• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of sleepGo To SLEEPAuthorsCurrent IssueSubscribeGo To SLEEP
Sleep. Mar 1, 2010; 33(3): 349–354.
PMCID: PMC2831429

A Case-Crossover Study of Sleep and Work Hours and the Risk of Road Traffic Accidents


Study Objectives:

Sleepiness, prolonged wakefulness, and extended work hours have been associated with increased risk of injuries and road accidents. The authors' objective was to study the relation between those factors and road accidents using a case-crossover design, effective in estimating the risk of acute events associated with transient, short effect exposures.


Five hundred seventy-four injured drivers presenting for care after road accidents to the Emergency Room of Udine, Italy, were enrolled in the study from March 2007 to March 2008. Sleep, work, and driving patterns in the 48 h before the accident were assessed through an interview.

Measurements and Results:

The relative risk (RR) of accident associated with each exposure was estimated using the case-crossover matched pair interval approach. Sleeping ≥ 11 h daily was associated with a decrease of the RR, as was sleeping less than usual. Being awake ≥ 16 h and, possibly, working > 12 h daily were associated with increases in the RR.


Extended work hours and prolonged wakefulness increase the risk of road accidents and suggest that awareness should be raised among drivers. The findings regarding acute sleep amount are less clear, possibly due to an effect of chronic sleep loss.


Valent F; Di Bartolomeo S; Marchetti R; Sbrojavacca R; Barbone F. A case-crossover study of sleep and work hours and the risk of road traffic accidents. SLEEP 2010;33(3):349-354.

Keywords: Case-crossover, driver, sleep, traffic accident, work

SLEEPINESS AND FATIGUE MAY DECREASE ATTENTION AND INCREASE THE LIKELIHOOD OF ERRORS WHEN PERFORMING TASKS. IN A GROUP OF AUSTRALIAN nurses, the frequency of errors at work and of near accidents when travelling home after work was increased when they were tired.1 Lack of sleep has been shown to increase the risk of unintentional injuries among children,2,3 adolescents,4 and adults.5 In addition, probably because of sleepiness, obstructive sleep apnea/hypopnea is associated with an increased rate of motor vehicle crashes, particularly those causing personal injury.6 Increased risk of road traffic accidents (RTA) was associated with drivers' acute7 and chronic8 sleepiness. Extended work shifts were also found to increase the risk of occupational injuries9,10 and motor vehicle crashes.11 The majority of those studies on fatigue and injuries, however, have assessed the risk of accidents among groups of subjects exposed to different amounts of sleep or work hours; therefore, the results may have been distorted by the effects of potential interpersonal confounders and inappropriate selection of comparison groups.

A study design that has repeatedly proved to be effective in estimating the risk of acute events, e.g., injuries, associated with transient exposures with short effect, such as acute sleep deprivation or overwork, is the case-crossover,3,1221 in which each case acts as his/her own control. Using this design, a significant increase in the risk of RTA was found by Barbone et al.20 among drivers taking benzodiazepines. Another case-crossover study recently found little evidence for associations between sleep and injury21; however, caution in the interpretation of its results is warranted, according to Rupp and Balkin, because of methodological limitations.22 Case-crossover studies have shown an increased risk of injury associated with overwork9,16; however, to our knowledge, this design has never been used to evaluate whether long work shifts (> 8 h/day) influence the risk of RTA in particular.

We conducted a case-crossover study to assess the relation between sleep and work hours, and the risk of RTA in a population of Italian drivers.


Study Subjects

Subjects were recruited at the Emergency Room (ER) of the Hospital of Udine, North-Eastern Italy, from March 12, 2007, to March 11, 2008. They were a sample of all the injured drivers (including motorcyclists and cyclists) who presented for care to the ER after an RTA. Drivers were eligible if they were ≥ 14 years of age, were alive at the time of arrival at the ER, and had sufficient knowledge of Italian to be interviewed. Subjects were included in the study if they (or their parents in case of drivers < 18 years of age) provided written consent to participate and if an interview was possible within 36 h from the time of accident.

Potential study subjects were identified by ER personnel (triage nurses) who alerted trained interviewers who systematically covered selected 12-h (weekends and nights) or 6-h (days Monday to Friday) shifts at the ER (covered hours = 3432). The interviewers approached the eligible drivers and proposed the participation in the research, without hindering or delaying diagnostic and care activities. If possible, participating subjects were interviewed directly at the ER. The study was approved by the ethical committee of the hospital.

