This Article Extract Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (8) Citing Articles via Google Scholar Google Scholar Articles by Davidoff, F. Search for Related Content PubMed PubMed Citation Articles by Davidoff, F. Related Collections Ethical Issues Personal Experience ("Narrative Matters") Pharmaceuticals

Narrative Matters

Between The Lines: Navigating The Uncharted Territory Of Industry-Sponsored Research

And ignore them we could not. In 1995 the report of a large, industry-supported drug trial submitted to Annals of Internal Medicine forced us to rethink our approach to handling industry-project relationships in the clinical research we published. In practice, Annals had always taken conflicts of interest seriously, but an author reading the journal’s information for authors in 1995 found only a brief section on the topic that stated, "Financial support for the work, including equipment and drugs, should be listed on the title page. Authors should describe in the cover letter any financial interests, direct or indirect, that might affect the conduct or reporting of the work they have submitted." Spurred by our 1995 experience, and after much deliberation, we changed our policy. Since 1997 authors conscientious enough to read our "Information for Authors" have found fairly extensive information on conflict of interest in a new section called "Dual Commitment," calling for additional disclosure from authors. How did we come to reframe the issues in this way and demand this additional detail?

Catalyst For Change

 
The title page of the study we received in 1995 made the fact of industry support quite clear; the trial was well conducted, and the drug it tested, potentially important. The problem lay buried in the text. It seemed to me, my fellow editors, and our statistician that the authors had gone well beyond the data in stating the drug’s efficacy and safety. I suggested alternative wording that we felt was more appropriate, but the authors’ revision still contained the original wording. I tried again, but the second revision also came back unchanged. I then called the lead author to find out what was going on.

He made no bones about the fact that the drug company sponsoring the research had reserved the right to review the manuscript before it was submitted—something that had not been disclosed to us. When I pushed him about how much control the company had over the paper’s wording, things got a lot murkier. The principal researchers had nominally retained control, he said, but the company’s opinion did very likely influence the report’s language.

That did it. The problem for me was not that the trial’s sponsor had an interest in how the study was conducted and reported—that was natural enough, given its sizable financial investment. The problem was that details of the sponsor’s involvement in and control over research done by "independent" investigators weren’t being made known to editors, reviewers, and readers.

The primary motivation behind the pharmaceutical industry’s support of clinical research is, understandably, to meet regulatory requirements; the discovery of scientific fact is only a part of its agenda. But, importantly, research pointing to a drug’s efficacy and safety is also used for marketing purposes, and the driving force behind marketing is persuasion rather than objectivity. For both of these reasons, commercial influence is inevitably one potential source of bias in clinical research.

Here was an irony, indeed: Over the years we editors had become hawks about transparency in the reporting of study methodology—the technical side of things. Yet we had been willing to leave readers in the dark about key nontechnical factors (including the role of outside sponsors) that lie behind the way a project is carried out. These same factors can affect a study’s conduct and reporting as much as if not more than its purely technical aspects can.

The back-and-forth with the author of this paper led us to one conclusion: We had an obligation to improve our disclosure of nontechnical factors—particularly conflicts of interest—making it at least as detailed and specific as the disclosure we required on the technical side. But, we wondered, what information should we be asking for, and how should we ask for it?

Parsing Relationships

Top Catalyst For Change Parsing Relationships Easier Said Than Done Forming Editorial Consensus Truth Finding

 
From an editor’s point of view, the elements at play in conducting and reporting original clinical research are analogous to those of continuing medical education (CME). Before my editorship at Annals, I had spent eight years running CME programs and so was influenced considerably by what I had learned from that experience. CME programs, usually taught by faculty from academic medical centers, focus on the latest advances in clinical disease management, usually including drug therapy. These programs are often underwritten by drug companies through educational grants to the program’s organizers and inducements to the participants that include everything from free prescription pads to travel to resort locales. Like CME, clinical research involves three main entities: the research project (analogous to an educational program in the case of CME), the researchers (analogous to CME faculty), and, through the journal, the readers of the research reports (analogous to CME attendees). Industry sponsorship of clinical research thus creates a complex of relationships: with the reported study as a whole, with individual researchers, and with readers.

The pharmaceutical industry’s interest in clinical research studies of particular drugs was heightened by passage of the Kefauver-Harris Amendments in 1962. In addition to the prior requirement to demonstrate safety, these amendments now mandated the Food and Drug Administration (FDA) to require that drug companies also demonstrate the efficacy of a drug before it can be marketed. But demonstrating both efficacy and safety are demanding and expensive tasks, even more so in recent years. Since the very survival of a new drug depends on the results of clinical trials, it’s hardly surprising that control over the conduct and reporting of such trials can become a source of tension between independent researchers and sponsors. The search for objective scientific fact is a complex process, considerably more complex than CME. Thus, while the issue of control in CME involves simply who chooses what’s presented, the issue of control in original clinical research is correspondingly more complex, ranging from who defines the research hypothesis to who designs the study, collects and analyzes the data, draws the inferences, and makes the decisions about publication. Industry control of clinical research works itself out in the details of the interaction between the sponsor and the researchers, some of whom, to complicate things further, may be salaried employees of the sponsor.

