III. CHARACTERISTICS OF DROWSY DRIVING CRASHES As noted in section II, unlike the situation with alcohol-related crashes, no blood, breath, or other objective test for sleepiness currently exists that is administered to a driver at the scene of a crash. No definitive criteria are available for establishing how sleepy a driver is or a threshold at which driver sleepiness affects safety. If drivers are unharmed in a crash, hyperarousal following the crash usually eliminates any residual impairment that could assist investigating officers in attributing a crash to sleepiness. As a result, our understanding of drowsy-driving crashes is based on subjective evidence, such as police crash reports and driver self-reports following the event, and may rely on surrogate mea- sures of sleepiness, such as duration of sleep in a recent timeframe or sleep/work patterns. Some researchers have addressed the problem by analyzing only those crashes known not to be caused by alcohol (because alcohol can cause sleepiness and affect other performance variables), mechanical problems, or other factors and by looking for evidence of a sleepiness effect in categories of inattention or fatigue. Thus, reports on drowsy driving are often inferential. The strength of the inferences is increased when different types of studies reach similar conclusions. The characteristics of drowsy-driving crashes reported below resemble the inclusion criteria that some researchers have used to define a crash as having been caused by drowsiness. This similarity suggests the possibility that the researchers' initial assumptions influenced the determination of crash characteristics. Despite these caveats, a fairly clear picture emerges from studies conducted to date of the typical crash related to sleepiness. The problem occurs during late-night hours. Drowsy-driving crashes occur predominantly after midnight, with a smaller secondary peak in the midafternoon (Studies of police crash reports: Pack et al., 1995; Knipling, Wang, 1994; New York State GTSC Sleep Task Force, 1994; New York State Task Force on Drowsy Driving, 1996; Langlois et al., 1985; Lavie et al., 1986; Mitler et al., 1988; Horne, Reyner 1995b; Studies based on driver self-reports: Maycock, 1996; McCartt et al., 1996). Studies of commercial drivers show a similar pattern (see figure 3). According to a 1996 report, time of day was the most consistent factor influencing driver fatigue and alertness. Driver drowsiness was markedly greater during night driving than during daytime driving, with drowsiness peaking from late evening until dawn (Wylie et al., 1996). Nighttime and midafternoon peaks are consistent with human circadian sleepiness patterns. The risk of a crash related to sleepiness increases during nighttime hours among both younger drivers (25 years of age and younger) and drivers between the ages of 26 and 45. However, younger drivers have no increased risk during the afternoon, when the predictable circadian sleepiness peak is expected. Drivers ages 45 through 65 have fewer nighttime crashes, with a peak at 7 a.m. Drivers ages older than 65 are more likely to have fall-asleep crashes during the midafternoon (Pack et al., 1995; Wang, Knipling, Goodman, 1996). Fall-asleep crashes are likely to be serious. The morbidity and mortality associated with drowsy-driving crashes are high, perhaps because of the higher speeds involved (Horne, Reyner, 1995b) combined with delayed reaction time. In North Carolina, more of these crashes resulted in injury compared with other, nonalcohol-related crashes-fatalities occurred in 1.4 percent and 0.5 percent, respectively (Pack et al., 1995). Pack (1995) and Maycock (1996) both conclude that a higher proportion of the most serious crashes are sleepiness related. A single vehicle leaves the roadway. An analysis of police crash reports in North Carolina showed the majority of the nonalcohol, drowsy-driving crashes were single-vehicle roadway departures (Pack et al., 1995). Among New York State drivers surveyed about their lifetime experience with drowsy driving, almost one-half of those who had a fall-asleep or drowsy-driving crash reported a single-vehicle roadway departure; about one-fourth of those who had fallen asleep without crashing also reported going off the road (McCartt et al., 1996). NHTSA General Estimates System data reflect the same trend but also suggest that sleepiness may play a role in rear-end crashes and head-on crashes (Knipling, Wang, 1994). The crash occurs on a high-speed road. In comparison with other types of crashes, drowsy-driving crashes more often take place on highways and major roadways with speed limits of 55 to 65 mph (Knipling, Wang, 1994; Wang, Knipling, Goodman, 1996). Pack and colleagues (1995) found that most sleepiness-related crashes occur at higher speeds, attributing this finding to the effect of sleep loss on reaction time. NHTSA figures show that most drowsiness- or fatigue-related crashes occur on higher speed roads in nonurban areas. However, Maycock (1996) found that a greater absolute number occur in built-up areas. Panel members noted the possibility that more crashes occur on high-speed roads because more long-distance nighttime driving occurs on highways. The driver does not attempt to avoid crashing. NHTSA data