Long work hours are a time-honored tradition in most residency programs. Not only are demanding schedules cited as necessary for adequate learning and professional development, but also coverage of patient care responsibilities by resident physicians has become embedded in the economics of our teaching hospitals. Recently, such long hours have been called into question and suggested as a possible contributing factor to medical errors. In conflict with the traditional model of medical training are more than a decade of scientific observations of negative effects of sleep loss and fatigue on performance and learning.1–13
Physicians-in-training are particularly susceptible to fatigue given their exposure to prolonged work hours and rotating shift work schedules. Several reviews of approximately 30 studies of reduced sleep among physicians and physicians-in-training support the hypotheses that performance is impaired and errors occur more frequently following acute and chronic sleep loss than when subjects are well-rested.14–16 Additionally, recent studies have shown that internal medicine residents are experiencing high levels of stress,17 increased burnout,18 and depressive symptoms19 that ultimately compromise patient care. A recent conference on sleep, fatigue and medical education set research priorities that could inform possible interventions.20 In the United States, the focus has been on imposed schedules and hours worked.21,22 The Accreditation Council for Graduate Medical Education (ACGME) recently put forth a position statement calling for across-the-board limitations of residents’ work hours.23 However, the relationship of imposed schedules to actual sleep quantity is unknown. Specifically, trainees’ activities external to program requirements such as caring for a dependant, pursuing community service, recreational activities or moonlighting may also contribute to residents’ fatigue.
Sleep deprivation has been shown to alter perceptions of sleepiness and performance. It is well documented that subjective sleepiness plateaus and may even decrease as objective sleepiness increases and performance declines.3–7,24,25 However, the medical community continues to debate the evidence regarding the impact of sleep deprivation and fatigue on resident trainees.20 If compliance with work hour restrictions is to be maximized and a cultural change is required, the varying (mis)perceptions and their magnitude should be delineated prior to attempts at interventions.
Accordingly, the purposes of this study were to describe the state of sleepiness and attitudes about sleep and performance of work- and non–work-related tasks among incoming and current housestaff; and to assess how rotation, call cycle, and call status are related to acute and chronic sleep deprivation and perceptions of sleepiness among current housestaff. We expected that incoming interns would be less sleepy than current housestaff as they had not yet been assigned to a clinical rotation. However, given the pervasive vision of residency as a time of sleepless nights and long days, we expected few differences between interns and seasoned housestaff in their attitudes and perceptions. Regarding the second purpose of our study, we expected rotation, call cycle, and call status to be related to real and perceived sleepiness in predictable ways (e.g., housestaff on service rotations with overnight call would report increased sleepiness compared to those on a shift-work or noncall rotation, housestaff on an every third night call cycle would be more sleepy than those on an every fourth night cycle, and postcall housestaff would be more tired than those on call).
We conducted our study in the Internal Medicine Residency Program at the University of Pennsylvania School of Medicine, a university-based internal medicine residency program in Philadelphia. In June 2001, the program had 154 residents and rotated to three clinical sites. The program staffed five intensive care units with call every third night and every sixth day off, as well as 13 acute medicine services with call every fourth night and an average of one day in seven off. In addition, eight residents per month rotated in the Emergency Department of the University Hospital site. The Emergency Department sees over 70,000 patients per year and residents work eighteen 12-hour shifts per month. The average number of on call months per year was three, four to five, and five to six for postgraduate year three (PGY-3), two (PGY-2), and one (PGY-1), respectively.
The Department of Medicine conducts a mandatory intern orientation as well as transitional retreats for first- and second-year residents. In the early afternoon at the various retreats, attendees were asked to complete the instrument. In June 2001, 53 of 58 incoming PGY-1s, 43 of 47 current PGY-1s, and 36 of 49 PGY-2s attended their respective retreats. We anonymously surveyed all 53 incoming PGY-1s, 43 current PGY-1s, and 36 PGY-2s.
