Demographic factors, fatigue, and driving accidents: An examination of the published literature
Introduction
Developed and developing economies rely on transportation systems that operate well beyond the ‘normal’ 8-h workday to convey people and freight to meet personal and business needs. However, around-the-clock operational requirements of transport systems may exceed the human capacity to work efficiently and safely. This inability to optimally perform around the clock is typically attributed to fatigue that results from sleep homeostasis, circadian rhythms, and workload (Williamson et al., 2011). We expand upon these factors by highlighting the additional contributions to fatigue and accidents from demographic variables, work arrangements, and other individual differences. A more thorough understanding of fatigue in transport systems is paramount given that some 9.1 million Americans (Bureau of Labor and Statistics: www.bls.gov/news.release.Dec2008) and 8.2 million Europeans (Eurostat, 2007) are employed in this sector. Thus, fatigue in the transport sector has implications for the well-being of large numbers of transport workers and, in the case of a major safety mishap, for the community at large (Mitler et al., 1988). Of major concern is the fact that night work, itself, is associated with a heightened accident risk (Folkard and Tucker, 2003), including automobile (Langlois et al., 1985), truck (Kecklund and Åkerstedt, 1993), and train crashes (Hildebrandt et al., 1974, Torsvall and Åkerstedt, 1987).
There is general consensus among industrial experts on the ‘main’ determinants of fatigue, and some of these underpin various bio-mathematical models used to predict it (Dawson et al., 2011, Williamson et al., 2011). However, there are a great many additional endogenous and exogenous variables that contribute to or influence fatigue, for example, as depicted in Fig. 1. In this review, we first briefly discuss some difficulties in defining fatigue. Second, we focus on the demographic variables that have been linked with fatigue. Third, we examine the role of the less frequently discussed types of working arrangements in generating fatigue. Fourth, we discuss the potential role of personality and chronotype in fatigue and accident involvement. In the final section, we outline gaps in our knowledge and address those areas requiring future research prerequisite for future policy.
Several published studies indicate some 20% of the total working population report being fatigued (Pawlikowska et al., 1994, Hickie et al., 1996, Loge et al., 1998). In lay terms, fatigue is readily understood as an outcome state, feeling tired or sleepy. However, among professionals and academics, the situation is different; there is much debate between and within the many involved disciplines, but as yet no agreed definition (Noy et al., 2011). The inability to develop a shared understanding of fatigue has resulted in investigators employing a wide range of surrogate measures and experimental designs to assess and research it. We raise these issues because they pervade the entire literature on fatigue and thus make it difficult to clearly address it. Our ability to completely comprehend the construct of fatigue can contribute to the development of technologies that will assist people to better cope with 24/7 operations and reduce fatigue-related risks (Balkin et al., 2011, Dawson et al., 2011).
One of the most basic distinctions is whether fatigue refers to a physical or mental state, or both. In the field of industrial medicine, fatigue is often construed as a physical state phenomenon. As such, fatigue is a state of resource depletion that can be objectively measured. From this perspective, fatigue has been inferred by changes in: heart rate (Duchon et al., 1997), blood pressure (Iwasaki et al., 1998), hand strength (Volle et al., 1979), and actigraphic (Baulk et al., 2008) and electroencephalographic (Balkin et al., 2011) validated sleep alterations. Fatigue has also been assessed in terms of changes in objective performance on simulated tasks, such as the Psychomotor Vigilance Test (PVT; Dinges et al., 1997), logical reasoning, and numerical aptitude (Rosa, 1991). Poor performance on the PVT has been reliably linked with short sleep duration (Baulk et al., 2008, Van Dongen, 2006). The identification of a gold-standard to distinguish and quantify fatigue is advantageous, but it has two key limitations. First, it ignores individual differences in the response to fatigue, and, second, it assumes that change in the indicator is not due to other factors. For example, Duchon et al. (1997) showed that heart rate did not increase on an extended work shift. However, this was attributed to workers purposely reducing their output to meet the demands of the longer shift. Moreover, changes in performance may be the consequence of illness and/or adverse effects of medications (Smolensky et al., 2011)
Brown (1993) defined fatigue as “a subjectively experienced disinclination to continue performing the task at hand because of perceived reductions in personal efficiency (p. 240).” An advantage of this position is its recognition of the individual variability in the manner people experience fatigue, but it too has limitations. It provides no clue as to what may explain or cause the ‘disinclination,’ and it raises the question of whether fatigue is a motivational state. Indeed, Soames-Job and Dalziel (2001) argue the disinclination may be the consequence of one having achieved some minimum performance or a change in reward structure. A second limitation is that in the absence of an a priori indication of fatigue, it can only be inferred from its impact on a dependent variable. Thus, a fatigue-related crash is defined as one in which all other potential causes of the accident have been excluded (e.g., Horne and Reyner, 1995). A third limitation of Brown's (1993) definition is that it fails to address whether fatigue describes a physical or mental state. Given that Brown considers fatigue to be a subjective phenomenon, distinguishing between these two states is important. From an applied perspective, these differences do matter, since they have implications for recovery time and, hence, fatigue management (Dawson and McCulloch, 2005). Furthermore, the literature typically fails to differentiate between acute versus chronic states of fatigue (Tepas and Price, 2001).
Williamson et al. (2011) defined fatigue as ‘a biological drive for recuperative rest’. This definition is advantageous in that it recognizes that fatigue may take multiple, i.e., mental and physical forms, and that the remedy is also dependent on the form of the fatigue. Thus, sleep or rest is sufficient to alleviate fatigue. What should be clear from this discussion is that fatigue is a multi-dimensional construct with psychological, physiological, and behavioral outcomes.
