Managing Conflicts of Interest: General Models and Approaches
The management of conflict of interest might be approached from two different models. One is based on a presumption against any relationships that might present a conflict. This we call a ''prohibition model," although such a prohibition might be overcome with a demonstration of sufficient social benefit. The other is based on a presumption for such relationships with a provision for disclosure and review. We call this model "disclosure and peer review."
TWO MODELS OF CONFLICT OF INTEREST MANAGEMENT
The prohibition model discourages any arrangement, particularly financial, that might create a conflict of interest unless that arrangement provides a sufficiently countervailing social benefit. Implementing this approach requires that those who advocate it establish a framework within which certain conflicts of interest may be acceptable. Within this framework the first consideration is whether the activity presenting the conflict of interest has any redeeming social value. If it does not, a process is required for prohibiting the activity and hence the conflict of interest. Second, if there is redeeming social value, does it clearly outweigh the risk of biased or incomplete conduct or reporting of research? If not, a process is needed to examine the arrangement or activity further and probably to prohibit it. If the putative benefits do sufficiently balance the conflict-of-interest risks, a process is needed to minimize the risks through disclosure and management.
At least four examples of social benefits have been seen as outweighing the risk of bias: improved transfer of medical innovations to
the bedside, creation of jobs, furtherance of economic development generally, and facilitation of private support of research programs and public universities. Secondary benefits include efficiencies in the conduct of specific research studies.
Many investigators have unique expertise. If senior researchers are prohibited or severely restricted by federal or university regulations from such activities as serving on an advisory board, consulting with commercial entities, or holding stock in a firm related to their research, they might decline involvement in federally supported research. Conversely, they might decline to provide that expertise to industry where it might be in the public interest, another loss of social benefit. The prohibition model may therefore promote loss of social benefit, another factor that must be weighed in deciding on approaches to managing conflicts of interest.
Disclosure and Peer Review Model
The disclosure and peer review model holds that conflicts of interest are unavoidable and that financial conflicts of interest are only the most visible and perhaps the least scientifically dangerous. Acknowledging potential sources of bias promotes an awareness of different points of view and the possibility of developing some kind of balance within PORTs—provided also that a strong peer-review process is in place and that there are opportunities for secondary data analysis. Because PORTS are part of a political process created by Congress with a number of expectations that may engender conflicts of interest, some observers believe that disclosing and balancing biases to the maximum extent possible is a more realistic goal than trying to eliminate them. According to this view, blanket prohibitions are not needed. Strong peer review within the PORTs and by journal reviewers and public evaluation by other researchers would counteract or detect unacceptable biases. Knowing that their most valuable asset is credibility in the field, this perspective argues that it would be irrational for individual PORT researchers to jeopardize this credibility for a given study.
Many biases, however, may be exceedingly subtle and hard to detect (as discussed in Chapter 4). This factor may preclude dependence on disclosure and peer review alone to control the bias resulting from certain (especially nonfinancial) conflicts of interest.1 Such sources of unintended bias include determining what materials are to be included in meta-
analysis; "cleaning" and transforming of insurance claims data tapes; choosing and constructing outcome measures, data analyses, and presentation of data; and developing clinical practice guidelines.
GENERAL APPROACHES TO DEALING WITH CONFLICTS OF INTEREST
Assuming that under both models attention is given to the climate and support for ethical standards in the PORT environment, what implications might we draw about managing conflicts of interest under these two models? Seemingly, both the prohibition and the disclosure and peer review models would permit PORT studies to continue despite real or apparent conflicts of interest. Where they significantly differ is in their underlying presumptions and in where they draw the line between prohibition and management, though not necessarily in the means used to deal with conflicts of interest. These means, which are discussed in the remainder of this chapter, include mandatory disclosure, financial distancing, self-regulation, defining categories of acceptable activities and implementation of oversight mechanisms, defining unacceptable activities and implementing prohibitions and where necessary, sanctions. The two models would rely on different mixes of these mechanisms in attempting to forestall, control, or manage conflicts of interest.
