Occupational Health - Public Health Poster Session



Materials & Methods


Discussion & Conclusion



INABIS '98 Home Page Your Session Symposia & Poster Sessions Plenary Sessions Exhibitors' Foyer Personal Itinerary New Search

Sleep Deficit, Fatal Accidents, and the Spring Shift to Daylight Savings Time

Contact Person: Stanley Coren (scoren@cortex.psych.ubc.ca)


Recent data suggests that insufficient sleep and disrupted circadian rhythms contribute to major public health problems. For example, the U.S. National Commission on Sleep Disorders commissioned a study to determine the direct cost of accidents which result from sleep debt (Leger, 1997). The results were staggering. In the test year (1988) in the United States, the cost of motor vehicle accidents caused by sleepy drivers was $37.9 billion. Sleep debt related accidents in the public transportation accidents cost $720 million dollars. Work-related accidents caused by sleepiness added up to $13.4 billion. Falls and other accidents in public places that are directly due to sleepiness came to $1.3 billion. Finally, accidents around the home due to sleep debt resulted in a cost of $2.7 billion. The grand total came to over 56 billion dollars.

Beyond the monetary losses, the human cost of sleep related accidents is also astonishing. The year 1988 saw a total of 24,318 deaths from accidents related to sleepiness. There were also 2,474,430 disabling injuries resulting from accidents where the decreased mental efficiency and attentiveness due to sleep loss was the major underlying factor. Injuries in the work place due to sleep debt resulted in the loss of 29,250,000 work days, with 13,650,000 days lost directly due to the accident itself, while 15,600,000 days were lost due to complications and long term effects of the accident within the first year. Accidents outside the work place, took a toll of 23,400,000 work days lost. This means that in one year, 52,650,000 work days were missed from work plus non-work related accidents which all occurred because of too little sleep.

Sleep deficit has also been implicated in many major public catastrophes (Mitler, Carskadon, Czeisler, Dement, Dinges, & Graeber, 1988). Coren (1996a) has reviewed data suggesting that the oil spill of the Exxon Valdez, the destruction of the space shuttle Challenger, the nuclear accident at Chernobyl (which cost over 50,000 lives) and the near nuclear accidents at the Three Mile Island and Peach Bottom reactors, were all associated with sleep deficits on the part of personnel involved.

The average young adult, today, reports sleeping about 7 to 7 hours each night. When we compare this to sleep patterns in 1910, before Edison's modern coiled tungsten filament light bulb, was introduced, we find that the average person slept 9 hours each night. This means that today's population sleeps 1 to 2 hours less than people did early in the century (Webb & Agnew, 1975). Based upon data like this, some researchers have claimed that society is chronically sleep deprived, and even small additional reductions in sleep time may have consequences for safety (see Coren, 1996a for a review). Coren (1996b) apparently confirmed this by showing that the shift to Daylight Savings Time (DST) had an impact on accident rates. The spring shift to daylight savings time results in a loss of one hour of sleep while the fall shift provides an additional hour which can be used for sleep. Using data from two years of Canadian traffic accident records, he found that on the Monday following the shift to DST in the spring, there was an increase in traffic accident rates of about 7 percent, while in the fall there was a decrease in accident rate of about the same magnitude.

Coren (1996c) attempted to confirm this finding using accidents other than those associated with vehicle operation. He did this by looking at every accidental death in the US reported to the National Center for Health Statistics for the years 1986 through 1988. Since over 80% of accident induced mortality occurs within 4 days of the accident, data for analysis was restricted to the first 4 workdays in the weeks preceding, immediately following, and one week after, the DST change. While he found a significant increase in deaths following the spring shift (6.6%) he could not confirm the fall rebound, which only showed a nonsignificant (1.5%) decrease. Other studies that have used traffic accidents have found the increase following the spring shift to DST but have also failed to find any decrease following the fall time change (e.g. and Hicks, Lindseth & Hawkins, 1983; Monk, 1980).

If we accept the fact that there there is a spring increase in accidents and no corresponding fall decrease in accidents following the DST shift, sleep deficit may still explain such results. The reasoning is that work schedules enforce the loss of sleep following the spring DST shift while it is possible that many people may not take advantage of the additional hour in the fall to extend their sleep. However, explanations, not involving sleep, are also possible. For instance, following the spring DST shift people must drive to work under lower levels of illumination resulting in reduced visibility of road conditions and other vehicles. Alternatively, people may simply forget the DST time change, and may not adjust their clocks. These people then find find themselves late for work on Monday morning and may end up driving with reckless haste as a result. To test these possibilities the following study was conducted.

Back to the top.

Materials and Methods

The data were drawn from the Fatal Accident Reporting System, which is maintained by the U.S. National Highway Traffic Safety Administration and the National Center for Statistics and Analysis. This data base maintains full records on all traffic accidents in the U.S. that result in fatalities in the participating states. Thirty-three states participate in this data base fully, and several additiona states provide partial information. Most importantly for this research, data is coded by the date of accident (not the date of the fatality) and further information is given about the actual time of day that the accident occured. The data set available to this researcher went from 1986 through August 1995. There were a total of 366,910 traffic deaths during that period.

Back to the top.


