authors listed employment, affiliation, funding and grants received or pending, consultancies, honoraria or payment, speakers' bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, and patents planned, pending or issued; journals vary in their definitions, and many extend them to a spouse or first-degree relative. Researchers rarely provide blatantly false data for money, and the effects of these relationships are usually subtler: letting industry decide what is studied and how it is analysed, choosing which data to include or exclude, or believing so honestly in an intervention's superiority as to be blind to a balanced review of the alternatives. In May 2017 JAMA devoted an entire issue, of 23 articles, to conflicts of interest. Transparent disclosure helps editors publish objective findings that keep the literature credible, lets readers evaluate possible bias and lets clinicians judge whether findings apply to their patients.
In 2018 most journals limited disclosure to financial interests, and controversies over non-financial interests were argued head to head in the BMJ. An expert's decisions are influenced by political, ideological, individual and religious interests, such as past or current disputes between researchers and differences between schools of thought; people are driven at least as much by non-financial motives as by financial gain, and fame may be more seductive than gain. Declarations of non-financial interests have to be handled with discretion. Every expert has non-financial interests that cannot be eliminated, while financial conflicts of interest can be eliminated or avoided.
The Nature research journals updated their policies in 2018 to ask authors of research articles, reviews, commentaries and research analyses to declare non-financial competing interests, which they describe as taking different forms including personal or professional relations with organisations and individuals, and encourage authors and referees to declare any unpaid roles or relationships that might have a bearing on the publication process. Their examples are unpaid membership of a government or non-governmental organisation, unpaid membership of an advocacy or lobbying organisation, an unpaid advisory position in a commercial organisation, writing or consulting for an educational company, and acting as an expert witness.
In a 2020 discussion of the transparency record of Dr Raoult's chloroquine trial, one participant noted that its authors supplied supplementary data at submission and registered the trial in the EU Clinical Trials Register, but that the lead author did not disclose that he was the editor-in-chief of the journal that published the paper, which the participant identified as a major conflict of interest.
Industry ties and bias
Financial ties to industry are pervasive. In a survey of orthopaedic surgeons presenting or serving in another role at a 2008 annual meeting of the American Academy of Orthopaedic Surgeons, 79.3% reported directly related and 50.0% indirectly related payments from manufacturers of joint prostheses, and after the National Cholesterol Education Program of the National Institutes of Health (NIH) released stricter serum cholesterol guidelines in 2004, most members of its panel turned out to have financial ties to the companies that stood to gain from increased use of statins. Disclosure is the fundamental way to deal with a possible conflict of interest, and in a grey area full disclosure is a strong defence against later criticism.
A survey of randomized clinical trials conducted over a decade found that trials funded by for-profit organisations were more likely to report positive findings than those funded by not-for-profit organisations, raising questions about the design and conduct of industry-funded trials and ethical concerns about potential violations of clinical equipoise.
In the late 1990s many published trials had selectively included or excluded data, underplaying harms and overselling benefits, and multiple studies of bias in the literature suggested that financial relationships between authors and industry were at least partly responsible. Among them, a 2003 systematic review in JAMA (144 articles retrieved, 37 meeting the inclusion criteria, 8 of them accounting for 140 individual studies) found industry-sponsored studies 3.6 times more likely than non-industry studies to reach pro-industry conclusions, and its investigators concluded that financial relationships among industry, investigators and academic institutions are widespread and that the conflicts arising from them can influence biomedical research in important ways. Educational efforts and ethical statements followed from the American Medical Association, the Pharmaceutical Research and Manufacturers of America (PhRMA) and the Office of the Inspector General, but participation was voluntary, adherence unregulated, there were no consequences when the guidelines were not followed, and compliance was repeatedly poor.
