FMCSA Contact: Albert Alvarez, MC-RTR, 202-385-2387; TTY: 800-877-9339

The goal of the Federal Motor Carrier Safety Administration (FMCSA) is to reduce the large truck fatality rate by 41% from 1996 to 2008. This reduction translates into a rate of 1.65 fatalities in truck crashes per 100 million miles of truck travel.

FMCSA's Research and Technology programs encompass a range of issues and disciplines, all related to motor carrier and bus safety and security. FMCSA defines a "research program" as any systematic study directed toward fuller scientific discovery, knowledge, or understanding that will improve safety, and reduce the number and severity of commercial motor vehicle crashes. Similarly, a "technology program" defines those programs that adopt, develop, test, and/or deploy innovative driver and/or vehicle best practices, and technologies that will improve safety and reduce the number and severity of commercial motor vehicle crashes.

Currently, FMCSA's Office of Research and Technology is conducting programs in order to produce safer drivers, improve safety of commercial motor vehicles, produce safer carriers, advance safety through information-based initiatives, and improve security through safety initiatives. The study described in this Tech Brief was designed and developed to support the FMCSA Research & Technology strategic objective to produce safer drivers. The primary goals of this initiative are to ensure that commercial drivers are physically qualified, trained to perform safely, and mentally alert.

Background

The primary objectives of the study were to:

• Obtain additional and more meaningful crash data by linking the University of Pennsylvania (UPenn) sleep apnea database to the FMCSA Motor Carrier Management Information System (MCMIS) crash database,
• Understand the impact of sleep apnea and driver impairment on crash involvement, the number of crashes, and the severity of crashes, and
• Gain insight into how crash rates are impacted before and after drivers are diagnosed with sleep apnea.

The study used a modeling approach to analyze specific data regarding truck drivers and sleep apnea. The hypothesis tested was that "A driver diagnosed with sleep apnea is more likely to be involved in a motor vehicle crash than a driver with no history or symptoms of sleep apnea, after controlling for differences in the other predictor variables included in the model." The data gained in this study did not support this hypothesis.

Study Design and Methodology

University of Pennsylvania Sleep Apnea Study

• Estimate the prevalence of sleep apnea among a sample of commercial truck drivers living in Pennsylvania within 50 miles of the University of Pennsylvania,
• Examine the relationship in commercial truck drivers between severity of sleep apnea and decreased function related to driving performance, and
• Develop a profile of an overall sample of commercial truck drivers with regard to their sleep apnea-related characteristics and risks.

The subjects were chosen after UPenn sent a multivariable apnea prediction questionnaire to a random sample of 4,826 commercial driver's license (CDL) holders residing in Pennsylvania. From the 1,391 responses that were returned, 406 drivers were selected to participate in the overnight laboratory study. The overall findings from this study revealed that 28 (6.9 percent) of the 406 participants were diagnosed with severe sleep apnea, 32 (7.9 percent) had moderate sleep apnea, 86 (21.2 percent) had mild sleep apnea, and 260 (64 percent) had none.

The results of the study showed that the prevalence rates of sleep apnea among commercial truck drivers are similar to sleep apnea rates found in other general populations. The study also revealed that the prevalence of sleep apnea depends on the relationship between two major factors - age and degree of obesity as measured by body mass index (BMI) - with the prevalence of sleep apnea increasing with increasing age and BMI. Another meaningful study finding showed that the prevalence of sleep apnea depends on the average duration of sleep over consecutive nights at home. Short sleep duration, six hours or less per night, results in an increase in the prevalence of sleep apnea.

