What do we really know about the benefits and harm of restricted resident duty hours? Implications for medical error and physician burnout
sola dosis facit venenum (The dose makes the poison) Paracelsus
This blog was inspired by a Medical Grand Rounds presented by Dr. Samir Hazra, from our Cardiology Division. Samir summarizes the history that led to modern duty hour restrictions for doctors in training (residents) and then critically appraises the impact.
Dr Samir Hazra, Assistant Professor. Samir is interested in cardiac imaging, general cardiology and medical education.
The goal of the blog is to critically evaluate whether restricting work hours has improved the quality of patient care or the satisfaction and/or the wellness of the residents themselves. I do not long for the imaginary “good old days”, nor would the absence of evidence of benefit from work hour restrictions necessarily mean it was all for naught. Societal expectations change and medicine as part of society must adapt. However, in understanding any problem its important to get the facts straight, frame a hypothesis and then evaluate it with an open mind using scientific rigor. The hypothesis is that tired residents are not happy, healthy or safe, a reasonable hypothesis to assess.
The “fact” that started the change in work hours for trainees was the death of a young woman in at New York Hospital (now New York Presbyterian Hospital) in 1984. The assumption was that her death was due primarily to physician fatigue. Without critical evaluation (based on the belief that it was a self-evident truth that fatigue was the cause), changes to the training of house staff were made that were substantial and which, until recently, have largely been unexamined. Dr Hazra will tell you the story of Libby Zion, the person whose tragic death became the burning platform upon which changes to resident work hours was made.
Before reading his contribution though consider this, if excessive work hours are the root cause of medical errors and/or physician burnout then less work should make for fewer medical errors and lower rates of burnout (presumably). Conversely, if work hours are not the major cause they may still require redress; but we may need to consider and evaluate other potential causes if we wish to reduce medical error and physician burnout. Anchoring on excessive work hours as the core flaw in medical care delivery and medical training can be misleading, much like anchoring on a single disease as an explanation for a patient’s symptoms, instead of constructing a proper differential diagnosis.
It is clear burnout has not decreased amongst physicians or residents despite work hour restrictions. In a 2018 CMAJ article by Lauren Vogel, survey results revealed the following concerning statistics “Of the 2547 physicians and 400 medical residents surveyed, 30% reported high levels of burnout, meaning they experienced symptoms of emotional exhaustion and depersonalization at least weekly. Thirty-four percent met criteria for depression. Nearly one in five reported having thoughts of suicide at some point in their lives; 8% thought about suicide in the past 12 months.”
The rate of medical error (or at least adverse medical events) has only modestly decreased during the period in which changes in work hours has occurred.
Blog by Robert Gorsky suggest rates of adverse medical events has modestly decreased
Although the topic of mental health is sensitive (too often taboo) it is worth considering the possibility that work hours may not be a primary determinant of burnout. For example, many professions with high rates of mental health issues don’t involve long work hours (e.g. pharmacy and dentistry). In addition, in Medicine burnout is occurring in medical students who have not yet assumed a clinical load. Perhaps burnout has to do with other stressors in the system and with the health, balance and coping skills of people entering the profession. We are (on average) overachievers and perfectionists working in an imperfect health care world with unlimited amounts of work to be done and little possibility of fully controlling all outcomes.
When one considers work hours and adverse outcomes, suicides are often mentioned; indeed Dr. Hazra began his lecture by noting that 3 of his classmates had fallen to suicide. This is tragic, but consider that professions other than medicine that have increased risk of suicide (values over 1 reflect the fold increase in risk relative to the normal population). These professions do not involve the hours of work associated with the practice of Medicine but share other stressors that could account for this finding, including high risk/consequence nature of the work, shift work, access to means of suicide (toxic drugs), as well as the possible susceptibility factors, such as the brittleness that can occur in perfectionists.
Fold risk in suicide by profession
- Marine engineers (1.89X)
- Medical Doctors (1.87X)
- Dentists (1.67X)
- Veterinarians (1.54X)
- Financial services (1.51X)
- Chiropractors (1.5X)
Indeed in the Medscape National Physician Burnout, Depression & Suicide Report for 2019, work hours was the 2nd most cited cause of physician burnout. (it's the paperwork and charting that is the leading burnout factor by 2-fold).
