How Much Do Residents Work?

Before deciding whether current resident work hours are excessive, it is useful to clarify how much residents actually work. Since 1988, the Accreditation Council for Graduate Medical Education (ACGME) has required that accredited internal medicine residencies comply with the following guideline:

[R]esidents must not be required regularly to per-form excessively difficult or prolonged duties. When averaged over 4 weeks, residents should spend no more than 80 hours per week of patient care duties in the residency program. Residents at all levels should be on call no more often than every third night and, on average, have the opportunity to spend at least 1 day out of 7 free of patient care duties in the residency program. [24].

Officially, almost all ACGME-approved internal medicine residency programs adhere to the work-hour limitations. According to the most recent available data, the mean number of hours of duty per week for first-year residents in 1992 was 71.2 in the 407 of the 418 accredited programs who responded to an American Medical Association survey [25]. Similarly, a 1990 survey of chief residents and program directors in 30 California teaching hospitals found that internal medicine residents worked an average of 74.4 hours per week [26]. However, some authors have questioned whether the reported compliance with ACGME guidelines is accurate [27]; in addition, this and similar surveys are probably affected by reporting bias. In fact, the chief resident survey itself found that 42% of internal medicine residents had violated ACGME requirements by working more than 80 hours per week [26].

Because more precise data are not available, it will be assumed that work hours for most internal medicine residents comply with the guidelines. However, the question remains: Is an 80-hour work week, with 24- to 36-hour shifts every third or fourth night, acceptable? On what grounds should work hours be shortened?

Effects on Patients

The most persuasive argument for reducing resident work hours would be that the practice harms patients. If residents' long work hours cause avoidable injuries, illnesses, and deaths, the practice would be professionally and ethically wrong. To date, however, neither the legal system nor the medical community has determined whether extended residency work hours actually harm patients; indeed, some even contend that they are beneficial.

The recent verdict in the Zion case is an example of this confusion. Libby Zion was an 18-year-old patient who unexpectedly died within 24 hours of admission to New York Hospital in 1984 [28]. Critics charged that her death could have been avoided if the residents caring for her had not been sleep-deprived and overworked, but a Manhattan grand jury did not indict the physicians or the hospital. It did, however, issue a report critical of the incident; this report led to the 1987 establishment of the Ad Hoc Committee on Emergency Services (known as the Bell Commission), whose recommendations ultimately led to legislative reform of residents' work hours in New York State. In February 1995, jurors issued a verdict in the civil lawsuit Zion vs. New York Hospital. The jury found New York Hospital not responsible for Ms. Zion's death but concluded that the hospital was negligent in assigning an intern too many patients, even though this negligence was not found to be the proximate cause of her death. The jury also concluded that the hospital was not negligent with regard to the manner in which its housestaff was supervised [29].

Ironically, the Zion case had its strongest effect on work hours, although neither the Bell Commission nor the 1994 jury who heard the Zion case ever concluded that sleep deprivation was the main cause of death. The Bell Commission concluded that inadequate supervision was the root of the problem, whereas the jury cited a too-great workload assignment [29, 30].

Research findings on the manner in which sleep deprivation affects resident performance have been well documented in several excellent reviews [1, 3, 12]. Well-rested residents have outperformed their sleep-deprived peers in many cognitive tasks, such as tests of rote memory, language, and numerical skills [31]; interpretation of electrocardiograms [4]; monitoring of anesthesia [32]; retention of information and problem solving [33]; tests of visual attention, short-term memory, and coding ability [34]; and tests of concentration [35]. Other studies, however, have shown no correlation between sleep deprivation and performance on tests of activities such as recall, concentration, and manual dexterity [11, 36].

Even if sleep deprivation adversely affects performance in a laboratory setting, it is not clear that such findings correlate with performance on the job. Indeed, one study that examined patient outcome rather than physician test performance reported that shorter hours (a decrease from 105 to 80 hours per week) could worsen patient care in the form of an increase in clinically insignificant complications and test delays [20].

Another way to understand the effect of residency training practices on patient care is to study the frequency and circumstances of mistakes [37]. In a survey, Wu and colleagues [38] asked 254 internal medicine residents about the most serious medical mistake they made during the past year. Of the 114 residents who responded, 51% believed that the mistake was caused by too many other tasks, and 41% attributed the mistake to fatigue. Similarly, another survey found that physicians-in-training attributed ethical errors to fatigue and overwork [39]. These self-reported surveys do not indicate whether mistakes are actually caused by sleep deprivation, but they do provide insight into residents' own assessment of the reasons for some medical errors.

In sum, although intuition may suggest that exhausted residents cannot function as well as rested ones, research has not proved this. Neither, however, does research show that long hours improve patient outcome by promoting continuity of care. Given such uncertainty, deciding whether it is wrong to require long work hours cannot be done solely on the basis of empiric evidence; other arguments for reform must be proposed.

