Sleep Deprivation, Physician Performance, and Patient Safety FREE TO VIEW

Extended work hours are rooted in the around-the-clock nature of the hospital medical practice, especially during internship and residency training. Although educational, professional, and economic arguments have been made to justify these demanding schedules, evidence increasingly reveals their potential negative impacts on the well-being of doctors and their patients. We summarize the processes that regulate sleep and wakefulness as they relate to health-care providers, describe newer evidence of risks posed by physicians' sleep loss, and discuss the countermeasure of resident/intern work-hour restrictions.

Sleep is a biologically imperative but physiologically enigmatic state. The ventrolateral preoptic nucleus of the anterior hypothalamus is critical for the initiation of sleep,¹ whereas wakefulness relies on a system of ascending brainstem and hypothalamic neurons.² The ventrolateral preoptic nucleus and arousal system inhibit one another, thus acting like an electronic “flip-flop” switch,³ resulting in rapid and stable vacillations between wakefulness and sleep.

The position of the sleep/wake switch and accompanying neurobehavioral states are governed by the interplay of two variables: the sleep-dependent homeostatic process and the sleep-independent circadian system⁴ (Fig 1). The homeostatic drive to sleep builds as time from initial wakefulness ensues, which results in progressively increasing sleep propensity and concomitant degradations in alertness and performance. The circadian system, centered in the suprachiasmatic nuclei of the anterior hypothalamus, oscillates over a period of approximately 24 h and determines the rhythmicity of sleep/wake states with respect to its main synchronizer, the environmental light-dark cycle. Personal tendencies to wake up early or stay up late (morningness or eveningness traits) underscore intrinsic circadian preferences between individuals.⁵ Under conventional circumstances, the interaction of the sleep homeostatic and circadian processes results in consolidated and behaviorally optimized periods of wakefulness and sleep.⁶

A third process, sleep inertia, may be another important feature of the organization of wake/sleep states.⁷ This entity is characterized by the desire to return to sleep on awakening, and it can be accompanied by varying degrees of grogginess, slurred speech, impaired cognition, and automatic behavior. The magnitude of sleep inertia is greatest within the first minutes of awakening, but its impact may extend several hours.⁸

Acute continuous and chronic partial sleep deprivation, ubiquitous among health-care providers, intensifies the impact of the sleep homeostatic process and impairs neurobehavioral functions. Acute sleep deprivation results in dose-dependent impairments of mood, cognitive performance, and motor skills well within 30 h of continued wakefulness,^9,10 the current shift cap for US resident physicians. After 10 and up to 26 h of wakefulness, hourly declines in cognitive psychomotor performance are similar to those observed with a progressive 0.004% rise in blood alcohol concentration, such that by 17 h the performance impairment is equivalent to that observed with alcohol intoxication in healthy subjects.¹¹ Long, vigilant-dependent tasks involving newly learned skills are most vulnerable to acute sleep loss, with accuracy maintained at the expense of efficiency.¹²

In comparison, chronic partial sleep deprivation (or insufficient recovery sleep) results in dose- dependent attention lapses, depressed mood, and reduced cognitive performance.¹³ In < 2 weeks, restricting sleep to 6 h per night results in vigilance deficits that are equivalent to 24 h of total sleep deprivation.¹⁴ Particularly disquieting is the observation that while performance measures steadily decline, subjective sleepiness ratings quickly plateau, suggesting that subjects perceive they are acclimating to sleep loss, despite objective evidence to the contrary.¹⁴

The impact of sleep loss on sleep/wakefulness can be modified by intrinsic and extrinsic factors.¹⁵ With respect to the former, there are significant interindividual differences in the global response to sleep loss, as well as significant intraindividual variations in the degree to which different domains of neurobehavioral function (eg, vigilance, subjective sleepiness, and cognitive performance) are affected.¹⁶ A variety of contextual factors, including motivation, environmental light and noise, posture, and drugs, can also temporarily modulate sleep homeostasis and circadian rhythmicity.¹⁵