show that sleepy drivers are less likely than alert drivers to take corrective action before a crash (Wang, Knipling, Goodman, 1996). Anecdotal reports also suggest that evidence of a corrective maneuver, such as skid marks or brake lights, is usually absent in fall-asleep crashes. The driver is alone in the vehicle. In the New York State survey of lifetime incidents, 82 percent of drowsy-driving crashes involved a single occupant (McCartt et al., 1996). Conversely, respondents who reported having fallen asleep without crashing were less likely to have been alone in the automobile. Wilkins and colleagues (1997) confirmed that crashes attributed to driver fatigue have characteristics similar to those cited above regarding driver age, time of day, crash type, and severity. But, in addition, when alcohol involvement was combined with fatigue or sleepiness, the patterns became more pronounced. For example, "asleep with alcohol" crashes involved a higher percentage of young males than did crashes in which the driver was asleep with no evidence of alcohol. IV. RISKS FOR DROWSY-DRIVING CRASHES Although its conclusions were based on a limited body of knowledge, the panel identified a number of chronic predisposing factors and acute situational factors that increase the risk of drowsy driving and drowsy-driving crashes. These include sleep loss, driving patterns that disregard the normal sleep-wake cycle or represent driving increased time or miles (exposure), the use of sedating medication, sleep disorders such as sleep apnea syndrome (SAS) and narcolepsy, and the increased drowsiness and performance impairment that result from consuming alcohol when drowsy. All factors may interact, and with the exception of medical disorders, all factors may have either chronic or acute effects. SLEEP LOSS As noted in section II, external and internal factors and current lack of knowledge and attitudes about sleep cause many Americans to get inadequate sleep either occasionally (acute sleepiness) or routinely (chronic sleepiness). Those who suffer chronic sleep restriction and sleepiness may also combine this lifestyle pattern with situational acute sleep loss, aggravating their risk of drowsy driving. Chronic sleepiness. In a recent Gallup survey, approximately one-half of U.S. adults reported experiencing sleeping difficulties sometimes, with about 1 in 10 saying the difficulties are frequent (National Sleep Foundation, 1995). In a 1997 followup survey, three of four Americans who reported getting as much or more sleep than they "need" said they were sleepy during the day. One in three of the adult public was deemed "significantly" sleepy on the Epworth Sleepiness Scale (ESS), and 1 in 20 scored at the "severe" sleepiness level (National Sleep Foundation Survey, 1997). In the New York State survey, the reported frequency of drowsy driving in the past year was associated with the quantity and quality of sleep obtained. For example, those who reported having fair or poor sleep quality were more likely to have driven drowsy sometimes or very often than were those who said their sleep was good or excellent (McCartt et al., 1996). In addition, Maycock (1996) found that higher scores on the ESS were positively associated with crashes. Drivers who reported having trouble staying awake during the day were more likely to report having sometimes or very often driven drowsy (McCartt et al., 1996). Acute sleep loss. As discussed in section II, the loss of even one night of sleep may cause extreme sleepiness. Short-term work demands, child care, socializing, preparing for a trip or vacation, and "pulling all nighters" are common causes of acute sleep loss. Sleep-restrictive work patterns. Working the night shift, overtime, or rotating shifts is a risk for drowsy driving that may be both chronic and acute. In the New York State survey, nearly one-half the drowsy drivers who crashed (and more than one-third of those who drove drowsy without crashing) reported having worked the night shift or overtime prior to the incident. In addition, a higher reported frequency of driving drowsy was associated with working a rotating shift, working a greater number of hours per week, and more frequently driving for one's job (McCartt et al., 1996). In the British study (Maycock, 1996), respondents said that working the night shift led to sleepiness while driving, and in many studies a majority of shift workers admit having slept involuntarily on the night shift. The return to day work and morning shifts starting between 4 a.m. and 7 a.m. also may lead to sleepiness. EEG studies of sleep in rotating shift workers in both the natural environment and the laboratory have shown that day sleep after night work and early night sleep before morning work (e.g., going to sleep at 7 or 8 p.m. before a 4 a.m. shift) is 2 to 4 hours shorter than night sleep (Åkerstedt, 1995a). In addition, a study of hospital house staff working around the clock (Marcus, Loughlin, 1996) found higher levels of sleepiness and crashes following on-call periods. In a survey of hospital nurses, night nurses and rotators were more likely than nurses on other shifts to report nodding off at work and at the wheel and having had a driving mishap on the way home from work (Gold et al., 1992). (For more on this