Sunday, October 11, 2009

The Tangled Web Woven by Industry-Sponsored Research

Our investigative-journalist friends, Jeanne Lenzer and Shannon Brownlee, have capped off a year-long investigation with this article on the Huffington Post:

This case has been getting tossed around since 1999. It defies a quick summation. An anesthesiologist then on the faculty at Columbia University, Elliott Bennett-Guerrero, received funding from Abbott Laboratories to do a study of several blood-expanding intravenous solutions--fluids that were supposed to treat blood loss and shock without the need for a blood transfusion. One ingredient in one of the fluids was hetastarch, which had long been known to trigger serious bleeding. Abbott apparently hoped that a new formulation of hetastarch would minimize the bleeding risks. However, the consent form administered to research subjects failed to reveal the known bleeding risks associated with hetastarch. A number of research subjects then developed serious bleeding problems during or after open-heart surgery.

Columbia conducted several internal investigations, and has just been ordered by the US government Office of Human Research Protections to go back and notify all surviving subjects of the risks they had faced.

As I say, the case is a convoluted one, but I would offer the following observations to accompany the investigative work of Lenzer-Brownlee:

  • The process of informed consent is supposed, at a minimum, to include mention of those harms and benefits the reasonable patient would wish to know about prior to deciding whether or not to accept the offered treatment. An increased risk of bleeding, to the extent of possibly needing a transfusion, would seem clearly to be among those harms. The standard for consent requires that for (nonexperimental) medical treatment. If the treatment is administered within the context of a research study, the threshold for adequate disclosure of information about harms and benefits is increased. This is because the fact that one’s personal physician is prescribing treatment for the patient’s benefit, is presumed to act as a safeguard to the patient’s welfare in the therapeutic setting. In the research setting, where the treatment might be assigned via an impersonal protocol randomly, this safeguard is absent; hence the need for even stricter disclosure of risks.
  • Commercial sponsorship of a research trial raises a couple of questions in this setting. If Abbott paid for the study via an “unrestricted” grant, did they also supply the consent form language? If so the language might have been slanted so as to increase the rate at which subjects were enrolled and to complete the trial faster, a common concern in the industry; the language would not necessarily have been slanted to maximize subject protection. Reportedly Dr. Bennett-Guerrero was conducting 25 trials at the same time. Can a single physician properly supervise that many trials at once, or was this an example of a faculty member trying to maximize revenue on behalf of the institution, or in pursuit of his own interests, or both?
  • Finally, when asked about the harms suffered by some patient/subjects, the investigator reportedly said that this could not be proven, because the study was underpowered. (That is, there were too few subjects in each category, to show a risk of being harmed at a level that reached statistical significance.) Why was an underpowered study being done in the first place? A study that enrolls too few subjects in toto to answer the scientific question it is supposed to answer is scientifically worthless, and therefore the subjects who were put at risk ran that risk or no good scientific reason. No research review board should approve such a study. Alternatively, was Columbia merely a single site of a multisite trial (so that there were too few subjects at Columbia, but that the total number of subjects was sufficient)? If it was, what arrangements were made to assure that investigators (and review boards) at one site, learned of harms to subjects that might occur at another site? How much of the study data were available for review by the investigator at each individual site? In other trials that have attracted criticism, the industrial sponsor has viewed the trial data as its proprietary information, and has refused to release it to individual investigators; this has compromised the ability of the investigator to protect the interests of subjects as well as to interpret the data and to assure that what is eventually published is true.

No comments: