Controversy: Informed consent and cluster-randomized trials
Two takes on a thorny issue
In this question-and-answer session, Mark Schreiner, MD, chair of the Committees for the Protection of Human Subjects (IRB) at the Children's Hospital of Philadelphia (CHOP) and an associate professor of anesthesia in pediatrics at the University of Pennsylvania in Philadelphia, PA, discusses the issue of informed consent in cluster-randomized clinical trials.
Schreiner, who also is an editorial advisory board member of IRB Advisor, believes that regulations under the Food and Drug Administration (FDA) preclude waiving individual informed consent when cluster randomized clinical trials involve the comparison of one or more FDA-regulated drugs, devices, or biologics. His comments are in response to papers published on cluster randomized trials in the Hastings Center Report and Pharmacoepidemiology and Drug Safety, as well as an article on the topic in the January, 2010, issue of IRB Advisor.1,2
IRB Advisor: We have seen the email you received from Christine Oliver, PharmD, of the FDA's division of drug information. In response to your question about whether FDA regulations would permit a waiver from individualized informed consent in cluster randomized clinical trials involving FDA-regulated substances, she wrote to you: "You are correct that waiver of the informed consent requirement is not permitted for FDA-regulated studies. While the study may be exempt from 312, it specifically states in 312.2(b)(l)(iv) that both Parts 50 (regarding informed consent) and 56 (IRB review and approval) are applicable even for studies with approved drugs used as described in the approved labeling. Informed consent requirements cannot be waived for studies considered to be FDA-regulated except in emergency conditions (50.23) or for emergency research (50.24). For studies with approved/cleared devices used according to approval/clearance, no comparable statement requiring adherence with Parts 50 and 56 exists in the exemptions section of Part 812 (812.2.(c)), but we still consider most such studies to be FDA-regulated and therefore Parts 50 and 56 are applicable.
What is the cause of this controversy over informed consent?
Schreiner: Many people don't understand that studies of approved drugs that don't require an IND must still comply with the FDA's IRB regulations. So even when an IND is not required to study a drug that is approved, the study must still comply with Parts 50 and 56.
That's where the problem comes in: Part 50 (§50.23 and §50.24) only allows waiver of informed consent from the individual for certain types of emergencies, emergency conditions, and emergency research.
What the articles proposed was that it was okay to do a cluster randomized trial where permission was obtained from the practices, without getting individual consent from the subjects. This is okay if the intervention isn't a drug, biologic, or device regulated by the FDA and where the IRB could waive consent. For example, an IRB could waive individual informed consent for a cluster-randomized trial of two different diets to determine if there was a difference in weight loss between the two groups because the study doesn't involve any FDA-regulated test articles.
Studies of approved drugs conducted at institutions with an FWA must comply with both the common rule and FDA regulations, and that's where the snag comes in. I just don't think most people are aware that when they're studying an approved drug that they have to follow the FDA regulations, which are slightly different than the Common Rule.
IRB Advisor: So how should institutions and IRBs handle reviews of cluster randomized clinical trials?
Schreiner: First, they should determine if the intervention is regulated by the FDA. If it's not, then the IRB can decide whether or not to waive consent.
We have a number of cluster-randomized trials underway at CHOP where we've waived consent. Most are studying the impact of electronic health records on improving care. The Obama administration is very keen on getting electronic health records, but it's not so clear exactly what the individual benefits will be for that. So a number of investigators are looking at whether some of the services available in electronic health records, mainly warnings and reminders, improve care. For example, if a patient is eligible for a flu shot, the screen will flash and remind doctors to give the patient a flu shot. Some studies look at whether or not there is a difference in the percentage of kids who get what are considered the guideline practices between the doctors who receive an electronic reminder and doctors who do not. Since we have about 30 practice groups that are part of our health care network it is possible to randomize them into two groups to conduct this sort of trial.
Another example is a study to see if antibiotic usage and prescribing habits can be improved by providing electronic health record reminders and educational materials. Will it work? Physicians are very busy, and they could ignore the messages.
IRB Advisor: Why would cluster-randomized drug comparison research be that different from what you've described?
Schreiner: It is important to remember that the studies I've described are not comparing two drug treatments where the practice is randomized to prescribe drug A or drug B. In the antibiotic reminder study, the investigators are just giving doctors information. The practices are randomized to different means of receiving information; one gets the electronic health record reminders, and the others just get standard practice.
The whole point of cluster randomization is to make large scale trials feasible. The randomization is not at the level of the patients, it's the practices that are being randomized. If the intervention is not regulated by the FDA and the IRB decides it meets the requirements for waiver of consent (45 CFR 46.116(d)) then they can waive consent.
