You are on page 1of 5

Vikram R. Comondore, MD, Joshua B. Wenner, BA, Najib T.

Ayas, MD, MPH

The impact of sleep deprivation


in resident physicians on
physician and patient safety:
Is it time for a wake-up call?
Sleep-deprived residents are at increased risk for motor vehicle
collisions and hospital-related injuries, while their patients are at
higher risk for adverse outcomes resulting from medical errors.

leep deprivation is a logical Quantitative sleep loss can be due

S
ABSTRACT: Long work hours are a
tradition in the medical profession, consequence of excessive work to acute continuous loss or chronic
and work schedules are especially hours and poses potentially partial loss.1 Acute sleep loss occurs
intense among postgraduate resi- significant problems for physicians when a person does not sleep for an
dent physicians. However, because and patients. Inadequate sleep may extended period of time. Erosion of
of the sleep loss and fatigue that lead to poor health and could adverse- performance has been shown by a
result, these intense work schedules ly affect the medical care physicians number of studies. For example, Daw-
may pose threats to both physician deliver. Medical residents typically son and colleagues studied 40 subjects
and patient safety. Understanding work shifts between 24 and 36 hours and found that the decline in hand-eye
the potential impacts of fatigue on in duration on minimal sleep, and are coordination after 28 hours of wake-
resident physician performance and at heightened risk for motor vehicle fulness was similar to that resulting
safety and using this knowledge to collisions, hospital-related injury and from a blood alcohol concentration of
optimize shift schedules may reduce infection,1 and compromised mental 0.10%.2 In British Columbia, this con-
risks to both staff and patients. health. Indeed, procedural, adminis- stitutes a level of impairment above
trative, and evaluation errors may be the legal driving limit. Similarly, in a
linked to physician sleep deprivation, prospective two-session within-sub-
and reducing these could lead to ject study of 34 pediatric residents,
improved patient safety and mortality. subjects assessed after a shift on-call
had sustained attention, vigilance, and
Determinants of alertness driving abilities comparable to those
Alertness in a normal subject affects resulting from a blood alcohol level
performance and is determined by between 0.04% and 0.05%.3
quantity of sleep, circadian effects, Chronic partial sleep loss occurs
and sleep inertia. when a person consistently obtains
less sleep than that person would if
Quantity of sleep given sufficient opportunity. Chronic
Typically, an appropriate quantity of sleep deficits lead to a dose-dependent
sleep makes a person feel refreshed decrease in cognitive performance
and capable of functioning well with-
out effort, even in monotonous situa- Dr Comondore is a resident in the Internal
tions. Easy alertness is important and Medicine Program at the University of
should be differentiated from the pres- British Columbia. Mr Wenner is a medical
sured alertness of people who sleep student at UBC. Dr Ayas is an associate
little. professor of medicine at UBC.

560 BC MEDICAL JOURNAL VOL. 50 NO. 10, DECEMBER 2008


The impact of sleep deprivation in resident physicians on physician and patient safety