Extended work shifts, fragmented sleep due to actual¹⁷ or threatened¹⁸ interruptions, moonlighting (at times, to repay education loans), concurrent primary sleep disorders, and insufficient recovery sleep contribute to acute and chronic sleep deprivation in physicians, particularly those in training. Night shifts force physicians to work at the circadian acme of sleep propensity, which may jeopardize waking function, and rotating work shifts disrupt circadian alignment and sleep consolidation. Naps may be an effective countermeasure,¹⁹ but when accomplished in the setting of sleep deprivation and/or increased circadian sleep propensity, sleep inertia may impair performance on awakening.²⁰

Several excellent reviews,^12,21,22 published on the eve of the Accreditation Council for Graduate Medical Education's (ACGME) decision to limit resident duty hours, summarized the existing literature on the effects of sleep loss on trainee performance, learning, errors, and personal well-being. and included some eye-opening reports. For instance, 41% of 254 internal medicine residents cited fatigue as a cause of their most significant medical mistake, with a third resulting in a patient fatality.²³ Surgical residents were found to make up to twice as many errors during simulated laparoscopy post-overnight call than after a night of sleep, demonstrating the deterioration of fine motor skills with sleep deprivation.^24,25 Anesthesia residents demonstrated sleepiness similar to narcoleptics even with no call duty over the preceding 48 h.²⁶ Nevertheless, some studies generated conflicting data, and many were methodologically flawed, perhaps not surprising given the inherent complexities of conducting research in this area.

The following studies are used to highlight newer endeavors to examine whether occupational perturbations of the sleep homeostatic and circadian systems are detrimental to physicians and their patients. The studies are grouped according to their relative emphases on acute sleep deprivation, chronic partial deprivation, or circadian issues, recognizing that it is very difficult to isolate these issues in the complex milieu of medical training. All studies involve trainees and, despite their pervasive limitations, suggest deleterious effects of long work hours on numerous assessed variables.

Howard and colleagues²⁷ studied the responses of anesthesiology residents under rested and acute sleep deprivation conditions to psychomotor testing, randomly presented vigilance probes, and planned clinical events during a 4-h simulated case. Sleep-deprived residents demonstrated progressive psychomotor deficits in vigilance and memory, they were slower to respond to vigilance probes, and nearly one-third of residents fell asleep during cases. Notwithstanding, there were no group differences in clinical skills or error rates, indicating that extrapolation from psychomotor testing into the complex realm of clinical performance is challenging and the use of group data may obscure individual differences in sleep loss vulnerability.

Landrigan and colleagues²⁸ prospectively examined the effects of a modified schedule (shifts no longer than 16 h) vs a traditional extended shift schedule on intern errors in the ICU setting, as detected by a comprehensive multidisciplinary observation system. Greater than 2,000 patient-days were monitored. The intervention schedule resulted in 19.5 h less work and 6 h more sleep per week, including more sleep in the 24 h that preceded each working hour.²⁹ Interns on the traditional schedule made 36% more serious medical errors, including 21% more serious medication errors, and 5.6 times more serious diagnostic errors compared with the intervention schedule. Neither the number of medications ordered nor number of tests interpreted differed between the two schedules. Procedural error rates were similar. Interns in the traditional schedule experienced twice the rate of EEG-documented attentional failures while working during on-call nights.²⁹ There were no differences in adverse event rates between the groups,²⁸ indicating that efforts to discern undesirable patient outcomes from physician sleep deprivation must account for the multidisciplinary support on-call providers receive from colleagues and allied staff.

