Issue Navigator

Volume 15 No. 11
Earn CME
Accepted Papers

Scientific Investigations

Effect of Positive Airway Pressure Therapy on Drowsy Driving in a Large Clinic-Based Obstructive Sleep Apnea Cohort

Harneet K. Walia, MD1; Nicolas R. Thompson, MS2; Maeve Pascoe, BS1; Maleeha Faisal, MD1; Douglas E. Moul, MD, MPH1; Irene Katzan, MD, MS3; Reena Mehra, MD, MS1; Nancy Foldvary-Schaefer, DO, MS1
1Sleep Disorders Center, Neurological Institute, Cleveland Clinic, Cleveland, Ohio; 2Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, Ohio; 3Cerebrovascular Center, Neurological Institute, Cleveland Clinic, Cleveland, Ohio


Study Objectives:

Drowsy driving related to obstructive sleep apnea (OSA) represents an important public health problem with limited data on the effect of positive airway pressure (PAP) therapy. We hypothesize that PAP therapy will reduce self-reported drowsy driving in a large clinic-based OSA cohort.


Drowsy driving (self-reported near-accidents/accidents) incidents from baseline to after PAP therapy (stratified by adherence) were compared in a cohort of 2,059 patients with OSA who initiated PAP therapy from January 1, 2010 to December 31, 2014. Multivariable logistic regression models evaluated the dependence of change in drowsy driving incidents on other factors, including change in Epworth Sleepiness Scale (ESS) and Patient Health Questionnaire-9 (PHQ9) scores.


In the entire cohort (age 56.0 ± 13.1 years, 45.4% female, 76.0% white, average follow-up 124.4 ± 67.3 days), drowsy driving incidents reduced from 14.2 to 6.9% after PAP therapy (P < .001). In subgroups, drowsy driving incidents reduced from 14% to 5.3% (P < .001) in patients who self-reported adherence to PAP therapy and 14.1% to 5.3% (P < .001) in patients objectively adherent to PAP therapy. For each one-point improvement in Epworth Sleepiness Scale score, the odds of drowsy driving decreased by about 14% (odds ratio 0.86, 95% confidence interval 0.82 to 0.90).


In this clinic-based cohort, drowsy driving improved after adherent PAP usage, with greater drowsy driving risk for those with greater sleep propensity. This highlights the importance of and need for routine drowsy driving assessments and careful clinical attention to PAP adherence and sleep propensity in this population. Our findings should be confirmed and may be used to provide support for initiatives to address the public health issue of drowsy driving.


Walia HK, Thompson NR, Pascoe M, Faisal M, Moul DE, Katzan I, Mehra R, Foldvary-Schaefer N. Effect of positive airway pressure therapy on drowsy driving in a large clinic-based obstructive sleep apnea cohort. J Clin Sleep Med. 2019;15(11):1613–1620.


Current Knowledge/Study Rationale: Drowsy driving risk due to untreated obstructive sleep apnea (OSA) has been recognized by the American Academy of Sleep Medicine as a significant public health issue. Positive airway pressure (PAP) therapy has been shown to reduce motor vehicle accident risk in people with OSA in limited populations; our analyses were conducted to evaluate broad drowsy driving risk in a large, United States clinical cohort with OSA in relation to multiple measures of PAP adherence and self-reported sleepiness.

Study Impact: Our study, demonstrating the associations of PAP treatment and drowsy driving in patients with OSA, informs recommendations for clinical practice. Clinical care paths for patients with OSA should include more frequent drowsy driving screenings and greater efforts to optimize PAP adherence in those who report a history of drowsy driving.


Obstructive sleep apnea (OSA) is a highly prevalent disorder, estimated to affect up to 49% of men and 23% of women in the general population.1 OSA is accompanied by sleep fragmentation, increased arousals, and intermittent hypoxia, causing adverse consequences spanning from cardiovascular morbidities to decreased vigilance.2 The decline in vigilance can cause delayed reaction times in those with OSA, placing patients at a threefold to 13-fold increased risk of motor vehicle accidents (MVAs),36 and making OSA the third most common risk factor for MVAs.7 In the year 2000, OSA-related MVAs cost nearly $16 billion and 1,400 lives.8

