Subjective and predicted sleepiness while driving in young adults

Abstract

Sleepiness is a significant contributor to car crashes and sleepiness related crashes have higher mortality and morbidity than other crashes. Young adult drivers are at particular risk for sleepiness related car crashes. It has been suggested that this is because young adults are typically sleepier than older adults because of chronic sleep loss, and more often drive at times of increased risk of acute sleepiness. This prospective study aimed to determine the relationship between predicted and perceived sleepiness while driving in 47 young-adult drivers over a 4-week period. Sleepiness levels were predicted by a model incorporating known circadian and sleep factors influencing alertness, and compared to subjective ratings of sleepiness during 2518 driving episodes. Results suggested that young drivers frequently drive while at risk of crashing, at times of predicted sleepiness (>7% of episodes) and at times they felt themselves to be sleepy (>23% of episodes). A significant relationship was found between perceived and predicted estimates of sleepiness. However, the participants nonetheless drove at these times. The results of this study may help preventative programs to specifically target factors leading to increased sleepiness when driving (particularly time of day), and to focus interventions to stop young adults from driving when they feel sleepy.

Introduction

Sleepiness is regarded as a significant contributor to motor-vehicle crashes (Mahowald, 2000, NTSB, 1990, NTSB, 1995, Horne and Reyner, 1995a, Horne and Reyner, 1995b, Philip et al., 1999). Most commentary suggests that the proportion of all vehicle accidents related to sleepiness is in the order of 20% (Garbarino et al., 2001, Sagberg, 1999, Horne and Reyner, 1995a), although a systematic review has noted that current evidence for causation is weak (Connor et al., 2001). Most importantly though, sleepiness related crashes are likely to be more severe than other crashes and are more often fatal (Horne and Reyner, 1995a, Åkerstedt, 2001; Åkerstedt, 2000; Maislin et al., 1995; Pack et al., 1995).

Young adults crash more than older drivers, and more of their crashes are related to sleepiness (Lyznick et al., 1998). For example, young adults (aged 15–24) make up 15% of the Australian population but account for 31% of crash fatalities (Triggs and Smith, 1996). Further, young adults are four times more likely to be involved in fall-asleep crashes than older drivers, and account for almost two-thirds of all sleepiness related crashes (Horne and Reyner, 1995a, Maycock, 1996, Knipling and Wang, 1994, Pack et al., 1995). There are a number of reasons why young adults are likely to experience chronic increased sleepiness. These reasons include social factors (Breslau et al., 1997), maturational changes, and disrupted sleep patterns (Carskadon and Roth, 1991). Sleepiness-related driving risk has been extensively investigated in truck drivers (for example, Hakkanen and Summala, 2000), and particularly in those with sleep apnea (for example, Vorona and Ware, 2002, George and Smiley, 1999, Turkington et al., 2001, Risser et al., 2000 and many others). While truck crashes are typically serious and expensive, Lyznick et al. (1998) suggest that truck drivers are involved in only 3% of sleep-related crashes. Young drivers are a much larger population also at great risk of sleepiness-related crash.

Retrospective estimates of sleepiness while driving have been used to identify an association between sleepiness and traffic accidents (e.g. Connor et al., 2002, Lloberes et al., 2000), and a high proportion of drivers report that they had driven while sleepy at some point in the past (Powell et al., 2002). Drivers are also aware of feeling sleepy while driving in laboratory simulations (Horne and Reyner, 1995a, Reyner and Horne, 1998) and can be aware of their diminished performance (Fairclough and Graham, 1999). However, Reyner and Horne (1998) noted that some subjects did not perceive the relationship between sleepiness and incipient sleep. That is, they continued to drive until they crashed. These data suggest that drivers can identify that they are sleepy while driving, but that they nevertheless continue to drive.

It seems clear that while driving when sleepy has a high associated risk, the level of exposure to that risk in young adult drivers is not known. Exposure is a function of total trips taken, and trips taken at times of increased sleepiness. For example, a roadside survey by Connor et al. (2001) found that relatively few drivers were sleepy during that specific driving episode, suggesting low exposure to risk. Thus, the proportion of all trips that young adult drivers make while sleepy needs to be determined.

Time of day is a strong influence on driver sleepiness and subsequent crash risk. Crashes attributed to the driver having fallen asleep occur primarily in the hours from midnight to 7.00 a.m., and during the mid-afternoon. These are periods of increased sleepiness during the 24 h day/night cycle (Pack et al., 1995). However, the effect is much greater during the night than during the day, with drowsiness peaking in the hours from late evening until dawn. Further, the increased risk of sleepiness related crash during the nighttime hours is more evident among younger drivers than older drivers (Pack et al., 1995, Wang et al., 1996).

Models that attempt to predict alertness/sleepiness and performance, having been developed from earlier sleep regulation models, incorporate sleep homeostasis and circadian components of sleep propensity (Folkard and Åkerstedt, 1992, Acherman and Borbély, 1994, Czeisler et al., 1994, Jewett et al., 1999a, Jewett and Kronauer, 1999). For example, the subjective alertness model of Folkard and Åkerstedt (Åkerstedt and Folkard, 1997, Åkerstedt and Folkard, 1995, Folkard et al., 1999, Folkard and Åkerstedt, 1992) involves a homeostatic component that rises during sleep and falls during wake (Process S) and a sinusoidal 24 h circadian component (Process C) and an additional component representing sleep inertia (Process W). Alertness (sleepiness) at a particular time can be predicted by the arithmetic sum of these three processes. Åkerstedt and Folkard (1997) presented a simplified ‘alertness nomogram’ incorporating Processes S and C, to enable ready estimation of alertness. This nomogram does not include a sleep inertia component and does not account for accumulated sleep debt. As such, this model may provide an upper estimate of alertness in young adults.

In summary, previous studies suggest that drivers are able to perceive that they are sleepy while driving, but may not alter their estimates of the risk of crash related to falling asleep, or their decision to keep driving. Young adults are more likely than older adults to be sleepy during in the day (NSF, 2000), and are much more likely to die in a sleepiness-related crash. It is important to know then, how frequently young adults drive when they are sleepy, and how sleepy they think they are when driving. This distinction can be formalized as the relationship between sleepiness predicted by models of alertness/sleepiness, and self-reported sleepiness at the same time point. This prospective study aimed to determine the relationship between perceived sleepiness while driving, and sleepiness levels predicted by an existing model incorporating known circadian and sleep homeostatic factors. A second aim was to estimate exposure to risk of sleepiness-related crash in this group.