Concerns often arise that the commercial goals of a sponsor can overrule the design, execution, analysis and interpretation of a trial, and few researchers have tried to compare industry-funded and non-industry-funded trials objectively. A 2012 commentary held that the evidence so far suggested that industry-funded trials were, on average, of higher quality. An analysis of randomized controlled trials across several disease areas, drawn from a sample of Cochrane reviews, found that while conclusions tended to favour the experimental drug in industry-funded studies, those studies also had larger sample sizes, more complete recording of adverse events, more frequent use of placebo or no-treatment controls and of double blinding, and were more likely to appear in high-impact journals. A paper on long-term randomized trials in obesity found reporting quality significantly higher for industry-funded trials, and a systematic review found more complete reporting of safety data in them.
The same commentary held that evidence that industry-funded studies are of higher quality does not excuse publication biases such as selective reporting or the downplaying of negative outcomes, nor cases of alleged or real misconduct in industry-funded research. It suggested that the multiple levels of scrutiny, often by external bodies, to which industry-funded trials are submitted may be partly responsible for their high-quality clinical and reporting practices, and that there is still room for improvement, the first step being to make datasets publicly available to encourage multiple independent interpretations.
How journals and other bodies ask for disclosure
The World Association of Medical Editors (WAME) position of 2016 sets out the key elements of a journal's conflict-of-interest policy. A policy should state clearly what the journal counts as a conflict of interest and who is captured by the definition, the types of competing interests to be declared, what is to be declared, when, to whom and in what format, the journal's role in seeking clarification, the consequences of failing to disclose before or after publication, and how the journal will manage conflicts. Its sample definition holds that a conflict of interest exists when a participant in the publication process (author, peer reviewer or editor) has a competing interest that could unduly influence, or be reasonably seen to influence, their responsibilities in that process, and its examples of competing interests to declare are financial ties, academic commitments, personal relationships, political or religious beliefs and institutional affiliations, with the recognition that competing interests can coexist. Journals rely on disclosure because routine monitoring or investigation is not possible, which places a particular onus on the declarer to report carefully and comprehensively, and on the journal to ask in a way likely to elicit relevant conflicts. A journal's policy should state that all relevant disclosures, or the declaration of none, will be published with the article, and which conflict-of-interest situations will result in a manuscript not being considered, and its rules for dealing with conflicts and disclosures need to be made known to everyone involved in the publication process.
A financial disclosure form derived from ICMJE documents tabulates, against the study or manuscript sponsor and any relevant competing or associated entity, employment by the sponsor, compensation for coauthoring the paper, equity and options (major at $10,000 or more, or minor), principal investigatorship, personal monies, grant support from entities that could be affected financially by the published work, academic support in the last 36 months, royalties and expert testimony, with further columns for government support and patents.
Journal conflict-of-interest policies were rare in 1997 and common by 2008. In 1997 only 16% of 1,396 highly ranked scientific and biomedical journals had conflict-of-interest policies, fewer than 1% of their articles mentioned a conflict for any author, and two-thirds of the journals with policies listed none that year; by the end of 2008, after a widely publicised disclosure scandal, 89% of 256 highly ranked journals across 35 disciplines had policies, of which 77% defined conflicts and 54% asked authors to sign disclosure statements, yet not all conflicts appeared in print, and definitions still varied. In 2019 failure to disclose, incomplete or inconsistent disclosure, and misunderstanding of what counts as relevant persisted, and the trend was toward transparency about the interests of reviewers and editors too.
Most researchers make time-consuming, detailed disclosures throughout the year, each one slightly different. A uniform disclosure format may be impractical because journals serve different readerships and what counts as relevant varies. One alternative is CONVEY, a secure global database introduced by the Association of American Medical Colleges four years after the Institute of Medicine asked it in 2012 to simplify the process, in which researchers and universities list and modify relationships free of charge, with metadata such as type, subject and dates for targeted retrieval by subscribers; pilots were under way at the New England Journal of Medicine and the University of Wisconsin in September 2018, and growth had been slow. Editors admit they lack the resources to check every contributor's disclosure and depend on authors' professionalism, universities treat conflict-of-interest training both as a prerequisite for NIH grants and as a defence against reputational damage, and coauthors and reviewers have their own reason to care, since one author's scandal reflects badly on everyone involved.
