Road accidents caused by drivers falling asleep

https://doi.org/10.1016/S0001-4575(99)00023-8Get rights and content

Abstract

About 29 600 Norwegian accident-involved drivers received a questionnaire about the last accident reported to their insurance company. About 9200 drivers (31%) returned the questionnaire. The questionnaire contained questions about sleep or fatigue as contributing factors to the accident. In addition, the drivers reported whether or not they had fallen asleep some time whilst driving, and what the consequences had been. Sleep or drowsiness was a contributing factor in 3.9% of all accidents, as reported by drivers who were at fault for the accident. This factor was strongly over-represented in night-time accidents (18.6%), in running-off-the-road accidents (8.3%), accidents after driving more than 150 km on one trip (8.1%), and personal injury accidents (7.3%). A logistic regression analysis showed that the following additional factors made significant and independent contributions to increasing the odds of sleep involvement in an accident: dry road, high speed limit, driving one’s own car, not driving the car daily, high education, and few years of driving experience. More male than female drivers were involved in sleep-related accidents, but this seems largely to be explained by males driving relatively more than females on roads with high speed limits. A total of 10% of male drivers and 4% of females reported to have fallen asleep while driving during the last 12 months. A total of 4% of these events resulted in an accident. The most frequent consequence of falling asleep—amounting to more than 40% of the reported incidents—was crossing of the right edge-line before awaking, whereas crossing of the centreline was reported by 16%. Drivers’ lack of awareness of important precursors of falling asleep—like highway hypnosis, driving without awareness, and similar phenomena—as well as a reluctance to discontinue driving despite feeling tired are pointed out as likely contributors to sleep-related accidents. More knowledge about the drivers’ experiences immediately preceding such accidents may give a better background for implementing effective driver warning systems and other countermeasures.

Introduction

Fatigued or drowsy drivers have long been acknowledged to constitute a potential traffic safety hazard, and several research studies have addressed various aspects of the problem. Brown (1994) has presented a comprehensive review and discussion of the research literature on the nature of fatigue and its effects on driver behaviour and traffic accidents, on the basis of which he concluded that ‘fatigue is insufficiently recognised and reported as a cause of road accidents’.

In discussing the effects of fatigue, Brown further points out that the main effect is ‘a progressive withdrawal of attention from road and traffic demands’. A most extreme form of withdrawal of attention is obviously the closing of eyes due to sleepiness.

To prevent accidents related to drowsiness and sleeping behind the wheel, it is important to acquire precise knowledge about the extent of the problem as well as its preconditions and consequences. The following issues, all of which are investigated empirically in the present study, are considered relevant for a better understanding of sleep-related accidents.

  • 1.

    What is the proportion of accidents caused by fatigue or sleep behind the wheel, and what are the most likely types of accidents to occur under these conditions?

  • 2.

    How prevalent is the problem of actually falling asleep while driving?

  • 3.

    What are the most frequent consequences of falling asleep while driving?

  • 4.

    To what extent is sleeping behind the wheel related to characteristics of the driver, the road and traffic conditions, and the trip?

Quite a few studies have estimated the proportion of accidents caused by drivers who were fatigued or actually were dozing off behind the wheel. Several studies before the mid 1980s were reviewed by Nairn (1987), showing estimated proportions of crashes related to sleep or fatigue varying from 1% for all accidents to 48% of fatalities.

Most typically, these studies were based on official crash report statistics. This holds true also for later studies. For example, a Norwegian study of police reports from frontal crashes (Amundsen and Christensen, 1986) showed that falling asleep at the wheel was noted as causal factor in 16 out of 286 accidents (5.6%).

In the US, analyses of larger databases have been reported. Thus, Pack et al. (1995) found that the driver was judged to be asleep in less than 1% of the crashes reported in North Carolina over the years 1991–1992. Lyznicki et al. (1998) report on the basis of a literature review that driver sleepiness is involved in 1–3% of all US motor vehicle crashes. They point out, however, that surveys of the prevalence of sleepy behaviour indicate that sleepiness is a more common causative factor that shown by these estimates. Similar estimates have been reported also from New Foundland (see Taoka, 1998), and from Switzerland (Laube et al., 1998).