Because of the dangers of drowsy driving relating to untreated OSA, the American Academy of Sleep Medicine (AASM) recommends routine clinical screenings of drowsy driving in patients with OSA as one of the most impactful process measures to improve patient outcomes and reduce the public health burden of OSA.2 The American Thoracic Society echoes the AASM’s directive, emphasizing the importance of treating OSA to mitigate drowsy driving risk.9 In order to improve patient outcomes, effective treatment of OSA is necessary.10 In recent meta-analyses of MVA risk, global and sleep-specific quality of life measures improved significantly after positive airway pressure (PAP) therapy in patients with OSA and MVA risk reduced by 72%.1114 However, limited data are available for PAP adherence in relation to MVA risk.15,16

We therefore investigated self-reported drowsy driving incidents after PAP therapy in a large clinical cohort presenting to a United States sleep center for a large clinical catchment area. Our specific aims were to evaluate: (1) change in drowsy driving incidents from baseline to follow-up after PAP therapy stratified by PAP adherence; and (2) potential modulators of the effect of PAP adherence on change in drowsy driving incidents. We hypothesize that self-reported drowsy driving incidents would decrease after PAP therapy usage and adherence. We also postulate that change in Epworth Sleepiness Scale (ESS) score would affect self-reported drowsy driving at follow-up.


Patient Population

Patients with OSA age 18 years or older who reported to be on PAP therapy at the Cleveland Clinic Sleep Disorders Center in the electronic health record (EHR) (Epic Systems Corporation, Verona, Wisconsin, USA) between January 1, 2010 and December 31, 2014 were included. Patients were required to have a baseline visit with no self-reported PAP usage prior and a subsequent visit with self-reported PAP use. The follow-up visit closest to 6 months after the baseline visit was used for analysis. Follow-up visits were required to be at least 30 days but no more than 1 year after the last date where no PAP use was reported. Patients who did not complete both the ESS and the drowsy driving question (described in the next section) at both baseline and follow-up visits were excluded. The study was approved by Institutional Review Board of Cleveland Clinic.

Electronic Patient-Reported Data Collection

The Cleveland Clinic’s Knowledge Program was used to electronically collect patient-reported outcome data. These data, collected at each clinic visit, is entered into the patient’s EHR at the point of care.17 For this study, accident/near-accident data, ESS scores, Patient Health Questionnaire (PHQ-9) scores, and PAP usage were recorded through the Knowledge Program. Drowsy driving incidents were assessed using the question, “How many times have you had any near accidents/accidents due to drowsy driving or falling asleep when driving over the past 4 weeks?” (Similar to other reports using independent questions to assess drowsy driving via near-accidents/accidents).18 Reports of ≥ 1 were considered as drowsy driving. The ESS measures self-reported daytime sleep propensity with scores ranging from 0 to 24, and scores > 10 indicating excessive daytime sleepiness (EDS) with a sensitivity of 94% and specificity of 100% compared to findings from the Multiple Sleep Latency Test.19,20 The PHQ-9, a self-administered, validated questionnaire, was used to detect depressive symptoms.21 Self-reported PAP adherence was determined by asking the patient to report number of days used per week and hours used per day. PAP adherence was defined as use for ≥ 5 days per week with ≥ 4 hours of use per day in the past 4 weeks, consistent with the Centers for Medicare and Medicaid Services adherence criteria.22 Objective PAP adherence was assessed in patients with available data obtained from their durable medical equipment company.

Sleep Study Data

Sleep study data were collected from overnight clinical polysomnography using the Polysmith system (Nihon Kohden Corporation, Tokyo, Japan). Initial measurements included body mass index (BMI, kg/m2) and neck circumference (cm). Oronasal thermistor and nasal cannula were used to measure nasal airflow and pressure, which were used to calculate the apnea-hypopnea index (AHI); airflow ≥ 50% in the nasal pressure channel for ≥ 10 seconds resulting in an arousal or ≥ 3% oxygen desaturation constituted a hypopnea, whereas a decrease in amplitude of oronasal thermistor signal by 90% for ≥ 10 seconds constituted an apnea.23

Statistical Methods

Descriptive statistics (eg, means, standard deviations, medians, interquartile range, frequencies, and percentages) were computed for the entire sample and stratified by drowsy driving self-report. Comparisons were made using t tests or Mann-Whitney U tests for continuous variables and chi-square tests for categorical variables. To handle missing data, multiple imputation was used to create and analyze 10 imputed datasets.24 Incomplete variables were imputed under fully conditional specification using the default settings of the mice 2.25 package.25,26 Model parameters were estimated with multivariable logistic regression applied to each imputed dataset separately. All computations were done in R version 3.5.0 (The R Foundation, Vienna, Austria) and values of P < .05 were considered statistically significant.27