Disclosure is also required outside journals. Since 1999, marketing applications to the US Food and Drug Administration have had to disclose relationships between industry and the investigators of studies showing a drug's or device's efficacy or safety, and filing may be withheld without it; FDA employees and their spouses and minor children may not hold financial interests in companies with more than 10% of their income regulated by the agency. Most grant applications carry extensive disclosure requirements, and the NIH, which awards almost 50,000 competitive grants, is obliged both to advance medical science and to steward taxpayer money responsibly: it introduced its reporting guidelines in 1995 and restructured the system in 2011 to define significant financial conflicts better, instruct researchers on what to disclose to their institutions, guide institutions in managing their conflict databases and outline investigator training, with universities monitoring their research staff's conflicts, reporting at application, yearly and on any change, and clarifying discrepancies through dialogue, and a reporting failure able in principle, though it had never happened, to cost an institution all its NIH grants.
In 2009 the International Committee of Medical Journal Editors (ICMJE) created a single form for its member journals to use in requesting disclosure of authors' potential financial conflicts of interest, placed it in the public domain with a blank version and a completed sample published, and ran it in beta until April 2010 with a comment facility for reporting problems; the hope was that authors would keep a partly completed form on their computers and fill in the manuscript-specific information as needed. Medical writers could do a public service by educating editors and authors about the form. In the same period PhRMA revised its principles on the conduct of clinical trials and communication of trial results, effective from 1 October, addressing disclosure of industry sponsorship, standards for authorship, acknowledgment of medical writers and sponsor review of clinical trial reports, and published them at no cost.
Also in 2009, a new association of clinical researchers and educators set out to promote responsible physician-industry collaborations and, on its own website, opposed what it called increasingly onerous regulations championed by an anti-industry movement, naming substantial disclosure, censorship on writing, speaking or advising, and restrictions on physician-industry associations, actions and rewards, on the argument that these only serve to decrease medical education and innovation and will eventually have a negative effect on patient health.
The September 2009 issue of Mayo Clinic Proceedings carried a commentary by a former manager of a pharmaceutical company's medical communications department for clinical research publications on conflicts of interest, authorship and disclosures in industry-related scientific publications, subtitled for the tort bar and the editorial oversight of medical journals, alongside an editorial by the journal's editor-in-chief asking who will champion integrity in bidirectional conflicts of interest involving industry and medical journals.
In one medical writer's experience, conflict of interest is a vague term in a manuscript, even when a template such as the ICMJE's is used. In January 2020 the ICMJE proposed a new author disclosure form, in an editorial published in the Annals of Internal Medicine and simultaneously in other member journals, to be adopted by all the committee's members (the Annals, the BMJ, the Bulletin of the World Health Organization, Deutsches Ärzteblatt, the Ethiopian Journal of Health Sciences, JAMA, the Journal of Korean Medical Science, the New England Journal of Medicine, the New Zealand Medical Journal, The Lancet, Revista Médica de Chile and Ugeskrift for Laeger, with the World Association of Medical Editors) and likely to spread to the many non-member journals that say they follow the recommendations. Its premise is that the perception of a potential conflict of interest is as relevant as an actual one in judging research: a potential conflict exists when professional judgement about a primary interest, such as the validity of research or its benefit to patients, might be influenced by a secondary interest such as financial gain. Public trust in the scientific process and the credibility of published articles rely on how accurately authors disclose relationships and activities directly or indirectly associated with a work, and transparent, consistent and complete disclosures are critical for editors, peer reviewers, clinicians, patients, educators, policymakers and the public to make decisions and build trust.
Journals began adding conflict-of-interest requirements to their instructions in the 1980s, as the American press noticed the commercialisation of biomedical science and Congress held hearings on federal research funds and conflicts of interest, but the ways journals collected and reported relationships were not uniform, so ten years before the proposal the committee adopted a uniform form for disclosing potential conflicts. Problems remained: opinions vary over which relationships to report, so one author omits a relationship as irrelevant or by oversight while a reader takes any listed relationship as a sign of problematic influence; and the software behind the form had become difficult or impossible for a growing number of authors to use.
