An Empirical Method for Materiality: Would Conflict of Interest Disclosures Change Patient Decisions?
Roy G. Spece Jr.
University of Arizona - James E. Rogers College of Law
David V. Yokum
University of Arizona - James E. Rogers College of Law; University of Arizona - College of Science
University of Arizona
Christopher T. Robertson
University of Arizona - James E. Rogers College of Law; Harvard University - Petrie-Flom Center for Health Law Policy, Biotechnology, and Bioethics
July 6, 2015
40 American Journal of Law and Medicine 253 (2014)
Arizona Legal Studies Discussion Paper No. 14-02
The law has long been concerned with the agency problems that arise when advisors, such as attorneys or physicians, put themselves in financial relationships that create conflicts of interest. If the financial relationship is “material” to the transactions proposed by the advisor, then non-disclosure of that information may be pertinent to claims of malpractice, informed consent, and even fraud, as well as to professional discipline. In these sorts of cases, materiality is closely related to the question of causation, roughly turning on whether the withheld information might have changed the decision of a reasonable advisee (i.e., patient). The injured plaintiff will predictably testify that the information would have impacted his or her choice, but that self-serving testimony may be unreliable. The factfinder is left to speculate about the counterfactual world in which the information was disclosed.
This Article shows how randomized vignette-based experimentation may be a valuable form of evidence to address these questions, for both litigation and policymaking. To demonstrate this method and investigate conflicts of interest in healthcare in particular, we recruited 691 human subjects and asked them to imagine themselves as patients facing a choice about whether to undergo a cardiac stenting procedure recommended by a cardiologist. We manipulated the vignettes in a 2×3 between-subjects design, where we systematically varied the appropriateness of the proposed treatment, which was described in terms of patient risk without the procedure (low or high), and manipulated the type of disclosure provided by the physician (none, standard, or enhanced). We used physician ownership of the specialty hospital where the surgery would be performed as the conflict of interest, disclosed or not, and the “enhanced” disclosure included information that such relationships have been associated with biases in prescribing behavior.
We found that the mock patients were significantly less likely to follow the cardiologist's recommendation of surgical implantation of a drug-eluting stent when he disclosed a financial conflict of interest, regardless of whether the disclosure was standard or enhanced. We also found that the mock patients were more likely to choose the treatment when they faced greater risk without it. We did not, however, find that the disclosure made patients more discerning about the appropriateness of the procedure.
We discuss the implications for law and policy. Mock patients seem likely to act upon such information, declining the low-value healthcare when conflicts are disclosed. This finding suggests that the information is material to such transactions, and that disclosures may be salutary for medical decisions. Arguably, therefore, physicians already have a duty under the common law to disclose the financial relationships they choose to accept. Other regulators and policymakers should recognize and clarify this duty, and courts should embrace this form of evidence. Methodologically, although this empirical approach has limits, it reduces speculation by factfinders and policymakers, focusing their attention on the right questions at least.
Number of Pages in PDF File: 23
Keywords: physicians, doctors, conflict of interest, financial, bias, material information, non-disclosure, disclosure, duty to disclose
Date posted: January 13, 2014
© 2015 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo1 in 0.203 seconds