Public Release: 

Predicting Accident Times On The Job

Human Factors and Ergonomics Society

We are all aware of daily ups and downs in our abilities. Such normal fluctuations in performance have been documented for many years and are not random throughout the 24-hour day. Knowing when industrial errors, incidents, and accidents are most likely to occur across the day/night cycle can help to focus job design and redesign activities for safety-sensitive jobs.

This article describes a mathematical model that may be used to predict the timing of errors, incidents, and accidents caused by sleepiness and inattention.

Bjerner et al. reported on the daily performance fluctuations for employees at a Swedish gas company whose work was monitored over a 20-year period. In the distribution throughout the 24-hour day of 74,927 meter reading errors made by these workers, most errors occurred during the night, with a major peak between 1:00 and 3:00 a.m. A smaller afternoon peak in errors occurred between 1:00 and 3:00 p.m.

Since these early observations, other, more destructive human error events have also been shown to occur with this same two-peak pattern--for example, the 24-hour distribution of 6052 vehicle crashes attributable to fatigue (that is, crashes in which no mechanical failure and no alcohol or substance-related causal factors were found). These data were originally compiled and plotted by Mitler et al. A two-peak pattern was apparent in this distribution, and it was similar to that of the gas meter- reading errors. The number of crashes was elevated between about midnight and 6:00 a.m. and again between about 1:00 and 4:00 p.m.

The evidence shows unequivocally that there are critical periods before dawn and in the middle of the afternoon when sleepiness and inattention may lead to errors that cause industrial incidents and accidents. How can we use this information about sleepiness to improve workplace safety?

First, we must recognize the huge increase that has occurred during the last century in the amount of energy a typical worker controls. For example, truck drivers today control several orders of magnitude more potential and kinetic energy than did stagecoach drivers of the 1800s. When the truck driver falls asleep on the job, there is much more at risk than when the stagecoach driver fell asleep. Nevertheless, the biology and sleep needs of both drivers are identical.

Second, we should recognize that many jobs and systems have not been designed to use human operators effectively. Any such design should exploit natural human strengths, such as pattern recognition and decision making, and it should protect the system from natural human weaknesses, such as vulnerability to attentional lapses in boring surroundings, especially during critical periods of the 24-hour day and under conditions of sleep deprivation.

For example, in the transportation industry, efforts are being made to introduce structured cockpit napping for transoceanic flight crews, who must deal with long duty days and extreme jet lag. Also, efforts are being made to take circadian rhythms into account in hours of service regulations for interstate commercial drivers.

Third, we should recognize that in the modern era, it is impossible to avoid assigning a significant portion of the workforce to night work. Emergency services such as hospitals and law enforcement, and utility services such as electrical power, natural gas, sewer, telephone, and water, require around-the- clock staffing. In conjunction with this awareness, we should also recognize that night work always compromises a person's ability to get enough sleep.

Fourth, we should recognize that the methods currently available to laboratory researchers for identifying unacceptable sleepiness do not lend themselves to widespread use in the workplace, even if some level of sleepiness could be agreed on as unacceptable for job performance. However, a number of laboratory-originated performance tests have been adapted for daily workplace use and are becoming available.


Once we accept these facts of life, what can we do with our current information and technology? We now know when during the 24-hour day industrial errors leading to incidents and accidents are most likely to occur. Therefore, it may be possible to focus job redesign and staffing redesign on those workers whose sustained alertness is required for safety. We can then integrate fundamental principles of the sleep research and human factors disciplines with the findings reviewed here. Consider the following possibilities:

Examine, through task analyses and other appropriate methods, those safety-sensitive jobs that are most vulnerable to loss of attention. Some of these jobs require vigilance, or watchkeeping. For example, some important inspection tasks and many security positions require sustained attention in boring environments. People generally have problems remaining alert for rare but important occurrences in boring situations. Other safety-sensitive jobs require the time-sharing of attention among several dynamic cues. Vehicle operators--especially drivers and pilots in high-traffic areas--fall into this category.

