1

Introduction

Approximately 4,000 fatalities due to truck and bus crashes occur each year, up to 20 percent of which are estimated to involve fatigued drivers. Commercial motor vehicle (CMV) drivers may decide to drive when they are fatigued as a result of either external demands (e.g., delivery schedules) or internal decisions (e.g., a need to get to the desired location). Consequently, the Federal Motor Carrier Safety Administration (FMCSA), an agency of the U.S. Department of Transportation, has established hours-of-service (HOS) regulations that limit the number of hours CMV drivers can drive and the number of hours in which they can engage in other tasks, such as loading and unloading, on both a daily and weekly basis. While HOS regulations are based on a substantial research literature on how they impact fatigue, fatigue and HOS regulations are two among many possible causal factors implicated in crash risk. A better understanding and mitigation of the risk posed by fatigue in commercial driving could be obtained through the acquisition of more relevant data and the use of more targeted quantitative methods.

To help answer questions about the linkages among hours of service, fatigue, highway safety, and the long-term health of CMV drivers, FMCSA requested that the National Academies of Sciences, Engineering, and Medicine, through its Committee on National Statistics, convene the Panel on Research Methodologies and Statistical Approaches to Understanding Driver Fatigue Factors in Motor Carrier Safety and Driver Health; see Box 1-1 for the panel’s detailed statement of task. This request came at an opportune time because many new and anticipated data sources on motor

vehicle crashes and their causes either are now available or soon will be. Examples include on-board electronic recording systems and video data capture and vehicle telematics, often associated with naturalistic driving studies. In addition, to help analyze the more detailed and complex data derived from these new sources, new statistical and computer science methodologies have been developed. In particular, methods for assessing causal relationships have been improved substantially in the past 20

years. Techniques for analyzing data with a complex time and spatial correlation structure, including mixed linear models, also have been developed and are relevant to the analysis of data in this area.

STUDY CONTEXT

Fatigue and Performance in Safety-Sensitivity Occupations

HOS regulations in regulated industries typically specify both the limits of time at work and minimum rest time in order to maintain safety and reduce risk resulting from fatigue. Fatigue generally refers to a subjective sense of weariness, but in work environments, it refers primarily to the objective decline in performance resulting from physical exertion and/or behavioral effort over time, as well as from inadequate time for recovery. The concept of degraded performance with time on task has its origins in the physical meaning of fatigue relative to biology (e.g., continued stimulation leading to temporary loss of responsivity) and physics (e.g., increasing tendency of a material to break down after repeated physical stress). When applied to human performance, fatigue refers to increasing performance variability and instability in behavioral alertness and vigilance due to continued time on task without breaks (Lim et al., 2010) and/or during night work (Neri et al., 2002) and/or after sleep loss (Basner and Dinges, 2011; Lim and Dinges, 2008). These effects are captured in the definition of fatigue as “a biological drive for recuperative rest” (Williamson et al., 2011).

The contribution of inadequate sleep to fatigued performance in safety-sensitive occupations has been of increasing concern as a result of mounting scientific evidence over the past 20 years that obtaining sufficient sleep is essential for optimal brain function and performance (Banks and Dinges, 2007; Lim and Dinges, 2008; Xie et al., 2013; Yang et al., 2014). Population studies reveal that many people experience insufficient sleep for medical and occupational reasons (Basner et al., 2014; Ford et al., 2015). The U.S. Centers for Disease Control and Prevention (CDC) reports that the number of American adults sleeping 6 hours or less per day increased from 38.6 million to 70.1 million between 1985 and 2012, an increase the CDC characterizes as a “public health epidemic” (Centers for Disease Control and Prevention, 2015; Ford et al., 2015). An estimated 37 percent of the adult U.S. population sleeps less than 7 hours per day, which recent evidenced-based consensus reviews and epidemiological studies have found to be the minimum sleep duration to prevent cumulative deterioration in performance and health and increased risk of mortality (Ferrie et al., 2007; Watson et al., 2015a, 2015b). Studies suggest that the majority of short sleepers do not require less sleep than other adults;

rather, these individuals gradually accrue sleep debt over time and report a greater tendency to fall asleep unintentionally (McKnight-Eily et al., 2011; Punjabi et al., 2003). Individuals who obtain less sleep than needed because of work typically sleep longer on nonwork days—a pattern seen in CMV drivers (Dinges et al., 2005b; Hanowski et al., 2007) that suggests many accrue a sleep debt on workdays.