Survey results from ~15,000 physicians practicing in the United States of America
Moreover, there are unanticipated consequences of restricted work hours for residents that might paradoxically promote medical error, such as lack of continuity of care for patients. In addition, the practice of Medicine requires, practice. At some tipping point, duty hour restrictions could even reduced residents’ experience, and in this era of competency based medical education prolong the time required for them to become competent and transition from trainee to independent practice. Ironically the anxiety relating to delay in becoming a “medical expert” can also cause stress and burnout. I encourage you to read this with an open mind, rather than presupposing the value of restricted duty hours is established or assuming that questioning the relationship between work hours and burnout makes one a luddite. For work hour reduction, as with most interventions the magnitude matters. For example, the previously mentioned Medscape survey showed a dose response in the relationship between work hours and self-reported burnout.
To paraphrase Paracelsus on the toxicity of medications, all drugs are poisons (at a high enough dose). A sidebar note that confounds the implementation of reduced resident work hours is a poor or absent change implementation program. In other words, we reduced resident coverage of hospitals without figuring out first who would care for the patients! The council of academic hospitals of Ontario (CAHO), which negotiates the duty hours agreement with the professional association of residents of Ontario, PARO (the house staff union), does not have front line responsibility for hospital budgets or patient care. Their negotiations have gradually granted residents more time off, more protected time post call and have created a system of duty hours restriction that is complex to manage and somewhat detached from the patient care imperative. This raise two issues-one for the residents and one for the system. For the resident trainees there is an implicit tension in the issue of work hours, namely: Is residency about service and leaning by doing or is it primarily educational, with service being a by-product? I would argue it is both; education goes hand in hand with service. Whatever one’s views on this balance, and its effects on competency there is the second issue of patient coverage. As duty hour restrictions were brought on, no provision was made within the healthcare system to replace the invaluable service provided by residents. To this day, there is no funding line in most Canadian hospitals to pay for the supplementary hospitalists, ANPs, and PAs that are required to run inpatient services that were traditionally formerly performed by residents. We are not going back in terms of work hours so it is clearly past time to have health care funding include the cost of these alternative providers. The faculty cannot absorb the care gap. We currently rely heavily on international trainees to cover the service gap in Canada. I have previously made the case we should be paying for additional Canadian trainees and/or alternative providers to meet the care gap left by reduction in resident coverage through a government funding mechanism (see: Why Canada should fund its own medical education system).
With that introduction, Dr. Hazra will reprise his discussion of residents work hours.
Medical education is complex with myriad moving parts. It has a profound impact on the care patients receive now and in future generations. Medicine itself has become progressively more evidence-based over time; however, medical education, has largely remained an art with little scientific evidence to guide us. Training has been based upon tradition, custom, emotional considerations and public perception. In this blog, I reprise a recent grand rounds I delivered to the Department of Medicine at Queen’s University regarding one small aspect of medical education, that of resident duty hours and its impact on patient outcomes, educational outcomes and resident well-being.
How resident duty hours became one of the first major focuses of research in the field of medical education is an interesting point to ponder. It would seem that public perception had more to do with it than scientific plausibility. From the New York Times to 60 Minutes, sensational headlines describing throngs of residents working endless hours before presenting themselves to care for your loved ones precipitated a battle cry with ripples all the way up to the United States Congress demanding that resident work hours be regulated with no evidence guiding exactly how to do so. In fact, public opinion resulted in regulations that reduced trainee work hours before rigorous studies had been undertaken. Now that we have data from some such studies, you might be surprised to see how the scientific data compares to the preconceived perceptions.
Before delving into the story of resident work hours itself, I would like to present a few disclosures. Firstly, I am a product of the evolving medical education system that we are set to discuss and as a result, have an inherent bias to suggest that despite its problems, the system has trained me well. That being said, you should know that I graduated in the class of 2008 and in the very short period of time since, have lost 3 classmates to suicide. While their deaths have not been linked directly to the issue of resident work hours, their loss has a profound impact on me and colors my perspectives. I also served on the PARO general council for 6 years between 2008 and 2014, the period of time over which the current contract framework that governs work hours was negotiated. Little did I know at the time that fast forward a few years to today, I would be directing the cardiology training program at Queen’s University, serving as faculty on the national trainee review program and sitting on the Royal College of Physicians and Surgeons of Canada’s (RCPSC) subspecialty committee in cardiology. It is my hope that as a result, I can offer both the perspective of the resident advocate, looking to optimize their overall learning experience, as well as that of a faculty member, responsible for delivering world class training. The views expressed in this blog represent my personal reflections and review of the data and certainly are not those of Queen’s University, Kingston Health Sciences Centre, our training programs, PARO, or the RCPSC.