Data Collection

Data were collected through a semi-structured questionnaire administered by trained interviewers. Within few days after the interview, the forms were checked for completeness and quality, and interviewers or subjects were re-contacted if clarifications were needed. The questionnaire collected descriptive information on sociodemographic characteristics of the driver and driving habits, characteristics of the vehicle and accident, usual alcohol and drug consumption, work shifts and hours, and sleeping patterns and their modifications during the last month. For the case-crossover analysis, work and sleep were assessed in each of the 48 h before the accident; alcohol and drug consumption were assessed in each of the 24 h before the accident. Each interview lasted approximately 30 min. At the end of the interview, reliability of responders was rated by the interviewers as excellent, good, questionable, or poor. It was excellent or good in most cases; among 574 subjects, reliability was questionable in 22 cases (3.8%) and poor in 5 (0.9%). Results did not change after excluding those subjects from the case-crossover analyses.

Statistical Analysis

The analyses were conducted according to the matched pair interval approach, in which for each subject transient exposures in a case window are compared with those in a control window.

Definition of the control windows

We adopted 2 different definitions of the case and control windows (Figure 1):

  1. a case exposure window defined as the 24 h immediately before the accident and a control window defined as the corresponding 24 h the day before (fixed distance in time from the accident—fixed window). These were used to assess the effect of sleep deprivation, wakefulness duration, and extended work hours. Only subjects who reported driving at the same hour on the day before the accident were included in these analyses, as previously done by McEvoy et al.19 and by Redelmeier and Tibshirani,23 to account for driving opportunity. In fact, as emphasized by Maclure and Mittleman,13 in collision studies the target person-times at risk are driving times, being physically impossible for drivers to be involved in collisions when not driving.
  2. a case exposure window defined as the 16 h immediately before the accident and a control window defined as the 16 h that preceded the most recent episode of driving from 16 to 32 h before the accident (variable distance in time from the accident—variable window). These were used in an additional analysis focusing on wakefulness duration (sleep and work hours were not included in this analysis). This second approach was used to maximize the number of subjects eligible for the analysis, since any driving episode occurring from 16 to 32 h before the accident (and not only those occurring at exactly the same time of the accident) could be chosen as the index driving.
Figure 1
Enrolment of subjects in the study, inclusion in the case-crossover analyses, and description of the case and control windows.

Analysis in the case of fixed windows

Given the non-normal distribution of the variables according to Kolmogorov-Smirnov test, work and sleep hours in the 2 windows were compared through Wilcoxon signed rank tests.

Conditional logistic regression was used to estimate the relative risk (RR) of accident. Exposure was first treated as a continuous variable and then in categories. The exposure categories were obtained dichotomizing daily work and sleep hours. Various cut-offs were used to assess how the risk of accident changed as a function of the cut-off chosen. Then, as proposed by Maclure,12 the cut-offs that maximized the RR were chosen as those associated with the minimum non-differential exposure misclassification. Based on those cut-offs, a 3-level classification was also used for sleep amount (< 7 h, 7-10 h, ≥ 11 h/day). In addition, the 3-level classification of sleep amount proposed by Edmonds and Vinson21 was adopted for comparison reasons (< 5 h, 5-9 h, ≥ 10 h/day). Various categories of work h were used (i.e., 0 h, 1-5 h, 6-12 h, > 12 h/day; and 0 h, 1-7 h, 8 h, > 8 h/day).

For workers, mutual adjustment was done by including both a term for daily work hours and a term for sleep amount. In addition, day of the week can be associated both with the incidence of RTAs and with sleep and work patterns; therefore, as in other studies,3 all our models included a term for day of the week (weekend vs non-weekend) to adjust for its potential confounding effect. The analyses regarding sleep hours were also adjusted for wakefulness duration in the 24-h case and control windows. An adjustment for the potential confounding effect of alcohol consumption in the previous 6 h, as done by Edmonds and Vinson,21 was not possible because, due to difficulties in recall, we could only collect information on alcohol in the 24 h before the accident and not 48. Stratified analyses were conducted to evaluate whether the effect of sleep and work hours was modified by self-judged culpability (full or partial culpability/ none), alcohol usual intake (any intake/ none),5 vehicle driven at the time of accident (4-wheel motor vehicle/ 2-wheel motor vehicle/ bicycle), time of accident (07:00–12:59/ 13:00–06:59), sex (male/ female), age (< 25/ ≥ 25 years), and sleep deprivation in the past month (much or a little less than usual/ the same as or a little or much more than usual).