Asking authors to disclose details of the financial sponsor’s role in the study’s conduct, just as we ask for the technical details of research methodology, was the obvious way to minimize the double reporting standard for technical versus nontechnical factors. And since the two factors share the potential for introducing bias, it made sense to ask authors to put the information about nontechnical factors into the methods section, where they traditionally put the technical details, rather than relegate it to a footnote in fine print. To reflect the change, in 1997 we created the following language for our "Information for Authors": "The role(s) of the funding organization, if any, in the collection of data, its analysis and interpretation, and in the right to approve or disapprove publication of the finished manuscript must be described in the Methods section of the text."

And what to do about industry and individual researcher links? Clinical researchers often establish relationships with commercial companies that vary all over the map in intensity and timing. At one extreme, for example, a researcher who submits the report of an industry-supported study might long ago have given a few lectures sponsored by the company. At the other extreme, a researcher might hold a sizable amount of stock in a company whose product is being tested in the research study in question. Should authors provide editors with all such information, and should editors publish it?

In developing Annals’ new policies on conflict of interest, we started from the premises that as with CME, all relationships between industry and individual researchers involve a commitment of the involved parties to more than one goal—dual commitment—but dual commitment doesn’t necessarily create true conflict of interest (that is, potentially introduce bias) into a study. The result of our thinking was the 1997 revision of our "Information for Authors," with its general heading of "Dual Commitment." Within that section we created a new "For the Individual" subsection, in which we defined dual commitment much as we had earlier defined conflict of interest: "any financial interests, direct or indirect (dual commitment), that might affect the conduct or reporting of the work...submitted." But we now distinguished dual commitment from conflict of interest by advising authors that if "in the Editors’ judgment, a dual commitment represents a potential conflict of interest, information concerning the relationship may be published."

Adopting this policy raised the question of how editors can make such judgments. Few editors, reviewers, and researchers have a hard time recognizing egregious conflicts of interest. The judgment problem comes in the gray zone. No simple solution exists, but judgments are more consistent and credible if based on explicit, previously defined criteria. The quantitative thresholds established by various organizations and agencies can be enormously helpful here. For example, although the U.S. Public Health Service defines significant financial interest as "anything of monetary value," the agency explicitly excludes from its disclosure requirements such things as salary and royalties from the researcher’s own institution and equity interests that don’t exceed $10,000 or 5 percent ownership of a single entity. When criteria like these didn’t settle the issue, we fell back on the advice we were giving to authors: Err on the side of full disclosure. (More recently, Annals has begun publishing all researcher-industry relationships that authors reveal.)

Easier Said Than Done

Top Catalyst For Change Parsing Relationships Easier Said Than Done Forming Editorial Consensus Truth Finding

 
Some time after we published the 1995 paper that lacked information on the role of the industry sponsor in the research writeup, we invited an editorial commenting on what we felt was an important new diagnostic approach. We chose an editorialist who came recommended as being highly knowledgeable on the topic. The author accepted our standard telephone invitation; we sent our usual confirmatory letter with explicit instructions to disclose any potential conflicts of interest. The editorial recommended the use of this new diagnostic maneuver in what we felt was reasonable fashion; nothing was said to us about financial interests. All seemed to be in order.

A few days later I got a call from the associate editor who was handling the manuscript and who happened to know the editorialist professionally. "I guess the editorial is okay," he said, "but did you know that the person who wrote it has a substantial financial interest in a company that does this kind of testing?" A phone call confirmed the fact; the editorialist had "simply neglected" to include that information when submitting the manuscript. We debated whether to publish the editorial but eventually did so, with a note disclosing the author’s financial interest.

The lesson is that when it comes to extracting information from authors, a whole family of devils lies in the details. First is the problem of silence. Annals had for years collected information about individual researchers’ potential conflicts by simply asking them to describe any financial interests that "might affect the conduct or reporting of the research." We interpreted silence to mean that no such interests existed. The problem is that a researcher can forget to mention a major financial interest, or feel it doesn’t matter, or decide that it’s none of the editor’s business. Silence, however, is ambiguous; when it comes to disclosure, it is not acceptable. This is why we decided to start asking all authors, editorialists, and reviewers to indicate whether they do or do not have a conflict of interest. Now if an author says explicitly that he or she has no conflicts but conflicts later come to light, it is unequivocally the author who is answerable for the missing information.

A second devil is the long-standing practice at many journals of having authors put dual commitment information in the cover letter that accompanies the manuscript. The problem is that generally the cover letter isn’t sent to peer reviewers, so they are unable to factor the dual commitment element into their evaluation. Including that information in the manuscript itself solves that problem.

A third devil: Annals authors have generally been conscientious about listing the sources of financial support for the research study on the title page of their initial manuscript. But our request to describe in the methods section the sponsor’s exact role in the study, added to the "Information for Authors" in 1997, was largely ignored. Getting that missing information required us to ask for it explicitly in our provisional acceptance letter.