Fatigue is a demonstrated risk to safety in transport and occupational settings (Williamson et al., 2011). In addition to compromising alertness and vigilance, prolonged time on task, inadequate breaks, reduced sleep time, and limited time off duty, individually and collectively, can slow reaction time and cognitive processing speed and degrade working memory, situational awareness, and impulse control. The result is an increase in both performance errors of omission, which involve the failure to respond in a timely manner, and errors of commission, which involve responding prematurely or incorrectly (Lim and Dinges, 2010). Subjective ratings of fatigue generally underestimate the magnitude of performance deficits due to fatigue (Banks et al., 2010; Van Dongen et al., 2003a), which limits the utility of self-reported fatigue relative to safety risk.

In current commercial transportation systems, the most common causes of fatigue relate to the interaction of eight temporal domains: (1) amount of time awake before work; (2) time of day work occurs (i.e., night versus day); (3) amount of time at work (i.e., on-duty duration); (4) amount of time working specific tasks (e.g., driving versus not driving); (5) time for acute rest during work periods (e.g., number and duration of breaks); (6) time not working (i.e., off-duty duration within and between days); (7) daily time asleep (i.e., both acute and cumulative); and (8) degree of sleep disruption (e.g., by untreated sleep apnea, pain, and other factors). HOS typically involve limitations in a subset of these temporal domains.

Because the concept of fatigue was first associated with deteriorating performance due to time on task, before the critical contributions of sleep need and circadian timing were understood, time working (on duty) and time on task (driving) have received much of the attention relative to fatigue mitigation through HOS regulations for CMV drivers. However, scientific studies in the past 25 years have established that driver fatigue and performance also are dynamically influenced by the regulation of sleep need and endogenous circadian rhythms, including the need to obtain sufficient sleep to ensure recovery from work schedules that might induce either acute or chronic sleep deprivation. The major advance in understanding of work-related fatigue has derived from scientific evidence that the neurobiology underlying human performance is such that changes in performance are not simple functions of amount of sleep, circadian cycle, and other factors.

Fatigue and Commercial Motor Vehicle Operators

An example of this nonlinearity is illustrated by the findings of a naturalistic study of CMV drivers. This study found that approximately 30 percent of all observed instances of driver drowsiness occurred within the first hour of the work shift, and that drowsiness was twice as likely to occur between 6 AM and 9 AM compared with baseline or nondrowsy driving (Barr et al., 2011). This finding appears to be the opposite of the expected adverse effects of driving time on alertness. However, a database study of all fall-asleep crashes in North Carolina found that they peaked in frequency between 6 AM and 9 AM (Pack et al., 1995). Also important, scientific studies of the interaction of sleep and circadian dynamics have established that these are among the hours of the day when sleep propensity is especially elevated as a result of little or no sleep the night before (Lim and Dinges, 2008), after repeated days of sleep restriction (Cohen et al., 2010; Mollicone et al., 2010), or during sleep inertia from just having awakened (Jewett et al., 1999). Therefore drowsy driving would be expected to be higher at 6-9 AM than at 6-9 PM (the circadian peak in daily alertness), even though 6-9 AM represents the beginning hours of a drive (Burke et al., 2015). Thus, an understanding of fatigue-related accidents and errors among CMV drivers needs to be based on the best available information on how time on task, time awake, sleep time, and circadian time interact relative to risk, in addition to well-recognized contributors to risk (e.g., road conditions, traffic density, environmental factors).

Fatigue interacts with other aspects of the lifestyles of CMV drivers, including unhealthy diet and insufficient exercise. A substantial body of evidence indicates that a chronic reduction in sleep time—especially to 6 or fewer hours per day, which has been objectively documented among CMV drivers (Dinges et al., 2005b; Hanowski et al., 2007; Van Dongen and Mollicone, 2014)—is associated with many long-term health problems, including obesity, hypertension, diabetes, and cardiovascular disease, as well as performance deficits (Watson et al., 2015a, 2015b). Evidence that obesity is the avenue by which the other disorders occur was found in a recent study of 88,246 CMV medical examinations from 2005 to 2012. The cohort had a 53-percent prevalence of obesity (defined as body mass index [BMI] > 30.0 kg/m²), and obesity was linked to heart disease, hypertension, diabetes mellitus, nervous disorders, sleep disorders, and chronic low back pain in these individuals (Thiese et al., 2015a). This study also found that between 2005 and 2012, the prevalence of morbid obesity (defined as BMI > 35.0 kg/m²) increased 8.9 percent, and the prevalence of three or more co-occurring medical conditions associated with obesity that limit driving certification increased fourfold.

