Assessing the risk of crashes in patients with obstructive sleep apnea

November 24, 2008

ABSTRACT: Patients with obstructive sleep apnea (OSA) are at increasedrisk for motor vehicle crashes as a result of excessivesleepiness. However, a number of factors complicate risk assessment.For example, self-reported sleepiness and the severityof OSA do not appear to be good predictors of accident risk.Many persons with OSA do not accurately perceive their levelof drowsiness-self-reported sleepiness does not correlate wellwith objective measures, such as results of the Multiple SleepLatency Test. Moreover, it is not clear whether objective testscan reliably predict accident risk in the real world, as opposedto during simulated driving. The indications for-and benefitsof-restricting driving in patients with OSA have not been established.However, there is good evidence that the use of continuouspositive airway pressure significantly reduces the riskof crashes in these patients. (J Respir Dis. 2008;29(12):459-464)

ABSTRACT:Patients with obstructive sleep apnea (OSA) are at increased risk for motor vehicle crashes as a result of excessive sleepiness. However, a number of factors complicate risk assessment. For example, self-reported sleepiness and the severity of OSA do not appear to be good predictors of accident risk. Many persons with OSA do not accurately perceive their level of drowsiness-self-reported sleepiness does not correlate well with objective measures, such as results of the Multiple Sleep Latency Test. Moreover, it is not clear whether objective tests can reliably predict accident risk in the real world, as opposed to during simulated driving. The indications for-and benefits of-restricting driving in patients with OSA have not been established. However, there is good evidence that the use of continuous positive airway pressure significantly reduces the risk of crashes in these patients. (J Respir Dis. 2008;29(12):459-464)

Excessive daytime sleepiness is a common symptom of obstructive sleep apnea (OSA) and is associated with increased risk of motor vehicle crashes (MVCs). A patient's involvement in a drowsiness-related accident resulting in serious injury or death is the most likely situation in which a physician could face charges of legal negligence in the management of OSA. The AMA policy H-15.958 (Fatigue, Sleep Disorders, and Motor Vehicle Crashes) states that physicians should "inform patients about the personal and societal hazards of driving or working while fatigued and advise patients about measures they can take to prevent fatigue-related and other unintended injuries."1

State regulations governing physicians' reporting of patients with medical conditions that may render them unfit for driving vary from no requirement to mandatory reporting of all patients with any diagnosis listed as reportable. Some states require reporting of patients with listed conditions only if the physician believes the condition significantly impairs function. The AMA policy also states that physicians should "become familiar with the laws and regulations concerning drivers and highway safety in the state(s) where they practice."1

A task force of the European Respiratory Society on public health and medicolegal implications of sleep apnea concluded that "whatever the legislation, the clinician has a responsibility to inform his patients of the risks related to sleepiness and to discourage him from driving as long as he is not effectively treated."2 In 2 instances in Canada, physicians were found liable because of failure to report patients potentially medically unfit to drive before the patients were involved in MVCs.3 However, these were not sleep-related conditions and, as of this writing, I am not aware of any instance in the United States in which a physician was held liable for failure to report a patient with a sleep disorder who subsequently had an MVC. There is clearly a potential for this to occur. Physicians have an ethical, if not legal, responsibility to inform their patients with OSA about the increased risk of crashes that are associated with this condition and to counsel them to refrain from driving while drowsy.

In this article, I will review the current understanding of the medical aspects of the risk of crashes for patients with OSA and the effect of therapy on that risk. I also will discuss the implications of making judgments about restricting driving in these patients.


Up to 20% of MVCs have been attributed to inattention associated with excessive sleepiness.4,5 Stutts and colleagues6 interviewed about 1000 persons who had been in MVCs and about 400 persons who had not been in an MVC (control group). Only 1.3% of those classified by law officers as being asleep at the time of the crash had previously received a diagnosis of a sleep disorder; however, 25% admitted to having driven while drowsy more than 10 times and 91% admitted to having fallen asleep while driving at least once.

The self-rated score on the Epworth Sleepiness Scale (ESS) for persons who had been asleep at the time of the accident was 7.6 (greater than 10 is considered abnormally sleepy), compared with 5.6 in those classified as neither asleep nor fatigued.6 The mean score was 5.1 in the control group. Thus, most persons with probable drowsiness-related crashes did not have a diagnosed sleep disorder, and most did not perceive themselves to be excessively sleepy despite admitting to previously driving while drowsy.