The above photograph is taken in April of 1889. It shows Sir William Osler with the Johns Hopkins Hospital’s first group of interns. This was the birth of residency education, a course of training that at the time, was grueling and all consuming and essentially available only to unmarried white men. Residents of that era lived (resided) in the hospital and were expected not to marry.
By the early 1900s, the American Medical Association provided a listing of approved residency programs in which house staff were expected to work every day and every other night. This tradition continued right through the mid 1900s when residents and interns were expected to be on call for 36 hours starting every other night, totaling more than 100 hours per week. It was not until 1975, 86 years after the above photograph, that residents in New York City finally took it upon themselves to go on strike calling for fewer duty hours. They returned to work only after hospitals agreed to reduce on-call frequency from every other night to every third night. This was the setup of resident work hours until March 4th 1984 and the case that shook medicine.
There are many varying accounts of exactly what happened to Libby Zion, a college freshman with an ongoing history of depression who presented with fever, dehydration and uncontrollable shaking. Unable to diagnose the condition definitively, the ER MD admitted her for hydration and observation. Raymond Sherman, a senior clinician known to the Zion family, approved the decision by phone. On the ward, Libby was evaluated by intern Louise Weinstein and resident Gregg Stone and was given the diagnosis of a viral syndrome with hysterical symptoms. Meperidine was given to control shaking (also approved by Dr. Sherman over the phone at 3am), after which Dr. Weinstein left to care for 40 other patients assigned to her overnight. As the night progressed, Libby became increasingly agitated. Nurses contacted Dr. Weinstein twice and she ordered physical restraints and haloperidol. The patient herself was not re-evaluated. By 6:30am, Libby’s temperature had risen to 107F. Dr. Weinstein was again called. Libby went into cardiac arrest and died.
Libby’s father Sidney Zion, a well-known lawyer and journalist, was convinced that inadequate staffing and supervision was to blame for his daughter’s death. He was famously quoted as saying, “they gave her a drug that was destined to kill her, then ignored her except to tie her down like a dog” and referred to her death as “murder”. He subsequently wrote in a New York Times (NYT) op-ed: “You don’t need kindergarten to know that a resident working a 36 hour shift is in no condition to make any kind of judgment call, forget about life and death”. He was incensed that the intern was covering a large number of patients, that the resident was never awakened, and that the attending was not called during Libby’s deterioration. Pictures of overworked residents were plastered all over the NYTs, Washington Post, Newsweek and a tired intern interviewed on 60 minutes even forgot one of Mike Wallace’s questions.
Clearly, the optics were very poor despite the fact that serotonin syndrome, an adverse effect of the medications and the likely cause of death, was a rather obscure condition in 1984. Indeed, there was no proof that exhaustion was actually the cause of the error. Despite this, in May of 1986, the Manhattan district attorney let a grand jury consider murder charges. While the grand jury declined to indict, they strongly criticized the supervision of interns and residents. In response, the New York State Health Commissioner established the Bell Commission to evaluate the training and supervision of doctors in the state. In 1987, the Bell Commission recommended an 80-hour limit on weekly resident duty hours, with no more than 24 consecutive hours on duty. By 1990, the Accreditation Council for Graduate Medical Education (ACGME) set an 80 hour workweek for certain specialties and limited on-call duties to every third night with at least one 24-hour period off every 7 days. This was insufficient to quell public opinion on the topic and in 2001, petitions led to proposed federal legislation to limit resident work hours. Fearing a complete loss of control over residency training, the ACGME responded in 2003 by extending the 1990 work hours restrictions to all specialties and added a maximum shift length of 24 hours with 6 additional hours for education and handoffs. This system remained in place until 2011, when the ACGME restricted interns to a maximum of 16-hour shifts, in response a 2007 report from the Institutes of Medicine commissioned by Congress. The 2011 ACGME requirements also encouraged the use of alertness management strategies and strategic napping.