Effects on Residents

Even if patients are unharmed by long resident work hours, other questions regarding the effect on residents remain. Does overwork and sleep deprivation harm residents? What are the physical and psychological effects? Is it always wrong to cause harm during training, or can some harms be justified by an overriding benefit?

Physical, Psychological, and Educational Effects on Residents

Ill effects of sleep deprivation and stress on residents during their training have been well described [1, 2, 5, 7, 40]. Most problems occur during internship, when as many as one in three interns experiences depression [41]. Interns commonly show substantial levels of anger and hostility [42], and sleep loss is associated with anxiety, tension, confusion, fatigue, irritability, inappropriate affect, memory deficits, and difficulty in thinking clearly [4, 43-45]. Female residents working more than 100 hours per week may be at increased risk for preterm delivery [46]. In one study, 1 in 20 of the 415 residents who took leaves of absence from their residency programs from 1979 to 1984 attempted suicide [47]. Sleep deprivation and long work hours strain family relations [10], and many partners of residents express frustration and despair about the unavailability of their spouses [14]. These negative aspects of residency seem to turn some medical students away from internal medicine; a national survey of medical students found that resident dissatisfaction and perceived overwork greatly contributed to students' decisions to switch from internal medicine to another specialty [48, 49].

Sleep deprivation also interferes with some educational objectives of residency. "Education," wrote educational philosopher John Dewey, "means the enterprise of supplying the conditions that insure growth" [50]. Common sense as well as established principles of adult learning suggests that education is optimized in a climate in which learners are physically and psychologically alert, comfortable, and rested [51]; this claim was supported by a study showing that protected sleep improves learning [52].

Benefits

Many physicians have defended long work hours because they believe such hours benefit the resident. For example, long hours and overnight shifts allow a resident to independently care for patients, an unusual opportunity during shorter day shifts. Even more basic is the precept in medicine that learning by doing is at least as important as learning by studying, and the more time a resident spends at work, the more he or she can see and do.

In 1981, Cousins [53] wrote an article criticizing the hardships of internship as a hazing ritual that "has long since outlived its usefulness" [53]. After this article was published, the journal's editors received "an avalanche of thoughtful commentary" [54], most of which supported long work hours. Supporters wrote that long and sleepless hours of solitary on-call work promoted responsibility, self-reliance, and confidence. Many noted that, in retrospect, their internships were rewarding and valuable. Unlike those who wonder whether "it is all worth it" [55], these physicians referred to their internships as "a fantastic learning experience" [56], and "one of the most rewarding experiences that I have ever been through" [57]; others said that long hours had helped them to become better physicians.

Some authors have tacitly accepted Cousins' characterization of residency as a hazing ritual and then defended it as a rite of passage into the medical profession [5, 53, 58], where self-sacrifice and suffering provide a meaningful way to formalize the change in status that occurs as novices become experts. It is argued that, similar to military boot camp, sacrifice in residency is integral to the process because it promotes group cohesion, emphasizes collegiality and bonding, solidifies social identity, and teaches humility in preparation for powerful social roles [58]. These characteristics are also common to other rites of passage [59, 60] and may be a ubiquitous aspect of initiation into healing professions across cultures [61].

However, it is important not to overstate purported benefits. One downside of a rite that promotes self-sacrifice as a virtue is that it may unduly promote the impression that physicians are somehow "martyrs to virtue" [58], which entitles them to prestige and economic privilege. This impression bolsters the belief that the profession is self-serving [62] and may contribute to the now commonplace "doctor-bashing" [63]. If the rite of passage primarily serves to teach shared responsibility, collegiality, and bonding, then less destructive means, such as requiring residents to spend a week together in the wilderness, could work equally well.

The argument that sleep deprivation prepares residents for future practice may also be overstated. In the past, when contagious diseases pervaded hospital wards and physicians were scarce, routine and heroic sacrifices may have been needed. But physicians now face more chronic and less curable illnesses; to care for patients with these illnesses, compassion and empathy are at least as important [64] as stamina and self-sacrifice. Obviously, there are times when we as a society encourage and expect self-sacrifice and heroic acts, and some self-sacrifice is probably important and necessary for physicians-in-training. However, it is an open question whether the daily practice of medicine during residency training should stand on the same footing as heroism, particularly when the grounds for justifying the sacrifice is that it serves the residents' own best interests.

Yet, it is unclear that harms to residents are typically permanent or extraordinary. Sleep deprivation affects persons differently, and one person's cause for depression may be another's source of self-esteem. Many residents claim to find the process rewarding, and some prefer longer hours [65]. Therefore, although the defense for long work hours is unconvincing, it also remains to be shown that the varied and temporary harms to residents are the best reasons for reforming residency training practices.