A recent Internet survey supported the finding of increased risk of crashes associated with driving while drowsy.7 Persons who reported 4 or more sleep-related near-misses were 1.87 times as likely to have had an MVC as those without any near-misses.

Multiple studies of data from state motor vehicle driving records, insurance company records, and self-reported crashes and experimental studies on driving simulators have found that the risk of MVCs is increased in noncommercial drivers who have OSA. Although the estimated risks have varied, 2 meta-analyses yielded overall estimates of 2.5 to 3.1.8,9

A retrospective 3-year study of insurance company data for 60 patients with OSA and 60 controls demonstrated that the rate of crashes was higher in patients with OSA (odds ratio, 2.6) and that a greater proportion of patients with OSA had had 2 or more crashes.10 However, the number of crashes was not associated with ESS score or the severity of OSA as measured by the apnea-hypopnea index (AHI) on polysomnography.

The lack of correlation between symptoms of daytime sleepiness or severity of OSA with MVC risk was noted by Ellen and coauthors.9 In their meta-analysis, only 20% of the higher-quality studies showed a positive correlation between MVC risk and sleepiness, and only 30% found a correlation between MVC risk and the severity of OSA. Only 1 of 3 studies of commercial drivers that used self-reported crashes as the outcome measure found even a weak association between OSA and MVC risk.9

Because of the potential financial consequences of restricted driving, self-reported information from commercial drivers may be less reliable than that from noncommercial drivers. However, in commercial drivers, impaired performance on a test of vigilance has been found to be related to the duration of the previous night's sleep and not to the AHI on polysomnography.11

Despite the overall elevated risk, most persons with OSA have not had a drowsiness-related MVC, and neither the severity of OSA nor selfrated sleepiness appear to be good predictors of MVC risk. Thus, which patients with OSA should be restricted from driving and how much benefit would result from restrictions are not clear.12


The physician has a responsibility to detect conditions that may impair driving and increase the risk of harm to the patient or the public. As mentioned above, the legal requirements for reporting such patients to authorities vary by state.

Some states do not require any reporting but allow it even without patient authorization if the physician considers the patient or the public to be at significant risk ("permissive reporting"). A North Carolina statute gives physicians limited immunity from civil and criminal liability for such reporting "as long as he was acting in good faith and without malice." If you practice in a jurisdiction with permissive reporting, factors bearing on your decision may include the following:
• The ethical standards you and  your medical community hold regarding the importance of patients' privacy.
• The amount of responsibility for public health that should be borne by physicians versus patients.
• Your judgment about the effectiveness of reporting in reducing overall risk-whether reporting will lead to concrete actions that alter the patient's driving behavior.

There is some evidence that patients may hide symptoms if they fear loss of driving privileges. Findley13 found that 21 of 30 patients being evaluated for OSA stated that they would have avoided medical evaluation if reporting to the Department of Motor Vehicles (DMV) were mandatory.


Patient perceptions
Can physicians rely on trustworthy and motivated patients to restrict their driving when sleepy? This depends on whether patients have an accurate perception of their level of drowsiness-their propensity to fall asleep. Unfortunately, many persons with OSA do not accurately perceive their drowsiness. Self-reported sleepiness on the ESS and objective measures of sleep propensity in the laboratory, such as the mean sleep-onset latency during 4 or 5 nap opportunities on a multiple sleep latency test (MSLT), are not closely correlated. This inaccurate perception of sleepiness was demonstrated in the above-described findings on drowsiness-related MVCs reported by Stutts and colleagues.6

Driving simulator studies also suggest that many persons do not adequately perceive their MVC risk when drowsy. Banks and coauthors14 had healthy persons perform a 70-minute driving simulator test at 1 AM after partial sleep deprivation (limited to 5 hours in bed) on 2 nights, one with and one without alcohol consumption before testing. Crashes were tabulated for 4.5-minute epochs during the trials.