During 2002 to 2003, > 2,700 interns nationwide (15% of the students matched via the National Residency Matching Program), mostly from internal medicine programs, completed monthly reports via the Internet that included questions about work, work hours, and sleep. Sleep and work-hour reports were validated by daily diaries, direct observation, and polysomnographic recordings in a randomly selected subgroup, and corroborative information was sought about accidents and injuries. Approximately 17,000 monthly reports were completed. Interns reported an average of 71 h of hospital work per week and four extended shifts per month of a mean duration of 32 h. The survey resulted in three publications^30–32 relating work hours to motor vehicle crashes, occupational injuries, and self-reported medical errors. The odds of a motor vehicle crash or a near-miss incident during the home commute after an extended work shift was increased by twofold and fivefold, respectively, compared with those working nonextended shifts.³⁰ Every extended work shift scheduled per month increased the risk of any motor vehicle crash by 9%, and the risk of crash on the home commute specifically increased by 16%. The odds ratio of sustaining a percutaneous injury the day after working overnight was 1.6, compared with an identical time frame on the previous day.³¹ Lapses in concentration and fatigue were the primary reasons cited by interns for these occupational injuries. Self-reported fatigue-related medical errors were reported during 4% of months with no extended shifts vs 10% of months with one to four extended work shifts and 16% of months with five or more extended work shifts.³² In comparison to months with no extended shifts, the odds ratio for fatigue-related preventable adverse events was 7 for months with five or more extended work shifts. The probability of falling asleep during educational activities and duty, including during surgery, increased in proportion to the number of extended shifts to which one was exposed. The uncertainty of the degree to which acute and chronic partial sleep loss are contributory poses limitations in interpreting these data.

In an effort to summarize the impact of acute sleep deprivation on physicians, Philbert³³ performed a metaanalysis of the existing literature and concluded that sleep loss in the range of 24 to 30 h reduces overall and clinical performances by nearly 1 and 1.5 SDs, respectively, comparable to the 15th percentile capacity of a rested group.

Chronically reduced sleep duration in trainees reflects societal trends. The average self-reported sleep duration among adults in the United States is 6.7 h on weekdays,³⁴ an amount that has been trending downward.¹³ A large national survey of residents across specialties (before the implementation of the ACGME duty hour restrictions) revealed mean self-reported nightly sleep durations of < 6 h.³⁵ These results are concerning, given the mounting evidence linking chronic sleep curtailment with obesity³⁶ and increased mortality.³⁷

Surveys have been frequently used to assess the degree to which chronic insufficient sleep influences the personal and professional lives of residents. The resident survey by Baldwin and colleagues³⁵ indicated that those working extended hours were more likely to report serious conflicts with attending physicians, other residents, and nurses, in addition to increased alcohol use and instances of unethical behavior. Using semistructured focus groups and questionnaires on trainees from multiple specialties and institutions, Papp and colleagues³⁸ found that the majority of residents described profound adverse effects of sleep loss on job performance, personal life, learning ability, and motivation. Sixty-four percent of residents agreed or strongly agreed that “Sleep loss and fatigue have a major impact on my personal life,” yet just 46% agreed or strongly agreed with the statement “Sleep loss and fatigue have a major impact on my work.” Rosen et al³⁹ surveyed internal medicine interns at baseline and at the end of internship, and they demonstrated a strong association between the development of chronic sleep deprivation and moderate depression in those who endorsed neither at the start of training. Admittedly, survey data are subject to recall biases and cannot readily determine the impact of other occupational and life stressors on interpersonal and professional difficulties.

Seeking to quantify the effects of work rotations with repeated night calls, Arnedt and colleagues⁴⁰ examined the neurobehavioral performances of post-call residents during an inpatient rotation. Study participants performing call every fourth or fifth night were compared with those involved in an outpatient rotation with limited call. Performance results were described relative to the impact of alcohol intoxication. Impairments in sustained attention, vigilance, and simulated driving during the heavier call rotation were comparable with those associated with a 0.04% to 0.05% blood alcohol concentration during a light call rotation, and residents' ability to judge this impairment was limited. The relevance of these findings to actual clinical performance is not clear.