Residents as Cheap Labor

Perhaps the most plausible reason for requiring residents to work such long hours is that they are the least expensive physicians in teaching hospitals. Residents earn far less than staff physicians, and they provide 24-hour patient care while relieving staff physicians of many emergency and overnight responsibilities. Is it wrong to use physicians-in-training to serve the economic interests of hospitals and medical staff, when it might not serve the best interests of either patients or residents?

It is not clear whether residents save or cost hospitals money. Some have argued that residents cost more than alternative providers, particularly in a capitated medical setting. Compared with more experienced practitioners, residents order more tests and procedures, are less efficient, and contribute to longer hospital stays [66-69]. Such inefficiencies are one reason the same care is estimated to be 33% more costly at university teaching hospitals than at nonteaching hospitals [70].

However, the extra costs associated with teaching hospitals may be related to the fact that residents provide such varied services, many of which do not even require the expertise of a physician. In a time-motion study of internal medicine residents at two urban hospitals in New York City, Knickman [71] found that 19% of internal medicine residents' time was spent on activities that could be done by nurses, laboratory technicians, or other staff and that a meager 3.1% of their time was spent exclusively interacting with patients. When residents do function as physicians, much of their role is in the capacity of "complements"—performing tasks and services for which the attending physician (or hospital) receives the fee [72]; in this role, residents are relatively cheap [73, 74].

Residents could, of course, be replaced by alternative providers, but doing so would be expensive. Stoddard [75] calculated that the use of substitute practitioners to provide services currently provided by resident physicians would cost approximately $58 000 to $77 000 annually (1993 dollars) to replace each full-time equivalent resident.

Regardless of whether resident labor is cost-effective, it can be argued that it is not wrong to require long work hours because the residents provided informed consent to the terms of their employment [76]. The doctrine of informed consent maintains that free and informed choices by competent persons should, in general, be respected, and is grounded in the principle of respect for autonomy. This doctrine is the main reason the government can exercise only limited restraint on persons who engage in voluntary hazardous activities or on physicians who offer risky or experimental therapies to patients. One interpretation of informed consent is that consent justifies subsequent harm: Volenti not fit injuriae—with consent, there can be no moral injury.

When such logic is used, little is wrong with depriving residents of sleep and free time if they voluntarily sign a contract agreeing to work long hours. Perhaps the hazards of residency are similar to those of professional football: Participation involves risks, but the risk-taker has the opportunity to learn about the hazards, to choose not to participate, and to walk away. Of course, informed consent would not justify sleep deprivation during residency if it harms third parties such as patients, nor would it suffice if the consent results from coercion or when the hazards are egregious. However, as previously discussed, sleep deprivation does not appear to harm patients, becoming a physician is not compulsory, and medical students can learn about the stresses before starting residency.

Even if informed consent offers a plausible defense for using residents as inexpensive labor, what follows from this is fairly narrow because most objections to residency work hours transcend concerns about fairness in labor practices. Although one might say that tired residents should not complain about the working conditions to which they voluntarily agreed, consent should not be the last word in the debate. What matters most is whether such practices are a good idea, not whether residents have agreed to overwork. This question falls outside the purview of consent.

Overwork and Virtue in Medicine

The chief problem with residency overwork is not that it harms patients or physicians, nor that it is exploitative. Rather, it is wrong because it interferes with the promotion of virtuous professionals. Without question, one of the most important tasks of residency is to learn the basic values that underlie the clinical encounter, including compassion, empathy, honesty, integrity, and a commitment to serve the interests of patients. Physicians have aspired to learn such values since the time of Hippocrates [77], and professionalism requires this of physicians [78]. So central are these values and behaviors to the practice of medicine that the American Board of Internal Medicine requires residency program directors to evaluate the humanistic qualities of their residents [79]; an influential 1984 report stated that every physician should be caring, compassionate and dedicated to patients [80].

None of these mandates is new. Many years ago, Francis Peabody [81] wrote simply, "The secret of the care of the patient is in caring for the patient." The message was echoed by Roscoe Pullen [82]:

[T]he greatest single quality which the intern should develop is that of compassion for the sick, the afflicted and the suffering ... No single attribute of medical practice is more demanding, more difficult to acquire, and more exacting to maintain than the bond which exists between the patient and the doctor.

If the primary noncognitive training objective for residency is to promote virtue and professionalism in a physician's interaction with his or her patients, then long work hours must be measured against this goal. In this regard, several observations are appropriate. First, patients value humanistic traits in their physicians but believe that physicians lack such attributes [83]. Second, compassion in medicine appears to have declined as technological medical care has increased [84]. Third, the physician-patient relationship tends to become increasingly negative during the internship, even though medical schools have tried to enroll students who have humane core values [85, 86]. Fourth, residents report that sleep loss hurts their relationship with patients [87]. Although many factors contribute to these trends (including information overload, lack of free time, inadequate role models, and new responsibilities for patient care [88]), the effects of sleep deprivation on attitudes and values should not be underestimated.