The frequency and percent of drowsy driving incidents were computed at baseline and follow-up in the full sample and stratified by PAP adherence. McNemar test was used to determine if the percentage of drowsy driving incidents changed from baseline to follow-up after PAP therapy. Multivariable logistic regression models evaluated the association between PAP adherence and change in drowsy driving. Two models were created: one that included all patients and one that included only those with objective PAP adherence data. Odds ratios and associated 95% confidence intervals were computed. The models were adjusted for baseline drowsy driving incidents as well as age, sex, race (Caucasian, African American, other), smoking status (current, former, never), median income by ZIP code (based on 2010 United States Census data), AHI (5–14.99, 15–29.99, ≥ 30 events/h), BMI, neck circumference, antidepressant use, self-reported average sleep time (≥ 7 hours versus < 7 hours), baseline ESS score, change in ESS score from baseline to follow-up, and PHQ-9 score. The following comorbidities, gathered from the EHR, were also included as covariates: cancer, chronic renal failure, diabetes, depression, coronary artery disease, hypertension, stroke, and atrial fibrillation. A binary indicator for PAP adherence was also included. In the model including all patients, self-reported PAP adherence was used whereas objective PAP adherence was used in the model for patients who had objective adherence data. As a sensitivity analysis, we also ran a model in the patients with objective adherence data but used self-reported PAP adherence as the predictor.


The flow diagram for the study sample is shown in Figure 1. The initial data query resulted in 2,963 patients with OSA age 18 years or older. After exclusions, of 1,995 patients with self-reported PAP adherence data at follow-up, 1,614 (80.9%) were adherent. Objective PAP adherence data were available in 1,029 patients, of whom 715 (69.5%) were adherent. Agreement between objective and self-reported adherence was 80.7%. Cohen kappa statistic was 0.48 (95% confidence interval 0.42 to 0.55), indicating moderate agreement.

Flow diagram for study sample.

The final cohort included 2,059 patients, with self-reported PAP adherence data from 1995 respondents and objective PAP adherence data from 1,029 respondents. ESS = Epworth Sleepiness Scale, PAP = positive airway pressure.


Figure 1

Flow diagram for study sample.

(more ...)

Sample Characteristics

Sample characteristics are shown in Table 1. Mean age was 56.0 years (SD 13.1) with 45.4% female, and 76.0% Caucasian. Average follow-up was 124.4 days (SD 67.3). Patients who reported any drowsy driving at baseline were younger (53.1 versus 56.4 years, P < .001), less likely to have cancer (15.0% versus 20.1%, P = .050), more likely to be on antidepressants (47.1% versus 40.3%, P = .033), and had higher ESS score (14 versus 8, P < .001) and PHQ-9 score (11 versus 6, P < .001).

Sample characteristics for entire sample and stratified by drowsy driving at baseline.


table icon
Table 1

Sample characteristics for entire sample and stratified by drowsy driving at baseline.

(more ...)

At baseline, 293 patients (14.2%) reported drowsy driving. The difference in baseline self-reported drowsy driving between self-reported PAP adherence (14.0%, 226/1,614) and nonadherence (16.0%, 61/381) was not significant (P = .356). Similarly, no difference in baseline drowsy driving was found between objective PAP adherent (14.1%, 101/715) and nonadherent (14.6%, 46/314) groups (P = .901).

Drowsy Driving Outcomes After PAP Therapy

At follow-up, drowsy driving incidents declined, with only 142 of all patients (6.9%) reporting drowsy driving (P < .001) including reduction from 14.0 to 5.3% in self-reported PAP adherent patients (P < .001, Figure 2A). However, in self-reported nonadherent patients, the change in drowsy driving incidents from 16.0% to 13.1% was not significant (P = .215). In objective PAP adherent and nonadherent groups, drowsy driving incidents reduced from 14.1 to 5.3% (P < .001) and from 14.6% to 8.6% (P = .005), respectively (Figure 2B).

Accidents and near accidents.

Graphs of accidents/near accidents (drowsy driving incidents) before and after positive airway pressure treatment, stratified by adherence/nonadherence for self-reported (A) and objective (B) measures of adherence.