The instrument contained items about respondents’ demographics, current rotation, call cycle frequency, and call status (see Appendix). Questions regarding current rotation assessed the various types of inpatient services including ward services and intensive care unit responsibilities, various outpatient services, emergency room service time as well as noncall inpatient consultative services and vacation time. Call cycle frequency was inherent to the various rotation assignments. At our institution, all housestaff on an inpatient medical ward services take call every fourth night and residents in some of our intensive care units have call on an every third-night basis. Of note, in order to reduce the length of the instrument, we omitted from the final version items regarding patient care responsibilities external to the education program (i.e., moonlighting). For individuals who were on a rotation with in-house call responsibilities, information regarding their call status on the day the survey instrument was completed was obtained. Possible choices included responses of on-call, postcall, one day precall, two days precall, and so on. Acute and chronic sleep deprivation were assessed with single items asking for hours of sleep the prior night and prior seven days, respectively. Analytically they were defined post hoc as less than six hours in the prior night and <42 hours in the prior week.6,25
Subjective sleepiness was measured using the Stanford Sleepiness Scale26 and the Karolinska Sleepiness Scale.27 The Stanford Sleepiness Scale is a single seven-option item in which respondents indicate how sleepy they feel at the moment. The Karolinska Sleepiness Scale is a single nine-option item asking respondents to indicate how sleepy they felt in the past five minutes. Scores for both scales have been reported to be both valid and reproducible.26,27 Other questions on our survey instrument related to attitudes about and perceptions of the effects of sleepiness on work and nonwork performance. Many items were drawn from the Epworth Sleepiness Scale.28 Items regarding the perceptions of the effect of sleepiness on work-related tasks were answered with a response format in which 0 = would never doze, 1 = slight chance of dozing, 2 = moderate chance of dozing, and 3 = high chance of dozing. Response options for attitude statements were on a rating scale where 1 = strongly agree, 2 = agree, 3 = no opinion, 4 = disagree, and 5 = strongly disagree. These items required use of rating scales. An expert panel of four sleep medicine physicians and researchers as well as two medical educators reviewed the content. Several drafts were reviewed, pilot-tested on non–internal medicine housestaff and revised before the 47-item instrument was finalized. The University's Institutional Review Board approved the study.
We used standard univariate statistics (frequencies and percentages) to characterize the sample. Chi-square and Fisher exact tests were used to test for differences between subgroups based on PGY, call cycle frequency, and clinical rotation. Spearman's correlations were used to look at relationships among sleepiness indicators.
All 132 participants (100%) completed the survey instrument, representing 86% of all PGY-1s and PGY-2s in the residency at the time of the study.
Degree of Sleep Restriction and Perception of Sleepiness
A total of 34% of the current housestaff experienced acute sleep deprivation, defined as less than six hours of sleep in the prior 24 hours,6 compared to 18% of the incoming interns (p = .12). Chronic sleep deprivation was a significant problem for current housestaff; 65% of active residents got less than 42 hours of sleep in the previous week,6,25 compared to 6% of incoming interns (p < .0001). Results for the subjective sleepiness scales are shown in Table 1. On the Stanford Sleepiness Scale, only 50% of current housestaff were alert or functioning at a high level compared to 89% of the incoming interns (p < .0001). Similarly, on the Karolinska Sleepiness Scale, only 53% of current house officers were at normal levels or better compared to 89% of the incoming interns (p = .0012). The Stanford and Karolinska scores were correlated (r = .71, p = .0001).
Perceptions and Attitudes of Performance Effects of Sleepiness
Table 2 shows a summary of response to survey items that asked how likely participants were to doze when performing non–work-related and work-related tasks. Both incoming interns and current housestaff believed they were more likely to doze on non–work-related tasks than on work-related tasks. Current house officers were almost always more likely to doze than incoming interns. More than 40% of the current housestaff admitted to the possibility of dozing while performing various work-related tasks such as writing notes in charts, reviewing medication lists, interpreting labs, and writing orders. Even when interacting with patients, sleepiness was a problem: 39% thought dozing was likely with taking a medical history.
As shown in Table 3 there were few differences among incoming interns and current housestaff in attitudes about sleep. Notably, current housestaff were more likely than incoming interns to feel they had learned to function well without sleep and to react quickly in a life-or-death situation, regardless of the amount of sleep. Additionally, current housestaff were more likely to feel they had learned to tolerate sleep deprivation. Almost everyone agreed that medical errors often result from sleep deprivation, but the majority felt their patients got good care despite sleep restriction and that they were able to react effectively in critical situations regardless of the amount of sleep. Although nearly everyone recognized that they learn better and feel less stressed when they get more sleep, and that that are more irritable when they are sleep deprived, half of all respondents felt that sleep deprivation was a necessary part of residency training.
Relationship of Sleepiness to Call Cycle and Rotation Among Current Housestaff
Nearly half (48%) of current housestaff rotating on a ward service reported acute sleep deprivation compared to 26% of those rotating in various intensive care units and 26% rotating on other services (p = .055) such as emergency department, outpatient electives, and other noncall rotations. Chronic sleep deprivation was more likely to be reported by residents rotating in the ICUs (83%) and on wards (74%) than on other rotations (36%, p = .01). Despite such reports, subjective sleepiness as measured by the Stanford and Karolinska scores did not differ based on rotation (p = .32 and .49, respectively).