Analysis in the case of variable windows

In the analysis of wakefulness duration using the variable control windows, exposure was first treated as a continuous variable and then dichotomized (≥ 16 vs < 16 h) and categorized into 4 levels (0 h, 1-8 h, 9-15 h, ≥ 16 h). To account for the fact that the control window was not at a fixed distance from the case window (and therefore time of day could be an intrapersonal confounder), we adjusted for time of the accident or of the driving episode, using 23 indicator variables coding for the 24 h in the day, as done by Mittleman et al.24 The agreement of index driving and accident time bands was assessed through the κ statistic.


During the 12 months of the study, 877 injured drivers presented to the ER during our shifts (Figure 1). Of these, 574 (65.5%) were enrolled; 100 subjects were injured too seriously to be interviewed, 95 refused to sign the consent, 40 were lost because of the contemporary arrival or more than one driver, 24 had no time to be interviewed, 22 did not understand Italian, 5 were minors not accompanied by any parent, 3 suffered from acute alcohol intoxication, and 14 did not meet other inclusion criteria. Most interviews took place in the ER (N = 546; 95.1%) soon after the subject's arrival. Table 1 illustrates the main characteristics of participants and their RTAs.

Table 1
Characteristics of the subjects and of the road traffic accidents1

Results in the Case of Fixed Windows: Sleep, Work, and Wakefulness

Of 574 subjects enrolled, 230 (40.1%) reported driving at the same time on the day before the accident. Table 2 shows their reported usual sleep duration and the duration of sleep and wakefulness in the 24-h case and control windows.

Table 2
Characteristics of reported daily sleep amount, wakefulness duration, and work hours among drivers included in the analyses with fixed 24-hour control windows

When sleep hours were treated as a continuous variable in the regression model, after adjusting for weekend, the RR of accident for each sleep hour was 0.98 (95% CI: 0.86-1.12). Table 3 shows the distribution of concordant and discordant window pairs and the RR of accident using different cut-offs for dichotomizing sleep amount. Among the cut-offs which defined sleep deprivation, only sleeping < 5 h was associated with a RR > 1, although this result was not statistically significant. Sleeping ≥ 11 h was associated with a significant reduction of the risk of accident.

Table 3
Distribution of discordant pairs, relative risk (RR) of accident and 95% confidence intervals (95% CI) using different cut-offs for daily sleep amount (total N = 230)

The results did not change after adjustment for wakefulness duration (data not shown), which was not associated with the risk of RTAs (RR = 0.99; 95% CI: 0.88-1.12) in the analysis based on 24-h case and control windows. Analyses of sleep duration and RTAs stratified by self-judged culpability, drinking habits, vehicle driven, time of accident, sex, age, and sleep deprivation in the past month resulted in very imprecise estimates (data not shown).

Of the 230 subjects who reported driving at the time of accident on the previous day, 182 (79.1%) were employed; of them, 141 (81.5%) reported working only during the day without shifts; shifts including nights were reported by 15 people (8.7%); 3 people (1.7%) always worked during the night; 13 (7.5%) worked in shifts during the day. Half of the workers (N = 94) reported having experienced overwork: 30 reported it happened daily, and 32 at least once a week. Weekly hours of work and work duration in the 24-h case and control windows are illustrated in Table 2. More than half of the RTAs among the 182 workers occurred while commuting (N = 108; 59.3%) or while working (N = 9; 4.9%). Table 4 reports the distribution of concordant and discordant window pairs and the RR of accident among employed drivers using different work hours cut-offs. A strong increase in the RR was noticed for > 12 daily h of work. The results remained unchanged when only the subset of workers who worked on both the day of the accident and the previous day (N = 121) were included in the analysis (data not shown). When working hours were analyzed as a continuous variable, however, there was no association with the risk of accident (after adjusting for weekend, RR = 1.04, 95% CI: 0.91-1.18). When the exposure was classified into 4 levels, the following results were obtained, after adjusting for weekend: the RR was 0.90 (95% CI: 0.43-1.89) for not working, 1.09 (95% CI: 0.49-2.39) for working < 6 h, and 4.67 (95% CI: 0.47-45.94) for working > 12 h compared with working 6-12 h a day; the RR was 0.65 (95% CI: 0.22-1.96) for not working, 0.67 (95% CI: 0.21-2.13) for working 1-7 h, and 0.73 (95% CI: 0.26-2.06) for working > 8 h as compared with working 8 h a day. For working hours, no stratified analyses were conducted because the number of discordant pairs was too small. Among employed subjects, mutual adjustment for daily sleep amount and work hours was conducted using 3 different models. However, these analyses did not significantly change the results (data not shown).