The underestimation of the involvement of sleep and fatigue in police reports has been pointed out also by Fell (1994). He reports on the development in new South Wales, Australia, of a set of criteria including crash trajectories and conditions of occurrence to determine whether a crash is related to sleep. On the basis of these criteria, which however have not been validated, it was estimated that sleepiness accounts for as much as 6% of all crashes. For fatal crashes the estimate was 15%, and for fatal crashes on rural roads as high as 30%. Knipling and Wang (1995) have pointed out that these criteria may be over-inclusive and capture some crashes not necessarily related to sleep or drowsiness. Their estimate for all US police-reported crashes, based on a series of criteria to ascertain drowsiness involvement in cases where this was not explicitly stated in the report, was that between 1.2 and 1.6% of all crashes are related to drowsiness or fatigue.

As indicated by some of the estimates above, sleep tends to be more prevalent as contributing factor in high-severity crashes. According to Pack et al. (1995) data from the US Fatal Accident Reporting System shows that falling asleep accounts for 3–4% of fatal crashes, an estimate that is higher than those given above for all accidents.

Probably more valid evidence on sleepiness involvement in fatal accidents comes from Finland, where special investigation teams routinely collect data from all fatal accidents. Based on the database from these investigations, Summala and Mikkola (1994) found that about 7% of the fatal accidents were due to a driver falling asleep, and an additional 3% were caused by fatigue.

The higher involvement of sleep and fatigue in fatal accidents is most likely related to a higher prevalence of driver sleepiness on major roads, especially in rural areas, where driving speeds are higher. It is notable that Horne and Reyner (1995) found that sleepiness was involved in as much as 16% of police-reported accidents on major roads in England and over 20% on motorways.

Even if the large discrepancies between the various estimates of driver sleepiness involvement in accidents cited above can be explained by differences regarding the accident categories included, there are good reasons for suspecting that this problem is underestimated in police reports, partly because of verification problems. An alternative source of information is therefore the drivers themselves. There have been published a few reports of self-reported sleep involvement in accidents, which generally tend to result in somewhat higher estimates than the police reports.

For example, in a British study, Storie (1984) collected reports from about 15 000 drivers who had been involved in an accident, and found that at ‘the time just before the accident happened’ 3% reported to have been sleepy and 8% bored (not mutually exclusive categories). It cannot, however, be concluded from that study that the mentioned conditions were causally related to the accident.

More recently, Maycock (1997) reported that out of some 1000 accidents reported by a sample of 4621 male UK drivers over a 3-year period, 9–10% were associated with tiredness.

An additional difference between police reports and self-reports of accidents is that the latter are likely to include more accidents, especially in the low-severity category. And since there seems to be a positive relationship between severity and sleepiness involvement, this may tend to deflate the proportion of sleep-related accidents in self-report data as compared to data based on official reporting systems. This implies that the underreporting of sleep as accident cause in official data is even larger than suggested by a direct comparison in terms of proportion of reported crashes.

In the present study, a self-reported approach was used for collecting data about sleep and fatigue as causal factors in accidents. A sample of Norwegian drivers who had reported a crash to their insurance company and were judged by the insurance company to be at fault for the event, were asked in a questionnaire whether they considered sleep or fatigue (among several factors listed) to have contributed to the accident. In addition, information was collected regarding crash type and severity, making it possible to consider the role of sleep/fatigue for different kinds of accidents.