Disclosure of relevant interests and activities (whether on a mandatory, periodic basis or as studies are initiated by the investigator) was once considered too intrusive or simply "impolite," but it is now virtually universally endorsed as a key means of coping with conflicts of interest. As one workshop participant said,
We have to change the nature of the discourse to make it clear that it is not only polite, but essential, to understand people's financial interest in areas that affect their work, just as it is essential to know where they came from and who they did their work with because of other subtle biases. You have to get that information so that others can judge [it] in the context of their work (C. K. Gunsalus).
Some state laws and some university policies require disclosure of certain categories of activities. Typically these include service as an officer or director of any commercial entity, investment of more than a given amount in any one company whose product is related to the individual's work, ownership interest of more than a given percentage in a partnership or corporation, and consulting arrangements that result in remuneration greater than a set amount (AAHC, 1990). Disclosure might include all financial interests of the investigator and his or her immediate family. The National Research Council requests disclosure of any prior public statements, including publications, relevant to a topic under study (NRC, 1989).
Regarding financial conflicts of interest, one possible method of reducing the influence of corporate money on research is to establish financial pools or mechanisms that increase the distance (real or perceived) between the funding source and the PORT or its members. There is considerable disagreement over this approach, as evidenced by this exchange between two participants at the IOM workshop:
MR. HUTT: . . . You said, "What if it [the funding] comes from the foundation versus the company?" I would hope people would realize there is no difference between those. What has troubled me . . . is the idea that laundering it through the university as opposed to giving it directly to an investigator somehow makes a difference . . . . Everybody knows where the money is coming from and where it is going . . . . I have always been offended by the thought that this laundering process somehow magically converts tainted money into clean money. We ought to get rid of that fiction and understand that the money is going to go . . . from Mega Pharmaceuticals [a fictional corporation used in the scenario] for the purpose of first-rate academic research. It is then the responsibility of the university to supervise the doctor and to make sure there is adequate peer review within the university and that something untoward does not occur . . . . If you cannot stand before the public and defend an academic researcher receiving money from whatever the source may be, whether it is a pharmaceutical or any other company, and you have to pretend it is being laundered . . . then you shouldn't be doing it . . . The question of where
the money came from is not the issue. The question is whether the science was valid, good, strong, academic, unbiased, straightforward, peer-reviewed science.
PROF. CAPRON: Some people would suggest that there is no such absolute animal. . . . As a dean, Dr. Korn, do you agree with Mr. Hutt's statement?
DR. KORN: . . . I don't understand laundering because it is not a concept that I use. When I was talking before about the Mega Foundation's money, I was thinking of it as grants that were coming into very specific people for very specific research projects, but coming in as grants through the university. That is not a laundering phenomenon. It is simply a matter of tracking the research portfolio that is going on within a place at any one time and assuring that whatever institutional assurances have to be met by federal and other regulations are in conformity. That is all. There is no laundering involved. There is no deception involved.
PROF. CAPRON: Would you feel differently if it were a no-strings-attached annual grant?
DR. KORN: You mean a gift?
PROF. CAPRON: Yes.
DR. KORN: Sure. We get gifts. Everybody gets gifts. That is the difference between a gift and a grant. A gift is a general award of funds for some very general use, and a grant is much more specific and targeted, usually with a named investigator and expected outcomes. They are different and equally acceptable.
PROF. CAPRON: What happens to the adage, "Don't bite the hand that feeds you?"
DR. KORN: It is a good one [laughter].
Such "financial distancing" by foundations may be quite variable. The Lilly Foundation, for instance, will support no research in the health area; other corporations, however, use their foundations to promote their own views.
Another form of financial distancing is the blind trust, in which control of equity is transferred to a fiduciary for the course of a researcher's involvement in a study. Such an arrangement might lessen, at the margin, the likelihood of insider trading. It would only affect personal financial gain, however, not prevent biased research, and it is the latter that reflects the intent of financial distancing. For instance, a blind trust would not provide a solution for the investigator who has a substantial holding in a closely held company whose product is involved in PORT studies (or is competitive with such a product) because the investigator is not really "blind" to this holding in the hands of the fiduciary.
