The impact of sleep deprivation and alcohol on driving: a comparative study

Assesses driving strategy and manoeuvring ability by examining the ability of participants to promptly react to a stimulus whilst correctly judging whether to brake or change lane [34] using a 5-lane simulated roadway with overhead signage prompts. It generates a test score based on reaction time (milliseconds) and correct decision making in response to speed or lane change prompts.

Assesses operational skills by testing steering and lateral vehicle control [35]. Participants, when prompted by signage, were required to perform multiple deliberate lane changes as quickly and efficiently as possible along a 6 km route whilst obeying the indicated speed limit [36]. Driver impairment was quantified by calculating mean deviation from the model lane change path (MDLT) in response to signage alongside the ability to maintain lateral control of the vehicle.

A 9-point Likert scale requiring participants to select the statement that best describes their current state of sleepiness, where a score of 1 corresponds with feeling extremely alert and 9 with feeling very sleepy and the individual having to make a great effort to fight sleep and remain awake [37, 38].

A 5-point Likert scale, where 1 indicates the driver considers that they are unable to drive and 5 as being completely capable.

Statistical analyses were conducted using IBM SPSS Statistics version 24 for Windows [39].

Sleep deprivation produced a significant detrimental impact on lateral vehicle control (P < 0.0001), braking reaction time (P < 0.0001) and lane change reaction time (P < 0.0001). Cohen’s d effect size for the impact of sleep deprivation = 1.22. Males were affected to a greater degree than females compared to their rested performance (male mean difference in impairment 6.07, female 1.97, P < 0.0071). An example of the detrimental impact of sleep deprivation on Adaptive Mean Deviation of Lane Tracking (MDLT) is illustrated in Figs. 2 and 3.

Although participants reported feeling significantly less impaired (mean SDIS increase of 0.87, P < 0.0001) after drinking coffee when sleep deprived, in reality it was not found to reduce driving impairment (mean combined score 9.03 before coffee vs. 10.16 following coffee; P = 0.01, Cohen’s d = 0.21). Driving performance worsened, with significant deterioration in lane tracking (mean; 0.62 vs 0.71, P = 0.03) and no improvement in braking and lane changing reaction times.

Alcohol intoxication produced significant deterioration in lateral control (P < 0.0001) and lane changing time (P = 0.0058, Cohen’s d = 1.57) Braking reaction times were also impaired, although not to a significant degree (Table 2). Participants reported feeling sleepier and that their driving was impaired by alcohol, reflected in increases and decreases in KSS and SDIS scores, respectively (Table 2). Coffee did not significantly alter alcohol-related driving impairment across any of the objective domains, although intoxicated participants reported feeling significantly more confident in their driving abilities after consuming it (mean SDIS 2.83 vs. 3.37; P < 0.0001).

Sleep deprivation was found to produce a greater degree of overall driving impairment than alcohol intoxication when compared to performance when fully rested, with a mean difference of 3.47 (P < 0.0001) in driving score when sleep deprived and 2.22 (P < 0.0001) when intoxicated (Fig. 4). Deterioration in lateral tracking, braking reaction time (P < 0.0001) and lane changing time (P = 0.0003) were all more pronounced when sleep deprived compared to intoxication, with sleep deprivation producing a mean retardation of 0.63 s for combined braking and lane changing reaction times compared to only 0.12 s when intoxicated. Participants reported feeling that their driving performance was equally impaired by sleep deprivation or by alcohol intoxication, with the degree of perceived impairment demonstrating a positive correlation with measured impairment (P < 0.0001). However, following coffee ingestion, sleep deprived participants reported feeling less impaired, despite further deterioration being observed in their driving performance.

Our alcohol intoxication results mirrored the observations of previous studies [40‐44], with a demonstrated reduction in lateral control when changing lane whilst intoxicated and a progressive deterioration in the ability to recognise and assess the degree of impairment with an increasing state of intoxication [45]. Additionally, our study also found that this mismatch between real verses perceived impairment was further enhanced following coffee, as participants reported feeling less impaired despite still being in excess of the BrAC limit for driving and exhibiting driving impairment.

Furthermore, the differences in reaction times following sleep deprivation were even more pronounced. Our results support current hypotheses regarding driving whilst sleep deprived. Individual response to sleep deprivation varied widely, with some subjects profoundly affected, as observed through episodes of microsleep and demonstrated through objective and subjective assessment, whilst others were impaired to a lesser degree.

The ability to keep the car centred within the lane is critical to safe driving and a particularly sensitive measure of driving impairment. Our results are consistent with previous research [46], although interestingly, our study found females to be significantly more resilient to impairment, particularly in relation to lateral control when changing lanes. However, whether this difference in impairment between the sexes is due to differences in their sleep behaviour or in their susceptibility to the resultant impairments remains unclear. Braking reaction rate was also detrimentally affected, although to a lesser extent than lane changing rate, which may reflect the more demanding complex decision-making nature of the latter skill, which requires the driver to not only correctly distinguish whether to brake, but also select and manoeuvre into the correct lane.

