Just how "hot" is the topic of conflict of interest in psychiatry? The answer was brought home to me dramatically this past May at the APA meeting in Toronto. During the meeting, I had the opportunity to chair a symposium titled "Pharmaceutical Industry Influence in Psychiatry." My copresenters and I showed up well ahead of time to meet and prepare introductions. As we gazed out at the empty seats, we joked that there would be at least 5 people in attendance since, after all, there were 5 presenters.
How wrong we were. As the time to begin approached, a slow trickle rapidly became a rush of attendees. Every available seat filled quickly, and a large crowd stood in the rear and hovered in the hallway outside the room.
In retrospect, I shouldn’t have been surprised. Over the past several months, psychiatry has been subject to several embarrassing exposés related to conflict of interest. These events provided painful lessons on what constitutes financial conflict of interest and what should be disclosed.
In this article, I will begin by discuss ing the notion of conflict of interest in medicine and then zero in on 2 examples of conflict of interest in psychiatry. These examples provide good jumping-off points for delving into major controversial issues concerning how much authors and researchers should disclose regarding their financial relationships with commercial enterprises.
Defining "conflict of interest"
The New England Journal of Medicine published the classic article defining conflict of interest in medicine in 1993. A clinical ethicist at Harvard University, Dennis Thompson, defined conflict of interest as "a set of conditions in which professional judgment concerning a primary interest (such as a patient's welfare or the validity of research) tends to be unduly influenced by a secondary interest (such as financial gain).1
In medicine, a typical scenario involves the physician who receives money from a pharmaceutical company for research, consulting, or speaking. The conflict is between 2 different interests or desires. One desire is to produce scientifically valid research or, in the case of a speaker, to communicate accurate and relevant medical educational material. The potentially conflicting interest is the desire to make money from the pharmaceutical company.
Obviously, there are times when these 2 interests coincide. Often, for example, speakers will choose to speak strictly on behalf of medications that they sincerely believe are superior. Even if this is true, however, a potential conflict of interest still exists because the audience will never know whether the promise of a check at the end of the dinner program has altered the speaker’s opinion of the featured product.
We usually focus on financial conflicts of interest (hereafter referred to as COI) because they are easy to identify and quantify. Other common COI in medicine include the desire for academic advancement and for media coverage. In all cases, COI can bias the content of research articles, review articles, and educational lectures. In recent years, several studies have been con ducted, both in psychiatry and in medicine at large, that documented the ways in which financial COI appear to influence the outcome of medical research.
Recently, for example, Perlis and colleagues2 examined all clinical trials that were published in 4 of the major psychiatric journals from 2001 to 2003. Of 162 randomized placebo-controlled studies, those that reported COI were 4.9 times more likely to report positive results for the sponsored product than other studies. Other reviews have documented similar relationships between sponsorship and positive research outcome in several other medical specialties.3,4
To be fair, however, the fact that industry sponsorship leads to results favorable to the sponsor does not necessarily mean that the research has been biased. It may be, as suggested recently in a Newsweek interview with Dr Ezekiel Emanuel, head of the department of clinical bioethics at the NIH Clinical Center, that drug companies invest their resources preferentially in clinical trials that are most likely to be positive.5 In this view, the correlation between sponsorship and outcome has less to do with bias and more to do with choosing promising drugs to study, as well as having the resources to conduct the large clinical trials that are needed to yield significant results. Indeed, in the Perlis study,2 industry-sponsored studies generally enrolled higher numbers of subjects than nonsponsored studies, implying greater statistical power to measure any differences between treatments.
On the other hand, a recent review6 of industry-sponsored studies of anti-psychotics detailed specific techniques used by companies to manipulate re search design in order to ensure positive results. The techniques included underdosing competitors’ medications, devising study inclusion criteria so that the subjects enrolled were more likely to respond to the sponsor’s product, and selectively reporting data so as to minimize the side effects.
Thus while still controversial, potential sponsorship bias is seen as an important issue in academic medicine. Over the past 2 decades, medical journals have responded to this problem by requiring authors to disclose their financial COI when they submit manuscripts. However, there continues to be much variation and debate about how much information authors should disclose and when they should disclose it.