comparable to acute deprivation. For nounced with sleep deprivation and shift. The odds ratio for having a col-
example, in a randomized study of 48 particularly when awakenings occur lision after working an extended shift
healthy adults aged 21 to 38 with vary- during performance nadirs such as rather than a non-extended shift was
ing degrees of sleep deprivation over during the early morning hours.10 In a 2.3 (95% CI, 5.4–6.3). Residents who
14 days, lapses in behavioral alertness study of nine healthy volunteers by worked more than five extended shifts
were related to a cumulative duration Wertz and colleagues, cognitive per- in 1 month had an odds ratio of 2.39
of wakefulness in excess of 15.84 formance measured upon awakening (95% CI, 2.31–2.46) for falling asleep
hours. Subjects who slept less than 6 was worse than performance mea- while driving, or 3.69 (95% CI,
hours per night for 2 weeks had cog- sured at all times with 26 hours of 3.60–3.77) for falling asleep while
nitive abilities similar to individuals sleep deprivation.11 stopped in traffic.13 A study by Kowa-
with 1 night of total sleep deprivation. The impact of sleep inertia is par- lenko and colleagues reviewed sur-
Subjects who slept less than 4 hours ticularly relevant to residents since veys from 697 emergency medicine
per night for 2 weeks had cognitive they are frequently required to com- residents and found that before these
abilities similar to individuals who plete complex processes immediately residents started their residency, only
had 2 nights of total sleep deprivation. after waking at night. Patient assess- 4.1% reported being involved in a
Subjects’ ratings of their sleepiness ment involves a significant degree of motor vehicle collision caused by
did not change in accordance with evaluative thinking and often necessi- falling asleep, compared with 19.3%
increased sleep deprivation and objec- tates quick, high-impact decision who reported being involved in the
tively measured performance deficits, making. A physician must accurately same type of accident during their res-
suggesting that individuals who are order medications and tests, and may idency years (P<.001).14
chronically sleep deprived may not be be called upon to quickly perform
completely aware of their impairment.4 invasive procedures that require con- Percutaneous injuries
siderable concentration and skill. The Percutaneous injuries with exposure
Circadian effects cognitive impairment attributed to to potentially contaminated blood or
A group of neurons in the hypothala- sleep inertia negatively affects the body fluids through a needlestick or
mus form the circadian pacemaker, ability to complete these tasks and laceration are also more common
which regulates sleep-wake cycles. poses a serious barrier to effective among sleep-deprived physicians. In
The fluctuating activity of these neu- care. The impact of sleep inertia may a cohort of 2737 interns studied
rons enables maximal drive for cogni- be compounded by the likelihood that between July 2002 and May 2003,
tive performance and alertness in the a resident has been sleeping lightly in Ayas and colleagues evaluated the risk
day, and maximal drive for sleepiness an attempt to diminish the effects of factors for percutaneous injury and
at night, thereby consolidating noc- sleep inertia upon waking, only to be found that fatigue was a contributing
turnal sleep. These circadian effects worse off at work the following day factor in 31% of incidents. Further-
mean that a performance peak occurs because of suboptimal rest overnight. more, injuries were more likely on a
during the day and a performance nadir day after working the previous night
occurs in the early morning hours Resident physician safety than on a day after not being on call,
(3 a.m. to 5 a.m.).5 Not surprisingly, an Sleep deprivation presents many with an odds ratio of 1.61 (95% CI,
increased potential for error has been potential risks to physician safety. 1.46–1.78).15
demonstrated among nightshift work-
ers across a range of occupations.6,7 Motor vehicle collisions Mental health and stress
Sleep deprivation is the second lead- Many studies suggest that prolonged
Sleep inertia ing cause of car and truck accidents.12 fatigue adversely affects physician
Cognitive performance is typically Many studies have found an increased mental health. In 2005, Fletcher and
submaximal immediately upon awak- risk of accidents among residents colleagues undertook a systematic
ening because of a phenomenon working long hours. In a prospective review of publications assessing the
termed sleep inertia.8 Sleep inertia’s cohort study of 2737 medical interns impact of work hour legislation on res-
effects are most apparent during the followed for 1 year, the rates of motor idents. Four of the 50 studies reviewed
initial 10 to 15 minutes after awaken- vehicle collisions after working an showed that symptoms of stress or de-
ing, but they may take hours to dissi- extended shift (> 24 h) were compared pression decreased when work hours
pate.9 These effects are more pro- to rates after working a nonextended were reduced, but two of the 50 stud-

VOL. 50 NO. 10, DECEMBER 2008 BC MEDICAL JOURNAL 561


The impact of sleep deprivation in resident physicians on physician and patient safety