Newer data also emphasize the importance of circadian factors with respect to health-care provider outcomes. Self-reported cutaneous injuries are twice as common in interns during the nighttime compared with the daytime.³¹ A survey of New Zealand junior doctors indicated night work and schedule instability were independent risks for fatigue-related medical errors, whereas total hours worked per week were not.⁴¹ However, the impact of circadian factors on patient safety is not fully known. Morales and colleagues⁴² reported that nighttime admission to a medical ICU was not associated with a higher mortality rate or longer hospital or ICU stays compared with daytime admission.

Under pressure from the American Medical Student Association and threat of federal intervention, the ACGME imposed nationwide work-hour limits in July 2003 (Table 1), similar to limitations enacted for medical trainees in New York State in 1989. The ACGME limits currently cover approximately 105,000 residents and fellows in > 8,300 training programs in the United States. During the 2006–2007 academic year, 227 (8.8%) of the 2,589 programs reviewed by the ACGME received one or more citations related to noncompliance, the highest annual percentage to date.⁴³ Family practice had the greatest number of programs cited,²⁵ and emergency medicine had the highest percentage of programs in violation (54%) (of specialties with more than one program reviewed). The most frequent area of noncompliance remains the “24 + 6” h limit on continuous duty.

Work-hour limits have fueled a vigorous discussion of the optimal balance of continuity of patient care, physician well-being, and trainee education. Do work-hour limits result in healthier, more energetic trainees who are more receptive to learning, have more opportunities for scholarly pursuits, and are more capable of providing safer patient care? Or do the work limits erode professionalism by the engenderment of a “shift work mentality,” the reduction of opportunities for didactic and longitudinal clinical education, the fragmentation of patient care, and the foisting of work on senior clinicians? In light of the limitations of suboptimal study designs and contradictory results^44,45 from the literature to date, these questions and concerns are bound to linger, but some common themes have emerged.

First, several large studies have revealed neutral or slightly positive effects on patient mortality and other select clinical parameters since implementation of the ACGME work-hour limits.^46–49 Second, surveys of trainees across specialties indicate improved quality of life but varied effects on perceived patient care and learning.^50–53 Third, surveys of program directors and clinical faculty reveal many negative views.^54–56 Faculty consistently perceive that duty hour limits have worsened residents' educational experiences, hampered continuity of care, and resulted in the faculty incurring more clinical work, with comparatively limited time for teaching and research. Objective outcome data against which to compare the perceptions judged by surveys are lacking in all of these studies.

Proponents of work-hour limits argue that the current ACGME rules do not go far enough. As already indicated, performance deficits are demonstrable within the current limits, and the ACGME rules are not as restrictive as those for workers in other safety-sensitive industries. For example, the maximal allowable flight time by the Federal Aviation Administration for a single pilot voyage is 8 h,⁵⁷ and the newly released “hours of service” rules by the Federal Motor Carrier Safety Administration limit driving time to 11 h/d and 60 h/wk for commercial motor vehicle drivers, effective January 2009.⁵⁸ In addition, the European Working Time Directive calls for a 48 h/wk limit for medical trainees, starting in August 2009.

Regardless, limiting work hours is not a singularly sufficient strategy to address sleep deprivation among physicians. To counteract chronic sleep restriction, trainees must use increased time away from work to increase their sleep duration. A Web-based nationwide survey⁵⁹ of residents in 2004 demonstrated an increase in mean nightly sleep duration of only 22 min (from 5.91 h to 6.27 h) despite falls in overall weekly work hours and durations of extended shifts. Even when a sufficient amount of total sleep time is achieved, the presence of concurrent sleep disorders can impair its restorative quality; these conditions should therefore be thoroughly assessed and properly treated if identified (eg, obstructive sleep apnea). When acute sleep deprivation is unavoidable, strategic caffeine use or napping are helpful countermeasures.