Current housestaff on call every third night did not have a higher prevalence of acute sleep deprivation compared to those on a different call schedule (21% versus 38%, p = .06)) but they did have a higher prevalence of chronic sleep deprivation (89% vs. 58%, p = .04). Call cycle frequency was not related to the Stanford (p = .35) or Karolinska (p = .60) scores.
As expected, house officers who were postcall reported a higher incidence of acute sleep deprivation than did those who had a different call status (81% versus 22%, p = .0001). Chronic sleep deprivation was similar for all groups on a call rotation (81% versus 61%, p = .30). Postcall housestaff did indicate more subjective sleepiness on the Stanford (69% versus 6%, p = .0001) but not the Karolinska (38% versus 13%, p = .06) scales.
Our results show that despite the careful regulation of days off, current housestaff suffered from sleep deprivation. Moreover, chronic partial sleep restriction was a problem for the majority of the current housestaff regardless of their rotation. Only postcall status related in a predictable way to sleep deprivation and subjective sleepiness. On all accounts, current house officers were more sleepy than incoming interns. Current housestaff admitted to being likely to fall asleep performing some tasks but not those that can be considered crucial to patient care. Finally, there are many areas where sleep education could serve to change attitudes and perceptions. Residents who had been in the program for at least a year believed they had learned to tolerate sleep deprivation and that it was an expected part of training.
Our study had several limitations. A single-institution study of a single residency program limits generalizability. The accuracy of self-reported sleep hours without the use of sleep diaries or actigraphy has not been studied in this population. Global estimates of sleep duration are usually less reliable than daily diaries, which in turn are less reliable than a combination of daily diary and actigraphy. Additionally, several of the attitude items on the survey instrument were new. Lastly, biases in responding may be related to social desirability and expectations. Sleepiness is an expected part of training and it may not be acceptable to say otherwise. In the spring of 2001, the issues surrounding sleep, fatigue, and residency training were not as high on the programmatic agenda as they are currently. However, it is conceivable that some current housestaff and incoming interns were already aware of the controversies that were soon to arise.
Contrary to popular thought the effects of chronic partial sleep loss are cumulative.6,25 Notably, perceptions of sleepiness are far less affected by sleep deprivation than are electrophysiological and objective measures of vigilance.25 Therefore, it is likely that many of the current housestaff responding to our survey had suboptimal cognitive function even when they perceived their performance was not impaired. Residents who suffer chronic sleep restriction have near pathologic levels of sleepiness as measured objectively by the Multiple Sleep Latency Test (MSLT).29 Other studies in this population have observed that impaired performance is related to sleep deprivation. For example, sleep deprivation is related to reductions in rote memory, verbal comprehension, and mathematical calculations in medical residents,30 and mistakes in ordering medications and documenting medical history.31 The inability of housestaff to subjectively perceive their true degree of sleepiness is of great operational significance. Patients’ safety is clearly at risk when individuals who suffer from sleep restriction are unable to determine their own true level of sleepiness. Personal safety for these residents is also a concern. A significant number of drowsy driving accidents are occurring in this population. Lastly, residents are unique in that they are actually a group of learners. The degree to which the sleep restricted work schedule affects the educational goals of residency training remains to be determined.
Our study recognizes the importance of chronic sleep deprivation in addition to acute sleep deprivation. Given that more than half of our current house officers reported chronic sleep deprivation, future work should examine factors that contribute to chronic sleep deprivation in this population including moonlighting and family responsibilities. Our findings that acute sleep deprivation is independent from call cycle frequency and type of call rotation suggests that implementation of widespread duty hour restrictions may not absolutely have the desired impact on performance and patient safety outcomes.
Actual sleep amounts and influences need to be better delineated so firm conclusions regarding their direct effects on performance can be determined. Objective measures of sleep duration including sleep diaries, actigraphy, and call-in systems should be used to quantify how duty hours and personal habits translate into sleep amounts. Key areas in which performance may be impaired include work-related tasks that lead to medical errors, residents’ education and ability to learn, and at-risk daily activities such as driving. Furthermore, studies of the health risks to house officers from sleep loss or fatigue will be necessary. Additional investigation might also assess the effectiveness of countermeasures such as increased sleep time, scheduled naps, lengthened call cycles, and caffeine. The notion that sleep deprivation is an expected part of residency and the widespread beliefs that patients get good care and that residents have learned to function without sleep suggest the need for a large-scale educational effort by both the sleep and medical education communities, perhaps before these countermeasures can even begin to be instituted.
1.Jewett ME, Kijk DJ, Kronauer RE, Dinges DF. Dose-response relationship between sleep duration and human psychomotor vigilance and subjective alertness. Sleep. 1999;22:171–9.
2.Dement WC, Mitler MM. It's time to wake up to the importance of sleep disorders. JAMA. 1993;269:1548–50.
