American Academy of Ophthalmology Web Site: www.aao.org
Original URL:

 
Opinion

Disclosing Conflict of Interest: Does It Really Mitigate Bias?
By Richard P. Mills, MD, MPH
Chief Medical Editor, EyeNet
 
 

The standard solution to the problem of conflict of interest in continuing medical education is disclosure. If a speaker lets the audience know about financial interests he or she has with specific companies, and the audience pays attention to the disclosure, ethical referees declare no foul. Armed with knowledge of potential bias, the audience simply discounts any positive things the speaker may say about products of the companies involved. Or does it?

Charlie M. Zacks, MD, chairman of the Academy’s Ethics committee, brought to my attention a disquieting study, “The Dirt on Coming Clean: Perverse Effects of Disclosing Conflicts of Interest” by social scientists Cain, Loewenstein and Moore from Carnegie Mellon University. Even more disturbing was that it appeared in the Journal of Legal Studies,1 a publication I try very hard not to read.

If I may be permitted some poetic license in the service of brevity, the study concerned the well-known game of guessing the value of coins in a jar. Undergraduate students served as either estimators or advisors, and were paid on a sliding scale. The estimators had only a 10-second look at the jar and were paid according to the accuracy of their estimate. But they had help from advisors, who were allowed to closely examine the jar and to suggest values to the estimators. The advisors were paid in three different ways:

  1. One group was paid according to the accuracy of their estimator’s guess.
  2. A second group was paid based on how much over the true value the estimator’s guess was; in other words, overestimates resulted in greater rewards to the advisors. Estimators in this group were not aware of these financial inducements to advisors.
  3. In a third group, advisors were also paid on the basis of overestimates, but the estimators knew the advisors made more money if their guesses were too high.

As one might expect, the advisors’ suggestions of value were higher in the second than in the first group, since the second group was being rewarded for overestimates by the estimators. Surprisingly though, the advisors’ suggestions of value were significantly higher still in the third group, in which they had disclosed their conflict of interest to the estimators. Stated differently, having been relieved of the moral dilemma of undisclosed conflict, advisors felt free to apply the “sky’s the limit” approach.

On the estimator’s side, estimates of value increased from the first through to the third group, as their advisors suggestions had. Notably, the disclosure by the advisor in the third group did not reduce the tendency of estimators to overestimate.

Continuing medical education isn’t about coins in a jar, though for me it might as well be when the complexities of Zernike polynomials are being discussed. Yet physicians are still influenced by gifts or incentives of minimal value.2 So the take-home message is clear. Disclosure may embolden speakers to introduce even more bias than they would otherwise, and listeners may not be able to discount biased information as much as they’d like.

So what to do? We can’t insist that presenters have no conflicts at all; to do so would cripple scientific advancement based on collaboration of industry and physicians. Thus, disclosure may be the best weapon we have to mitigate bias in presentations. But it sure isn’t perfect.

_______________________________

1 J Legal Studies 2005;34:1–25.
2 Brennan, T. et al. JAMA 2006;295:429–433.