The first issue to address is whether there are any differences in fatal accident rates following the shift to DST. The simplest and most sensitive measure which will address this question involves cumulating the traffic deaths over the full 10 year period. Next we compare the total of traffic fatalities for the Monday immediately following the DST shift with the pooled frequency of accidents for the previous and following Mondays. The spring DST shift (where one hour of sleep is lost) shows the expected increase in accidents with relative risk (RR) of 1.17 [95% CI=1.07/1.29, 2(1)=10.83, p < 0.001]. This 17 percent increase is larger than that observed in previous studies. The same analysis conducted for the fall DST shift, however, produces an insignificant reduction in traffic deaths [RR=0.97, 95% CI=0.89/1.07, 2(1)=0.29, ns.].

Given that the main effect of DST on accident rates was found, we can now address the issue of whether the obtained pattern of data is more consistent with the sleep loss hypothesis or the alternative set of hypotheses involving careless driving because people are dashing to work because they are late, or reduced morning illumination levels. This question can be answered because this data base includes the time of the episodes resulting in the fatality. Both the reduced light level and rushing to work explanations for the spring increase in accidents following the DST shift should involve incidents that occur early in the day. If sleep deficit is the cause, then the effects of sleep loss should increase as the day wears on, becoming more obvious later in the day. This means that the hypostheses can be simply tested by partitioning the day into early (arbitraily defined as before noon) and late (after noon). Obviously the comparison groups are the pooled accident frequencies for the Mondays preceeding and following limited to the same time periods. When we test these differences we find that the change in accident rates for the early half of the day following the shift to DST is not statistically significant [RR=1.07, 95% CI=0.92/1.24, 2(1)=0.75, ns.]. The increase in traffic fatalities in the later half of the day (where the effects of sleep should have their greatest impact) however, is statistically reliable [RR=1.13, 95% CI=1.01-1.28, 2(1)=3.89, p < 0.05].

Back to the top.

Discussion and Conclusion

Two major points are made by these data. The first is a confirmation of the fact that following the spring shift to Daylight Savings Time (when one hour of sleep is lost) there is a measureable increase in the number of traffic accidents that result in fatalies. Furthermore, it replicates the absence of any "rebound" reduction of accidents following the fall shift to DST (when the opportunity is present for an additional hour of sleep).

Of the two competing hypotheses for this increase in accidents, namely the one that suggests that it is the increased sleep deficit that causes the change in accident rate, versus notions based upon reduced illumination levels when driving to work, or suppositions that people forget the DST time change, fail to adjust their clocks, and find themselves rushing to work to avoid being late, the sleep hypothesis seems to be the most tenable. Hypotheses based upon haste and dim morning light both predict the bulk of the increased accidents to be confined to the morning hours. The sleep loss hypothesis would predict that individuals become more tired as the day wears on and hence the bulk of the accidents will appear later in the day. It is, of course, this latter pattern which appears with most of the accident fatality increase confined to the period after noon.

If the sleep loss hypothesis is correct, then why isn't there a reduction in the number of traffic accidents in the fall, when the shift back to standard time provides an extra hour for sleep. Although this was the pattern observed in one study (Coren, 1996b) it has not replicated in other studies. The failure of the "safety rebound" may simply have to do with human nature. Just because a person has the opportunity to sleep for an addition hour does not mean that people actually will go to sleep on time. Many may spend that extra hour socializing or watching television. In some instances, where individuals do go to bed at the appropriate time, their usual circadian rhythm may still wake them after 7 or 8 hours in response internal signals or the external morning increase in illumination. Contrast this to what happens in the spring, where an individual's work schedule will enforce the person's awakening on the new DST time in order to meet job committments.

Taken together then, these data are consistent with the hypothesis that as a society were are sufficiently chronically sleep deprived so that a small decrease in sleep duration, such as that which occurs with the spring shift to DST, can significantly increase accident susceptibility.


This research was supported in part by grants from the Natural Sciences and Engineering Research Council of Canada.

Back to the top.


Coren, S. (1996a). Sleep Thieves. New York: Free Press.

Coren, S. (1996b). Daylight savings time and traffic accidents. New England Journal of Medicine, 334, 924.

Coren S. (1996c). Accidental death and the shift to daylight savings time. Perceptual and Motor Skills, 83, 921-922.

Hicks, R.A., Lindseth, K., & Hawkins, J. (1983). Daylight savings-time changes increase traffic accidents. Perceptual and Motor Skills, 56, 64-66.

Leger D (1994). The cost of sleep-related accidents: A report for the National Commission on Sleep Disorders Research. Sleep, 17, 84-93.

Mitler, M.M, Carskadon, M.A, Czeisler, C.A, Dement, W.C, Dinges, D.F, Graeber, R.C. (1988). Catastrophes, sleep, and public policy: Consensus report. Sleep, 11, 100-109.

Monk, T.H. (1980). Traffic accident increases as a possible indicant of desynchronosis. Chronobiologia, 7, 527-529.

Back to the top.

| Discussion Board | Previous Page | Your Poster Session |
Coren, S.; (1998). Sleep Deficit, Fatal Accidents, and the Spring Shift to Daylight Savings Time. Presented at INABIS '98 - 5th Internet World Congress on Biomedical Sciences at McMaster University, Canada, Dec 7-16th. Available at URL http://www.mcmaster.ca/inabis98/occupational/coren0164/index.html
© 1998 Author(s) Hold Copyright