In a recent post in this space, I wrote about the iCOMPARE research study that is comparing the effects of increasing medical residents’ consecutive duty hours with observing the currently prescribed limits on their shifts. According to the study protocol, the primary hypothesis of the research addresses the safety of patients: that mortality under the increased residents’ duty hours will not significantly exceed patient mortality under the current mandated standard for medical residents on duty. I argued that the study design is unethical based on what is known about sleep deprivation. The study’s goal—seeing whether more patients die when residents get less sleep—is ethically unacceptable because research has definitively shown that residents working very long shifts commit more medical errors than their well-rested counterparts.
There is another reason why the study is unethical. Both the residents and the patients are considered research subjects, yet the researchers are not seeking informed consent from either group to participate in the research. The head of the Accreditation Council for Graduate Medical Education, one of the funders of the study, said that research ethics committees or Institutional Review Boards (IRBs) “at all the participating hospitals reviewed the trial protocols and determined that patients did not need to be informed.”
According to the study protocol, the IRB at the lead institution—the University of Pennsylvania—waived the informed-consent requirement, citing a provision in the U.S. federal regulations for the protection of human subjects in research. That provision is as follows: the IRB may “waive the requirement to obtain informed consent provided the IRB finds and documents that: (1) The research or demonstration project is to be conducted by or subject to the approval of state or local government officials and is designed to study, evaluate, or otherwise examine: (i) public benefit or service programs; (ii) procedures for obtaining benefits or services under those programs; (iii) possible changes in or alternatives to those programs or procedures; or (iv) possible changes in methods or levels of payment for benefits or services under those programs; and (2) The research could not practicably be carried out without the waiver or alteration.”
Who is responsible for iCOMPARE?
The first point to note is that this research is neither conducted by nor subject to the approval of state or local government officials. The University of Pennsylvania is a private school. Of the 63 iCOMPARE residency training programs, most are private institutions, and they include some of the most prestigious medical schools and hospitals in the country (for example, Brigham and Women’s Hospital, affiliated with Harvard Medical School; Brown University; the Cleveland Clinic; Johns Hopkins University; Stanford University; and Yale–New Haven Medical Center). It is therefore extremely puzzling that the waiver of consent mentions research “conducted by or subject to the approval of state or local government officials.”
Even more puzzling, paragraph 46.116(d) in the federal regulations governing human-subject research, which describes the criteria most often used by IRBs when approving a waiver of informed consent, states that an IRB may waive informed consent only if it finds that the research involves no more than minimal risk to the subjects. So if the waiver is inappropriate because the iCOMPARE clinical trial was not being conducted by or subject to the approval of government officials, the research would have to involve no more than minimal risk. But surely a study with the goal of measuring the number of patients’ deaths cannot be considered minimal-risk research. Nor can a study that exposes residents to the unacknowledged increased risks of motor vehicle accidents, needle-stick injuries that can result in infections and depression.
The website for this trial includes a section on Frequently Asked Questions (FAQs). FAQ 5 asks: What are the consent procedures? The response: “Trainees participating in iCOMPARE will consent by completing the study surveys.” But this is obviously not consent to participate in the study itself—that is, working the longer hours. The survey was used merely to collect data, not to elicit consent to participate in the study. And in any case, the residents at participating training programs assigned to the experimental group could not have opted out of the intervention itself because they were required to participate as long as they stayed in their training programs.
Questions of patient consent
As for the patients who happen to be hospitalized during the course of this clinical trial, they will not even know that they are participants in research. The Human Subjects section of the iCompare grant at the University of Pennsylvania provides the following reasons for not obtaining consent from patients: “First, since the trial will involve thousands of patients, it would not be feasible to consent patients.” This prompts the question whether it is ethically acceptable to conduct a study in which obtaining consent is not feasible.
Federal regulations do allow the IRB to waive consent in research when the committee finds that the research poses no greater than minimal risk. This brings us to the second reason stated in the iCompare grant—that the risk to patients is minimal: “Earlier retrospective analysis conducted by members of the current team demonstrated no difference in patient outcomes with even longer duty shifts than those tested in iCOMPARE.” This reasoning is a bit of sophistry. If the outcome for patient safety is already known, then why do the study? The answer can only be that a prospective, randomized controlled trial (RCT) is the “gold standard” in clinical trials. Such trials are conducted precisely because retrospective analyses are often flawed. The results of such studies are often overturned when a carefully designed RCT is carried out.
A third reason for not obtaining consent from patients is provided: “The data are not collected directly from patients; all patient data will be obtained from Medicare claims, requiring no active involvement by the patient.” While this is true, the patients are still subjects in the research since the effect on them of residents’ hours on duty is one of the study’s outcomes. When sleep-deprived residents enrolled in the study care for and perform procedures on Medicare patients, and data from those patients will be collected, surely those patients are research subjects.
Two medical residents at the University of Washington who were enrolled in the study co-authored an op-ed piece published in the Washington Post, criticizing the ethics of iCOMPARE. They wrote that they and their patients were not asked to give informed consent before being enrolled in the study, noting that this “violates the basic ethical principle of respect for persons.”
I am relieved that the residency programs at Montefiore and Einstein’s other affiliated teaching hospitals are not eligible to enroll in iCOMPARE. Since July 1989, teaching hospitals in New York State have been subject to regulations that limit the working hours of medical residents.
Comments on this entry are closed.
On the iCOMPARE website they outline the following standards by which they cap resident hours:
– 80 hours maximum duty per week
– 1 day off in 7
– In-house call no more frequent than q3 nights
Does the current evidence “[showing] that residents working very long shifts commit more medical errors than their well-rested counterparts” generalize to the work hours limits outlined above?
On a slightly different note, I disagree that the use of retrospective analyses is sophistry. Yes, RCT’s are the gold standard for evidence-based medicine, but when evaluating risk, we are taught to consider pre-test probabilities*. The retrospective analysis may not be completely accurate, but it helps us estimate the order-of-magnitude of risk that will manifest in the RCT. Since the waiver of consent requires us to establish minimal risk, that number has to come from somewhere and a retrospective analysis is better than an arbitrary estimate.
* I know this isn’t the actual context in which you use pre-test probabilities, but I think it’s a valid metaphor.
My answer to the first question is that I believe it does generalize, which is why I am critical of the study.
As for the point about sophistry, perhaps I was not clear. I did not intend to say that the use of retrospective analysis is sophistry. What I meant to say (and thought I did say) was the following. The iCompare grant states that “Earlier retrospective analysis conducted by members of the current team demonstrated no difference in patient outcomes with even longer duty shifts than those tested in iCOMPARE.” That statement is used to justify conducting the randomized, controlled trial with longer shifts than are now used in most places. But if the investigators are so certain that there is no difference in patient outcomes, based on the retrospective analysis, then why claim there is a need for an RCT, even if it places residents and patients only at the usual risks?
A cardinal principle of research ethics is that researchers are obligated to minimize risks to subjects. My contention is that the iCOMPARE study does not fulfill that obligation.