ies showed no significant change in effect on patient care. Ellman and col- patient safety, most institutions have
depressive symptoms.16After working leagues reviewed 6751 cardiac surg- not enforced shorter working hours.
30-hour shifts, residents have been eries, and found that mortality and sur- Concerns about discontinuity of care
found to have significantly elevated gical complication rates were no higher with patient handover has impeded
serum levels of cytokines and inflam- in surgeries completed by surgeons who changes in workshift duration.25 In a
matory markers such as IL-6 and CRP,17 had been awake the previous night review of staff physicians’ opinions at
suggesting that repeated sleep depri- than those completed by surgeons the Mayo Clinic, Keating and col-
vation episodes associated with ele- who had slept the previous night.22 In leagues noted that staff physicians felt
vated inflammatory markers may lead explaining these results, the authors discontinuity of care presented a
to vascular injury and atherosclerosis. postulated that certain procedural greater threat to patient safety than
tasks have an incentive for good per- physician fatigue. However, when
Patient safety formance, and that these tasks are less asked whether they would want a fam-
Sleep deprivation also presents many susceptible to retardation with fatigue. ily member cared for under a tradi-
potential risks to patient safety. Lockley and colleagues assessed tional call schedule or a shift sched-
the attentiveness of 20 interns work- ule, staff physicians preferred the shift
Medical errors ing in critical care units in a random- schedule.26 Many authors, including
In US hospitals, 50 000 to 100 000 ized crossover controlled trial, with Landrigan, have emphasized the im-
patients die annually from medical interns working both a traditional portance of formal evening rounds
errors, and inadequate sleep among schedule (e.g., one in three nights on and practical electronic signover sys-
physicians may be a factor.18 Many call) and an intervention schedule tems for enhancing patient care and
studies have shown that sleep dep- with a maximum shift length of 16 avoiding problems that arise when the
rivation is associated with poorer hours. Under the new schedule, resi- covering team is less familiar with
patient outcomes. In a recent web- dents worked 61 hours per week rather patient details.
based survey, Barger and colleagues than the traditional 77 to 81 hours per
showed that interns committed signif- week. Residents slept more with the Government policy on
icantly more fatigue-related medical intervention schedule (7.4 hours per physician work hours
errors resulting in adverse patient out- day vs 6.6 hours per day, P< .001). Sleep deprivation and fatigue are
comes during months with five or Furthermore, with the intervention often blamed for human error. Major
more overnight call shifts, compared schedule, residents were significantly disasters such as the Three Mile Island
with months with no extended shifts less likely to have attentional failures and Chernobyl nuclear power plant
(OR 7.0).19 A separate study evaluated during working hours.23 meltdowns and the Exxon-Valdez oil
surgical task performance before and Landrigan and colleagues evaluat- spill have been attributed to the poor
after a sleepless night. Surgical resi- ed the same cohort of residents work- judgment of sleep-deprived workers.27
dents who had been awake the previ- ing under the two call schedules, and As a result, maximum working hours
ous night made 20% more errors and assessed the incidence of medical have been defined in many jurisdic-
took 14% longer to complete a simu- errors identified by independent ob- tions for pilots,28 marine operators,29
lated laparoscopic task than their servers. Residents working under the and truck drivers.30 For example,
colleagues who had slept well the traditional rather than the intervention under the US Code of Federal Regu-
previous night.20 In an earlier study, schedule committed 35.9% more seri- lations, pilots who have flown more
Shanafelt and colleagues found that ous medical errors (136 vs 100 per 1000 than 8 consecutive hours must be
internal medicine residents meeting patient days, P<.001), 57% more non- given at least 16 hours of rest before
the criteria for burnout were more intercepted serious errors (45 vs 29 being assigned any further duties, and
likely than their colleagues to have per 1000 patient days, P <. 001), and may not exceed 100 hours per month
self-reported suboptimal patient care 5.6 times as many serious diagnostic or 1000 hours per year when working
practices.21 Burned-out residents were errors (18.6 vs 3.3. per 1000 patient for domestic air carriers. Similarly,
more likely than their colleagues to days, P<.001).24 truck drivers may drive a maximum of
report inadequate sleep and frequent 10 hours per day and 60 hours per
extended shifts as major stressors. Handover errors week. Drivers must additionally have
There have also been studies sug- Despite the evidence supporting at least 8 consecutive hours off after a
gesting that sleep deprivation has little shorter work hours for physician and 10-hour day, and marine operators

562 BC MEDICAL JOURNAL VOL. 50 NO. 10, DECEMBER 2008


The impact of sleep deprivation in resident physicians on physician and patient safety

