CTSI Blogs

The Research Ethics Blog by Bernard Lo, MD

Placebo Controls – Are they Ethically Acceptable in a Clinical Trial?

Every intern learns how to manage patients with massive upper GI bleeds: start two large-bore IVs and fluid resuscitation, send a clot to type and cross packed RBCs, call GI consult. Another component of this usual care is administration of proton-pump inhibitors. Evaluating the effectiveness of proton-pump inhibitors in GI bleeding is challenging because they are widely used in this condition.

In April 2009, a randomized controlled trial of intravenous esomeprazole versus placebo was published in a major journal. Participants had a clinical presentation of upper GI bleed and on endoscopy had a single ulcer with stigmata of recent bleeding. 767 participants were randomized at 91 hospitals in 16 countries between 2005 and 2007. The FDA and its European equivalent agreed that a placebo control should be used if the trial was to be used to support a new indication for the drug. An IRB or equivalent approved the study at each site, and all participants gave informed consent. This trial was sponsored by the manufacturer of the drug. All investigators on the published paper had financial ties to the company beyond participating in the research, and three were employees of the company. A steering committee (names not listed in the article) had responsibility for protocol design.

This trial illustrates an international controversy concerning placebo controls. The Declaration of Helsinki, an international statement on research ethics, states states that the use of placebo is acceptable in clinical trials if

  • no current proven intervention exists; or
  • for compelling and scientifically sound methodological reasons the use of placebo is necessary to determine the efficacy or safety of an intervention and the patients who receive placebo or no treatment will not be subject to any risk of serious or irreversible harm.

In 2008 the U.S. Food and Drug Administration (FDA) announced that it would no longer require that foreign clinical trials submitted to it as part of a new drug application be conducted in accordance with the Declaration of Helsinki. Instead, such trials need to adhere to Good Clinical Practice (GCP) guidelines set by the International Conference on Harmonization. The FDA has expressed reservations about tying U.S. regulations to guidelines that can be changed without its approval. In addition, the FDA has insisted on placebo controls of antidepressants and the use of proton-pump inhibitors for peptic ulcer disease because the percentage of participants responding to active drug or placebo varies widely in different trials, making it difficult to interpret the results of equivalency trials. The Good Clinical Practice guidelines, in contrast, do not discuss the appropriate use of placebos. Thus the FDA would be able to set its own guidelines, with public input, regarding placebo controls.

Placebo controls

  • What are your views about the placebo control in this trial? Your response might depend on what role you were playing.
  • What if you were a member of an IRB reviewing the study?
  • What if you were an official at the company manufacturing the drug? 
  • What if you were an official at the FDA who is charged with determining whether a drug is effective and safe for a particular condition?
  • What if you were a treating physician of a patient admitted to the ED with an upper GI bleed?
  • What if you were a patient with a GI bleed or the family member of such a patient?
  • What kind of evidence or argument would you require to be persuaded that a placebo control was ethically justified in this trial?

According to the bioethics literature, the ethical justification for randomization is that the arms are in “equipoise.” It’s not always clear what is meant by equipoise: the arms in a trial are never in exact balance, but the concept of equipoise means that there is reasonable uncertainty or disagreement over which arm is superior and by how much.

Informed consent may be particularly difficult in a clinical trial with seriously ill patients. What information about the two arms of the trial should be given to participants in this trial during the informed consent process? What concerns would you have about how well eligible participants understood the information that was disclosed to them and their alternatives to participation in the trial?

Comments

It seems like it would be difficult to assess the effect of ppi from observational data. Patients who were given ppi would likely be significantly different than patients not given ppi in a non-randomized study. One might not be able to adjust completely for disease severity or other confounders.

One of the options that past studies with similar intent have used is to compare a PPI with another PUD (peptic ulcer disease) treatment such as ranitidine (an H2-receptor antagonist), instead of using a placebo. However, as the current study cites, meta-analytic data suggests that H2-receptor antagonists 'are not efficacious' - and thus such a study may be questionable in terms of equipoise. An argument for using placebo is to obtain rigor in the study - as noted by both the FDA and the background section of the study itself, there is a history of wide variability in patient response to PPIs and placebo for PUD. Thus an alternative study design without placebo may not add much value to this study's field, even if the study results were positive. Further, the intervention used in the study was an adjuvant therapy to the standard therapy - which means that even with placebo treatment, the participants were receiving standard of care. Therefore, if I were a member of the local IRB committee reviewing the study, I would have compunctions only if I were in a resource-poor region in which the standard of care as cited in the study was not available. Then it may be unethical to carry out the study in my region - because the study did not mention the specific ED's or regions that participated, it is unclear to me whether this is an issue or not in this study. If I were an official at AstraZeneca (manufacturer of esomeprazole), I would be satisfied with the design, mainly because it is double-blinded. I might hesitate about involving the company in so much of the study design, data management, and manuscript writing. The FDA's stance is that placebo controls need to be involved in this kind of trial, so I would be willing as an FDA representative to endorse the study.

Establishing exclusion criteria is a way to enroll a target population with the right characteristics into a trial as well as to protect human subjects from potential adverse events. Although these measures may have undesirable consequences for the trial, this is a secondary concern and the safety of human participants should obviously take a higher priority. This is especially true for vulnerable populations like the one addressed in this ethics discussion. However, excluding patients at high risk from a trial is not always the best way to protect patients, yet all patients should be informed of possible adverse events per IRB protocol. What if a new drug was under investigation that could possibly reduce the risk of suicidality, then the investigators may not want to exclude these patients from the study. In terms of access to psychiatric care, this could be built into the follow up of the patient as a possible recommendation if they continue, or move, to a high risk characterization and a risk incurred by the investigators. For example, for patients of continued high risk of suicidality (ie, considered a threat to themselves) a referral to psychiatric care could be built into the protocol. How these patients are counted in terms of loss to follow up, drop out, or an unsuccessful case is up to the investigators