Motor Carrier Management Information System (MCMIS) Crash Database

The MCMIS crash database contains a census of trucks and buses involved in fatal, injury, and tow-away crashes. It is maintained by FMCSA and contains data on trucks and buses in crashes that meet the uniform crash data standards developed through the National Governors Association (NGA). An NGA-reportable crash involves a truck (a vehicle that is designed, used, or maintained primarily for carrying property and has at least two axles and six tires) or a bus (a vehicle with a seating capacity of at least nine people, including the driver). The crash must result in either at least one fatality, at least one injury for which the injured person was taken to a medical facility for immediate medical attention, or at least one vehicle that was towed from the crash scene as a result disabling crash damage. The crashes are reported by the individual states and transmitted to FMCSA through a microcomputer-based software system called SAFETYNET. While some states do not report all NGA-eligible crashes and the MCMIS crash file captures only approximately 60 percent of all NGA-reportable crashes per year, it was confirmed that Pennsylvania, the state in which the majority of crashes occurred involving the drivers used in this study, commonly reports about 96 percent of all truck crashes to MCMIS.

Study Database

Methods of Analysis

The data were also analyzed using logistic regression, which is a mathematical modeling approach that can be used to describe the relationship of several independent or predictor variables to a binary or dichotomous dependent variable, such as crash or no crash. Logistic regression models identify factors that affect the likelihood of an outcome and can be used to predict the outcome of an event. Because the outcome of interest for this study is a heavy truck-related crash, logistic regression models were used to predict the occurrence of a crash event based on a set of predictor variables, which included such driver characteristics as the presence and severity of sleep apnea, age, gender, alcohol and drug use, and annual miles driven.

Finally, crash rates were normalized by the number of miles driven annually for each driver, and a Poisson regression model was estimated. Poisson regression analysis is a regression technique used for modeling dependent variables that describe discrete data. The methodology of Poisson regression analysis assumes that the underlying distribution of the dependent variable under consideration is Poisson. The Poisson distribution is often used to model the occurrence of rare events, such as the number of motor vehicle crashes occurring either over time or as a function of the number of miles driven. Thus, Poisson regression is an appropriate statistical technique for analyzing crash rates.

Results

Looking at the seven year period following participation in the UPenn sleep study (1996 - 2003) in Table 1, crash involvement among sleep apnea patients was shown to be higher than in the comparison group - 14.4 percent vs. 11.2 percent. But the chi-square statistic, used in this case to test an association between crashes and sleep apnea, was not significant (p = 0.3). The results for all crashes over the entire 14 year period can be similarly interpreted. Therefore, there is no statistical evidence in these data to suggest that the presence of sleep apnea significantly increases the likelihood or the risk of motor vehicle crashes.

Table 1: 2x2 Contingency Table: Presence of Sleep Apnea vs. Crash Involvement (Post-Diagnosis)

Looking at the results in a logistic regression model, with odds ratios adjusted for potential confounding effects of various factors by including them as independent variables, the same results are revealed. Various factors included age, gender, alcohol use, body mass index, self-reported daytime sleepiness, presence or absence of high blood pressure, use or nonuse of medications causing drowsiness, unusual work/sleep schedule, and miles driven per year. The odds ratios included 95 percent confidence intervals, which indicated that the relationship between sleep apnea severity and motor vehicle crashes is not statistically significant. The null hypothesis being tested in the logistic regression analysis is that the two possible outcomes, crash or no crash, are equally likely (odds ratio = 1). Since the value of 1 is contained in every confidence interval at a 95 percent level of significance in this model, the null hypothesis of no association cannot be rejected, and, therefore, it was concluded that there is no statistical evidence in the data to suggest that drivers with sleep apnea are more likely than drivers without sleep apnea to have a commercial vehicle crash.

If the presence of sleep apnea in commercial drivers does not significantly influence the crash risk, the question arises as to whether the severity of sleep disturbances contributes to crashes. However, once again, the data lacked statistical evidence of a positive association between sleep apnea severity and post-diagnosis crash involvement, as indicated by an insignificant chi-square test (p = 0.7). Similar findings of no association of sleep apnea severity were obtained for pre-diagnosis crashes and all crashes, as well.

