Received from the Anesthesia Service, VA Palo Alto Health Care System, Department of Anesthesia, Stanford University School of Medicine, Palo Alto, California; the Department of Anesthesiology, Yale University School of Medicine, New Haven, Connecticut; and the Department of Anesthesiology, Emory University School of Medicine, Atlanta, Georgia.HEALTHCARE delivery takes place 24 h a day, 7 days a week, and is colloquially termed a “24/7” operation. Anesthesia providers are required to deliver critical around-the-clock care to a variety of patients. This parallels the situation in many other domains that provide such services, e.g ., transportation, law enforcement, communications, fire fighting, technology, manufacturing, and the military. Even “convenience” industries (e.g ., gas stations and grocery stores) now provide uninterrupted access. These continuous operational demands present unique physiologic challenges to the humans who are called on to provide safe operations within these systems. Human physiologic design dictates circadian patterns of alertness and performance and includes a vital need for sleep. Human requirements for sleep and a stable circadian clock can be, and often are, in direct opposition to the societal demand for continuous operations.Recently, patient safety has taken center stage in health care. The Institute of Medicine's report “To Err Is Human: Building a Safer Health System,” revealed that medical errors contribute to many hospital deaths and serious adverse events. 1The response to this report was widespread and included the Quality Interagency Coordination Task Force's response to the President of the United States, “Doing What Counts for Patient Safety: Federal Actions to Reduce Medical Errors and Their Impact.”∥2This report listed more than 100 action items to be undertaken by federal agencies to improve quality and reduce medical errors. One action promised by the Agency for Healthcare Research and Quality was “the development and dissemination of evidence-based, best safety practices to provider organizations.” In addition to the multiple recommendations to improve patient safety, the report from the Agency for Healthcare Research and Quality included a review chapter on sleep, fatigue, #and medical errors. **There is evidence that the issue of fatigue in health care is coming to prominence on a national level. In April 2001, Public Citizen (a consumer and health advocacy group) and a consortium of interested parties petitioned the Occupational Safety and Health Administration to implement new regulations on resident work hours (table 1). The primary intent of the regulations is to provide more humane working conditions, which the petitioners declare will result in a better standard of care for all patients. Also, the Patient and Physician Safety and Protection Act of 2001, which would limit resident physician work hours, was introduced in Congress. Recently, the Accreditation Council on Graduate Medical Education, the accrediting organization for residency training programs in the United States, has approved common program requirements for resident duty and rest hours that will take effect in July 2003. ††The potential impact of sleep loss and fatigue, specifically among anesthesiologists, has received only sporadic attention. 3,4The cognitive demands of intraoperative patient care requires an iteration of data collection, evaluation of its relevance to patient status, development and implementation of plans to maintain the desired patient status, and monitoring the outcome of interventions. These complex tasks require sustained attention or “vigilance” and are particularly vulnerable to the effects of fatigue. 5–8The purpose of this article is to review the physiology of prolonged work cycles and fatigue, to relate this to the work milieu of the practice of anesthesiology, and suggest economically feasible recommendations to mitigate the effects of fatigue.Sleep loss and disruption of circadian rhythm that result from arduous work schedules can lead to reduced safety, performance, and health. While some of these outcomes are well documented, much remains to be learned about the short- and long-term effects of sleep and circadian disruption. The following nonmedical examples of the safety, performance, and health risks associated with around-the-clock operations illustrate the increasing human and economic costs related to ignoring the effects of these physiologic disruptions.There have been several high-profile accidents where fatigue was identified as either causal or contributory. For example, although alcohol is often cited as the central reason in the Exxon Valdez marine grounding, the National Transportation Safety Board investigation identified fatigue as one of the probable causes of the accident. 9Similarly, circadian factors were identified as contributing to the errors that resulted in the nuclear accidents at Three Mile Island and Chernobyl. 