While all of the above regulatory changes were implemented with the explicit purpose of mitigating resident fatigue, they were not based on any scientific evidence that the changes would have a positive effect on patient outcomes, educational outcomes or on resident well-being. Thinking back to the Libby Zion case, was exhaustion really the cause of the negative outcome? What about lack of supervision? What if the drug interaction had been identified by a computer? Is there any proof that the death would have been avoided if the intern had taken a nap? In fact, none of these questions were answered or even asked when implementing the regulatory frameworks. It is this desire to act in the absence of evidence which is the true historical paradox of resident duty hours. Leaders in medical education ultimately began to openly question the rationale for such regulatory changes. Sanjay Desai, the internal medicine program director at Johns Hopkins was quoted as saying “Everybody says we’re done with duty hours and we can’t go back. That’s a defeatist attitude. This is the future of American medicine, and the risk is too great. Creating more regulation in the absence of data is not a tenable solution”.
One of the first robust studies looking at the effect of reducing interns’ work hours on serious medical errors was conducted by the Harvard Work Hours, Health, and Safety group. In 2004, Christopher Landrigan at Harvard medical school publish a randomized controlled trial of medical interns in an intensive care setting randomly assigned to a standard every-third-night schedule (30hr shifts alternating with 10hr shifts) or an intervention schedule limiting interns to a maximum of 16 consecutive hours of work. Interns in the intervention group made fewer serious medical errors (4 more error for every 100 patient days; most related to medication management) but there were no differences among groups in total rates of adverse events. The generalizability of the findings remains an issue due to the fact that implementation was in two high-acuity units in a single hospital that were extraordinarily well-staffed in the intervention arm. Attribution bias was also raised as an important issue with 3 residents from the intervention group writing a letter questioning the study’s conclusions. In it, they wrote “worried residents and attending physicians, aware that the interns on the intervention schedule were poorly informed, took a more active role in patient care, making the majority of decisions and more closely supervising the interns’ actions. This hypervigilance may have strongly biased the study towards a positive result”.
In 2013, Sanjay Desai and colleagues published a randomized trial comparing a call model comprising 30 hour call shifts every 4th day to one of 2 models consistent with the 2011 ACGME regulations (requiring interns to be scheduled for shifts no longer than 16 hours). Although residents assigned to both newer models had more sleep, they reported a decrease in educational opportunities, more frequent handoffs disrupting continuity of care and a perceived decrease in the quality of care. Nurses also reported worse perceived quality of care.
Many observational studies assessing patient safety outcomes as they relate to resident work hours have been published and while the overall results vary, most agree that reducing work hours does not, by itself, markedly improve patient safety. In fact, reducing resident duty hours has the potential to increase the risk of errors that arise from more frequent handoffs and reduced clinician presence. Similar observational studies have been performed to assess the relationship between resident work hours restrictions and competency and have suggested that reduced clinical exposure may require extending the length of training. Procedure-intensive specialties such as general surgery and neurosurgery have reported negative educational outcomes with restricted work hours including worse scores on licensing exams. Finally, observational studies on the effects of restricted work hours on resident wellness have shown no consistent evidence that reduced duty hours have a positive effect on resident mood, stress, or personal relationships. In fact, a survey of surgical residents in Quebec suggested that new work hour limits had a negative effect on quality of life with trainees feeling more sleep deprived, disconnected from the clinical environment and receiving less mentorship.
In 2013, a survey of program directors after implementation of the 2011 ACGME regulations suggested an increased faculty workload was associated with decreased time for teaching and a growing shift-work mentality in which restrictions of duty hours were emphasized over duty to the patient (Drolet et al NEJM).
At this point, there were many assumptions at play on the topic of restricted duty hours. The general feeling in the public was that longer hours led to less sleep, which in turn led to more errors. The difficulty was that while one could regulate duty hours, one could not regulate sleep. It was also possible that resident fatigue may be caused more by the compression of a large volume of clinical work, which may be exacerbated by work hour limits. In addition, shorter work hours led to more patient handoffs, less education, and development of a shift mentality possibly leading to less competent and less committed doctors for patients in the future. There was a true concern that the safety benefits of reducing resident fatigue would be offset by harms associated with disrupting continuity of care. Clearly, large, robust, randomized controlled trials were needed to guide future recommendations.