Without alcohol consumption, women reported that they would have stopped driving because of a perceived crash risk in 85% of the epochs in which a crash occurred.14 However, the corresponding figure reported by men was only 37%. With a mean blood alcohol level of 0.037 g/dL at the start of the driving trial, accurate risk perception dropped to 38% for women and 33% for men. Another finding in this study was that in the absence of alcohol consumption, no crashes occurred during the first 45 minutes of the simulated drive after partial sleep deprivation, but crashes began occurring in the first 5 minutes when alcohol was consumed.14

Another driving simulator study showed that the addition of alcohol (a blood alcohol concentration of about 0.04%) to sleep restriction worsened performance and increased the electroencephalographic manifestations of sleepiness, but the participants did not perceive the increased sleepiness.15

A study of performance and perception in a real driving situation was conducted by Philip and coworkers.16 They had 10 young men drive 600 miles on an expressway in France after an average of 8.25 hours of sleep and after an average of only 1.8 hours of sleep. Breaks were taken every 105 minutes for testing that included self-rating of sleepiness on the Karolinska Sleepiness Scale and measurement of reaction time.

After only 1.8 hours of sleep, the participants reported being more sleepy and fatigued at all times of the day, but this perception was not related to reaction time, which remained constant throughout both drives. However, the authors of this study reported that "in the rested condition the experimenter never had to take control of the car. In the sleep-restricted condition 7 out of 10 subjects had to be assisted by the experimenter."16

Objective tests of sleepiness
The lack of correlation between self-reported sleepiness and the propensity for sleep measured on an MSLT has been interpreted as evidence that objective measures of sleepiness must be used to accurately judge a patient's risk. However, findings on an MSLT do not correlate closely with the frequency of sleep-related crashes in the real world. For example, Aldrich17 found that MSLT results did not predict self-reported history of MVCs in 424 patients with OSA.

Some studies have found that sleep latency on MSLTs correlated with crash frequency during a driving simulator test.18 However, study participants are aware that no dire consequences occur if they fall asleep during simulated driving, in contrast to actual driving. The latter setting clearly provides more motivation to resist falling asleep. Philip and coworkers19 showed that the rate of inappropriate line crossings on a simulator drive after sleep restriction of 2 hours was 25 to 50 times higher than that observed in a real drive on an expressway in a car with dual controls.

Thus, patients may be able to resist sleep when motivated to do so despite a high propensity to fall sleep when allowed to do so. Indeed, MSLT results do not correlate closely with results of a maintenance of wakefulness test (MWT); the instructions for the MSLT are to "try to fall asleep," while the instructions for the MWT are to "try to stay awake."

The ability to resist falling asleep would seem to be a more appropriate measure of a person's accident risk, and the MWT has been advocated for use in assessing workplace safety.20 Unfortunately, MWT results do not correlate closely with overall performance on a driving simulator after sleep restriction in healthy persons. Banks and associates21 found that although sleep latency on MWT after sleep restricted to 5 hours was correlated with reaction time, it explained only about 20% to 30% of the variance of driving performance and was not significantly correlated with the number of crashes. However, sleep latency did correlate with the number of crashes on the simulator when alcohol ingestion (mean blood alcohol content was 0.037 g/dL at the start of the drive) was combined with sleep restriction.21

Hack and coworkers22 found that the performance on a driving simulator of persons with untreated OSA did not differ from that of healthy persons after 24 hours of sleep deprivation or after alcohol consumption (blood alcohol level of 0.072 g/dL). Some studies of patients with untreated OSA have found that sleep latency on an MWT was associated with the mean deviation from the center line of the road during a driving simulator test23; however, as noted above, simulator performance may not predict real performance19 and does not predict actual crash history for patients with OSA.

Turkington and colleagues24 found that the severity of OSA, as measured by the respiratory disturbance index on polysomnography, was not associated with accident history or performance on a driving simulator. Patients who performed well on the driving simulator were unlikely to have had an accident in the previous year; however, poor performance was not associated with accident history.

Some of these inconsistencies may be the result of individual differences in susceptibility to sleep deprivation. In a laboratory evaluation of neurobehavioral performance, including tests of vigilance, Van Dongen and coworkers25 found that some persons performed well after 36 hours of sleep deprivation that followed 1 week of sleep restriction (about 4.6 hours), while others performed poorly after the acute deprivation despite having had 8 hours of sleep per night for the previous week. Similarly, Philip and coworkers26 found a significant difference in performance among 22 healthy men during a 625-mile drive after sleep was restricted to 2 hours-the number of line crossings during the drive ranged from 0 to 94.

