A Review of Alcohol-Impaired Driving: The Role of Blood Alcohol Concentration and Complexity of the Driving Task
[Correction added on March 25, 2023, after first online publication: Re-use of this article is permitted in accordance with the Terms and Conditions set out at: http://olabout.wiley.com/WileyCDA/Section/id-828080.html]
Abstract
The operation of a motor vehicle requires the integrity of sensory, motor, and intellectual faculties. Impairment of these faculties following the consumption of alcohol has been studied extensively through laboratory, closed-course and on-road driving, and epidemiological studies. The scientific literature was reviewed critically, with a focus on low-to-moderate blood alcohol concentrations (BAC ≤ 0.100%), to identify the most reliable determinants of alcohol-impaired driving. Variables such as age, gender, driving skill, and tolerance were shown to have limited impact on impairment. It was concluded the most relevant variables are BAC and complexity of the driving task. The scientific literature provides a high degree of confidence to support the conclusion that a BAC of 0.050% impairs faculties required in the operation of a motor vehicle. Whether impairment is apparent depends upon the complexity of the driving task, which applies to both study design and actual driving.
It has been recognized for over a century that alcohol (the unmodified term “alcohol” refers to ethanol) produces deficits in the ability to operate a motor vehicle. The relation between alcohol and vehicular death and injury was identified in a 1904 editorial of the Quarterly Journal of Inebriety, “The general palsy and diminished power of control of both the reason and senses are certain to invite disaster in every attempt to guide such [automobile] wagons.” 1 The advent of chemical testing, along with increased understanding of the pharmacology of alcohol, subsequently allowed for more specific characterizations regarding the impairing effects of alcohol.
Over 60 years of alcohol research has examined the effects of alcohol on driving through (i) laboratory experiments and driving simulations, (ii) closed-course and on-road driving experiments and (iii) epidemiological studies. From these studies, it has long been observed that alcohol degrades skills required in the operation of a motor vehicle, as well as actual driving performance, culminating in an increased risk of being involved in a motor vehicle collision.
Past research has tended to focus on assessing the scientific merit of per se limits. The first per se limit was set by Norway in 1936 at a blood alcohol concentration (BAC) of 0.050% 2. In North America, it was more than three decades later in 1969 that Canada legislated a per se limit of 0.080% 2. Only recently has the U.S.A. adopted a national standard of 0.080%, replacing older per se legislation of 0.10% 3. This focus on BACs of 0.080% and 0.100% has resulted in a relative paucity of laboratory studies dealing with low-to-moderate BACs (i.e., ≤0.100%). As observed by Moskowitz et al. 4, “[l]aboratory examinations of skills performance at blood alcohol levels (BALs) below 50 mg/dL (0.050%) are uncommon in the literature.” Per se limits are based on a variety of influences including determinants that are not based on science (e.g., legal aspects, political pressure). Therefore, per se limits are not entirely scientific and should not be construed as the threshold BAC at which impairment in the ability to operate a motor vehicle occurs. As concluded by Moskowitz and Robinson 5, “…from a scientific point of view, there appears no lower BAC below which impairment cannot be said to exist. The legislature is free to prohibit driving at any BAC, since such a limit would not contradict the scientific data demonstrating no lower alcohol limit to impairment.”
A number of reviews have summarized the effects of alcohol on driving 6-8. Literature reviews focussing on driving impairment at low BACs have also been conducted 9-11. The advancement and refinement of technology, coupled with political will and societal pressure, has provided scientific research demonstrating impairment at progressively lower BACs. An overwhelming number of scientific investigations now provide convergent evidence supporting impairment of faculties required to operate a motor vehicle at low-to-moderate BACs. The objective of this review is to crystallize the most important concepts in determining the effects of alcohol on driving ability. The relation between BAC and psychomotor impairment is examined, with emphasis on experimental research concerning low-to-moderate BACs. In addition, the role of experimental variables (i.e., complexity of task) and subject characteristics (e.g., age, gender, functional tolerance) in predicting alcohol-related impairment of driving skills is critically evaluated.
Definition of Impairment
Legally, impairment is an issue for the trier of fact to determine. The current Canadian legal authority is provided by the Supreme Court of Canada in R. v. Stellato 12, which upheld that impairment comprises “…any degree of impairment ranging from slight to great.” Moskowitz 13 defined alcohol-induced impairment as “any change in performance level from that exhibited at zero BAC.” For the purposes of this review, impairment is defined as “a decreased ability to perform a task.” Specifically, impairment relates to a diminished ability to operate a motor vehicle and is based on an individual's abilities comparable to their own performance in an alcohol-free state. Furthermore, psychomotor impairment, as it relates to driving, is separate from the physical effects produced by alcohol consumption (e.g., slurred speech, diminished coordination, staggered gait). While such effects may be present in an alcohol-impaired driver, they are nonspecific and subjective attributes of “drunkenness” or “intoxication,” and are distinguishable from the complex behaviors required for the operation of a motor vehicle. Appreciation of this creates an important distinction between impairment and intoxication: While an “intoxicated” individual will definitely be impaired in their ability to operate a motor vehicle, an absence of intoxication should not be taken as an absence of impairment.