must have at least 10 hours of rest in may not schedule in-house call more leagues, no significant changes were
24 hours. than 1 in every 3 nights.33 observed in surgical patients.36
In some jurisdictions, governments A recent study by Shetty and Bhat-
have passed legislation on physician tacharya used hospital mortality as a Conclusions
hours to improve quality of care. For robust outcome to assess the utility of To function well, physicians require
example, the 1989 death of 18-year- US resident work-hour regulations in adequate sleep. Sleep deprivation
old Libby Zion in New York was par- enhancing patient care. With a before leads to cognitive decline, altered
tially attributed to poor supervision and after comparison based on admin- mood, and impaired motor skills.
and heavy patient loads for sleep- istrative survey data, work-hour legis- There is mounting evidence that sleep
deprived house staff; thereafter, the lation was associated with a 0.25% deprivation has long-term health con-
state passed legislation that limited decrease in absolute mortality rate sequences such as premature death,
physicians’ shifts to 24 consecutive (P = .043), and a 3.75% reduction in cardiovascular death, obesity, and dia-
hours, or 80 hours per week.31 New relative risk of death. Although large betes. Sleep-deprived residents are at
York State has been the only jurisdic- improvements in mortality were noted increased risk for motor vehicle colli-
tion in North America to enact legally in patients admitted for infectious dis- sions and percutaneous injuries, and
binding legislation concerning work ease (change -0.66%, P = .007) and in their patients are at higher risk for
hours; while there are similar recom- medical patients over 80 years of age medical error. In other parts of the
mendations in many other states and (change -0.71%, P = .005), this change world, shorter work hours for physi-
provinces, they may be disregarded (0.13%, P = .54) was not significant cians are becoming the norm and these
with impunity. In Europe, doctors for surgical patients, which comprised have often been enacted through leg-
follow the European Working Time one-fifth (or 243 207) of the patients islation. Physicians in British Colum-
Directive, which currently limits their studied. The authors suggested that bia govern themselves through the
work to a maximum of 56 hours per differences might be attributed to the College of Physicians and Surgeons.
week. In its final phase, the directive smaller number of surgical patients in Rather than waiting for a government
will enforce a reduction to 48 hours the sample, fewer surgical residents body to direct our actions through leg-
per week. While this directive was ini- working at a given time as a result of islation, we believe physicians should
tially introduced in 1993, it excluded the new schedule, handover errors, and address this problem and adequately
health care systems until 2004 and its poor compliance of surgical programs limit resident work hours to protect
gradual application is targeted for with the new legislation.34 Indeed, both our patients and our trainees.
completion in 2009. These restrictions Landrigan and colleagues show that
include time spent on call and are interns widely reported noncompli- Competing interests
enforceable laws, not just recommen- ance with ACGME guidelines in the None declared.
dations.32 first year of their implementation.35
In a similar study, Volpp and col- References
Work hours and patient leagues studied mortality outcomes 1. Mountain SA, Quon BS, Dodek P, et al.
mortality before and after ACGME regulations The impact of housestaff fatigue on occu-
In July 2003 the US Accreditation were set in place, using nonteaching pational and patient safety. Lung 2007;
Council for Graduate Medical Educa- hospitals as a control group. While no 185:203-209.
tion (ACGME) issued a set of stan- significant results were found in the 2. Dawson D, Reid K. Fatigue, alcohol and
dards to limit resident work hours. In first year after these regulations, dur- performance impairment. Nature 1997;
order to maintain accreditation status, ing the second post-reform year the 388:235.
residency programs must not exceed odds of mortality significantly de- 3. Arnedt JT, Owens J, Crouch M, et al. Neu-
80-hour work weeks, with a maximum creased in teaching hospitals for med- robehavioral performance of residents
of 24 hours per shift (this is excepted ical patients only. The odds of mortal- after heavy night call vs after alcohol
when more time is needed to ade- ity were reduced for patients with ingestion. JAMA 2005;294:1025-1033.
quately fulfill patient handover res- myocardial infarction, congestive 4. Van Dongen HP, Maislin G, Mullington
ponsibilities). Additionally, programs heart failure, gastrointestinal bleed- JM, et al. The cumulative cost of addi-
must allow residents 1 day off out of ing, or stroke with the work-hours leg- tional wakefulness: Dose-response ef-
every 7, a minimum 10 hours off islation (OR 0.74, 95% CI 0.61-0.89). fects on neurobehavioral functions and
between daily work activities, and As in the study by Shetty and col- sleep physiology from chronic sleep

VOL. 50 NO. 10, DECEMBER 2008 BC MEDICAL JOURNAL 563


The impact of sleep deprivation in resident physicians on physician and patient safety