Table 4
Distribution of discordant pairs, relative risk (RR) of accident and 95% confidence intervals (95% CI) using different cut-offs for daily work hours (total N = 182)

Results in the Case of Variable Windows: Wakefulness

The analyses of wakefulness duration based on 16-h case and control windows with the index driving at a variable distance in time from the accident included 407 subjects. Thirteen persons (3.2%) had been awake ≥ 16 h only in the case window (f10), whereas 6 (1.5%) had been awake ≥ 16 h only in the control window (f01). After adjusting for time of day, the RR of accident was 1.07 (95% CI: 0.98-1.17) for each additional wakefulness hour (as a continuous variable) and 12.11 (95%CI: 1.29-114.01) for being awake for ≥ 16 vs < 16 h (dichotomized analysis). When 4 categories were considered and being awake 1-8 h was used as the reference, the RR was 0.45 (95% CI: 0.17-1.20) for being awake < 1 h; 0.98 (95% CI: 0.50-1.93) for being awake for 9-15 h; and 8.83 (95% CI: 0.86-91.12) for being awake for ≥ 16 h.


We assessed the effect of acute sleep deprivation and extended work hours on the risk of RTA through a case-crossover study. In this study design, each case serves as his/her own control because each subject's exposures in a defined time interval before the event (case window) is compared with the exposures of the same subject in a comparable interval in the past (control window). There are several advantages of using the case-crossover design. First, factors such as a subject's sex, age, socioeconomic status, or behavioral characteristics, potential sources of confounding in case-control or cohort studies, do not change within matched pairs and are controlled for by design. In addition, there is no need to have a separate comparison group of subjects, simplifying the recruitment phase of the study and limiting the chances of selection bias.

In this study, we could not conclude that there was an association between working hours and the risk of RTA; in fact our findings were compatible with a wide range of effects, including the null. However, despite the fact that the results were imprecise and chance cannot be excluded as a possible explanation, the direction and magnitude of the RRs suggest, consistent with other studies,1,11 that working a high number (> 12) of hours a day could strongly increase the risk of RTA, possibly because of physical and/or mental fatigue associated with acute heavy workload, prolonged occupational exposure to noise, chemicals,25 or to a modification of daily habits due to extended work duration.

In the analyses of sleep and risk of accident using different sleep hours as cut-offs, the only significant finding was a risk reduction associated with sleeping ≥ 11 h a day. In all the other cases, the analyses returned imprecise RR estimates. Since sleep needs may vary widely,26 comparing recent sleep hours with fixed sleep amounts may be meaningless, whereas a comparison with a person's usual sleep duration may be more meaningful. Surprisingly, sleeping less than usual in the past 24 h but not on the day before was associated with reduced risk of RTA. It is possible that our findings are the result of a mix of sleepiness-related exposures. In fact, on one hand, sleeping more than usual on the previous day may indicate in some subjects better rest and lower levels of sleepiness, but, in others, it could indicate a compensatory response to accumulated sleep loss and, therefore, higher levels of sleepiness. In a case-control study, Liu et al.8 found that chronic, but not acute, sleepiness in car drivers significantly increased the risk of crash.

The analysis of the number of hours drivers had been awake, based on control windows located at a variable distance in time from the accident, showed that being awake for ≥ 16 h may represent a strong risk factor for RTAs. This result is consistent with a previous case-crossover study that demonstrated an association of the risk of injury with awake time among young children,3 but not with another study among adults.21 It is, however, biologically plausible because prolonged wakefulness results in increased fatigue. In addition, laboratory studies demonstrated that periods of prolonged wakefulness as those considered in our research can produce significant cognitive performance deficits.27

The main limitation of this study is that only small numbers of subjects could be included in the analyses. In fact, in case-crossover studies using the matched pair interval approach, only pairs of case and control intervals with discordant exposure status provide information usable for estimating the RR.12 In our study, subjects tended to sleep and work regularly, i.e., exposure crossovers in the 24 h before the accident and in the 24 h on the previous day were relatively rare. In addition, we focused our attention on RTAs and did not study injuries in general. An injury can occur any time in a day, in any place, and while performing any task, but the same is not true for collisions, which can only occur while drivers are actually driving. Therefore, driving in the control window, i.e., in our study, at the same time of the day on the previous day, was an additional requisite for a subject to be eligible for the analysis. Although 574 drivers were enrolled in the study, only 230 reported driving in the control window; at most, one-fourth of this group had discordant sleep exposures in the two paired windows (using sleeping ≥ 9 h as the exposure cut-off). As a consequence, the estimated measures of association were imprecise. Furthermore, an association between sleep deprivation and accidents could have been missed if a high proportion of drivers with acute sleep deprivation refused to participate because they were too tired to be interviewed.