Another issue is that alertness is not simply the opposite of sleepiness.27 The interplay between the sleep and arousal systems may influence sleep latency on the MSLT and the MWT.28 Individual differences in the usual intrinsic level of arousal (trait arousal) may partly account for the wide variation in sleep onset on the MWT for healthy persons who have no sleep-related complaints or clinical impairment of alertness.

Thus, the prediction of driving performance in a person with a sleep disorder may not be accurate even if it is based on objective tests of sleepiness. Bonnet28 has discussed the lack of understanding of how well the MSLT and the MWT can predict functional alertness in complex situations and has recommended against the use of these tests for the assessment of workplace safety. Similarly, the results of driving simulators may be of limited value in determining which patients with OSA should be restricted from driving.


Effects of treatment
There is good evidence that the treatment of OSA with continuous positive airway pressure (CPAP) significantly reduces the risk of MVCs. Based on DMV data, George29 evaluated MVC rates before and after CPAP therapy in 210 patients with OSA over a 3-year period. The crash rate was 0.18 per driver per year before treatment and 0.06 after treatment, a rate that is similar to that among persons without OSA.

Hack and coworkers30 demonstrated a beneficial effect of CPAP on simulator performance. The incidence of off-road events decreased from 17.8 to 9.0 after 1 month of therapy when CPAP was set at an effective level, but it did not change if CPAP was set at 1 cm H2O. A rapid onset of a benefit of CPAP was demonstrated by Turkington and coworkers,31 who found a marked improvement in off-road events by day 3 and significant improvement in tracking errors and reaction time by day 7.

In addition, Orth and coworkers32 found significant improvement in performance on a driving simulator after 2 days of CPAP therapy in 24 patients with OSA. The mean number of crashes during a 60-minute drive in which obstacles were presented randomly was reduced from 2.7 before CPAP to 1.5 after 2 days of treatment (P < .01). The accident rate remained significantly reduced at 0.9 in the 21 patients who were retested after 42 days of CPAP treatment.32

Neuropsychological tests of alertness and divided attention also improved after 2 days and after 42 days, but vigilance did not improve.32 Although the arousal index on polysomnography and ESS scores improved after treatment, no significant effect on total sleep time or slow wave sleep was found. The ESS score was not associated with polysomnographic results, such as sleep architecture and ventilatory measures, or with driving performance on the simulator.

Finally, Mazza and colleagues33 studied driving ability on a controlled course before and after 3 months of CPAP therapy in patients with OSA. The participants had to brake to avoid hitting a water spray obstruction that suddenly appeared without warning on the track. Before CPAP treatment, patients with OSA had significantly longer reaction times and more than twice the rate of collisions, compared with healthy controls.

However, Mazza and colleagues33 found that there was no correlation between driving performance and ESS score or any sleep variable measured. After treatment, reaction time and rate of collisions decreased significantly and were not different from those of the controls.33

Published guidelines
Recommendations for dealing with driving risks in patients with OSA were published by the American Thoracic Society in 1994.34 These recommendations included the following: "learn about applicable state laws, be alert for symptoms of excessive daytime somnolence or other symptoms of OSAHS [obstructive sleep apnea-hypopnea syndrome] especially in commercial drivers, assess risk of impaired driving in patients with OSAHS (risk is increased if patient has had recent accidents or ‘near-misses' or has severe OSAHS or severe EDS [excessive daytime sleepiness]), report high-risk drivers to DMV if they insist upon driving before treatment is initiated or fail to comply with treatment. High-risk truck, bus or other occupational drivers may be considered for reporting."34

More recently, recommendations for screening and evaluation of fitness for duty of commercial drivers with possible OSA were issued by a joint task force of the American College of Physicians, the American College of Occupational and Environmental Medicine, and the National Sleep Foundation.35 For patients being treated with CPAP, evaluation of adherence at 2 to 4 weeks is recommended. If the patient's AHI was documented to be 5 or lower during titration (or 10 or lower, depending on clinical findings) and the patient is adherent to therapy (objectively documented minimum acceptable average use, 4 hours in a 24-hour period), the driver may be allowed to return to work. However, as described above, Pack and coworkers11 found that lapses in vigilance in commercial drivers were not related to AHI but were associated with the mean duration of sleep per night at home during the previous week as measured by actigraphy.