Figure 2

Accidents and near accidents.

(more ...)

Among patients who reported drowsy driving incidents at baseline, 79.6% of self-reported PAP adherent and 62.3% of nonadherent patients reported zero incidents at follow-up; these percentages were significantly different (P = .008). Similar, but nonsignificant, results were observed for objective PAP adherence (80.2% versus 65.2%, P = .080). Among patients who reported zero drowsy driving incidents at baseline, 2.9% of self-reported PAP adherent and 8.4% of nonadherent patients reported drowsy driving incidents at follow-up (P < .001). Analogous values for objective adherence were 2.9% and 4.1% (P = .488).

Table 2 displays the results of the multivariable logistic regression models for follow-up drowsy driving. After adjustment for other covariates, the odds of reporting drowsy driving was about half for self-reported adherent patients compared to nonadherent patients (odds ratio [OR] 0.53, 95% confidence interval [CI] 0.35 to 0.82). The same difference was not observed in objectively adherent versus nonadherent patients (OR 0.99, 95% CI 0.53 to 1.86). Our sensitivity analysis model, which used self-reported adherence as the predictor instead of objective adherence within the subset of patients with objective adherence data, yielded an OR for adherence of 0.72 (95% CI 0.36 to 1.41, P = .338).

Results of multivariable logistic regression models for any drowsy driving incident by adherence.


table icon
Table 2

Results of multivariable logistic regression models for any drowsy driving incident by adherence.

(more ...)

Drowsy Driving and Self-Reported Sleepiness

There was a statistically significant association between improvement in ESS and drowsy driving at follow-up (Table 2). After adjusting for clinical characteristics, for each one-point improvement in ESS, the odds of drowsy driving decreased by about 14% (OR 0.86, 95% CI 0.82 to 0.90). A similar effect was observed in patients with objective adherence data (OR 0.86, 95% CI 0.80 to 0.92). The relationship between change in ESS score and self-reported follow-up drowsy driving is shown in Figure 3. The plot was created using the model fit in all patients and assumes a PAP adherent patient with a baseline ESS score of 15, median values for other covariates, and reference categories for categorical covariates.

Relationship between change in ESS score and percentage of patients reporting drowsy driving at follow-up.

Plot was created using the model fit in all patients and assumes a self-reported positive airway pressure adherent patient with a baseline ESS score of 15, median values for continuous covariates, and reference categories for categorical covariates. The dashed lines represent a 95% confidence band. ESS = Epworth Sleepiness Scale.


Figure 3

Relationship between change in ESS score and percentage of patients reporting drowsy driving at follow-up.

(more ...)

Baseline drowsy driving and higher baseline ESS score were associated with greater odds of follow-up drowsy driving. No other predictors were significantly associated with follow-up drowsy driving.


We investigated change in drowsy driving with PAP therapy in a large clinic-based cohort of OSA presenting to a United States sleep center for a large clinical catchment area. Our main findings were: (1) there was a significant reduction in drowsy driving after PAP usage; (2) the reduction in drowsy driving was more pronounced in those who were adherent to PAP therapy; (3) after adjusting for covariates, the odds of reporting drowsy driving was about half for self-reported PAP adherence than nonadherence; (4) reduction in drowsy driving was associated with reduction in ESS scores.

This is the largest clinical cohort from which self-reported drowsy driving has been compared both to self-reported and objective PAP adherence. Our findings are consistent with a recent Swedish study examining the effect of objective PAP adherence on MVA rates.14 Our study expands upon these data to include self-reported PAP adherence as well as all drowsy driving. Additionally, Sweden is a world leader in traffic safety and has the lowest MVA-related death rate in Europe,28 with only 2% to 4% of MVAs related to drowsy driving.29 Our study addresses a large clinical cohort in the United States, where drowsy driving is estimated to cause nearly 7% of all MVAs,30 and the National Transportation Safety Board has issued recommendations directly pertaining to reduction of drowsy driving risk for individuals with or at high risk of OSA in their 2019 Most Wanted List of Traffic Safety Improvements.31 Because drowsy driving has safety implications for both the drowsy driver and others on the road, by evaluating the effectiveness of current treatment interventions for a common condition, this study identifies the need for practical interventions to address a pertinent public health issue.