Rank these jobs using a scale or set of scales that reflect each job's vulnerability to attentional lapses and the potential for loss of life, health, or property if an attentional lapse occurs. The scale(s) might be devised by managers, inspectors, quality circles, or others who observe the work environment with a critical eye. There may be more than one dimension that should be used to characterize various jobs--for example, how far the job deviates, either high or low, from moderate mental workload, and how likely an error is to injure someone. The range of the scales might be 0 to 10, 0 to 100, -10 to +10, and so forth. Once scores are obtained on such scales, it might be best to reduce the overall results to rank orders of jobs within these dimensions.

During the hiring process for safety-sensitive jobs, use interview and screening methods to identify and treat or exclude persons with sleep disorders characterized by uncontrollable sleep. For example, we use a questionnaire that highlights possible sleep disorders and then follow up with an interview. The limitations of self-reporting must be considered. Additionally, all applicants must be informed of the performance requirements of the safety-sensitive job.

Examine the recent daily patterns of errors of commission, omission, and judgment that occur in the various safety-sensitive jobs in the workplace. In this examination, do not mix data from crews with different shift-change times, because their patterns of error may be somewhat out of phase with each other. If data that are out of phase are mixed together, their patterns of oscillation will obscure each other. For example, if shifts change at 6:00 a.m., 2:00 p.m., and 10:00 p.m. for one set of jobs and at 9:00 a.m., 5:00 p.m., and 1:00 a.m. for another set of jobs, keep the data separate.

Educate workers about the intelligent management of their principal sleep periods and naps. Most people view sleep as a passive state of the brain. It is not. The brain generates sleep actively, and it needs the proper environmental conditions and time of day to do that efficiently. Simply educating workers about the conditions the brain needs to generate good sleep can help. This approach is being used for commercial truck drivers by both government regulators and motor carriers.

When staffing levels are specified and rotating shifts are designed, consider the periods in the 24-h day when there is the greatest vulnerability to sleep and error. Those who schedule rotating shifts should learn the principles of chronohygiene and the fundamental arithmetic of shift scheduling. For example, using factors of the 168-h week and the 364-day year are essential for producing shift rotation patterns that both minimize the influence of the two-peak pattern on worker errors and are acceptable to workers.

Double-check or triple-check the safety-sensitive work performed during periods of vulnerability. If the work is so critical that it is usually double-checked, then it should probably be triple-checked in the predawn and midafternoon hours. Similarly, other important work should be double-checked during those periods.

Use daily (or more frequent) fitness-for-duty testing to prevent workers who are impaired for any reason--including sleep deprivation--from trying to perform safety-sensitive jobs. A number of workplace tests are available that are brief, reliable, and self-administered.

Introduce preplanned naps into the duty periods of selected workers by adapting the approach used for long-haul pilots by Rosekind and colleagues. For example, military managers assigned five people to work in a four-position workplace that required sustained attention throughout an 8-hour shift. Each member of the team rotated through eight 48-minute work cycles and two 48- minute rest cycles within each shift. During those rest breaks, workers often took 20- to 40-minute naps in a dark, quiet room set aside for that purpose.


Some or all of these ideas may be applied to safety- sensitive jobs that require around-the-clock operations or operations during midafternoon. Examples include the following:

  • Commercial motor vehicle operators, especially those carrying passengers or hazardous materials
  • Health care workers who monitor critically ill patients, respond to medical emergencies, and perform surgery
  • Maritime watchstanders, especially on ships carrying passengers or hazardous materials
  • Monitors of safety-sensitive systems, such as air traffic controllers
  • Pilots
  • Quality control inspectors whose detection capabilities determine the safety and quality of manufactured products
  • Security guards who are responsible for personal safety and the security of potentially hazardous work environments
  • Shift workers who control machines with high potential for personal injury.

Careful attention to the prevention of accidents always pays off. These ideas should help managers and workers to design and implement new preventive measures that will help to keep incident and accident rates low.

[Extracted from "Predicting Accident Times," by James C. Miller & Merrill M. Mitler, ERGONOMICS IN DESIGN, October 1997, Vol. 5, No. 4. Copyright 1997 by Human Factors and Ergonomics Society, P.O. Box 1369, Santa Monica, CA 90406-1369; 310/394-1811, fax 310/394-2410,, A copy of the complete article (with references) may be obtained from HFES on request.]

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.