In 2016, the FIRST trial was published in the New England Journal of Medicine (Table 1 below). In it, 117 surgical residency programs were randomly assigned to adhere to either standard ACGME duty-hour rules or flexible rules, under which limits on shift lengths and time off between shifts were waived. The study was conducted as a non-inferiority trial with the primary trial measure being patient outcomes (30 day rate of postoperative death or serious complications). Multiple other outcomes were measured simultaneously, including resident perceptions and satisfaction regarding well-being, education and patient care.
While there were differences between groups with respect to maximum shift length and minimum time off between shifts, the maximum hours worked per week, mandatory time free of duty, and frequency of on-call duty (1 in 3) were the same for both groups.
The study concluded that as compared with standard duty hour policies, flexible, less restrictive duty-hour policies for surgical residents were associated with non-inferior patient outcomes and no significant difference in residents’ satisfaction with overall well-being and education quality.
How to interpret these results continued to be debated with the accompanying editorial stating, “because the FIRST trial found no evidence that removing restrictions on resident shift length and time off between shifts was harmful to patients, programs should have more autonomy to train residents as they choose. I reach a different conclusion. The FIRST trial essentially debunks the concerns that patients will suffer as a result of increased handoffs and breaks in the continuity of care. Rather than backtrack on the ACGME duty-hour rules, surgical leaders should focus on developing safe, resilient health systems that do not depend on overworked resident physicians”.
None the less, the ACGME did respond in 2017 with new regulations maintaining an 80 hour per week cap but allowed this to be averaged over 4 weeks while extending permissible work shifts for interns from 16 to 24 hours. They also permitted more within-shift flexibility as long as weekly duty-hour limits were met.
On the internal medicine side of the resident duty hours debate, the iCOMPARE research group published their educational outcomes data in the NEJM in April of 2018. The trial was a cluster-randomized trial comparing 2 duty-hour structures (standard policies of ACGME 2011 vs. flexible policies with no limits on shift length or time off between shifts) and evaluated the attitudes, educational activities, and test performances of internal medicine residents across a wide variety of programs in the US. In total, 63 internal medicine training programs were randomized. Patient safety outcome data and sleep and alertness data were not reported in this initial study (Table 1, 3 an 4 below).
Residents in flexible programs spent more time on direct patient care and less time in handoffs than did patients in standard programs.
Interns reported that safety was negatively affected in flexible programs, but that continuity of care was negatively affected in standard programs. The ability to attend educational conferences and the relationship between interns and residents was also negatively affected in flexible programs.
Continuity of care, quality and ease of handoffs and transitions in care, ability to follow the clinical care of patients admitted by the intern and the number of admissions handled entirely by the intern were more satisfying in the flexible programs, but all other parameters listed were more satisfying in the standard programs. The study did have a number of limitations including desirability bias in survey responses, low response rates to some elements, difficulties characterizing true differences between the intervention group and control group in terms of hours worked, large variation in outcomes across programs and the fact that flexible programs applied flexibility only to a small minority of rotations.
In the end, medicine residents in flexible programs were substantially more dissatisfied overall. The negative effect on overall wellbeing, morale, personal health, interpersonal relationships, professionalism, job satisfaction, ability to attend educational programming, and perceived effects of fatigue on safety could not be justified by small positive effects on continuity of care. There were high rates of burnout with >66% of residents reporting high to moderate levels of emotional exhaustion, depersonalization, and low perceptions of personal accomplishment. Burnout is known to affect rate of errors, patient mortality, teamwork, malpractice suits, patient satisfaction, productivity, and costs. Burnout can be disastrous for individuals and programs as it can lead to depression, alcohol and drug abuse, and suicide. Since burnout tends to peak at mid-career, this degree of burnout among residents portends greater problems for IM residents in future years.