One way to deal with problematic arrangements is to establish rules internal to the research group. For instance, Healy and her colleagues (1989) described decisions by the key investigators in a new multicenter clinical trial of treatment after coronary-artery bypass graft surgery. Among their decisions were not to buy, sell, or hold stock or stock options in the companies manufacturing or distributing the medications they were testing and not to serve as paid consultants to these companies throughout the study (Healy et al., 1989). This arrangement has been hailed as sound protection against this form of financial conflict of interest (Relman, 1989). Another example is the decision by members of the Dartmouth assessment team not to accept honoraria or consulting fees and not to own stock whose value is affected by urologic treatment, the area under study by their team (Wennberg, 1990b).
Defining Categories of Acceptable Activities and Implementing Oversight
Several ways to define acceptable activities and the oversight procedures to permit those (and only those activities) might be outlined in theory. In practice, the approach of the Harvard University Faculty of Medicine offers some useful guidance.
The Harvard University Faculty of Medicine's (1990) new rules distinguish three relationships: (1) those requiring special attention and specific approval, (2) those permitted with oversight, and (3) those that are routinely allowable. Further, they require five actions by all faculty members. Specifically, faculty must (1) make a full annual disclosure of their potential conflicts of interest to university administration; (2) seek explicit approval before embarking on studies funded by companies in
which they or their families have a financial interest; (3) receive approval before sitting on a review committee that judges a technology in which they or their families have an interest; (4) receive approval to serve as a managing executive for a profit-making biomedical company; and (5) disclose to the public their financial interest in any subject that they discuss in a research publication, a formal presentation, or an expert commentary, and do so "simultaneously," that is, as they speak or publish.
As recommended by the guidelines, the dean of the medical school has appointed a standing committee of the Harvard Medical Center, the Committee on Conflict of Interest and Commitment, to review activities that are disclosed and implement procedures for approval and oversight.
Defining Categories of Unacceptable Activities and Implementing Prohibitions
Although differing in their presumptions, both the prohibition and the disclosure and review models recognize that when ameliorative approaches are insufficient to ensure adequate conflict-of-interest protection for researchers, prohibition may be required. Consequently, several schemes have been proposed to delineate permissible from impermissible activities. These range from simply advocating the use of common sense, developing only those prohibitions that are tied to a specific danger to the public, or creating entire categories of unacceptable behavior, such as those included in the (now withdrawn) September 1989 NIH Guide for Contracts and Grants (DHHS, 1989b). Other examples include the New England Journal of Medicine's policy not to accept reviews or editorials by authors with financial connections to the product being reviewed (Relman, 1990) and the Johns Hopkins University policy that permits researchers to hold paid consultancies but not equity interest in companies that support their research (Johns Hopkins University School of Medicine, 1990).
The American Medical Association's Councils on Scientific Affairs and on Ethical and Judicial Affairs (1990) have published a report entitled Conflicts of Interest in Medical Center/Industry Research Relationships. In this report the two councils strongly endorse full disclosure to medical centers, funding organizations, and journals. The report further asserts that researchers "cannot ethically buy or sell a company's stock until the involvement ends and the results of the research are published or otherwise disseminated to the public" (p. 2793). It affirms, however, that a researcher may ethically share in economic rewards that are commensurate with the value of his or her actual efforts (such as royalties), and
it recommends that medical centers develop policies to provide clear guidelines to clinical researchers.
Research sponsors can withdraw funding or apply penalties when their rules regarding conflict of interest are broken. Institutions also sometimes apply sanctions. For example, the Harvard University Faculty of Medicine's approach to conflicts of interest specifies the following sanctions in rough order of severity: formal admonition; inclusion in a personnel file of a letter from the Office of the Dean that an individual's good standing as a member of the faculty has been called into question; ineligibility for grant applications, institutional review board approval, or supervision of graduate students; nonrenewal of appointment; and dismissal from the faculty.
Two models provide the main approaches for managing conflict of interest. A prohibition model is based on a presumption against relationships that might present a conflict; a "disclosure and peer review" model rests on presumption that such relationships are unavoidable but manageable. Although the models differ in their underlying presumptions, in where the line is drawn between prohibition and management, and in the means used to deal with conflicts of interest, both models are likely to use one or more of the following mechanisms for dealing with conflict: disclosure, financial distancing, self-regulation, defining categories of acceptable activities, implementing oversight of those activities, defining categories of unacceptable activities, and implementing prohibitions and sanctions. Chapter 6 considers these procedures as applied to PORTs.