10,11Fatigue resulting from the work–rest patterns of managers was also acknowledged as an important component of the flawed decision-making that contributed to the space shuttle Challenger accident. 12Fatigue-related accidents have been identified in every mode of transportation and can be found in many around-the-clock operational settings. Clearly, there are a variety of adverse outcomes such as economic costs, disrupted service, injuries, and even fatalities that result from these accidents. For example, the Exxon Valdez grounding was associated with environmental cleanup operations and legal cases involving billions of dollars, and Space Shuttle operations were suspended for several years after the Challenger disaster.Fatigue-related safety risks affect us at both individual and societal levels. A recent poll by the National Sleep Foundation indicated that one of two drivers reported having driven while drowsy in the past year, ‡‡and one of five acknowledged having “nodded off” while driving. Fatigue contributes to 100,000 crashes annually that result in 76,000 injuries and 1,550 fatalities, according to estimates by the National Highway Traffic Safety Administration. 13Recently, an international group of scientists estimated that fatigue is causal in 15–20% of all transportation accidents, that official statistics underestimate the scope of the problem, and that fatigue exceeds the combined contribution of alcohol and drugs in transportation accidents. 14Fatigue caused by sleep loss and circadian disruption can degrade performance and reduce many aspects of human capability. 15Known performance effects include reduced attention–vigilance, impaired memory and decision-making, prolonged reaction time, and disrupted communications. 16–20These degraded performance outcomes create a situation where there is increased risk for the occurrence of errors, critical incidents, and accidents. 15Fatigue also creates increased performance variability, with cyclic reductions in alertness and performance. 21Fatigued workers have a tendency to slow down work processes to maintain accuracy, a classic effect known as the speed-accuracy trade-off. 22It takes only a moment of reduced performance during a critical task to have a negative outcome. Even if a lapse in performance occurs during a noncritical task, the system vulnerability shifts to a less safe state.Fatigue-related accidents are sometimes considered to be a result of falling asleep. Performance gaps can be the result of these “microsleeps,” which are brief, uncontrolled, and spontaneous episodes of physiologic sleep. 8There can be significant performance reductions that are sufficient to create safety risks prior to and immediately after the occurrence of a microsleep. 23,24Slowed cognitive throughput, reduced memory, slowed reaction time, lowered optimal responding, and attention lapses can create an increased opportunity for errors to occur. 25Consider the circumstance where an anesthesiologist's response to an alarm is slowed and an inappropriate decision guides an incorrect action. The practitioner may have been “awake,” but fatigue-related performance decrements could be contributory to the occurrence of any error, incident, or accident that resulted from the action.The decrement in psychomotor performance resulting from sleep deprivation have been correlated with those resulting from the impairments associated with ethanol ingestion. 26Performance on a hand–eye tracking task declined such that the impairment was equivalent to a blood alcohol level of 0.05% after 17 h of wakefulness. At 24 h of sustained wakefulness, the impairment in psychomotor function was equivalent to a blood alcohol concentration of 0.1%, at or above the legal limit for driving in most states. These data could be useful to help quantify fatigue-related effects with a drug that the public and policy makers better understand.Specific clinical skills of importance to the practice of anesthesiology deteriorate as a result of fatigue. On a simulated monitoring task where subjects were asked to monitor and record the time of significant deviation of clinical variables (e.g ., heart rate, blood pressure), Denisco et al . reported lower “vigilance scores” in the group that had been on call. 27The ability to interpret electrocardiographic changes and to do simple mathematical calculations is compromised among sleep-deprived house officers. 28The speed and quality of intubation was diminished among emergency department physicians working the night shift as compared with their performance while working during the day. 29,30Many of the fatigue-related decrements in performance identified in residents are potentially worse in older physicians. Aging is associated with a tendency toward early awakening, an exaggerated dip in arousal midafternoon, and a decreased tolerance of late-night and shift work. 31The unique demands of night call on older anesthesiologists are more onerous than those found in other specialties. 