The proposal removed language associating relationships with problematic influence or malpractice, retitling the form "The ICMJE Disclosure Form". It stopped asking authors to decide what might be perceived as a conflict: they disclose all their relationships and activities and readers decide what should affect their assessment, through a checklist table answered yes or no with comments, whose items include grants or contracts from any public, private, for-profit or not-for-profit entity, royalties or licences, consulting fees, and payment or honoraria for lectures, presentations, speakers' bureaus, manuscript writing or educational events. And it proposed accepting disclosures from web-based repositories such as Convey, which let an author keep an inventory and generate tailored disclosures instead of filling in a new form for each occasion, provided the repository collects and reports according to the committee's requirements, charges no fee to enter, store or export data, delivers disclosures to journals electronically with an alternative for journals without a digital interface, and complies with the General Data Protection Regulation.
Comments were invited until 30 April 2020 before the form was finalised, and those sent by researchers, pharmaceutical companies and readers of medical journals pointed to room for improvement. The terms "topically related" and "directly related" needed clarifying, plain language was a recurring request, and the wording was seen as a barrier to transparency. Forms of support might need subgroups, because disclosure of writing support is easily overlooked when grouped with the receipt of materials; and using one form for several publications contradicts the form's own instruction that disclosures cover the period from the initial conception and planning of the work to publication, so each should be tailored to the submission. One assessment at the time was that the changes were likely to increase transparency and help authors list everything relevant, while it remained to be seen whether the comments would be taken up.
Gathering disclosures
Prompt participants to disclose their financial interests and intellectual contributions prospectively as well as transparently, using forms such as those journals provide in their author guidelines.
Publication, regulatory, continuing education and grant writers and editors should take as much care with conflict-of-interest statements as with data, to protect both themselves and the people they work for, and the following procedure, aimed at publication writers, can be adapted to most kinds of medical communication. Study leads may not realise how much accurate and complete disclosure matters, and busy, slow-to-respond authors make writers hesitant to chase the information, but the interruptions could save a career: if an author refers the writer to an assistant, gather the disclosures promptly and compare them with the author's recently published disclosures, noting any discrepancy. Read the target journal's instructions for authors for a disclosure form or a required format for the statement; many journals use the ICMJE disclosure form, and the ICMJE keeps a list of participating journals. Where the journal has no form, create one that is painless to complete: the journal's name, its requirements copied directly from the instructions to authors, a table of the required components and the people they apply to (researcher, spouse, minor children), a request for clarification of any discrepancy found against recent publications, a preliminary draft of the statement, a request for deletions and additions, and the researcher's signature and date, which focuses the author's attention. If it is unclear whether a relationship is relevant, ask the editor. Plan ahead, because time is scarcest at the end of any project: the process is best completed within several weeks of the deadline rather than days before submission.
An example form for a fictitious journal and author quotes the journal's requirement to acknowledge all funding sources and all institutional or corporate affiliations of the authors and any domestic partner, including employment, royalties, consulting arrangements, stock or other equity, stock options, patent licensing, payments for conducting or publicising a product or study, and consulting for investment companies, as well as associations with makers of competing products; tabulates those components against author and domestic partner; asks a specific clarification (a 2015 article listed the author as a founder of a company that a 2018 article did not mention, so has she had financial ties to it in the last three years); offers a drafted statement naming grant support, a consultancy and a spouse's employer; and ends by asking whether anything is no longer accurate or should be added, above the signature and date.
Full disclosure is the key: disclose everything and let the reader decide what may be of interest. This holds although many researchers' ties to industry are productive and researchers with conflicts are not fundamentally greedy or unethical and can handle them professionally, while a few honestly believe they can take the money, let industry control the agenda and still conduct unbiased research. In the ICMJE's words, public trust in the scientific process and the credibility of published articles depend in part on how transparently conflicts of interest are handled, and the AMWA Code of Ethics likewise ties ethical behaviour to the integrity of the medical literature.