The inability to define fatigue has resulted in studies employing subjective and/or objective measures and a diverse variety of study designs. Self-report protocols raise concern over validity. For example, a large UK survey found that nearly 30% reported almost falling asleep while driving; yet, only 7% attributed sleep loss as resulting in an accident (Maycock, 1996). We do not wish to imply that all self-report measures are suspect. The Karolinska Sleepiness Scale (KSS; Åkerstedt and Gillberg, 1990) is extensively used (Di Milia and Bohle, 2009) and has been validated against physiological parameters (Kaida et al., 2006). Excessive sleepiness ratings from the KSS (≥7) have been linked with increased accidents in simulator studies (Åkerstedt et al., 2005).
Some studies have used both subjective and objective measures with mixed results. Rosa (1991) found long work shifts degraded subjective and objective performance tasks, while Dahlgren et al. (2006) found 12-h day shifts were linked with greater exhaustion, increased sleepiness, and reduced sleep (evaluated via actigraphy). However, salivary cortisol, heart rate, and blood pressure levels showed no adverse effect of the longer work-week. In a review of long working hours, van der Hulst (2003) concluded that subjective, rather than objective, measures are more commonly linked to fatigue.
Other evidence for fatigue-associated impairment comes from behavioral studies of driving performance. Simulator studies that examined performance in terms of lane drift (Åkerstedt et al., 2005, Otmani et al., 2005) have demonstrated poor performance is associated with excessive subjective sleepiness and elevated levels of EEG alpha and theta activity. While simulator studies offer a number of benefits, their main limitation is ecological validity; the consequences of driving and crashing in a simulator study environment are quite different from those of an actual road environment.
Finally, we wish to emphasize the study design affects the validity of research findings. Most studies employ cross-sectional designs to explore associations between selected variables. A major weakness of these is their inability to determine causal relationships between variables. Longitudinal designs are not necessarily better, since they are seldom able to randomly assign participants to groups, and they tend to have a high drop-out rate over time, resulting in small samples (Di Milia, 1998, Rosa, 1991). Epidemiological studies better underscore the complex nature of the link between the independent and dependent variable(s), but they also do not enable determinants of causality; the odds ratio reflects the variable most associated with the dependent variable, but not inference of cause (Peng et al., 2002). Moreover, an important limitation of certain epidemiological studies of driving accidents is that they are based on survey of survivor populations; therefore, they may not be representative of all drivers involved in at-fault driving and transportation accidents in which fatigue plays a role.
In closing this section, we also wish to emphasize that the vast majority of studies typically report the basic demographic variables of age and sex. Perhaps, due to space limitations in journals or simply the inherent focus of the investigators other demographic variables are either not assessed or, if they are, not considered as potential determinants or predictors of outcomes. We address some of these variables in the next section. Another limitation of understanding how demographic variables impact fatigue and driving accidents is that they are not directly used to assess the dependent variable. Rather, they are typically used as covariates in statistical analysis, i.e., to partition the variance attributed to them, to better determine the main effects of the study variables of prime interest according to stated hypotheses. The literature would greatly benefit by assessing all the demographic variables as primary ones in more broadly conceived hypotheses.
Section snippets
Demographic variables
A fairly large number of publications have explored the role of fatigue in relation to selected classic demographic variables such as age and sex. However, we are unaware of any study that has comprehensively explored the contribution of a large number of demographic factors simultaneously. In this and subsequent sections we summarize the literature linking fatigue with demographic factors, selected working arrangements, and personality traits/circadian chronotype. Since the number of
Industrial work arrangements and fatigue
In this section the focus is on some selected work arrangements that receive less attention than shiftwork but nonetheless may generate fatigue. Reviews of shiftwork can be found elsewhere (Åkerstedt, 2003, Costa, 2003). A discussion of this topic fits well into the broader context of understanding that demographic factors interact with work arrangements. Work schedules vary by a number of criteria, including the timing of work hours, shift length, shift schedule (fixed or rotating), speed of
Personality, circadian chronotype, fatigue, and accident risk
This section further broadens the scope of demographic variables to include the seldom considered factors of circadian chronotype and personality traits, since they constitute individual differences in the susceptibility, recognition, and response to fatigue. The findings of numerous studies clearly indicate these attributes are at least related to demographic variables, in particular, sex and age-related differences in circadian chronotype and the personality traits of conscientiousness,
Discussion
There is general agreement that fatigue in the workplace is the result of three interacting factors, i.e., circadian rhythmic, sleep homeostatic, and task/work – core processes (Dawson et al., 2011, Williamson et al., 2011), although other factors can be involved in this and other settings (see Smolensky et al., 2011). However, these processes only emphasize the understanding of fatigue at the global level. Much less is known about the role of demographic and other individual differences in the
References (229)
- et al.
Work load and work hours in relation to disturbed sleep and fatigue in a large representative sample
J. Psychosom. Res.
(2002) - et al.
The challenges and opportunities of technological approaches to fatigue management
Accid. Anal. Prev.
(2011) - et al.
What aspects of shiftwork influence off-shift well-being of healthcare workers?
Appl. Ergon.
(2008) - et al.
Impulsivity but not venturesomeness may be related to morningness
Psychiatry Res.
(2005) - et al.
Comparing three morningness scales: age and gender effects, structure and cut-off criteria
Sleep Med.
(2009) - et al.
Phase advance of sleep and temperature circadian rhythms
Neurosci. Lett.
(2002) - et al.
An analysis of the eveningness–morningness dimension in depressive college students
J. Affect. Disord.
(1999) - et al.
Association of sleep–wake habits in older people with changes in output of circadian pacemaker
Lancet
(1992) - et al.
Driving anger, sensation seeking, impulsiveness, and boredom proneness in the prediction of unsafe driving
Accid. Anal. Prev.
(2005) - et al.
Managing fatigue: it's about sleep
Sleep Med. Rev.
(2005)