32Among recently retired anesthesiologists, night call was identified as the most stressful aspect of anesthetic practice and the most important reason for retirement. 33,34Beyond the safety risks and performance decrements associated with sleep loss and circadian disruption, there are a variety of personal health concerns. Several studies have shown that long-term exposure to shift work represents an independent risk factor for the development of both gastrointestinal and cardiovascular diseases. 35–39A recent study found that women working the night shift had a 60% greater risk for breast cancer compared with women who never worked the late shift. 40There is evidence that some adverse pregnancy outcomes are related to working conditions. 41A meta-analysis of 29 studies, including more than 160,000 women, evaluated the association of physically demanding work, prolonged standing, long work hours, and cumulative “fatigue score” with preterm delivery, pregnancy-induced hypertension, and small-for-gestational-age infants. There was a positive association between physically demanding work and preterm births, pregnancy-induced hypertension, and delivery of small-for-gestational-age infants. Shift work alone was found to increase the incidence of preterm births. 41There is evidence that sleep restriction alters physiologic function. Significant detrimental effects on immune function can be found after a few days of total sleep deprivation or after several days of partial sleep loss. 42,43Sleep restriction of 4 h per night for six nights is associated with harmful effects on carbohydrate metabolism and endocrine function. 44This degree of sleep restriction resulted in abnormal glucose tolerance, decreased thyrotropin concentrations, increased evening cortisol concentrations, and increased sympathetic nervous system activity (as measured by heart rate variability). Sleep deprivation and circadian disruption affect cerebral metabolic and cognitive function. In a study of changes in regional cerebral glucose utilization (i.e ., positron emission tomography) during 85 h of consecutive sleep loss, decreases in cerebral metabolic rate were observed primarily in the thalamus and prefrontal and posterior parietal cortices. Alertness and cognitive performance declined in association with these brain deactivations. 45A recent study of aircrew members suggests there may be a linkage between long-term exposure to time-zone changes (i.e ., circadian disruption), temporal lobe atrophy, and deficits in learning and memory. 46Investigations using functional magnetic resonance imaging technology contradict some of the aformentioned findings and reveal compensatory changes of increased activation in the prefrontal cortex and parietal lobes during verbal learning after sleep deprivation. 47–50Studies show altered mortality with sleep loss and circadian disruption. Circadian disruption in hamsters and Drosophila reduce life span from 11 to 15%. 51,52A prospective investigation of more than one million individuals conducted by the American Cancer Society found that men who reported “usual” daily sleep times of less than 4 h were 2.8 times more likely to have died within a 6-year follow-up as men who obtained 7.0–7.9 h of sleep. 53The risk for women was increased by 48%. Conversely, men and women who reported sleeping 10 h or more per day had about 1.8 times the mortality rate of those who reported 7.0–7.9 h of sleep.The two primary determinants that underlie fatigue and interact in a dynamic manner are sleep homeostasis and circadian rhythms. 54An individual's level of alertness (e.g ., on the job) or potential for sleep (e.g ., during a rest period) will be determined by a complex interaction of these factors. Performance and alertness decrements may occur when either of these elements is disrupted. 55Factors other than fatigue, such as workload, environment, stress, boredom, motivation, and professionalism, also influence the ability to perform. 4In addition, there are large interindividual differences on the effects of fatigue. 56Sleep serves a vital physiologic need. 57Like other basic physiologic requirements such as food and water, sleep plays a fundamental role in survival. Sleep homeostasis is the balance between sleep need and quality and quantity of sleep obtained by an individual. On average, the adult human requirement for sleep appears to be greater than 8 h (8 h:14 min) per 24-h period. 58,59The range of sleep need varies from 6 to 10 h, and this requirement is probably genetically determined and cannot be “trained” to a different sleep need. 60Estimates suggest that most American adults obtain about 1–1.5 h less sleep than needed. §§This lost sleep accumulates to produce a “sleep debt.”8,58For example, an individual who obtains 1.5 h less sleep per night over a 5-day work week will begin the weekend with 7.5 h of sleep debt. This deficit is roughly equivalent to the loss of a full night of sleep and requires about two nights of at least 8 h of sleep for recovery. 