Recent studies have also shown that sleep propensity, as measured by the ESS, is predictive of MVAs.3,14,32 The relationship between sleep propensity and drowsy driving, independent of OSA severity, was demonstrated in our study: at baseline, AHI was not significantly different between drowsy and nondrowsy drivers, and interestingly, baseline AHI was not associated with follow-up drowsy driving in our model. This may be attributable to contribution of confounding influences of other covariates in our model (eg, baseline ESS score, baseline drowsy driving incidents, habitual sleep duration) associated with both AHI and follow-up drowsy driving, which reduced the association of AHI on follow-up drowsy driving. But nonetheless, drowsy drivers had a significantly higher ESS. Clinically significant drowsy driving improvements have been observed in patients with self-reported sleep propensity who use PAP therapy,33,34 and our study highlights the relationship between adherent PAP usage and ESS improvements.

Our study is strengthened by the use of real world, clinical effectiveness data that directly align with the AASM’s call to action to assess drowsy driving in clinical OSA evaluations. These evaluations allowed for assessment of near-accidents, which are not available with police reports,35 but may be subject to recall and reporting bias. Additionally, our analyses were strengthened through accounting for potential confounding comorbidities that could affect drowsy driving incidence.10 Our study was limited by the use of a clinical population and limited objective PAP adherence data, which pose the risk of selection bias. Furthermore, although epidemiologic studies suggest that sleep-related MVAs constitute 10% to 20% of all MVAs,36 and we saw a baseline prevalence of 14% in our sample, we do not have comparative data to ascertain selection bias or whether our sample falls toward or away from the mean of drowsy driving events in our broader clinical population of patients with OSA, which may be different from national averages. Examination of a clinical population, however, has direct clinical relevance and implications, perhaps more effectively informing patient care in terms of patient education about drowsy driving risks and benefits of OSA treatment. Another limitation is that objective adherence data were available to us only as a binary indicator, based on Centers for Medicare and Medicaid Services adherence criteria, which did not allow us to evaluate the effect of PAP usage that was near to, but did not cross the threshold of, adherence.

Additionally, there was a difference in the association of adherence on follow-up drowsy driving for self-reported and objective adherence, with self-reported adherence showing a fairly strong association (OR 0.53) and objective adherence showing essentially no association (OR 0.99). There are several possible explanations for this counterintuitive result. Objectively nonadherent patients may be partially adherent and still derive benefit from PAP usage. It is also possible that partially objective PAP-adherent patients self-report as being adherent. In our sample, agreement between objective and self-reported adherence was 80.7%, but the Cohen kappa statistic (0.48) revealed only moderate agreement. Another likely explanation for the differing result is that bias may have been introduced by restricting analysis to those patients who had objective adherence data. To partially address this, we performed a sensitivity analysis to examine the effect of self-reported adherence on follow-up drowsy driving in the subset of patients with objective adherence data. The OR in this model was 0.72, which is closer to the null value of 1 than that observed in the full cohort (0.53). To further assess the possibility of selection bias, we compared patient characteristics of those who did and did not have objective adherence data. Patients who had objective adherence data were older, more likely to be a never smoker, had slightly higher BMI, were less likely to have diabetes, coronary artery disease, hypertension, and atrial fibrillation, and had lower baseline PHQ-9 scores (data available upon request). Other studies have shown that MVAs negatively correlate with objective PAP usage,11,37 so there is reason to believe that drowsy driving is negatively correlated with both self-reported and objective PAP usage. However, this should be further investigated in future studies.

Given the association between OSA treatment and drowsy driving, the integration of routine drowsy driving assessments into clinical care paths for patients with OSA is recommended. Furthermore, given that the effectiveness of PAP therapy for reducing drowsy driving is related to adherence, further steps should be taken to optimize PAP adherence particularly in those who report a history of drowsy driving. Because those with greater sleep propensity may be more susceptible to drowsy driving, clinical care paths for these patients should include more frequent drowsy driving screenings and particular attention to the treatment of their sleep-related symptoms.


The authors have seen and approved the manuscript. Work for this study was performed at the Cleveland Clinic. Harneet Walia reports grant support from Resmed for a different study. Dr. Mehra reports receiving National Institutes of Health funding support from the National Heart, Lung, and Blood Institute [U01HL125177, UG3HL140144] and the American Heart Association. Her institution has received positive airway pressure devices and equipment from Philips Respironics, ResMed, GE Healthcare, and Natus for research. R. M. has received honorarium from the American Academy of Sleep Medicine for speaking; serves as a consultant for Respicardia, Enhale and Merck; received funds for service on the American Board of Medicine Sleep Medicine Exam test writing committee and received royalties from UpToDate. Nancy Foldvary-Schaefer is a consultant for Jazz Pharmaceuticals. Nicolas Thompson has received grant support from Novartis Pharmaceuticals. The other authors report no conflicts of interest.