When comparing the FIRST and iCOMPARE data, it is clear that both surgical and IM residents subject to flexible hours were more dissatisfied with general well-being and amount of time for rest and were less dissatisfied with continuity of care and ease of handoffs. IM residents, however, were unhappy with the current training environment whether flexible or standard and there was a suggestion that what works for surgery may not be appropriate for IM (Table 1 below). In fact, standard programs were better in IM for all but continuity of care and transitions, whereas flexible programs were better in surgery for all but amount of rest and overall wellbeing.
In March 2019, the iCOMPARE group published their patient safety outcome data and sleep and alertness data in the New England Journal of Medicine.
They found that 30-day mortality among patients in flexible programs was non-inferior to that in standard programs.Differences in changes between flexible programs and standard programs in the unadjusted rate of readmission at 7 days, patient safety indicators, and Medicare payments were also below the pre-specified non-inferiority margin and the non-inferiority criterion was not met for 30-day readmission or prolonged length of hospital stay.
Sleep duration in flexible programs was non-inferior to that in standard programs as was the score on the Karolinska Sleepiness Scale. Non-inferiority was not established for alertness according to the brief computerized psychomotor vigilance test.
In conclusion, the FIRST Trial was a national, prospective, randomized trial involving surgical residency programs that found no significant differences in rates of death and surgical complications between standard and flexible programs.Residents in flexible groups were less dissatisfied with continuity of care and handoffs and there were no meaningful differences between the 2 groups in terms of quality of training or morale.
The iCOMPARE Trial was a randomized trial of internal medicine residency programs assessing patient safety, educational outcomes, sleep and alertness. While program directors were more satisfied with flexible work hours, there were no significant differences in the proportion of time interns spent on patient care and education. Interns in the flexible group were more dissatisfied with educational experience and had lower levels of overall wellbeing. 30 day mortality, 7 day and 30 day risk-adjusted readmissions and Medicare payments were non-inferior in the flexible as compared to standard groups and sleep duration was non-inferior in flexible as compared to standard groups with interns making up for lost sleep on extended shifts by sleeping more during their time off.
The complexity of the current PARO contract: Given these contemporary randomized trials, does the current PARO contract provide a framework for optimal patient outcomes, educational outcomes and resident wellbeing? The 2016-2020 PARO contract defines Maximum Duty Hours as in house call of maximum 24 hours plus adequate handover, for a maximum of 7/28 nights, and 2/8 weekend days over the same 28-day period. An additional 3 calls over 6 months can be given in case of emergency, after having asked for volunteers. Home call maximums are 10/30 nights with no consecutive weekends and must take into account days away. Maximum handover times are 1.5 hours in the ICU/CCU and 2 hours on other medicine services. If written advanced notice is given to PARO, residents may stay until an absolute maximum of noon on the day after a 24 hour call, but circumstances must be reviewed by a joint committee. Call conversion occurs if 4 consecutive hours are worked with at least 1 hour after midnight. For shift work, the ICU maximum is 60 hours and in the ER, one can work a maximum of five 12 hour shifts per week. Shift workers must have 2 weekends off per month and be off 12 hours between shifts. Educational rounds count toward maximums and absences must be taken into account. For blended call models, the following formula applies: (Home call x 3) + (In-hospital call x 4) cannot exceed 30 over a 28 day period.
In the end, resident work hours are but one aspect of the overall entity that is medical education today, but it is one that has the potential to have a profound impact on patient outcomes, educational outcomes and resident wellbeing. While regulations governing resident work hours started out being derived based upon public opinion and emotional considerations, much work has been done to develop robust data to guide future modifications. Many questions still remain, including “whether the existing work hour constraints, somewhat arbitrary in their derivation, are ideally suited to the diverse training requirements across specialties. Can we be sure we are preparing trainees to handle the challenges of practice in the real world where such protections clearly do not exist? And how might we design a system capable of fostering the morale of its workforce, while simultaneously sustaining the relationships that remain fundamental to the wellbeing of both patients and their doctors?”
In conclusion: Thanks Dr. Hazra for a very thoughtful appraisal of the emerging literature on the topic of resident work hours. The conversation around duty hours is (or at least should be) nuanced. To paraphrase Paracelsus, Work like medication can at high doses be toxic; conversely work hours restrictions can have unintended adverse effects!
I look forward to your feedback and thoughts.