Thursday, November 12, 2009

COI: Define It Narrowly Enough and It Goes Away

Thanks to a couple of friends for alerting me to Dr. Michael A. Weber's editorial in the Journal of Clinical Hypertension, entitled, "Academic Physicians Confront a Hostile World: The Creation of ACRE." You can get a sense from the "poor us" tone of the title what this defense of the Association of Clinical Researchers and Educators is going to be like. I could go on for a while about the "hostile world" that expects that academic physicians might actually manage to live on their salaries, that are generally several times greater than those of any other academics, without also stuffing their pockets with industry largesse. But of greater interest is the discussion of the definition of conflict of interest (COI).

In a section labeled "Conflict of Interest: A Disturbing Misnomer," Dr. Weber quotes with approval a speaker at the inaugural ACRE meeting, Prof. Lance Stell of Davidson College, whom he described as a "nationally recognized medical ethicist." I consider Lance a good friend of many years standing and a fully legitimate philosopher-ethicist, but the fulsome description leaves out that on this particular issue, Lance's stance is probably quite atypical of bioethicists (though no one has done a survey so I could never prove that).

The reason Dr. Weber likes Lance Stell's definition so much, and no doubt why Lance was the only bioethicist I am aware of that was invited to speak at the ACRE conference, is that he is said to define COI as what "occurs when practitioners accept personal return for actions that could violate their professional obligations. In essence, to accuse a physician of conflict of interest would require empirical proof that, in return for a reward, an action was taken that resulted in diminished care or even harm to patients." Given that starting point, Dr. Weber then says very reasonably, "Clearly such occurrences are extraordinarily rare..."

This definition contrasts quite markedly with the definition of COI that I offer in HOOKED, which in turn is based on a definition constructed by philosopher Edmund Erde. By the definition I favor, a person may be involved in a COI if she becomes involved in certain social arrangements, which would cause a reasonable observer to believe that a person of normal human psychology would (under those arrangements) be tempted to forsake her professional obligations. Erde, in constructing his definition, explains that he did so with the idea that the core concept that COI needed to be grounded in was trust in a social role. I believe that judgment to be exactly correct-- it is precisely the concerns we have about loss of public trust in physicians and in medicine as a whole that motivates our present concerns about COI.

It should be no great surprise that we could make COI go away if we engaged in the right sort of definitional gerrymandering. The question is whether that sort of definition seems genuinely to enlighten us about the core ideas and behaviors, or whether it rather serves to obfuscate the issues.

I am going to stick my neck out here as I am not schooled in the laws regarding bribery. But my current state of understanding is that you could arrest a public official on a charge of bribery if, say, he was observed to have been offered money in exchange for his vote on some pending legislation, and on accepting the money he indicated his willingness to vote the way that the payer wished him to.

Now, imagine that we were to apply the logic Lance Stell uses to define COI to the bribery case. The first thing we would note is that we cannot arrest the corrupt politician just because he takes the bribe and agrees to change his vote. We have to wait until the vote occurs and see that he actually does vote the way the bribe required. Nor are we done yet. We also have to see the outcome of the vote. Let's imagine that despite the politician voting "no," as the person paying the bribe wished, the legislation passes anyway. Then there was no harm, associated with the bribe; and so by the Stell approach, there would have been no crime of bribery. I would not be surprised under those circumstances to be told that the cases in which public officials are bribed are vanishingly rare.

I suggest that the bottom line is:
  • What we are concerned about in the COI issue is public trust in medicine.
  • Public trust can be lost just as much by the appearance or suggestion of compromising behavior as by the behavior itself.
  • Therefore things that create appearances or suggestions that would reasonably cause public trust to be lost ought to be included in the definition of COI.
  • This is especially true when the social arrangements that give rise to the suggestions or appearances are not essential to the conduct of medical practice or medical research, and consist of extra perks for the academic physician. (I gather that the ACRE folks argue that taking money from industry is essential to medical research, beause absent the good ol' entrepreneurial spirit and profit motive, many fewer discoveries will be made. That debate will have to occur on a different occasion.)

Incidentally, you might wonder why a hypertension journal seems to be going out of its way to provide a platform for ACRE and its preaching. You can get a sense of why this is so by reading more about where some of the ACRE leaders get their money, in Danny Carlat's recent blog on the topic:

Weber MA. Academic physicians confront a hostile world: the creation of ACRE. Journal of Clinical Hypertension 11:533-36, 2009.

Erde EL. Conflicts of interest in medicine: a philosophical and ethical morphology. In: Speece RG, Shimm DS, Buchanan AE, eds. Conflicts of interest in clinical practice and research. New York: Oxford University Press, 1996:12-41


Bernard Carroll said...

When I read Dr. Weber's piece today myt reaction was, can you spell P-A-N-G-L-O-S-S?

Joe said...

Perhaps the hostility is deserved.

Anonymous said...

Re bribery of public officials, I have not reviewed the relevant statutes and case laws, but my understanding, which is explored in some detail in Andrew Stark's wonderful book, is precisely that federal laws are intended to proscribe the relationships that might lead to behavior of partiality, regardless of whether the latter actually occurs.

My understanding is that both impermissible contact between officials and private actors, as well as impermissible conduct may qualify as independent criminal acts.