restriction and total sleep deprivation. 16. Fletcher KE, Underwood W 3rd, Davis et al. Catastrophes, sleep, and public pol-
Sleep 2003;26:117-126. SQ, et al. Effects of work hour reduction icy: Consensus report. Sleep 1988;11:
5. Bjerner B, Holm A, Swensson A. Diurnal on residents’ lives: A systematic review. 100-109.
variation in mental performance: A study JAMA 2005;294:1088-1100. 28. Operating requirements: Domestic, flag,
of three-shift workers. Br J Ind Med 17. Zheng H, Patel M, Hryniewicz K, et al. and supplemental operations. 14 CFR
1955;12:103-110. Association of extended work shifts, vas- 121.471. 26 January 1996. Available from
6. Santhi N, Horowitz TS, Duffy JF, et al. cular function, and inflammatory mark- GPO Access www.gpoaccess.gov/index
Acute sleep deprivation and circadian ers in internal medicine residents: A ran- .html (accessed 26 March 2008).
misalignment associated with transition domized crossover trial. JAMA 2006; 29. Manning of vessels. 46 USC 8104. 18
onto the first night of work impairs visual 296:1049-1050. August 1990. Available from GPO Ac-
selective attention. PLoS ONE 2007; 18. Kohn LT, Corrigan JM, Donaldson MS cess. www.gpoaccess.gov/index.html
2:e1233. (eds). To Err Is Human: Building a Safer (accessed 26 March 2008).
7. Dembe AE, Delbos R, Erickson JB, et al. Health System. Washington, DC: Nation- 30. Hours of service of drivers. 49 CFR.
Associations between employees’ work al Academy Press, 1991. 395.3/395.5. 25 August 2005. Available
schedules and the vocational conse- 19. Barger LK, Ayas NT, Cade BE, et al. Impact from GPO Access. www.gpoaccess.gov/
quences of workplace injuries. J Occup of extended-duration shifts on medical index.html (accessed 26 March 2008).
Rehabil 2007;17:641-651. errors, adverse events, and attentional 31. Wallack MK, Chao L. Resident work
8. Downey R, Bonnet MH. Performance failures. PLoS Med 2006;3:e487. hours: The evolution of a revolution. Arch
during frequent sleep disruption. Sleep 20. Taffinder NJ, McManus IC, Gul Y, et al. Surg 2001;136:1426-1431.
1987;10:354-363. Effect of sleep deprivation on surgeons’ 32. Pickersgill T. The European working time
9. Jewett ME, Wyatt JK, Ritz-De Cecco A, et dexterity on laparoscopy simulator. Lan- directive for doctors in training. BMJ
al. Time course of sleep inertia dissipa- cet 1998;352(9135):1191. 2001;323:1266.
tion in human performance and alert- 21. Shanafelt TD, Bradley KA, Wipf JE, et al. 33. Accreditation Council for Graduate Med-
ness. J Sleep Res 1999;8:1-8. Burnout and self-reported patient care in ical Education. Common program re-
10. Dinges DF, Orne MT, Orne EC. Assessing an internal medicine residency program. quirements. 2004. www.acgme.org/
performance upon abrupt awakening Ann Intern Med 2002;136:358-367. acwebsite/dutyhours/dh_dutyhours
from naps during quasi-continuous 22. Ellman PI, Law MG, Tache-Leon C. Sleep commonpr.pdf (accessed 10 January
operations. Behav Res Methods deprivation does not affect operative 2008).
1985;17:37-45. results in cardiac surgery. Ann Thorac 34. Shetty KD, Bhattacharya J. Changes in
11. Wertz AT, Ronda JM, Czeisler CA, et al. Surg 2004;78:906-911. hospital mortality associated with resi-
Effects of sleep inertia on cognition. 23. Lockley SW, Cronin JW, Evans EE, et al. dency work-hour regulations. Ann Intern
JAMA 2006;295:163-164. Effect of reducing interns’ weekly work Med 2007;147:73-80.
12. Mitler MM, Dinges DF, Dement WC. hours on sleep and attentional failures. N 35. Landrigan CP, Barger LK, Cade BE, et al.
Sleep medicine, public policy, and public Engl J Med 2004;351:1829-1837. Interns’ compliance with accreditation
Health. In: Kryger MH, Roth T, Dement 24. Landrigan CP, Rothschild JM, Cronin JW, council for graduate medical education
WC (eds). Principles and Practice of Sleep et al. Effect of reducing interns’ work work-hour limits. JAMA 2006;296:1063-
Medicine. 2nd ed. Philadelphia, PA: WB hours on serious medical errors in inten- 1070.
Saunders; 1994;453-462. sive care units. N Engl J Med 2004; 36. Volpp KG, Rosen AK, Rosenbaum PR, et
13. Barger LK, Cade BE, Ayas NT, et al. 351:1838-1848. al. Mortality among patients in VA hospi-
Extended work shifts and the risk of 25. Petersen LA, Troyen AB, O’Neil AC, et al. tals in the first 2 years following ACGME
motor vehicle crashes among interns. N Does housestaff discontinuity of care resident duty hour reform. JAMA 2007;
Engl J Med 2005;352:125-134. increase the risk for preventable adverse 298:984-992.
14. Kowalenko T, Kowalenko J, Gryzbowski effects? Ann Intern Med 1994;121:866-
M. Emergency resident related auto acci- 872.
dents—Is sleep deprivation a risk factor? 26. Keating RJ, LaRusso NF, Kolars JC. Per-
Acad Emerg Med 2000;7:1171. ceived impact of duty hours limits on the
15. Ayas NT, Barger LK, Cade BE, et al. fragmentation of patient care: Results
Extended work duration and the risk of from an academic health center. Am J
self-reported percutaneous injuries in Med 2005;118:788-793.
interns. JAMA 2006;296:1055-1062. 27. Mitler MM, Carskadon MA, Czeisler CA,

564 BC MEDICAL JOURNAL VOL. 50 NO. 10, DECEMBER 2008