Wording a disclosure statement
An acknowledgment that only thanks people for preparing and reviewing the manuscript is woefully deficient. A refined model acknowledgment states that the sponsor supported the study and the report financially, provided the study medications and had a role in study design, data acquisition and analysis and the decision to report the findings; that the sponsor reviewed the manuscript for accuracy but the authors had final autonomy over its content; each author's financial interests, such as paid consultancy, honoraria, lecture payments and speakers' bureaus, travel support, board membership, grants, expert-witness work, employment, minor shareholdings and fees for data safety monitoring or endpoint committees, or that the author has none; and who assisted with manuscript preparation, with whose support.
Sponsors that review a manuscript should, and increasingly do, vet it for factual accuracy, while key opinion leaders typically wish to disclose that they were not paid to author a single-sponsored review, and the two often meet in a disclosure statement on the title page or in the Acknowledgments: that the sponsor financially supported named writing and editorial services, and reviewed and approved the final manuscript for factual accuracy and integrity, but the authors had complete autonomy over its intellectual content.
Failures to disclose
Providing incorrect information to journals is an act of scientific misconduct that frequently goes unnoticed, usually an omission to disclose conflicts of interest or a failure to declare financial support or collaborative help, typically from the pharmaceutical or biotechnology industry, and financial pressures on universities have encouraged it. Disclose all pertinent information in a manuscript intended for publication, and in a thesis or dissertation.
Disclosure has been measured among doctors and scientists linked to off-label marketing. Off-label marketing is the promotion of a drug for an unapproved indication, age group, dosage or route, which companies are prohibited from doing directly but have paid physicians and researchers to endorse. A study reported in 2012 collected whistleblower complaints alleging it from the US Department of Justice, media reports and other public sources, then examined the publications of the individuals named over the following three years. Twenty-six complaints yielded 91 doctors and scientists, of whom 39 (43%) had authored 404 publications related to the drugs they were linked to in the complaints, and those 39 were alleged to have had 42 relationships with the company: paid speaker, writing articles, consultant or advisory board member, recipient of gifts or honoraria, recipient of support funds.
Only 15% of those 404 publications had adequate disclosures. Of the publications, 43% had no disclosure statement at all, 4% carried statements denying any conflict of interest, 40% disclosed something but not the pharmaceutical company in question, and 13% named the company without expressing the nature of the relationship. Adequacy also varied by article type, commentaries being less likely to carry adequate disclosures than reports of original research. The study's authors argued that the results show authors' inadequacy in preparing conflict-of-interest statements and suggested that journal practices need to improve.
By the end of 2018 the debate over financial and non-financial conflicts of interest was running in the mainstream media as well as in journals. In September 2018 the New York Times, with ProPublica, reported that Jose Baselga, chief medical officer of Memorial Sloan Kettering Cancer Center in New York and one of the world's top breast cancer doctors, had failed to disclose millions of dollars in payments from drug and health care companies, omitting his financial ties from research articles in journals including the New England Journal of Medicine and The Lancet. The Open Payments federal database recorded nearly $3.5 million paid to him by nine companies between 2013 and 2017, while more than 100 of his publications in that period disclosed no industry relationship. From 2014 to 2017 he had not provided complete disclosure in more than 60% of his more than 100 publications, with the omissions growing each year to 87% in 2017; he called it an unintentional lapse.
His ties were extensive. He served companies on their boards of directors, on advisory boards or as a paid consultant, owned or part-owned several start-ups, and was president of the American Association for Cancer Research and editor-in-chief of its journal Cancer Discovery; as chief medical officer he had input into the drugs and equipment the cancer centre bought, though steps had reportedly been taken to keep his relationships from influencing its business decisions. His interpretation of results at speaking engagements also appeared biased: when a Roche compound improved disease-free survival by 1% rather than the anticipated 20%, he told investors the results were very meaningful, the data far from mature, and that he was confident the benefit would get better.