American Academy of Sleep Medicine


apnea-hypopnea index


body mass index


excessive daytime sleepiness


electronic health record


Epworth Sleepiness Scale


motor vehicle accident


obstructive sleep apnea


positive airway pressure therapy


Patient Health Questionnaire 9



Heinzer R, Vat S, Marques-Vidal P, et al. Prevalence of sleep-disordered breathing in the general population: the HypnoLaus study. Lancet Respir Med. 2015;3(4):310–318. [PubMed Central][PubMed]


Aurora RN, Collop NA, Jacobowitz O, Thomas SM, Quan SF, Aronsky AJ. Quality measures for the care of adult patients with obstructive sleep apnea. J Clin Sleep Med. 2015;11(3):357–383. [PubMed Central][PubMed]


Arita A, Sasanabe R, Hasegawa R, et al. Risk factors for automobile accidents caused by falling asleep while driving in obstructive sleep apnea syndrome. Sleep Breath. 2015;19(4):1229–1234. [PubMed Central][PubMed]


Vakulin A, D’Rozario A, Kim JW, et al. Quantitative sleep EEG and polysomnographic predictors of driving simulator performance in obstructive sleep apnea. Clin Neurophysiol. 2016;127(2):1428–1435. [PubMed]


Terán-Santos J, Jimenez-Gomez A, Cordero-Guevara J. The association between sleep apnea and the risk of traffic accidents. N Engl J Med. 1999;340(11):847–851. [PubMed]


Powell NB, Riley RW, Schechtman KB, Blumen MB, Dinges DF, Guilleminault C. A comparative model: Reaction time performance in sleep-disordered breathing versus alcohol-impaired controls. Laryngoscope. 1999;109(10):1648–1654. [PubMed]


Todea D, Herescu A. Modern and multidimensional approach of sleep apneea as a public health problem. Clujul Med. 2013;86(1):10–15. [PubMed Central][PubMed]


Sassani A, Findley L, Kryger M, Goldlust E, George C, Davidson T. Reducing motor-vehicle collisions, cost, and fatalities by treating obstructive sleep apnea syndrome. Sleep. 2004;27(3):453–458. [PubMed]


Mukherjee S, Patel SR, Kales SN, et al. An official American Thoracic Society statement: the importance of healthy sleep: recommendations and future priorities. Am J Respir Crit Care Med. 2015;191(12):1450–1458. [PubMed Central][PubMed]


Strohl KP, Brown DB, Collop N, et al. An official American Thoracic Society clinical practice guideline: sleep apnea, sleepiness, and driving risk in noncommercial drivers - an update of a 1994 statement. Am J Respir Crit Care Med. 2013;187(11):1259–1266. [PubMed Central][PubMed]


Tregear S, Reston J, Schoelles K, Phillips B. Continuous positive airway pressure reduces risk of motor vehicle crash among drivers with obstructive sleep apnea: systematic review and meta-analysis. Sleep. 2010;33(10):1373–1380. [PubMed Central][PubMed]


McDonald AD, Lee JD, Aksan NS, Dawson JD, Tippin J, Rizzo M. Highway healthcare: how naturalistic driving data index adherence to CPAP therapy in obstructive sleep apnea. Proc Hum Factors Ergon Soc Annu Meet. 2013;57(1):1859–1863. [PubMed Central][PubMed]


Antonopoulos CN, Sergentanis TN, Daskalopoulou SS, Petridou ET. Nasal continuous positive airway pressure (nCPAP) treatment for obstructive sleep apnea, road traffic accidents and driving simulator performance: a meta-analysis. Sleep Med Rev. 2011;15(5):301–310. [PubMed]


Karimi M, Hedner J, Häbel H, Nerman O, Grote L. Sleep apnea-related risk of motor vehicle accidents is reduced by continuous positive airway pressure: Swedish Traffic Accident Registry data. Sleep. 2015;38(3):341–349. [PubMed Central][PubMed]