Efficiency of collision case-crossover studies could be improved by choosing different control windows for different cases, based on the individual driving pattern. In this way, however, intrapersonal confounding may be introduced, due to potential circadian changes in a person's lifestyle, and appropriate adjustment is needed. In our case, despite adjustment for time of day, residual confounding may be present.

Other than power limitations, there are a number of methodological issues related to the case-crossover design. Since information on exposures and previous driving was provided by the cases themselves soon after the accident, recall bias could exist, despite the fact that responders seemed reliable in general, as a consequence of differential exposure misclassification (i.e., different between the case window, closer to the time of interview, and the control window, farther from the interview). On the contrary, selection bias should not have been an issue because we matched two windows (the 24 h before the accident as the case and the 24 h before that as the control window) that we considered reasonably comparable, except for day of the week, that we adjusted for in the analyses. On the other hand, despite the fact that stable participant-specific covariates such as sex, age, personality, driving ability are controlled for by design, residual confounding due to transient exposures which may change over time within a subject cannot be excluded. In fact, in our study we only took into account sleep and work hours, however, other transient exposures (e.g., use of mobile phone, listening to music, traffic, and weather conditions when driving, route, number of passengers, alcohol intake, rush, anger, or distraction), could have been correlated with sleepiness and fatigue and not controlled for, possibly representing a source of within person confounding.28

It must be noted that the most seriously injured drivers, those who arrived at the ER under the influence of alcohol or drugs, those who were slightly injured with no need of hospital care, and those who considered themselves culpable for the accident were less likely to participate in the study, either because they could not or because they did not want to be interviewed. This may limit the generalizability of the results.

In conclusion, our findings seem to confirm the role of extended work schedules as a risk factor for RTAs and suggest that awareness should be raised among workers: if prolonged shifts cannot be avoided, then at least particular attention should be paid by workers while driving afterwards. Analogous considerations apply to prolonged wakefulness. In fact, Jones et al. found that overconfidence in one's own ability to drive after sleep loss may be a concern in terms of road safety.29 Our findings regarding sleep amount are less clear. They reflect in some way the controversial literature and suggest that the drivers' sleepiness and fatigue, which can influence the risk of accident, may derive from both acute and chronic sleep loss and that acute sleep amount may be related in different ways to the level of chronic sleepiness. Further case-crossover studies of accidents with a specific focus on sleep should investigate both short- and long-term patterns and the relations between them.


This was not an industry supported study. The authors have indicated no financial conflicts of interest.


The work was supported by a grant from the Ministry of Education, University and Research within the program 2005 and within the project entitled ‘Studio dei fattori umani degli incidenti stradali e della sicurezza di guida' coordinated by Prof. Gianfranco Vivoli, University of Modena.


confidence interval
emergency room
relative risk
road traffic accident


A commentary on this paper appears in this issue on page 283.