An explanation for the differences in driving impairment following sleep deprivation and intoxication has yet to be described, but differences in the mechanisms fundamental to both and in strategies employed by subjects to counteract the impairment generated by these two stressors have been proposed [47]. Alternatively, the possibility that both stressors share a common mechanism has also been considered. For instance, intoxication – like fatigue – may enhance the habituation process. This results in the brain becoming progressively susceptible to environmental and circumstantial influences to maintain performance. When these influences are monotonous this habituation might accelerate, “driving” the brain to sleep [48]. However, this is a complex relationship, influenced by several factors such as: demographic differences, emotional states, circumstantial factors and personality traits [49]. Neurologically, one can consider the role of the main inhibitory neurotransmitter in our brain, GABA, as a potential common mechanism. For example, alcohol is well known to be a positive allosteric modulator at GABA receptors and interestingly, Tadavarty et al. found that as little as 12 h of sleep deprivation can alter the expression of GABA_B receptors [50].

The beneficial effects of caffeine on non-sleep deprived subjects are well documented and supported by this study; reaction times improved significantly in the control condition after coffee [51]. However, far from improving, driving impairment when sleep deprived or intoxicated continued to deteriorate despite the consumption of coffee. Factors which may have contributed to this outcome include reduced concentration and fatigue due to the additional time that testing the effects of the countermeasure added to the session, or that the interval between coffee consumption and re-testing was of insufficient duration to allow for the onset of effects. Whilst it has been previously shown that an 80 mg dose of caffeine, delivered in a similar timeframe to our own study produced driving improvement during a prolonged 4-h monotonous simulated drive [52], it may be that coffee has less of a beneficial effect on shorter drives such as commutes or when driving in urban areas with heavy traffic and its associated increased demands in relation to braking and lane changing.

Another possibility is that the caffeine dose was insufficient to produce an observable improvement, as noted in other studies [9, 53]. Many of the experiments that have reported improvements employed significantly larger doses of caffeine than the moderate amounts that would be encountered in real life situations, such as suggested in the Highway Code guidance [25]. 150–250 mg of caffeine (comparable to two/three cups of coffee) has been shown to effectively counteract sleep deprivation (< 5 h spent in bed) when driving in early morning [54] and in the early afternoon [55]. This dose is larger than the one used in this project and the length of sleep deprivation is also shorter. It may be therefore, that coffee improves performance after a shorter length of sleep deprivation than tested in this project.

Two on-the-road driving studies on a motorway in France also suggest differing results – caffeine improved night-time driving in young and middle-aged drivers [56]. Additionally, one study showed that coffee (150-200 mg caffeine) and napping at night significantly reduced driving impairment without altering the subsequent sleep [57]. However, limited comparisons can be drawn from these; on-the road testing has been shown to produce different results from simulated driving. For instance, impaired performance has been noted to occur faster in simulated driving relative to actual driving; most likely as a result of the greater monotony associated with simulated driving [58].

The act of taking short periods of sleep (napping) has been demonstrated to be comparatively more effective in younger subjects, suggesting that the impact of sleep as a countermeasure to sleep deprivation varies with age [56]. The possibility that variance in the effectiveness of coffee as a countermeasure may have occurred, in part with the age of the subject, was eliminated in this study through selecting subjects from a narrow age range. Therefore, the possibility that coffee may be more effective as a countermeasure in an older population cannot be discounted.

In addition to objectively measuring driving impairment, the study sought to establish the subjects' awareness of an impairment and the impact that coffee had on the confidence they placed in their driving abilities.

Baranski found that actual performance and subjective performance were closely related during periods of prolonged wakefulness [59]. This is consistent with the majority of findings from our study, showing the participants’ perceived level of impairment significantly correlated with their actual driving impairment when sleep deprived and following alcohol. As with other studies [60], participants reported feeling that their driving was impaired and noted a deterioration in their driving performance when sleep deprived compared to when fully rested. A strong correlation was observed between driving impairment and sleep deprived KSS scores for each of the three different tests in all 30 study participants. This is consistent with a study which concluded participants were good at estimating their sleepiness if presented with a task involving some form of objective measure [61]. In contrast to alcohol intoxication, which has been demonstrated to increase a driver’s confidence in driving ability [3], sleep deprived drivers appear to have better insight into the associated impairment of their abilities, as indicated by their subjective driver impairment scale scores. Participants showed a mean 2.16 decrease in subjective impairment when sleep deprived but showed a 1.86 decrease when intoxicated.

We observed that sleep deprivation-associated driving impairment continued to deteriorate following coffee ingestion, despite drivers reporting an enhanced state of alertness and increased confidence in their fitness to drive. This highlights a concerning false perception amongst drivers that coffee will effectively counteract any driving impairment caused by sleep deprivation.

The validity of driving simulator studies will always be challenged because they are limited by how much they replicate real on-the-road driving. Although simulator studies may not completely represent real driving, they have been validated for assessing aspects of performance fundamental to driving, such as steering deviation, reaction time and vigilance [36, 62]. Additionally, the driving scenarios used demonstrate to a significant extent situations from which the project findings can be extrapolated. A simplistic scenario was chosen to allow generalising to motorway driving which is important because drivers experience the greatest risk during monotonous motorway drives [63]. The order of experimental conditions was not randomised or counterbalanced so the study cannot discount the possibility of order affecting the results. Participants were however, given a pilot session to familiarise themselves with the test battery and moreover, the tests themselves were simplistic to eliminate the risk of practice effects. If these had occurred, an improved performance would be expected after each repeated exposure to the test battery. Yet, it can be confidently said that the observed results were minimally influenced by practice because of the pattern they produced. If practice was a factor, if anything, the degree of driving impairment following sleep deprivation was probably reduced. Subjects may have aimed to demonstrate a more substantial impairment after sleep loss than alcohol – participant bias. Although, it is believed the results are valid because the hypothesis was not communicated to participants and they were unaware of how the scoring system worked, helping deter intentional poor performance.