Forming Editorial Consensus

Top Catalyst For Change Parsing Relationships Easier Said Than Done Forming Editorial Consensus Truth Finding

 
The struggles of editors over these details at many journals, including Annals, have contributed to important changes in the "Uniform Requirements for Manuscripts Submitted to Biomedical Journals" (URMs) and "Separate Statements" on editorial policy. The URMs are a set of model editorial standards first created some twenty years ago by the International Committee of Medical Journal Editors, sometimes referred to as the Vancouver Group—a self-appointed group of editors of about a dozen medical journals, Annals among them. Although these guidelines have no official standing, many medical journals throughout the world use them as the basis for their own editorial policies.

Since 1997 the URMs’ recommendations on disclosure have been dramatically tightened along the lines of those adopted at Annals. Several years of vetting these issues culminated in an unprecedented editorial, signed by most of the members of the Vancouver Group, that appeared in the 18 September 2001 issue of Annals, as well as in eleven other leading medical journals worldwide. In it the editors go so far as to state: "We will not review or publish articles based on studies that are conducted under conditions that allow the sponsor to have sole control of the data or to withhold publication."

Truth Finding

Top Catalyst For Change Parsing Relationships Easier Said Than Done Forming Editorial Consensus Truth Finding

 
Most editors and researchers continue to believe that, in principle, the statistician John Bailar is right: "Disclosure is almost a panacea." Disclosure does seem to be an exceedingly powerful tool, but for what? Disclosure is a process variable, a surrogate for what we’re really interested in: the successful search for scientific truth. In the case of clinical journals, this translates into well-designed, well-conducted clinical research, appropriately analyzed and objectively reported. To what extent does disclosure actually make a difference in achieving these outcomes?

Having such information, as Annals editors, reviewers, and statisticians now do, helps in judging both internal and external validity of clinical studies, in much the same way as knowing sampling strategy, sample size, and study population characteristics. Moreover, in principle, disclosing the role of financial supporters in the methods section of clinical journals benefits readers. It is, after all, the interpretation of published articles that converts study reports into clinical practice. And since it is often true that "what the Results section giveth, the Methods section taketh away," the information in the methods section can be crucial to fine-tuning a reader’s interpretation. That is, if readers actually read that section.

At the same time, it’s reasonable to question whether any single remedy, including disclosure, can be a perfect solution to a problem as subtle, complex, and pervasive as conflict of interest. Disclosure has its limits. It’s possible, for instance, that nobody will care that journals are publishing more extensive information about potential conflicts of interest. However, revelations about hidden conflicts of interest are generally seen as shameful, and shame, with its associated loss of trust and credibility, is a powerful emotional and social force. That said, individual thresholds for what’s shameful vary. Moreover, relationships between academia and industry are rapidly changing. Actions and interests once broadly viewed as compromising may over time come to be generally acceptable.

Further, neither disclosure of major conflicts nor failure to disclose them, for all of their social and emotional impact, has real teeth, no hard penalties—legal, financial, or otherwise. But despite not having a police function, science has protected its integrity reasonably well. There is good reason to believe that disclosure is an effective measure of social control in clinical research, just as the expectation of honesty is generally effective in regulating commerce.

Disclosure can be self-defeating if it requires unacceptably large and expensive bureaucracies and generates irrelevant, trivial information to the point where the "signal-to-noise" ratio becomes unacceptably small. It’s up to editors, reviewers, authors, readers, and the organizations that underwrite research collectively to keep that from happening.

Finally, we can gain some insight into the enduring importance of disclosure by recognizing that the secrecy surrounding morally doubtful acts has for a very long time been seen as a villain, almost as much as the acts themselves. Thus, the Oxford English Dictionary tells us that the word murder originally "denoted secret murder, which in Germanic antiquity was alone regarded (in the modern sense) as a crime, open homicide being regarded as a private wrong calling for blood-revenge or compensation." From that perspective, the act of disclosing "who did what" can be a deeply meaningful part of maintaining a civil society.

Editor's Notes

 
Following many years as a diabetologist and general internist in academic institutions, Frank Davidoff became the senior vice-president for education at the American College of Physicians in 1987, a position he held for eight years. From 1995 to 2001 he was editor of Annals of Internal Medicine. For a more detailed essay, see the Health Affairs Web site, <www.healthaffairs.org/WebExclusives/Davidoff_Web_Excl.htm>.

                      What's this?

This article has been cited by other articles:

				D. L. Nordstrom The Professionalization Of Editing Health Aff., May 1, 2008; 27(3): 890 - 891. [Full Text] [PDF]

				G DuVal Institutional ethics review of clinical study agreements J. Med. Ethics, February 1, 2004; 30(1): 30 - 34. [Abstract] [Full Text] [PDF]

				M. J. Tobin Conflicts of Interest and AJRCCM: Restating Policy and a New Form to Upload Am. J. Respir. Crit. Care Med., May 1, 2003; 167(9): 1161 - 1164. [Full Text] [PDF]