Conflicts of interest in medicine…
- Sometime leads to academic dishonesty
- Is pervasive
- Is probably unavoidable in the absence of infinite non-profit grants
- All of the above
Charles Denham is a well-known figure in patient safety. He is also involved in a conflict of interest (COI) scandal involving CareFusion paying his organization $11.6 million for software production. One of the results from incident is increased discussion about financial conflict of interest in the medical community. It is probably important to note that the presence of conflict of interest does not specifically pose a problem.
Instead, these discussions focused on its disclosure. The case in point is an article from these experts, who studied Denham’s incomplete disclosure. The authors also included 362 words (longer than most scientific abstracts) to disclose their own possible financial conflicts in the interest of informing the reader.
The question is now twofold:
- Is it the conflict itself, or the dishonesty – i.e. the failure to disclose – that creates problem in evidence-based medicine?
- Does proper disclosure of financial COI actually sufficiently inform the information’s consumer?
In discussion with my co-residents in a morning conference session, it became clear that academicians are probably aware that financial support from non-profit grants alone is probably insufficient to advance medicine. Industry support is sometimes necessary, and that the question is really how do we manage our own dedication to objectivity in the setting of the for-profit interests from these companies. In recognition of this occasionally obligatory collaboration, it has become standard practice for an academic to disclosure possible biases from financial incentives.
The paper in Journal of Patient Safety above, which was published after the incident and shortly after Dr. Denham resigns as its editor-in-chief, highlights the concept that the omission of disclosure is one of the major focuses. The article then outlines the various steps it aims to take to improve the process of declaring financial involvements for submission.
Therefore, the willful withholding of disclosure may be considered a far more serious offense than that breaches the trust we place on the authors of a paper or speakers at a conference.
The focus placed on financial COI disclosure may have been in part because it is a measurable intervention that logically should inform intelligent readers/audiences about how much they can trust the presented data.
A JAMA commentary in 2008 reviewed the ecosystem surrounding COI, including how disclosure affects it. The authors stated that “even where disclosure does not make advice worse, the huge body of research on anchoring (where advice is often disclosed as being randomly generated, yet it still impacts judgment) suggests that disclosure is often unlikely to serve as sufficient warning.” More on anchoring later.
This phenomenon has been shown under a controlled experiment where they asked randomly paired an adviser/advisee dyad to guess the value of coins in a jar. The experiment found that, in their controlled environment, when an adviser is reimbursed by how high the advisee guessed rather than by accuracy, he/she made more money with disclosure than with non-disclosure of the reimbursement mechanism.
The effect is twofold: first, the “paid by high guesses” advisers who disclosed sometimes made much higher recommendations than their non-disclosing counterparts. This effect of disclosure has been termed “moral licensing.” Second, the estimators who received the disclosure of the perverse reimbursement method did discount the advisers’ recommendations but generally fall short of sufficiently discounting the bias. This effect is sometimes called “anchoring bias,” a well-studied phenomenon of our minds being fixated on a presented set of data prima facie and inadequately correct for the biases.
Although this experiment is in a controlled environment and does not pertain to medicine, similar incomplete discounting can be inferred in medical education as well. Another JAMA research letter in 2012 showed that when medical students are exposed to full disclosure (versus no disclosure) from lecturers, they are indeed more likely to increase student interest in posing limitation for such industry relationships. However, their attitude towards commercial involvement in medical education remains statistically insignificant.
The Full Circle
If disclosure is the only reasonable intervention to curb unethical recommendations, but the mechanism is flawed and possibly backfires, do we then have a problem with no solution?
At least under experimental conditions a potential solution might exist. In another Journal of Legal Studies article from 2005, researchers repeated the adviser experiment but with the addition of a new “sanction” mechanics. The sanction mechanics allows the estimators to “blow the whistle” if they feel that the adviser is being dishonest, causing both players to make much less money. Simply by making negative feedback possible was effective in reducing the rate and magnitude of moral licensing.
This experiment is fairly removed from academic medicine. One may even argue that it focuses on the dishonesty related to disclosure rather than disclosure itself. In either case, it does make the point that all is not lost. The practice of disclosure may not have the power to eliminate bias in medical research, but the joint attitude of the medical community with zero tolerance against all types of dishonesty – plagiarism, data manipulation, and incomplete disclosure – may go a long way to making our patients safer. When a check-and-balance system exists to empower the individual, the full disclosure practice in medical journals and presentations also become more effective.
(This post is based on my own interpretation of the events and discussions that transpired and reflect my own opinions.)