A recent expert review of the reliability and validity of scoring respiratory events during sleep studies stated: "Case-control studies generally have confirmed the experimental data concerning the physiologic impact of sleepiness on driving ability, reaction times and self-report. However, the culpability surveys have yet to show what definition of SRBD [sleep-related breathing disorder] as classified by event type (apnea or hypopnea), AHI or desaturation indices predicts risk for a 'crash.'"36

George37 pointed out that we do not currently have an adequate metric for quantifying the severity of OSA in relation to the associated driving risk and suggested that it is not necessary to report or restrict the driving of patients who have mild OSA, defined by an AHI of less than 30, and do not have excessive sleepiness.


My recommendation is that you try to be fair to your patients while avoiding legal and moral liability- follow the law and your conscience. Clinically assess the patient's overall risk for unsafe driving, their perception of that risk, and their willingness to take appropriate measures to minimize risk. Even if you are not legally required to do so, strongly consider reporting a patient to the DMV if your assessment indicates a high risk and a low likelihood of adequate risk reduction in the absence of intervention, especially if the patient is a commercial driver.

Assess patients' adherence to CPAP clinically and with objective monitoring measures available on current CPAP devices. Be familiar with the guidelines for commercial drivers described above.

Despite the lack of validation of MWT as a predictor of crash risk and the uncertainty about its clinical significance, post-treatment MWT results that show an ability to maintain wakefulness throughout all nap periods provide evidence for risk reduction. The MWT may be required by some governmental licensing agencies for the medical evaluation of fitness to drive in persons with OSA.


OSA is associated with an increased risk of MVCs, although the absolute risk is relatively low. A high AHI is probably associated with increased risk, but other factors can greatly modify the risk.

The evidence indicates that patients' perception of their sleepiness is often inaccurate. When combined with reduced sleep, even small amounts of alcohol further increase risk and reduce the ability to correctly perceive the level of sleepiness. A propensity to fall asleep quickly when asked to do so during the MSLT does not accurately predict an inability to stay awake when motivated to do so. The ability to maintain wakefulness and vigilance depends on the situation and on inherent individual characteristics.

No currently available test is highly predictive of MVC risk. However, if patients with OSA adhere to appropriate CPAP therapy, excessive sleepiness and the associated risks can be effectively reduced.




The American Medical Association policy H-15.958: Fatigue, Sleep Disorders, and Motor Vehicle Crashes.

. Accessed May 9, 2008.


Krieger J, McNicholas WT, Levy P; ERS Task Force. European Respiratory Society. Public health and medicolegal implications of sleep apnoea [published correction appears in Eur Respir J. 2003;21:561].

Eur Respir J.



Weaver T, George C. Cognition and performance in patients with obstructive sleep apnea. In: Kryger MH, Roth T, Dement WC, eds.

Principles and Practice of Sleep Medicine.

4th ed. Philadelphia: Elsevier Saunders; 2005:1030.


Pack AI, Pack AM, Rodgman E, et al. Characteristics of crashes attributed to the driver having fallen asleep.

Accid Anal Prev.



Lyznicki JM, Doege TC, Davis RM, Williams MA. Sleepiness, driving, and motor vehicle crashes. Council on Scientific Affairs, American Medical Association.




Stutts JC, Wilkins JW, Vaughn BV. Why do people have drowsy driving crashes? Input from drivers who just did. AAA Foundation for Traffic Safety, Washington, DC, 1999.

. Accessed May 2, 2008.


Powell NB, Schechtman KB, Riley RW, et al. Sleepy driver near-misses may predict accident risks.




Sassani A, Findley LJ, Kryger M. Reducing motor-vehicle collisions, costs, and fatalities by treating obstructive sleep apnea syndrome.




Ellen RL, Marshall SC, Palayew M, et al. Systematic review of motor vehicle crash risk in persons with sleep apnea.

J Clin Sleep Med.

2006;2: 193-200.


Barbe F, Perics J, Munoz A, et al. Automobile accidents in patients with sleep apnea syndrome. An epidemiological and mechanistic study.