1. Dorrian J, Tolley C, Lamond N, et al. Sleep and errors in a group of Australian hospital nurses at work and during the commute. Appl Ergon. 2008;39:605–13. [PubMed]
2. Koulouglioti C, Cole R, Kitzman H. Inadequate sleep and unintentional injuries in young children. Public Health Nurs. 2008;25:106–14. [PubMed]
3. Valent F, Brusaferro S, Barbone F. A case-crossover study of sleep and childhood injury. Pediatrics. 2001;107:E23. [PubMed]
4. Lam LT, Yang L. Short duration of sleep and unintentional injuries among adolescents in China. Am J Epidemiol. 2007;166:1053–8. [PubMed]
5. Choi SW, Peek-Asa C, Sprince NL, et al. Sleep quantity and quality as a predictor of injuries in a rural population. Am J Emerg Med. 2006;24:189–96. [PubMed]
6. Mulgrew AT, Nasvadi G, Butt A, et al. Risk and severity of motor vehicle crashes in patients with obstructive sleep apnoea/hypopnoea. Thorax. 2008;63:536–41. [PubMed]
7. Connor J, Norton R, Ameratunga S, et al. Driver sleepiness and risk of serious injury to car occupants: population based case control study. BMJ. 2002;324:1125. [PMC free article] [PubMed]
8. Liu GF, Han S, Liang DH, et al. Driver sleepiness and risk of car crashes in Shenyang, a Chinese northeastern city: population-based case-control study. Biomed Environ Sci. 2003;16:219–26. [PubMed]
9. Ayas NT, Barger LK, Cade BE, et al. Extended work duration and the risk of self-reported percutaneous injuries in interns. JAMA. 2006;296:1055–62. [PubMed]
10. Dembe AE, Erickson JB, Delbos RG, Banks SM. The impact of overtime and long work hours on occupational injuries and illnesses: new evidence from the United States. Occup Environ Med. 2005;62:588–97. [PMC free article] [PubMed]
11. Barger LK, Cade BE, Ayas NT, et al. Extended work shifts and the risk of motor vehicle crashes among interns. N Engl J Med. 2005;352:125–34. [PubMed]
12. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol. 1991;133:144–53. [PubMed]
13. Maclure M, Mittleman MA. Should we use a case-crossover design? Annu Rev Public Health. 2000;21:193–221. [PubMed]
14. Kucera KL, Loomis D, Marshall SW. A case crossover study of triggers for hand injuries in commercial fishing. Occup Environ Med. 2008;65:336–41. [PubMed]
15. Lombardi DA, Sorock GS, Holander L, Mittleman MA. A case-crossover study of transient risk factors for occupational hand trauma by gender. J Occup Environ Hyg. 2007;4:790–97. [PubMed]
16. Vegso S, Cantley L, Slade M, et al. Extended work hours and risk of acute occupational injury: A case-crossover study of workers in manufacturing. Am J Ind Med. 2007;50:597–603. [PubMed]
17. Chow CY, Lee H, Lau J, Yu IT. Transient risk factors for acute traumatic hand injuries: a case-crossover study in Hong Kong. Occup Environ Med. 2007;64:47–52. [PMC free article] [PubMed]
18. Vinson DC, Arelli V. State anger and the risk of injury: a case-control and case-crossover study. Ann Fam Med. 2006;4:63–8. [PMC free article] [PubMed]
19. McEvoy SP, Stevenson MR, McCartt AT, et al. Role of mobile phones in motor vehicle crashes resulting in hospital attendance: a case-crossover study. BMJ. 2005;331:428. [PMC free article] [PubMed]
20. Barbone F, McMahon AD, Davey PG, et al. Association of road-traffic accidents with benzodiazepine use. Lancet. 1998;352:1331–6. [PubMed]
21. Edmonds JN, Vinson DC. Three measures of sleep, sleepiness, and sleep deprivation and the risk of injury: a case-control and case-crossover study. J Am Board Fam Med. 2007;20:16–22. [PubMed]
22. Rupp TL, Balkin TJ. An unfounded conclusion from a confounded study. J Am Board Fam Med. 2007;20:420. [PubMed]
23. Redelmeier DA, Tibshirani RJ. Association between cellular-telephone calls and motor vehicle collisions. N Engl J Med. 1997;336:453–8. [PubMed]
24. Mittleman MA, Maclure M, Robins JM. Control sampling strategies for case-crossover studies: an assessment of relative efficiency. Am J Epidemiol. 1995;142:91–8. [PubMed]
25. Rosa RR. Extended workshifts and excessive fatigue. J Sleep Res. 1995;4(S2):51–6. [PubMed]
26. Ursin R, Bjorvatn B, Holsten F. Sleep duration, subjective sleep need, and sleep habits of 40- to 45-year-olds in the Hordaland Health Study. Sleep. 2005;28:1260–9. [PubMed]
27. Williamson AM, Feyer AM. Moderate sleep deprivation produces impairments in cognitive and motor performance equivalent to legally prescribed levels of alcohol intoxication. Occup Environ Med. 2000;57:649–55. [PMC free article] [PubMed]
28. Sorock GS, Lombardi DA, Gabel CL, et al. Case-crossover studies of occupational trauma: methodological caveats. Inj Prev. 2001;7(Suppl 1):i38–42. [PMC free article] [PubMed]
29. Jones CB, Dorrian J, Jay SM, et al. Self-awareness of impairment and the decision to drive after an extended period of wakefulness. Chronobiol Int. 2006;23:1253–63. [PubMed]

Articles from Sleep are provided here courtesy of Associated Professional Sleep Societies, LLC
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...