20Sleep debts are not repaid hour for hour, but instead through an increase in deep sleep or nonrapid eye movement stages 3 and 4. 20A variety of factors can affect sleep quantity and quality. Perhaps some of the most dramatic changes in sleep occur as a normal function of aging. Approaching age 50 and beyond, sleep becomes more disrupted with frequent awakenings. There are reduced amounts of deep sleep, and sleep becomes less consolidated. 61Nocturia in men and menopausal symptoms in women are likely to contribute to sleep disturbances in older individuals. There are also age-related increases in complaints of insomnia and depression that negatively impact sleep. Sleep need does not necessarily decrease with age, and increased daytime sleepiness can be the consequence of reduced sleep quantity and quality. There have been no formal studies assessing whether these changes in sleep quantity and quality affect the performance of older anesthesia providers.There are approximately 90 known sleep disorders that have been described and classified in a diagnostic nosology. 62The causes for these disorders range from physiologic to psychological to environmental. Some sleep disorders are relatively prevalent in the population and have well-documented negative effects on waking alertness and performance. 63–65Often, the affected individual is unaware of their disorder, and the bed partner may be the first to identify the problem. Obstructive sleep apnea is a common example of a sleep disorder that has implications in operational settings. There are many health consequences associated with sleep apnea, but, in addition, it has been shown to be associated with a twofold to sevenfold increase in risk for automobile accidents. 66,67Consistent with this, Powell et al . demonstrated that individuals with mild to moderate sleep apnea had a decrement in performance equivalent to that of an individual with a blood alcohol concentration of 0.05–0.08 g/dl. 68Alcohol is the most widely used sleep aid, and its use is typically intended to provide relaxation or to promote sleep. 69However, alcohol is a potent suppressor of rapid-eye-movement sleep, especially in the first half of the night. 70As the blood alcohol concentration declines, there is a rapid-eye-movement rebound in the second half of the night, producing more rapid-eye-movement sleep with increased awakenings and a reduction in total sleep time. Therefore, although alcohol may be consumed as an aid to promote sleep, it actually has the potential to significantly disrupt it.Sleep can be measured both subjectively, using a variety of questionnaires, and objectively, using standardized physiologic measures. Generally, humans are inaccurate subjective reporters of alertness. 71,72Individuals can report being awake and alert, when physiologically they could be asleep in minutes. This discrepancy between self or subjective reports and physiologic levels of alertness can have significant operational implications. First, it indicates that verbal reports of subjective alertness are generally unreliable sources to determine an individual's fitness for duty. Second, an individual with the subjective experience and report of being alert might be less likely to engage an alertness strategy (e.g ., strategic caffeine or nap opportunity, as discussed in the section on alertness strategies) that could address the underlying disturbed physiologic state. It is important to note that when an individual reports a subjective experience at either end of the continuum (e.g ., extreme fatigue or sleepiness), it is more likely to reflect the actual physiologic status. 71The human circadian (circa = around, dies =a day) timekeeper is located in the suprachiasmatic nucleus (SCN) of the hypothalamus and is an active pacemaker for internal 24-h rhythms. 73The most powerful and well-studied synchronizer of the SCN is light, while melatonin, a complementary synchronizer of the SCN, is secreted by the pineal gland at night and is suppressed by light. 74A retino-hypothalamic pathway to the SCN provides direct access for light and dark exposure to affect the circadian clock. The daily light–dark cycle entrains the SCN to its 24-h pattern. The natural tendency of the circadian clock is to run slightly slower (24.18 h) than our 24-h day, 75which is the physiologic rationale to phase delay rather than phase advance work–rest cycles. In other words, rotation of shift assignments going from days to evenings to nights has a circadian physiologic justification, but this has not been a major solution to the problems of shift work. 76The SCN controls a broad range of physiologic, behavioral, and mood functions. For example, it drives the 24-h sleep–wake pattern, daily digestive activity, hormone secretions, and mood, as well as alertness and performance levels. 55The underlying mechanisms regulating the cellular neurobiology of sleep are important and complex but are beyond the scope of this article and are reviewed elsewhere. 