COI disclosure not a surefire remedy
Coming clean may make matters worse
IRBs may want to re-think their policies on evaluating research conflicts of interest in light of new studies indicating disclosure may not have its intended effect.
"Our research shows that disclosure, may, in fact make the impact [of conflicts of interest] worse," says Don Moore, PhD, assistant professor of organizational behavior and theory in the Tepper School of Business at Carnegie Mellon University in Pittsburgh. "In our study, participants felt less responsibility to provide unbiased advice, not more."
Moore, along with fellow Carnegie business school researchers Daylian Cain and George Loewenstein, is researching conflicts of interest and their impact on behavior.
In one particularly telling experiment, they divided a volunteer group of 146 Carnegie-Mellon students into two groups — advisors and estimators — and asked them to estimate the value of cash in six jars containing amounts ranging from $10.01 to $27.06.
The advisors were permitted to handle, but not open, each jar and study it for as long as they liked. They then recorded their estimate of the amount in each jar, which was delivered on a piece of paper as a suggestion to an estimator.
The estimators then got a 10-second look at each of the jars and made their own estimates.
Participants earned small sums of money based on their performance. The estimators’ pay solely depended on how accurate their guesses were, while the payoff to the advisors varied. The researchers used three different scenarios:
- Scenario 1: Advisors and estimators were told they would be paid based on the accuracy of the estimators’ assessments — conditions encouraging both groups to make their guesses as accurate as possible.
- Scenario 2: Advisors were told they would be paid based on how high the estimators’ guesses were, while estimators still were paid based on accuracy. The conflict was not told to the estimators.
- Scenario 3: Advisors and estimators were paid on the same basis as the second scenario, but, this time, the estimators were made aware of the advisors’ conflict of interest.
Before any of the scenarios were initiated, researchers asked advisors to write down their personal best estimates of the amount in each jar.
The researchers were surprised to learn that disclosure encouraged even greater distortion in the advisors’ accuracy than the addition of the conflicting incentive alone.
After the first scenario, advisors made suggestions that were, on average, within a dollar of their personal estimates. In the second scenario, the advisors’ suggestions were, on average, $3 greater than their personal estimates. But the average went up to around $7 greater in the third scenario, when both sides knew the advisor could profit from an inflated estimate.
While the estimators’ guesses differed slightly depending on the disclosures, those differences were not as substantial as the advisors’ behavior.
Many behavioral researchers already know that people have difficulty using information about conflicts of interest, Moore explains.
Psychological traits a factor
Certain psychological traits interfere with the average person’s ability to use disclosure information, he adds.
An observable phenomenon known as "anchoring" demonstrates that most people will focus on a given piece of information and allow that to influence future decisions even when told to disregard that information. And it’s obviously very difficult to how to discount information in light of a disclosure of a conflict.
"For example, your doctor tells you, You need to have surgery right away before this condition gets serious,’" Moore says. "But then he tells you that he owns a stake in the company making the particular device or instrument that will be used in the procedure."
So, do you proceed and have the surgery, or not? Moore asks. Is the physician influenced by his external relationship? Probably, one might reason, but to what extent?
Consumers of expert information — whether they are businesses consulting experts, patients consulting physicians, or the medical community reading about clinical research — have a very difficult time giving meaning to disclosures of conflicts, Moore notes.
Even more compelling is the indication that just disclosure of the conflicts, in addition to the conflicts themselves, may also adversely affect the quality of information given.
"It seems to suggest that the real problems are the conflicts themselves and we really ought to look more at eliminating them and not just disclosing them," Moore says.
1. Cain D, Loewenstein G, Moore D. The dirt on coming clean: The perverse effects of disclosing conflicts of interest. Available on the Social Science Research Network (SSRN) Electronic Library at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=480121.