topic, see section V on shift workers.) DRIVING PATTERNS Driving patterns, including both time of day and amount of time driven, can increase crash risk. As detailed in section III, the greatest proportion of drowsy-driving crashes occurs during the late-night hours. The biology of the sleep-wake cycle predicts sleepiness during this time period, which is a circadian sleepiness peak and a usual time of darkness. Other driving time patterns that increase risk include driving a larger number of miles each year and a greater number of hours each day (McCartt et al., 1996) and driving a longer time without taking a break or, more often, driving for 3 hours or longer (Maycock, 1996). THE USE OF SEDATING MEDICATIONS A number of studies indicate that using certain medications increases the risk of sleepiness-related crashes, particularly using prescribed benzodiazepine anxiolytics, long-acting hypnotics, sedating antihistamines (H1 class), and tricyclic antidepressants (Kozena et al., 1995; Van Laar et al., 1995; Ray et al., 1992; Leveille et al., 1994; Ceutel, 1995; Gengo, Manning, 1990). The risks are higher with higher drug doses and for people taking more than one sedating drug simultaneously (Ray et al., 1992). Younger males have higher risks than do females or other age groups across all drug classes. It appears that risk is highest soon after the drug regimen is initiated and falls to near normal after several months (Ceutel, 1995). Recreational drug use also may exacerbate sleepiness effects (Kerr et al., 1991). UNTREATED SLEEP DISORDERS: SLEEP APNEA SYNDROME AND NARCOLEPSY Untreated sleep apnea syndrome and narcolepsy increase the risk of automobile crashes (Findley et al., 1995; George et al., 1987; Aldrich, 1989; Alpert et al., 1992; Broughton et al., 1981; Broughton et al., 1984). No current data link other sleep disorders with drowsy-driving crashes. However, other medical disorders causing disturbed sleep and excessive daytime sleepiness could pose risks. In sleep apnea syndrome, brief interruptions of air flow and loss of oxygen during sleep disrupt and fragment sleep. The condition also is associated with loud, chronic snoring. Although people with untreated sleep apnea syndrome may not be aware of the brief disturbances, poor sleep quality often leads to daytime sleepiness. Narcolepsy is a disorder of the sleep-wake mechanism that also causes excessive daytime sleepiness. In untreated patients, involuntary 10- to 20-minute naps are common at 2- to 3-hour intervals throughout the day. Cataplexy, a sudden loss of muscle tone ranging from slight weakness to complete collapse, is another major symptom of narcolepsy that increases the risk of crash. These conditions are unrecognized and untreated in a substantial number of people (National Sleep Foundation Survey, 1997; American Thoracic Society, 1994). (See section V for more information on sleep apnea syndrome and narcolepsy.) CONSUMPTION OF ALCOHOL INTERACTS WITH SLEEPINESS TO INCREASE DROWSINESS AND IMPAIRMENT Although sleepiness and alcohol are distinct crash causes, the data also show some evidence of overlap. NHTSA found that drivers had consumed some alcohol in nearly 20% of all sleepiness-related, single-vehicle crashes (Wang, Knipling, Goodman, 1996). More than one in three New York State drivers surveyed in drowsy-driving crashes said they had drunk some alcohol (McCartt et al., 1996), and police-reported, fall-asleep crashes had a higher proportion of alcohol involvement than other types of crashes in that State. (New York GTSC Task Force, 1994; New York State Task Force, 1996). Laboratory studies explain and predict these patterns. Many researchers have shown that sleepiness and alcohol interact, with sleep restriction exacerbating the sedating effects of alcohol, and the combination adversely affecting psychomotor skills to an extent greater than that of sleepiness or alcohol alone (Roehrs et al. 1994; Wilkinson, 1968; Huntley, Centybear, 1974; Peeke et al., 1980). Driving simulation tests specifically show this effect, even with modest reductions in sleep, low alcohol doses, and low blood ethanol concentrations. In a driving simulation study, alcohol levels below the legal driving limit produced a greater number of deviations from the road after 4 hours of sleep than after 8 hours of sleep (Roehrs et al., 1994) It is possible that the effects of low levels of blood alcohol may have an interaction with circadian rhythms that produces sleepiness in the afternoon and evening (Roehrs et al., 1994; Horne, Baumber, 1991; Horne, Gibbons, 1991). The panel speculated that drinking alcohol before driving in the afternoon or at night might pose special risks given the circadian effects. INTERACTIONS AMONG FACTORS INCREASE OVERALL RISK Some of the crash-related factors have been studied more than others. The panel could not find evidence to determine whether chronic or acute situations pose the greater risk for crashes. However, it is clear that these factors are cumulative, and any combination of chronic and acute factors substantially increases crash risk. For example, people with chronic sleep loss who drive in the early morning hours are likely to be at greater risk than are early morning drivers who slept well the night before and usually get enough sleep.