3.Doran SM, Van Dongen HP, Dinges DF. Sustained attention performance during sleep deprivation: evidence of state instability. Arch Ital Biol. 2001;139:253–67.
4.Pilcher JJ, Huffcutt AI. Effects of sleep deprivation on performance: a meta-analysis. Sleep. 1996;19:318–26.
5.Blagrove M, Alexander C, Horne JA. The effects of chronic sleep reduction on the performance of cognitive tasks sensitive to sleep deprivation. Appl Cogn Psychol. 1994;9:21–40.
6.Dinges DF, Powell JW, Pack AI, et al. Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4-5 hours per night. Sleep. 1997;20:267–7.
7.Anderson CM, Maislin G, Van Dongen HP, et al. Effect of chronically reduced nocturnal sleep with and without daytime naps, on neurobehavioral performance. Sleep. 2000;23:A74.
8.Linde L, Bergstorm M. The effect of one night without sleep on problem-solving and immediate recall. Psychol Res. 1992;54:127–36.
9.Horne JA. Sleep loss and “divergent thinking” ability. Sleep. 1988;11:528–36.
10.Smith C, MacNeill C. Memory for motor task is impaired by stage 2 sleep loss. Sleep Res. 1992;21:139.
11.Smith C, Kelly G. Paradoxical sleep deprivation applied two days after the end of training retards learning. Physiol Behav. 1988;43:213–16.
12.Blagrove M, Akehurst L. Effects of sleep loss on confidence accuracy relationships for reasoning and eyewitness memory. J Exp Psychol App. 2000;6:59–73.
13.Harrison Y, Horne JA. Sleep loss and temporal memory. Q J Exp Psychol A. 2000;53:271–9.
14.Samkoff JS, Jacques CH. A review of studies concerning effects of sleep deprivation and fatigue on residents’ performance. Acad Med. 1991;66:687–93.
15.Veasey S, Rosen R, Barzansky B, Rosen I, Owens J. Sleep loss and fatigue in residency training: a reappraisal. JAMA. 2002;228:1116–24.
16.Gaba DM, Howard SK. Fatigue among clinicians and the safety of patients. N Engl J Med. 2002;347:1249–55.
17.Collier VU, McCue JD, Markus A, Smith L. Stress in medical residency: status quo after a decade of reform? Ann Intern Med. 2002;136:384–90.
18.Shanafelt TD, Bradley KA, Wipf JE, Back AL. Burnout and self-reported patient care in an internal medicine residency program. Ann Intern Med. 2002;136:358–67.
19.Bellini LM, Shea JA. Variation in mood and empathy during internship. JAMA. 2002;287:3143–6.
20.Buysse DJ, Barzansky B, Dinges D, et al. Sleep, fatigue, and medical training: setting an agenda for optimal learning and patient care. A report from the Conference “Sleep, Fatigue, and Medical Training: Optimizing Learning and the Patient Care Environment”. Sleep. 2003;2:218–25.
21.New York State Legislature. Bell Regulations. New York State Health Code, Section 405; 1989.
22.One Hundred Seventh Congress, House of Representatives, Conyers J Patient and Physician Safety and Protection Act of 2001. HR 3236. Washington, DC: Government Printing Office, 2001.
23.ACGME approves new proposed common requirements for resident duty hours. Chicago: Accreditation Council for Graduate Medical Education, 2001 〈http://www.acgme.org/new/residentHours602.asp
〉. Accessed 5 November 2002.
24.Williams HL, Lubin A, Goodnow JJ. Impaired performance with acute sleep loss. Psychol Monogr. 1959;73:1–26.
25.Carskadon M, Dement WC. Cumulative effects of sleep restriction on daytime sleepiness. Psychophysiology. 1981;18:107–13.
26.Hoddes E, Zarcone V, Smythe H, et al. Quantification of sleepiness: a new approach. Psychophysiology. 1973;10:431–6.
27.Akerstedt T, Gillberg M. Karolinska Sleepiness Scale (KSS), a highly sensitive subjective measurement scale for sleepiness. Int J Neurosci. 1990;52(1-2):29–37.
28.Johns MW. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep. 1991;14:540–5.
29.Howard SK, Gaba DM, Rosekind MR, Zarcone VP. The risks and implications of excessive daytime sleepiness in resident physicians. Acad Med. 2002;77:1019–25.
30.Hawkins MR, Vichick DA, Silsby HD, Kruzich DJ, Butler R. Sleep deprivation and performance of house officers. J Med Educ. 1985;60:530–5.
31.Gottlieb DJ, Parenti CM, Peterson CA, Lofgren RP. Effect of a change in housestaff work schedule on resource utilization and patient care. Arch Int Med. 1991;151:2065–70.