For many residents, fatigue cultivates anger, resentment, and bitterness rather than kindness, compassion, or empathy. Embittered physicians may still be able to accurately analyze chest radiographs, and they may do procedures with technical proficiency, but we would not say that they are virtuous. As the following resident's entry in an intensive care unit diary shows, sleep deprivation can contribute to attitudes that differ from those admired by the profession:

1 AM and I'm ready to go to bed: one should never be ready to go to bed in the ICU—you'll always be disappointed. Anyway, I'm on my way to the EW ... when there's a code (cardiac arrest). Get up there and find (a resident) trying to intubate a lifetime asthmatic who is as blue as this ink. I keep thinking—he's blue enough to go to the ICU. I keep hoping he's going to be too blue to go anywhere. Probably a nice man with a loving wife and concerned children, but I don't want that SOB to make it because I've got one special who is going to keep me up 2 more hours. I don't need an intubated, blue, pneumothoraceed SOB coming to my unit ... I don't want the asthmatic SOB to live if it means I don't sleep. I don't want the special to live if it means I don't sleep. I just want to sleep. [58]

How do residents learn to view patients as the objects of disdain? One suggestion is that such attitudes occur when meeting a patient's needs becomes incompatible with meeting the resident's own needs. As described by Mizrahi [89], this attitude often develops into a desire to "get rid of patients," a pervasive phenomenon during residency.

Do these observations prove that sleep deprivation and overwork cause dehumanized medical care? Obviously not. Some house officers who work the longest hours manage to maintain humanistic and caring relationships with patients, whereas others who are the least hardworking can be distant and aloof. The relative paucity of data specifically addressing the effect of sleep deprivation on the delivery of humanistic care amplifies the need for additional empiric research.

However, we also know from anecdotal and descriptive analyses that inadequate sleep interferes with some residents' abilities to "care" about patients [58, 61, 89]. Even if cynicism, anger, and bitterness are temporary, one study showed that moral reasoning is an accurate predictor of clinical performance [90]. If attitudes really matter, then those who are training residents must identify and remove barriers to desired attitudes. For this reason, it makes little sense to demand that residents be compassionate to others when they themselves receive so little compassion from their own colleagues [61, 91].

Some maintain that economic necessity nevertheless requires that residents continue to work long hours. But this line of reasoning is also problematic because it sends a particular moral message: that economic concerns justify overworking physicians, even if it embitters them to patients. If young physicians internalize that message, it will undoubtedly affect attitudes and behaviors after residency by encouraging them to value economics more than patients; this attitude is precisely the opposite of the values that should be encouraged.

Aristotle long ago suggested that the best way to cultivate desirable values is by habit and training. Because habits, both good and bad, are learned by example, residents must be exposed to exemplary mentors and role models [92, 93]. But this is not enough; residents must also be provided the working conditions that promote desired values. As Mark Twain ironically noted, Few things are harder to put up with than the annoyance of a good example.

Conclusions

What does this analysis offer the debate on residency reform? Arguments for shortening work hours have tended to rely on the assumptions that long hours harm patients and residents. Neither assumption seems to be consistently true, although in specific cases, adverse effects undoubtedly occur. The primary adverse consequence of prolonged work hours seems to be that overtired housestaff develop undesirable and unprofessional attitudes. I have argued that attitudes matter because they form the moral basis on which a physician's relationship with the patient rests.

What should be done? Reform should focus on changing the institutional ethos that promotes cynicism and dislike of patients, while addressing the root causes with openness and honesty. Reynolds [91] has argued that much of the problem is that long work hours have been defended as educational, when the real reason for continuing the practice is to meet the service needs of hospitals. If she is correct, this justification is a departure from articulated educational standards. It suggests that the overriding purpose of residency is not to train competent and compassionate physicians or to provide the best possible care to patients, but to staff hospitals with affordable physician labor. It is not self-evident that such a goal is the best one for residency training.

Residents, especially when fatigued, need adequate supervision; it is unfair and perhaps unsafe for attending physicians to shift all work and responsibility to them. Shortening work hours could help eliminate some attitude problems that develop during residency. However, as experience with reform in New York has shown [20, 94], efforts to solve existing problems can create new difficulties and changes should therefore not take place carelessly. Creative alternative staffing methods must be examined, including more widespread use of night floats, nonteaching services, nurse practitioners, physician assistants, and additional ancillary care to relieve some of the noneducational burdens of patient care. Implementing such changes will be complex, but it remains more desirable for the profession itself to initiate change rather than to wait for changes to be imposed from the outside. Keeping in mind the broad goals of residency training, it is hoped that thoughtful reform will lead to improved medical care while satisfying patients and residents.