77,78Humans are programmed for increased sleepiness at two approximate times each day: 3–7 am and 1–4 pm 55,79The circadian nadir, associated with the lowest levels of activity, alertness, and performance and greatest vulnerability to errors, incidents, and accidents, occurs at about 3–7 am. As an example, it has been well established that a peak in fatigue-related single car accidents, without alcohol involvement, occurs roughly between 3 and 5 am. 80–83The complementary periods of maximal alertness occur at approximately 9–11 am and 9–11 pm.Rotating to a different work schedule such as the night shift or crossing time zones disrupts the entrained circadian pattern. Jet lag will occur for days as the SCN synchronizes to the new local environmental cues (e.g ., the light–dark cycle) after traveling through several time zones. Night work creates a different challenge by its disruption of the circadian pattern. When individuals are working at night, circadian programming drives sleep, and when they attempt to sleep during the day, the circadian clock is programmed for wakefulness. Generally, studies have shown that “adaptation” does not occur despite prolonged exposure to night work. 84After an all-night shift, the individual returns home and is exposed to daytime light cues that maintain SCN programming for a day–active, night–sleep pattern. Social factors such as interacting with family and performing duties that can only be done during daytime hours also play a major role in the inability to readily alter the endogenous rhythm to night work. 85A study of unintentional dural puncture during epidural anesthesia has provided further support for a circadian difference in clinical performance among anesthesiologists. 86The risk of dural puncture was greater at night (midnight to 8:00 am) and among inexperienced practitioners. Although this investigation is supportive of a negative circadian effect on performance, it was limited by the low frequency of unintentional dural punctures as well as by not including important covariants such as patient body habitus and physician workload.Subjective data from surveys of anesthesiologists 87–89and other healthcare personnel 90,91reveal that fatigue is perceived by practitioners as creating a significant risk for patients. In two studies of anesthesia caregivers, more than 50% reported having committed an error in medical judgment that they attributed to fatigue. 88,89Cooper et al ., using the critical incident method of evaluating anesthetic errors, estimated that human error played a role in more than 80% of anesthetic mishaps and that fatigue was an associated factor in 6% of reported critical incidents. 92In a survey of New Zealand anesthesiologists, 58% reported that they exceeded their self-defined limit for safe continuous administration of anesthesia, and 86% reported having committed a fatigue-related error. 87Data from 5,600 reports of critical incidents to the Australian Incident Monitoring Study from 1987 to 1997 revealed that fatigue was listed as a contributing factor in 152 reports (3%). 93These data suggest that there is a specific association between fatigue and medication errors (syringe swaps) occurring at circadian low points (2–4 am). The conclusions from these studies are limited since they are based on retrospective, self-report data, but the majority of respondents consistently indicate that quality of care is compromised and that some errors are attributed to working while fatigued.A recently litigated case clearly demonstrates the effect of fatigue in the operating room. An anesthesiologist was accused of literally falling asleep while his patient, an 8-yr-old child, died. 94During the litigation, testimony was given that the defendant had been repeatedly warned by supervisors about falling asleep during operations. He was convicted of criminal medical negligence and acquitted of felony counts of reckless manslaughter and criminally negligent homicide. The conviction was later overturned on a technicality as the statute of limitations on the misdemeanor charge had expired. Interestingly, using NTSB methods of accident analysis (table 2), the majority of errors and accidents that occur in the healthcare environment are likely to have fatigue as a contributing factor based on work schedules alone. 95There is a well-documented association between long work hours or late work and an increased potential for injury from industrial accidents. The risk of an accident increases exponentially with each hour after the ninth consecutive hour of work. 96This effect is exaggerated when extended work hours occur on a late shift. Needle stick injuries, among the most frequent of the occupational injuries suffered by anesthesia providers, are usually self-inflicted, occur during disposal or recapping of sharp devices, and are associated with carelessness from fatigue. Among residents and medical students, there is a 50% greater risk of sustaining a bloodborne pathogen exposure during night work than during days. 97Studies have extended the previously described risks associated with drowsy driving to physicians. 