From a more general cultural standpoint, a greater emphasis must be placed on cultivating a team-based ethos with clearly defined roles, or a “New Professionalism,” in the words of Van Eaton and colleagues.⁶⁰ Crucial to this approach is a robust communication system between providers because transfers of care inevitably increase in an environment of work-hour limitations. Unfortunately, comprehensive sign-out procedures are lacking in many programs.⁶¹ Other proposed changes to facilitate the safe and expedient transfer of patient care include the increased use of nonphysician professionals to perform certain tasks and the scheduling of shift changes in a “circadian-friendly” manner (days → evenings → nights), to minimize provider fatigue.

The Institute of Medicine's Committee on Optimizing Graduate Medical Trainee (Resident) Hours and Work Schedules to Improve Patient Safety has issued a report with suggestions to amend the ACGME work hour limits in order to balance safety considerations for trainees and their patients.⁶² Although the committee recommends maintaining the 80-h workweek, it calls for a maximum work shift duration of 16 h, an uninterrupted 5-h break for sleep some time between 10:00 pm and 8:00 am during shifts that must last up to 30 h, no new admissions for residents after 16 h of work, variable time off between shifts in proportion to the timing and duration of work shifts, curtailment of moonlighting by counting both internal and external shifts against the 80-h weekly limit, and a mandatory 5 days off per month. The committee also calls for enhanced resident access to their supervisory physicians, scheduled shift changeovers to promote patient care transfers, and creation of medical specialty-specific guidelines detailing the number of patients that residents should care for based on resident competency and patient complexity. Tougher monitoring of duty hours by the ACGME is also called for, a move that seems prudent because 84% of interns reported at least one work-hour violation during ≥1 month in a Web-based survey,⁵⁹ a percentage markedly higher than that of programs annually cited by the ACGME for work-hour violations. The estimated cost of implementing the recommendations is $1.7 billion, a key limitation acknowledged by the committee.

A major group that is overlooked in the discussion of work hours and sleep deprivation is physicians in practice. At present, there are no national work-hour guidelines for physicians beyond training. Precisely how frequently established physicians exceed the ACGME work-hour limits is not known, but 14% of faculty within a single academic institution were errant in one survey study.⁶³ The impact of the work hours of established physicians on patient care is similarly unknown, as is the extent to which older physicians are more vulnerable to longer work hours, or whether recent residency/fellowship graduates continue to adhere to work-hour limits in practice. More data are sorely needed to address these issues.

Sleep and wakefulness are governed by homeostatic and circadian processes that are challenged by the number of consecutive work hours, inadequate recovery sleep, and overnight shifts of physicians. Acute and chronic partial sleep deprivation results in dose-dependent impairments of mood and cognitive performance, with limited personal insight. Circadian alterations disrupt sleep architecture and consolidation. Increasingly robust evidence demonstrates that sleep loss degrades physician performance and well-being. Prolonged work shifts are associated with more medical errors, greater attentional failure, and an increased frequency of self-reported physician motor vehicle crashes and occupational injuries. Nevertheless, the impact of physician sleep loss on patient safety has not been fully elucidated.

The ACGME has responded to mounting concerns about physician sleep loss by limiting work hours for physicians in training, but this mandate has been contentious. Opponents argue that work-hour limits threaten resident professionalism, education, and patient care delivery. Proponents argue for tighter limits. Attending faculty and residents have discordant views about the impacts of the work-hour limits on patient care and education, yet data so far do not reveal major changes in patient mortality under the new rules.

We view the work-hour limits as a necessary, plausible, and positive step in the evolution of graduate medical education. However, these limits are not above scrutiny and must be paired with changes in resident behavior and workplace safeguards to ensure quality of patient care, a view emphasized by the recent report of the Institute of Medicine.⁶²

Author contributions: Drs. Olsen, Drage, and Auger all contributed to the planning, writing, and editing of the manuscript.

Financial/nonfinancial disclosures: The authors have reported to the AACP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.