Bioethics Forum Essay
A FIRST-rate Oversight, and Other Problems with Studies of Medical Residents’ Work Hours
The rigors of medical and surgical training require long hours dedicated to providing clinical care. While long hours are necessary to obtain the experience to eventually practice independently, performing too many hours of continuous duty promotes fatigue, which can result in medical errors, adverse patient outcomes, and risks of personal harm for trainees.
Given these concerns, in 2003 the Accreditation Council for Graduate Medical Education (ACGME), the private accrediting organization that oversees American graduate medical education programs, imposed limits for the first time on resident physician work hours. One key requirement was that resident physicians’ work shifts be limited to no more than 30 hours of continuous patient care duty. However, many patient safety advocates argued that the 2003 ACGME requirements fell short in preventing fatigue-related harms to both resident physicians and their patients.
In 2008, at the request of Congress, the Institute of Medicine (IOM) published a report outlining the increased risks of work shifts up to 30 hours in duration for trainees–including increased risks of motor vehicle accidents, exposure to blood-borne pathogens (for example, from needle sticks), and depression—and the risks to their patients as a result of medical errors. The IOM report recommended that trainee work shifts, regardless of level of training, be limited to no more than 16 consecutive hours of patient care duty, after which a protected sleep period of at least five hours was recommended. As a result, in 2011 the ACGME imposed additional restrictions to limit trainee duty hours. According to the current requirements, first-year trainees are limited to performing no more than 16 continuous duty hours; upper level residents are limited to 28-hour shifts, during which “strategic napping” is encouraged. To comply with these duty hour restrictions, many programs had to adopt new models to provide patient care, with shifts requiring multiple handovers of patient care between trainees.
There has been increasing concern that the duty hour restrictions have adversely affected the educational experience of trainees by limiting the continuity of care they provide, and that they introduced a different source of medical errors due to the handovers of care. There has not been consistent evidence that shorter continuous duty hours for trainees translates into safer patient care by reducing fatigue, yet there is rising concern that shorter shifts are compromising the trainees’ educational experience. As a result, there has been interest in comparing patient outcomes and resident satisfaction resulting from the two different approaches to scheduling trainees’ duty hours.
The FIRST trial has attempted to provide this comparison. The trial was conducted at 117 general surgery residency programs across the country. Each participating residency program, and by extension the trainees in the program, was randomly assigned to either the control arm that conformed to the current duty hour restrictions as imposed by the ACGME or to an experimental arm that allowed a flexible duty hour schedule, permitting shifts of unlimited duration. As such, the surgical patients of participating hospitals were also randomly assigned to be cared for by resident physicians of varying duty hour schedules.
The ACGME provided funding for the study and granted a waiver to all participating programs assigned to the experimental arm, allowing flexible schedules that did not comply with the current ACGME limits on consecutive work hours; programs in the experimental arm were still required to adhere to a maximum duty limit of 80 hours per week, averaged over a four-week period. The study groups were compared with respect to patient outcomes, including the rates of 30-day postoperative death or serious illness, and resident outcomes, namely a self-reported level of satisfaction with the overall quality of resident education and well-being.
The FIRST trial has been criticized as unethical by the advocacy group Public Citizen, the American Medical Student Association (AMSA), and the Committee of Interns and Residents/SEIU Healthcare for inadequate protection of the human subjects involved in the study. The protections of the Common Rule require that risks to human research subjects are minimized; that the potential benefits to the subjects, if any, and of the knowledge expected to be gained by the study outweigh the risks posed to the research subjects; and that subjects provide informed consent prior to participation. In the case of the FIRST trial, however, the institutional review board (IRB) office of the lead institution, Northwestern University, determined that the trial was not human subjects research. The lead investigators therefore advised participating institutions, “we do not believe local IRB approvals are needed. . .” Thus, the standard protections for human subjects participating in research were not extended at some, and perhaps all, participating institutions (information that has not been made public). Herein lies a first-rate oversight and regulatory failure.