In addition to accidents, which represent the worst-case outcome of the driver falling asleep, there are likely to be numerous instances of near-miss situations, where the driver wakes up soon enough to avoid a serious situation. To get information about the prevalence of falling asleep while driving, Gårder and Alexander (1995) carried out a survey among a random sample of motorists as well as a sample of students in the US. They found that 52% reported to have fallen asleep at least once while driving a motor vehicle, and that 29% had fallen asleep during the last 12 months. As to the consequences, 7% reported an accident as a result of having fallen asleep. Gårder and Alexander (1995) also asked their drivers who had fallen asleep to state their lateral position at the time when they awoke, and found that 62% were off their driving lane with at least one pair of wheels. It was also shown that deviations to the right side of the lane are much more likely than to the left. Apart from these results, knowledge about other consequences than accidents is extremely scarce.

Somewhat lower prevalence estimates were reported by McCartt et al. (1996) based on a telephone survey among 1000 New York State license holders. A total of 23% reported to have fallen asleep some time while driving, and 2.8% had crashed when fallen asleep.

Information about the consequences of near-miss situations where the driver has fallen asleep can provide useful indications about accident risk. Since the number of near-miss incidents is large compared to accidents, use of such data is a good alternative to accident data for driver samples that are too small for meaningful analyses of accidents. For instance, the relative prevalence of crossing the right edge-line when falling asleep can be assumed to be related to the relative risk of running off the road, whereas crossing the centreline bears a similar relationship to the relative risk of head-on collision. One purpose of the present study was to collect information both regarding the proportion of drivers who had experienced episodes of falling asleep while driving, and what had been the most serious consequence of the incident.

A further objective was to investigate to what extent sleep-related accidents as well as the prevalence of falling asleep while driving can be predicted on the basis of background information about the driver, such as sex, age, and driving experience. For the accidents additional background factors to be investigated include characteristics of the road, the car and the trip where the accident occurred.

It was also of interest to see if the findings of Gårder and Alexander (1995) on US drivers could be replicated on a sample of Norwegian drivers. The risk of falling asleep during driving is probably related both to the type of road and to the traffic density, with higher risk on straight, monotonous, dual carriageway roads with little traffic, where boredom is likely to occur. Such roads make up a relatively large proportion of US roads, as compared to many other countries. It is therefore likely that the prevalence of falling asleep while driving is higher in the US than e.g. in Norway, where roads on the average are more hilly and curved, and where dual carriageway roads make up a relatively small part of the road network.

Section snippets

The driver sample

From the files of the two largest insurance companies in Norway a random sample of the most recent accident reports was drawn. The sample consisted of 29 600 accident reports received within the last few months. The car owners who had reported these accidents, received a postal questionnaire, to be filled in by the person who had been driving the car during the accident.

The questionnaire

Questions about sleeping behind the wheel were included as part of a more comprehensive questionnaire to accident-involved

Accidents and sleep/fatigue

Among the 3239 responsible drivers 125 (3.9%), reported that fatigue or sleep contributed to the accident. In Fig. 1 it is shown how the proportion of accidents related to fatigue or sleep varies with type of accident. Most notably, the highest proportion of sleeping or fatigued drivers is found for running-off-the-road accidents.

To find out how the proportion of accidents where fatigue or sleep contribute varies with characteristics of the accident, the traffic situation and the driver, we

Discussion and conclusions

The present study provides two kinds of evidence that automobile drivers who are drowsy or who actually fall asleep while driving constitute a very significant road safety hazard. First, the results regarding sleep or fatigue involvement in accidents showed that these factors were causally related to about one in every 14 personal injury accidents, and about one in 25 accidents in total. Second, the results on the prevalence of falling asleep among drivers in general showed that one in 12

Acknowledgements

This study was supported by grants from Telenor Mobil AS and The Ministry of Transport and Communications.

References (26)

  • Gårder, P., Alexander, J., 1995. Fatigue related accidents and continuous shoulder rumble strips (CSRS). Transportation...
  • J.A. Horne et al.

    Sleep-related vehicle accidents

    Br. Med. J.

    (1995)
  • J.S. Kerr

    Driving without attention mode (DWAM): a formalisation of inattentive states in driving

  • Cited by (231)

    View all citing articles on Scopus
    View full text