George CFP. Reduction in motor vehicle collisions following treatment of sleep apnoea with nasal CPAP. Thorax. 2001;56(7):508–512. [PubMed Central][PubMed]


Findley LJ, Smith C, Hooper J, Dineen M, Suratt PM. Treatment with nasal CPAP decreases automobile accidents in patients with sleep apnea. Am J Respir Crit Care Med. 2000;161(3):857–859


Katzan I, Speck M, Dopler C, et al. The Knowledge Program: an innovative, comprehensive electronic data capture system and warehouse. AMIA Annu Symp Proc. 2011;2011:683–692. [PubMed Central][PubMed]


Zwahlen D, Jackowski C, Pfäffli M. Sleepiness, driving, and motor vehicle accidents: a questionnaire-based survey. J Forensic Leg Med. 2016;44:183–187. [PubMed]


Johns MW. Sensitivity and specificity of the multiple sleep latency test (MSLT), the maintenance of wakefulness test and the Epworth sleepiness scale: failure of the MSLT as a gold standard. J Sleep Res. 2000;9(1):5–11. [PubMed]


Johns MW. Reliability and factor analysis of the Epworth Sleepiness Scale. Sleep. 1992;15(4):376–381. [PubMed]


Löwe B, Unützer J, Callahan CM, Perkins AJ. Monitoring depression treatment outcomes with the Patient Health Questionnaire-9. Med.Care. 2004;42(12):1194–1201. [PubMed]


Billings ME, Kapur VK. Medicare long-term CPAP coverage policy: a cost-utility analysis. J Clin Sleep Med. 2013;9(10):1023–1029. [PubMed Central][PubMed]


Iber C, Ancoli-Israel S, Chesson AL Jr, Quan SF; for the American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications. 1st ed. Westchester, IL: American Academy of Sleep Medicine; 2007.


Rubin DB. Multiple Imputation for Nonresponse in Surveys. New York, NY: John Wiley & Sons; 1987.


Van Buuren S, Brand JPL, Groothuis-Oudshoorn CGM, Rubin DB. Fully conditional specification in multivariate imputation. J Stat Comput Simul. 2006;76(12):1049–1064


van Buuren S, Groothuis-Oudshoorn K. Mice: multivariate imputation by chained equations in R. J Stat Softw. 2011;45(3)


R Foundation for Statistical Computing. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2016.


Åkerstedt T, Bassetti CL, Cirignotta F, et al. Sleepiness at the Wheel: White Paper. Paris, France: Institut National du Sommeil et de la Vigilance, Autoroutes & Ouvrages Concedes; 2013.


Phillips RO, Sagberg F. Road accidents caused by sleepy drivers: update of a Norwegian survey. Accid Anal Prev. 2013;50:138–146. [PubMed]


Tefft BC. Prevalence of motor vehicle crashes involving drowsy drivers, United States, 1999-2008. Accid Anal Prev. 2012;45:180–186. [PubMed]


National Transportation Safety Board. 2019-2020 Most Wanted List. Accessed September 20, 2019.


Ward KL, Hillman DR, James A, et al. Excessive daytime sleepiness increases the risk of motor vehicle crash in obstructive sleep apnea. J Clin Sleep Med. 2013;9(10):1013–1021. [PubMed Central][PubMed]


Batool-Anwar S, Goodwin J, Kushida C, et al. Impact of continuous positive airway pressure (CPAP) on quality of life in patients with obstructive sleep apnea (OSA). J Sleep Res. 2016;25(6):731–738. [PubMed Central][PubMed]


Aksan N, Marini R, Tippin J, Dawson JD, Rizzo M. Driving Performance and Driver State in Obstructive Sleep Apnea: What Changes with Positive Airway Pressure? Proceedings of the Ninth International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design. June 26-29, 2017. Manchester Village, Vermont.


Howard ME, Jackson ML, Stevenson M. Who needs sleep apnea treatment for safety critical tasks—are we there yet? Sleep. 2015;38(3):331–332. [PubMed Central][PubMed]


Smolensky MH, Di Milia L, Ohayon MM, Philip P. Sleep disorders, medical conditions, and road accident risk. Accid Anal Prev. 2011;43(2):533–548. [PubMed]


Andreu AL, Chiner E, Sancho-Chust JN, et al. Effect of an ambulatory diagnostic and treatment programme in patients with sleep apnoea. Eur Respir J. 2012;39(2):305–312. [PubMed]