Am J Respir Crit Care Med.



Pack AI, Maislin G, Staley B, et al. Impaired performance in commercial drivers: role of sleep apnea and short sleep duration.

Am J Respir Crit Care Med.



Pack AI, Pien GW. How much do crashes related to obstructive sleep apnea cost?




Findley LJ. The threat of mandatory reporting to a driver’s license agency discourages sleepy drivers from being evaluated for sleep apnea.

Am J Respir Crit Care Med.



Banks S, Catcheside P, Lack LC, et al. Low levels of alcohol impair driving simulator performance and reduce perception of crash risk in partially sleep deprived subjects.




Horne JA, Reyner LA, Barrett PR. Driving impairment due to sleepiness is exacerbated by low alcohol intake.

Occup Environ Med.

2003;60:689- 692.


Philip P, Sagaspe P, Taillard J. Fatigue, sleep restriction, and performance in automobile drivers: a controlled study in a natural environment.




Aldrich MS. Automobile accidents in patients with sleep disorders.




Pizza F, Contardi S, Mostacci B, et al. A driving simulation task: correlations with Multiple Sleep Latency Test.

Brain Res Bull.



Philip P, Sagaspe P, Taillard J, et al. Fatigue, sleepiness, and performance in simulated versus real driving conditions.




Arand DL. The MSLT/MWT should be used for the assessment of workplace safety.

J Clin Sleep Med.



Banks S, Catcheside P, Lack LC, et al. The Maintenance of Wakefulness Test and driving simulator performance.




Hack MA, Choi SJ, Vijayapalan P, et al. Comparison of the effects of sleep deprivation, alcohol and obstructive sleep apnoea (OSA) on simulated steering performance.

Respir Med.



Sagaspe P, Taillard J, Chaumet G, et al. Maintenance of wakefulness test as a predictor of driving performance in patients with untreated obstructive sleep apnea.




Turkington PM, Sircar M, Allgar V, Elliott MW. Relationship between obstructive sleep apnoea, driving simulator performance, and risk of road traffic accidents.




Van Dongen HP, Baynard MD, Maislin G, Dinges DF. Systematic interindividual differences in neurobehavioral impairment from sleep loss: evidence of trait-like differential vulnerability.




Philip P, Sagaspe P, Moore N, et al. Fatigue, sleep restriction and driving performance.

Accid Anal Prev.



Moller H, Devins G, Shen J, Shapiro C. Sleepiness is not the inverse of alertness: evidence from four sleep disorder patient groups.

Exp Brain Res.



Bonnet MH. The MSLT and MWT should not be use for the assessment of workplace safety.

J Clin Sleep Med.



George CF. Reduction in motor vehicle collisions following treatment of sleep apnoea with nasal CPAP.




Hack M, Davies RJ, Mullins R, et al. Randomised prospective parallel trial of therapeutic versus subtherapeutic nasal continuous positive airway pressure on simulated steering performance in patients with obstructive sleep apnoea.




Turkington PM, Sircar M, Saralaya D, Elliott MW. Time course of changes in driving simulator performance with and without treatment in patients with sleep apnoea hypopnoea syndrome.




Orth M, Duchna H-W, Leidag M, et al. Driving simulator and neuropsychological testing in OSAS before and under CPAP therapy [published correction appears in Eur Respir J. 2006;27:242].

Eur Respir J.



Mazza S, Pépin JL, Naëgelé B, et al. Driving ability in sleep apnoea patients before and after CPAP treatment: evaluation on a road safety platform.

Eur Respir J.



American Thoracic Society. Sleep apnea, sleepiness, and driving risk. Official statement of the American Thoracic Society.

Am J Respir Crit Care Med.

1994;150(5 pt 1):1463-1473.


Hartenbaum N, Collop N, Rosen IM, et al. Sleep apnea and commercial motor vehicle operators:statement from the joint task force of the American College of Chest Physicians, the American College of Occupational and Environmental Medicine, and the National Sleep Foundation.


2006;130: 902-905.


Redline S, Budhiraja R, Kapur V, et al. The scoring of respiratory events in sleep: reliability and validity.

J Clin Sleep Med.



George CF. Sleep apnea, alertness, and motor vehicle crashes.

Am J Respir Crit Care Med.