The investigators have claimed in their IRB form for determining whether a project involves human subjects research that “the project is limited to the use of existing and/or prospectively collected coded private information . . .” and that “the private information . . . were/are not collected specifically for the currently proposed research project through an interaction or intervention with living individuals.” The investigators adhere to the position that since the intervention occurred at the level of the residency program, and the outcomes data collected was confined to mining a de-identified, validated database of surgical outcomes and the anonymous responses of a nationally administered resident survey, that no interaction or intervention with individuals occurred. It is an intriguing premise, but flawed in the dehumanization of the subjects randomized by their association with the program.
The practice environment for the 4,330 residents and the care environment for the 138,691 patients involved were most certainly manipulated for research purposes; therefore the residents and patients were subject to the intervention of a systematic investigation and, in fact, were human subjects. While the study was not funded by federal entities, as Public Citizen’s Michael Carome describes, “the FIRST trial almost certainly was conducted at institutions that extend their OHRP-approved Federalwide Assurances (FWAs) to all research, regardless of funding.” The subjects of the trial therefore should have been protected under the Common Rule.
Public Citizen and AMSA contend that in assigning trainees to flexible duty hour schedules that were allowed to approximate “the old way” of 28-plus hours of continuous duty, the FIRST trial placed the trainees, particularly first-year residents in the intervention arm, at an unacceptable increased risk of motor vehicle accidents, occupational injury, and depression. Because the investigators were tracking patient safety outcomes, it can be assumed that the flexible-duty hours intervention had the potential to pose more than minimal risk to patient subjects. Yet the subjects of the trial, including both trainees and their patients whose outcomes were tracked, were not provided the opportunity to consent or withdraw from the study due to the “cluster randomization” procedure affecting all of the trainees and patients within a particular center.
In the ongoing iCOMPARE study, a similarly designed trial investigating flexible duty hours among internal medicine residency programs, the designated lead IRB at the University of Pennsylvania recognized the study as human subjects research, but determined that informed consent of both resident and patient subjects could be waived. The Penn IRB found that “iCOMPARE could not practicably be carried out without the waiver . . .” given the cluster randomization design. As with the FIRST trial, Public Citizen and AMSA have criticized iCOMPARE as unethical due to the risks posed to the subjects, and also have alleged inappropriate application of the waiver of informed consent requirements.
The FIRST trial investigators were transparent in their expectation (see “expected results”) that the study would demonstrate no difference in outcome between groups, and thus provide support for a movement to return to more flexible duty hour schedules with longer shifts. The investigators’ agenda, however, may have compromised the scientific validity of the study. The design of the intervention arm allowed for significant variability in implementing the experimental schedule, even allowing the “intervention group” to structure a schedule similar to the control group’s ACGME compliant schedule, and thus reducing the probability of finding a difference between groups. Indeed, the “Flexible-Policy Group” in the trial is reported to be “100% adherent” to the policy of allowing interns’ duty hours to exceed 16 hours, however no data regarding the actual length of time residents worked is provided. The questionable study design, lack of informed consent process, and exposure of subjects to increased personal risk are all problematic from an ethical perspective.
I imagine that institutions justified their participation in the FIRST trial without IRB approval by invoking what I shall call a “regulatory misconception.” Graduate medical education trainees are an inherently vulnerable population, with limited autonomy under the best of conditions. Trainees must conform to the standards of the program in which they are enrolled. They often have little to no control over their schedules or prescribed curriculum, which are regulated and standardized by the ACGME. Thus, when the ACGME waived the typical duty hour restrictions and provided funding for the trial, there was an appearance of a regulatory approval by the body whose primary role is “assessing and advancing the quality of residents’ education” (ACGME mission statement), despite the lack of regulatory approval by a body providing research oversight.
The FIRST trial attempted to answer an important question. With a paucity of evidence that duty hour restrictions have improved patient care, and with continued concern that the educational environment has been harmed by the same restrictions, the educational community needs a rigorous investigation of different paradigms to maximize both resident education and patient safety. It is only through high quality, evidence based research that we can move forward with meaningful educational reform. But we must recognize such research endeavors as human subjects research, and in so doing, promote the safety and autonomy of the subjects. The FIRST trial failed in this regard and we must not allow it to set a dangerous precedent of regulatory failure for research within medical education. We owe this to our trainees, our profession, and above all, to the patients we serve.
Nicole Chiota-McCollum, MD is a vascular neurology fellow at the University of Virginia.