What the recent papers proposed was to randomize practices A, B, and C to give one drug, and practices D, E, and F to give another drug in order to determine which of the two commonly-prescribed treatments was better.
The cluster-randomized trial design provides an easier path by avoiding the need for consent from individuals. But since the intervention is a drug regulated by FDA, the IRB must require that the investigators obtain the consent of every individual subject.
IRB Advisor: What about your example of a cluster randomized trial looking at antibiotic prescribing habits?
Schreiner: The IRB was able to waive informed consent because the investigators were able to demonstrate that the study was minimal risk, and it wasn't practical to get informed consent from 30,000 to 50,000 children coming in for office visits with ear infections. Clinicians will get information messages with reminders that these are antibiotics that should be first-line treatment, which is not any different from the way the insurance company says, 'These are the drugs on our approved formulary for an ear infection,' and the physician is free to read the message and prescribe based on this information or not. The other practices don't receive any reminders, just their usual update to review antibiotics.
Scientist responds to issue
Editor's note: IRB Advisor asked James E. Sabin, MD, director, ethics program at Harvard Pilgrim Health Care, and clinical professor in the departments of population medicine and psychiatry at Harvard Medical School in Boston, MA, to respond to Schreiner's comments about the need for individual informed consent.
IRB Advisor: Your papers on cluster-randomized trials discuss gray areas in interpreting the need for individual informed consent for the types of cluster-randomized trials you've described. Dr. Schreiner and the FDA email state there are no gray areas when the study involves comparison of drugs, biologics, or devices that are FDA-regulated. How do you respond to this point of view?
Sabin: 312.2(a) states that Part 312.2 applies to "all clinical investigation of products…" Therefore, the key question is — what is a "clinical investigation"? According to 312.3(b), a "clinical investigation" is "any use of a drug except for the use of a marketed drug in the course of medical practice."
In the form of cluster randomized trial we envisioned, the drugs were commonly used "in the course of medical practice," and were in clinical equipoise with regard to their comparative effectiveness. In the envisioned trials, one set of clusters would give preference to drug A while another set gave preference to drug B, but prescribing physicians retained the option of prescribing the non-preferred agent if, in their clinical judgment, it was preferable for a particular patient.
In contemporary practice physicians often receive memos or prompts encouraging them to choose a particular agent among the many that are in common use. In the cluster randomized trials that we were discussing, the assignment of the cluster creates the equivalent of a prompt, which the prescriber is not required to follow.
These are the conditions of ordinary medical practice. My colleagues and I concluded that in terms of 312.3(b) the kind of comparative effectiveness trial we were concerned with is not a "clinical investigation." Given that interpretation of 312.3(b), 312.2 does not apply, and, as a result, Parts 50 and 56 are also not applicable.
Even if we assumed Parts 50 and 56 to apply, the definition of "clinical investigation" remains crucial. Part 50.3(c) defines the term as follows: "any experiment…that either is subject to requirements for prior submission to the FDA…or is not subject to requirements for prior submission…but the results of which are intended to be submitted later to, or held for inspection by, the FDA as part of an application for a research or marketing permit."
The kind of comparative effectiveness studies we were discussing were not subject to prior submission, and involve no intention to obtain a marketing permit. On that basis we concluded that Part 50 would not apply even if Part 312.2(b) did.
IRB Advisor: If the federal government's new goal is to increase the number of drug comparison studies, as has been stated by the Obama Administration, do you believe the requirement of individual informed consent will have a significant impact on reaching this goal?
Sabin: If individual informed consent is required for the kinds of cluster randomized trials we envisioned it will drastically reduce the potential for drug comparison studies. Requiring individual informed consent for every drug comparison study would effectively kill the Obama Administration's comparative effectiveness initiative.
IRB Advisor: Do you believe there will continue to be a public debate over this issue, and how might it proceed?
Sabin: There will and should be public debate. FDA and OHRP regulations were not constructed with comparative effectiveness studies of commonly used products and common medical practices in mind. Debate should proceed by first asking what consensus can be reached about what is right from the perspectives of clinical practice, public policy, and medical ethics. When we're clear on these questions we should look at the regulations. Do they promote the ethically desirable outcomes? If so, fine. If not, change them.
- Sabin JE, Mazor K, Meterko V, et al. Comparing drug effectiveness at health plans. Hastings Center Report. 2008:39-48.
- Mazor KM, Sabin JE, Goff SL, et al. Cluster randomized trials to study the comparative effectiveness of therapeutics: stakeholders' concerns and recommendations. Pharmacoepidemiolog & Drug Safety. 2009;18:554-561.