The newspaper also accused journal editors of laxity for not checking conflict-of-interest statements, and journals responded that they do not routinely fact-check authors' disclosures. Under pressure he resigned as chief medical officer, which the two outlets reported on 13 September, and as editor-in-chief of Cancer Discovery, and he sent corrections to the journals declaring his conflicts. In January 2019 he was hired to head oncology research and development at AstraZeneca. In December 2018 the same newspaper revealed further cases in the cancer field, among them the president-elect of the American Society of Clinical Oncology, who had declared no conflicts of interest in more than fifty journal articles while drug companies had paid his employer nearly $114,000 for consulting and speaking and nearly $8 million for his research during the period for which disclosure was required.
These articles forced journal editors to react. A JAMA editorial signed by the editor-in-chief concluded that conflicts of interest are likely to become more challenging as more investigators and institutions enter financial relationships they benefit from, that it is critical for authors to report the information accurately, completely and transparently so that readers can evaluate whether the article could be biased, and that editors bear an equal or greater responsibility for ensuring published information is accurate and objective and for maintaining the integrity of the scientific record, with the final judgement resting on readers assessing the information and deciding its applicability. A viewpoint in the same issue proposed redefining research misconduct, on the argument that if leaders do not follow the rules then there are no rules, and that it is time to strengthen institutional conflict-of-interest policies by treating the intentional or negligent failure to disclose significant financial relationships relevant to the conduct of research as research misconduct. In December 2018 the ICMJE issued updated recommendations that added to its definition of scientific misconduct, alongside data fabrication, data falsification including deceptive manipulation of images, and plagiarism, the purposeful failure to disclose conflicts of interest.
What a declaration changes
Whether industry sponsorship biases papers has been argued at length, but until 2012 no one had looked at whether conflicts of interest influence how peer reviewers view and review manuscripts. A 29-question web survey was sent to 410 active reviewers for Annals of Emergency Medicine, one of the journals requiring authors to declare conflicts of interest, and 218 gave complete responses. Most felt that authors were influenced by their financial ties to industry. That belief did not clearly translate into different behaviour: a majority said they would read a paper with declared conflicts more carefully and felt its credibility was reduced, but considerably fewer would change the recommendation they made to the editor. Three-quarters doubted whether authors of industry-sponsored articles have full access to the data. A small majority, 54%, held that an honorarium of any size biases an author's judgement, which does little to support the survey authors' own proposal that authors should divulge the sizes of the payments they receive.
A smaller proportion of the reviewers who had themselves received such payments considered that payments cause bias, and why is open to speculation: the experience of consulting may fail to match the suspicions of those who have never consulted, or academics who do not believe honoraria bias may accept them more readily. Of the survey authors' three proposals, two were already in force: authors confirming full access to the study data is covered by the ICMJE recommendations, and peer reviewers disclosing industry payments had long been a stipulation of the very journal they surveyed. In the same period the editor-in-chief of the Journal of Paediatrics and Child Health published an editorial warning of the dangers of financial conflicts of interest and refuting the notion that declaring them does anything to prevent bias.
By 2012 the processes for disclosing authors' potential conflicts of interest had improved: one set of recommendations for improving the credibility of industry-sponsored research asked journal editors and sponsors to support the updated ICMJE conflict-of-interest reporting form, and raised a centralised, publicly accessible disclosure database as a future possibility. The PLoS Medicine editors nonetheless asked whether disclosure worsens bias. They held that disclosure is preferable to non-disclosure but may be a limited strategy for mitigating bias: disclosure converts a secret bias into an open one, a reader can be blinded by the amount of information provided, and declaring a conflict can be taken to mean that the declarer no longer needs to manage it. Loewenstein and colleagues demonstrated that patients think doctors would not intentionally mislead them and tend not to discount advice even when they know a conflict of interest exists.