Next, the hypothesis that drivers with various degrees of sleep apnea had a greater likelihood of being involved in multiple crashes over the 14 year period was investigated. No association was found between sleep apnea presence or severity and multiple crashes. This suggests that the commercial drivers in this study who were diagnosed with sleep apnea were not at increased risk for having more than one crash over the 14 year period prior to and following diagnosis.

In the final logistic regression analysis, the relationship between sleep apnea severity and the severity of motor vehicle crashes was assessed. In this analysis, a significant relationship was found between severe sleep apnea and severe crashes (those which include multiple injuries and vehicle towed from accident scene). Drivers with severe sleep apnea were 4.6 times more likely to be involved in a severe crash in the seven year period than drivers with no sleep apnea. However, it was also found that drivers with severe sleep apnea were no more likely to involved in moderate/minor crashes than those without sleep apnea. Moreover, no statistically significant relationship was found between either moderate or mild sleep apnea and crash severity.

A Poisson regression model was estimated to explore the impact of sleep apnea on crash rates, or crashes as a function of miles driven. The results indicated that none of the sleep apnea severity categories had a significant impact on whether or not a driver had a crash.

Conclusions

The results of this study contradict those found in several previous studies, which found a strong positive relationship between sleep apnea and motor vehicle crashes. While it appears that sleep apnea is not a risk factor for vehicular crashes among this sample of CMV drivers, the limitations of the data must be acknowledged, and these results must be interpreted with caution. For example, the data includes many confounding variables, which were based on self-reported information, and many drivers may have misinterpreted some survey questions. Incomplete crash and driving records add to the limitations of the study, as well. Finally, the large majority of subjects analyzed in this study were short-haul drivers who operated local routes, generally driving in dense urban environments requiring a higher level of alertness than long-haul truckers, who drive long stretches on interstate highways under monotonous driving conditions and who are more susceptible to fatigue and daytime sleepiness. These are just a few of the limitations of the data included this study.

Given the limitations in the data and the results of these analyses, some recommendations for future research are suggested. These include:

• (1) Obtain complete and accurate crash records for every subject in the study population.
• (2) To the extent possible, obtain prospective information regarding treatments advised and actually administered to the study subjects for the sleep disorders diagnosed during clinical testing. These data should include the recommended treatment, if any, and follow up on treatment receipt and effectiveness.
• (3) Future studies should target long-haul commercial truck drivers who operate heavy tractor-trailer vehicles. In over-the-road highway driving, the longer distances and monotonous routine may contribute to drowsiness, and the higher speeds may increase the danger to drivers.

With these additions, a more reliable database can be created and a stronger case can be made from the results of a future study.

References

Information

Researchers:
This study was performed at the Volpe National Transportation Systems Center, Cambridge, MA 02142-1093, by Lawrence C. Barr, along with Linda Boyle (University of Iowa), and Greg Maislin, (University of Pennsylvania Biomedical Statistical Consulting). PPA No. SA1J/WPB No. AB136.

Availability:
The study final report will be made available from USDOT's Information Services Online Publishing at http://isddc.dot.gov.

Key Words:
Apnea, commercial motor vehicle, crash, crash risk, sleep, sleep disorder, truck, truck driver.

Notice:
This Tech Brief is disseminated under the sponsorship of the Department of Transportation in the interest of information exchange. The Tech Brief provides a synopsis of the study's final publication. The Tech Brief does not establish policies or regulations, nor does it imply USDOT endorsement of the conclusions or recommendations. The U.S. Government assumes no liability for its contents or their use.

Web Site:
FMCSA Tech and Analysis Briefs may be accessed at: http://www.fmcsa.dot.gov.

Technical Writer:
Kirse Kelly, C2 Technologies, Inc.

[U.S. Department of Transportation Logo]
U.S. Department of Transportation
Federal Motor Carrier Safety Administration

September 2004
Publication No. FMCSA-RT-04-007