Obesity and related health conditions are of concern not only for driver health in general but also for driver retention and for the increased risk of fatigued driving. As a result, FMCSA joined with other U.S. and Canadian agencies to develop the North American Fatigue Management Program (NAFMP), an online information and educational program (described in Chapter 8) designed to help drivers understand the factors that contribute to their fatigue and its consequences and to encourage healthier habits.¹ In 2010, the National Transportation Safety Board (NTSB)² made the following recommendation:

However, the effectiveness of the NAFMP among CMV drivers is unknown, as is the case for alternative approaches to fatigue management. Consequently, the NTSB stated “. . . FMCSA, as the Federal agency responsible for motor carrier safety, must also be involved in the evaluation of the fatigue management programs used by carriers to determine whether they successfully mitigate fatigue.” The NTSB concluded that if fatigue management programs are to be successful, FMCSA oversight is needed; therefore, the NTSB made the following recommendation to FMCSA: “Develop and use a methodology that will continually assess the effectiveness of the fatigue management plans implemented by motor carriers, including their ability to improve sleep and alertness, mitigate performance errors, and prevent incidents and accidents.”

DATA LIMITATIONS

Upon reviewing the data available for developing an understanding of how driver fatigue relates to crash risk, the panel identified a number

___________________

¹ Available: http://www.nafmp.com/en [March 2016].

² National Transportation Safety Board Safety Recommendation Date: October 21, 2010. In reply refer to H-10-8 through -11 and H-08-13 and -14 (Reiteration).

of limitations. First, any analysis of fatigue, hours of service, and crash frequency must reflect the substantial heterogeneity in the types of jobs performed by CMV drivers. Differences exist in the size of the fleets with which they are affiliated, truck-handling characteristics, whether trucks are monitored electronically, drivers’ work schedules, how drivers are compensated, the length of time drivers are away from home, and other factors. Although considerable data are collected on drivers who work for large carriers, much less information is available on those who work for small carriers and, especially, on independent owner-operators.

Second, the data available to researchers on various causal factors are a patchwork, with some of the most essential variables being either not recorded, imperfectly captured, or recorded but without the information needed to evaluate their linkages to crash risk.

Third, the statistical methods used in analyzing the available data often fail to take adequate account of confounding influences. More extensive and sophisticated analysis methods are needed because analysis of the data sets now available, as well as those being collected in even larger naturalistic studies, needs to account for the effects of a wide range of factors, in addition to and in interaction with fatigue, to determine the extent to which they affect crash risk among CMV drivers. These factors can be grouped broadly into the characteristics of the driver (which include the temporal domains relevant to fatigue, as discussed above), the carrier, the driving environment, and the vehicle. Given the complexity of many types of confounding factors, different outcomes of interest (e.g., fatal crashes or all crashes), different subsets of the populations of CMV drivers, and different types of jobs, it is not surprising that efforts to date to examine the causal association among driver behavior, hours driving, driver fatigue, and crash rates have resulted in a variety of somewhat disparate findings.

Better quality data, including temporal synchronization across variables, and the use of advanced quantitative methods could provide an understanding of the multiple interacting factors that increase crash risk. This understanding would in turn inform future HOS regulations and the development of other types of countermeasures, and help ensure that they instantiate a more comprehensive picture of the relationship between driver fatigue and highway safety.

FMCSA also is responsible for overseeing the regular medical certification of CMV drivers through the efforts of medical examiners who certify that they are fit to drive. These examinations take place at least every 2 years and are intended to evaluate the risk of sudden or gradual impairment or incapacitation due to medical conditions or their treatment. Evidence on the effectiveness of these medical exams in evaluating drivers who are likely to have sleep apnea and those who are suscep-

tible to fatigue for other medical reasons is limited. Some evidence does indicate that, since the National Registry of Certified Medical Examiners was implemented, examiners have been inconsistent in the criteria they apply in evaluating the risk of medical conditions that may lead to driver fatigue. This inconsistency represents another avenue of opportunity for identifying ways to improve CMV drivers’ health and highway safety.

ORGANIZATION OF THE REPORT

Part I of this report provides background information on the problem of CMV driver fatigue and its relationship to drivers’ long-term health and highway safety. Chapter 2 describes the trucking and bus industries. Chapter 3 presents what is known about fatigue, operator performance, and safety. Chapter 4 then describes the HOS regulations and provides a short history and some international comparisons.

Part II of the report describes the relevant available data and the relevant newer methodologies for analyzing those data. Chapter 5 describes the data available from a variety of sources, while Chapter 6 details the types of methods that could be applied to analysis of the data.

Part III of the report provides an in-depth discussion of what is currently known about fatigue. Chapter 7 addresses the effects of fatigue on highway safety. Chapter 8 describes how fatigue relates to health. Chapter 9 describes countermeasures for dealing with the effects of fatigue. Chapters 8 and 9 also present the panel’s conclusions.

Part IV addresses research directions. Chapter 10 describes the research needed on fatigue and highway safety. Chapter 11 describes the research needed on the role of fatigue in long-term health and on management of fatigue. These two chapters present the panel’s recommendations.

Finally, it should be noted that, because of an imbalance in the available literature and knowledge, as well as the difference in the degree of prevalence of the problem, this report is somewhat more targeted to truck than to bus safety. However, both issues are covered in the chapters that follow.