98–100Fatigued physicians are at risk for accident and injury as they drive home after completing their duty cycle. In a study of an academic pediatrics department, 49% of residents (averaging 2.7 h of sleep while on call) reported falling asleep at the wheel compared with 13% of faculty members. These residents had almost twice as many traffic citations for moving violations than did the better-rested faculty members. 98Residents in pediatrics and emergency medicine have been reported to suffer twice the expected number of accidents, in many cases while driving home after being on call. 98–100In a more recent retrospective study of driving-accident risk among anesthesia trainees, only eight accidents were reported, which did not differ from the control. 101The authors attributed this finding to the “protective” circadian alerting effect during the drive home (8–10 am).The effect of work hours on pregnancy outcomes in female resident physicians has been evaluated. These data reveal that there is an increased incidence of pregnancy-induced hypertension, 102preterm labor, 103and small-for-gestational-age infants. 104Another study documented an association between preterm delivery and residents who worked more than 100 h per week. 105As in other “24/7” settings, health care has experienced some high-visibility tragedies where fatigue was identified as causal or contributory. 94,106–110The most often referenced example is the death of Libby Zion, which focused attention on work hours and supervision of resident physicians. Although there has been much debate 106–110as to whether her death was related to the fatigued healthcare providers who cared for her, a high-profile commission was formed in 1987 that issued recommendations to limit house staff work hours and to increase their supervision. These recommendations became part of the revised Section 405 of the New York State Health Code (table 3). 111The Accreditation Council on Graduate Medical Education and its Residency Review Committees developed institutional and program requirements for resident supervision, duty hours, and work environment. ∥∥Institutions were required to ensure that each training program established formal policies for resident duty hours that fostered education and facilitated care of patients. The Residency Review Committee for Anesthesiology specifies that in-house duty hours should not be excessive and suggested that, on average, residents should have “at least 1 day out of 7 free of routine responsibilities and be on call in the hospital no more often than every third night.” Finally, if these policies are followed, residents are prohibited from “administering anesthesia on the day after in-house overnight call.”Subsequent evaluation suggests that the regulation of resident duty hours may not be the panacea that alone improves patient outcome. Data collected before and after implementation of the New York regulations found that there were no differences in in-hospital mortality rates, rate of patient transfer to intensive care units, or length of stay, and that there were more patients having at least one complication. 112Petersen et al . demonstrated that preventable adverse events were more common when cross-covering house staff were caring for patients compared with times when a resident knowing the patient was involved with the care. 113A follow-up study revealed that improving the quality of communication during patient sign-outs improved the quality of care. 114This suggests that during some circumstances, use of cross-covering residents to relieve tired house staff may introduce the possibility of more medical errors, but that these errors might be mitigated in other ways.There is little consensus among studies on the effects of fatigue on the performance of healthcare personnel. 3,4,115–119Previous authors have detailed the methodologic problems inherent in most of the studies (table 4). 115–118It is not surprising that the interpretation of this body of data are contradictory given these methodologic problems.There is a tremendous inconsistency across studies in the definitions used for fatigue or sleep loss. Although studies of partial sleep deprivation consistently reveal that measurable performance decrements occur if sleep is restricted by as little as 1 h, 120Bartle et al . used 4 h of sleep on the night prior to performance testing to distinguish between fatigued and rested subjects. 121There is no scientific basis for the assumption that sleep times of greater than 4 h should be considered as “rested.” Many other investigations use similarly arbitrary study conditions. 27,122–130An additional source of inconsistency in previous studies is the lack of a standardized instrument to test performance. A