Laboratory Studies
Scientific understanding of the specific impairing effects of alcohol on driving performance has evolved primarily from studies conducted in the laboratory, which are far more prevalent when compared with either on-road or closed-course driving experiments. Laboratory studies assess the effects of alcohol on individual, albeit interrelated, skills and faculties relevant to the operation of a motor vehicle including simple and complex reaction time 14, tracking ability 15, and the ability to perform divided attention tasks 14, 16. Laboratory studies provide an opportunity for greater control over experimental conditions thereby maximizing internal validity. This promotes confidence that an observed decrement in performance is attributable to an effect produced by alcohol.
Divided Attention and Complex Reaction Time
A broad range of faculties related to driving have been examined through laboratory studies. The ability to divide attention and complex reaction time (i.e., choice reaction time) has been identified by many investigators as being particularly sensitive to the effects of alcohol. For example, impairment of these abilities has been demonstrated consistently at BACs lower than 0.050% 7, 17.
The ability of a driver to divide their attention between multiple operations and sources of information is essential to driving. When impaired, the available attention of the driver is allocated to fewer aspects of driving such as maintaining vehicle position or speed. In the laboratory, divided attention tasks typically combine both tracking skills and a visual search component. In one study, impairment in a divided attention task was observed in all subjects at the lowest BAC tested (0.015%) while impairment of visual information processing was observed in all subjects at BACs of 0.022–0.035% 4. The degree of impairment in both tasks increased as BAC increased. These findings were supported by a later study that demonstrated impairment in simulated driving and divided attention tasks at BACs as low as 0.020% 18. Other laboratory studies have yielded similar findings, providing reliable evidence that alcohol-related impairment of divided attention tasks occurs at a BAC <0.050% 19-24.
Complex reaction time refers to the time required to respond in a scenario when there are two or more stimuli and several possible responses. A short complex reaction time allows a driver to react quickly and appropriately to emerging situations. When demonstrated experimentally, statistically significant increases in complex reaction time may be measured as millisecond (ms) changes. These increases in response time might therefore be misinterpreted as inconsequential; however, a 100 ms increase in overall reaction time at a driving speed of 100 km/h would result in an additional 2.8 m of distance travelled before a response is elicited. Such an example illustrates the utility of complex reaction time as a measure of alcohol effects given the complex nature of the driving task.
Laboratory studies of complex reaction time following low-to-moderate doses of alcohol demonstrate an increase in both response time 14, 25-27 and the number of errors made during the task 19, 28, 29. The conclusion from such work is that alcohol both slows an individual's ability to respond to a situation and increases the likelihood of incorrect decision-making in the scenario. The role of task complexity in revealing psychomotor impairment relates to the increase in test sensitivity that follows. Accordingly, impairment becomes demonstrable at a lower BAC. One recent laboratory study, which examined response time to visual and auditory stimuli, measured significantly longer reaction times at ethanol doses corresponding to BACs ranging from c. 0.013% to 0.038% relative to an alcohol-free condition 27. In the same study, total response time was significantly slower for a complex reaction time task (where multiple stimuli were present) when compared with a simple reaction time task (single stimulus) during both visual and auditory testing, illustrating the importance of task complexity on impairment of this faculty.
Early recognition of this concept includes Mead 30 who postulated that when attempting to accurately describe the physiological effects of alcohol, “we must find valid and reliable measurements of performances of different complexities.” Experimentally, it was observed that a moderate dose of alcohol more readily affected performance on a complex task, requiring intellectual activity, when compared with a simple reflex functioning task 30. Other authors have also emphasized the impact of task complexity in detecting psychomotor impairment. Maylor et al. 14 reported that mean (±standard deviation; SD) BACs ranging from 0.060% (±0.019) to 0.080% (±0.020) significantly impaired speed of detection on an auditory stimulus detection task. When concurrently challenged with a tracking task during the stimulus detection test, the impairing effect of alcohol was even greater. Breitmeier et al. 26 examined the effects of a low dose of alcohol (0.3 g/kg) on cognitive functions in 16 healthy males using a double-blind, placebo-controlled design. The authors reported that at a BAC of c. 0.030%, cognitive functions, especially those relying on perception and processing of visual information, were significantly impaired. Neuropsychological impairment became increasingly apparent with increasing complexity and urgency of the task 26.
Tracking Ability
Tracking, as it relates to the operation of a motor vehicle, can be described as the ability to maintain position on a roadway and is determined by both within-lane lateral position and excursions out of lane. An early demonstration of the effects of alcohol on steering accuracy and lateral lane position was provided by Drew et al. 31 who showed deterioration in these tasks at a BAC as low as 0.020% and a linear increase in impairment with increasing BAC. There was no evidence for a threshold effect, with measurable increases in error at all measurable quantities of alcohol in the blood 31.
In a study of 14 male subjects, Evans et al. 15 observed BAC-dependent decreases in tracking ability, as determined by a pursuit meter task, at concentrations <0.100%. Once again, when the complexity of the task was increased through the addition of a simultaneous distraction task, further impairment was revealed: a mean (±SD) BAC of 0.077% (±0.008), produced impairment in 13 of 14 test subjects.
Tracking ability is most often assessed by calculating the standard deviation of lane position (SDLP): a composite index that is a measure of weaving, swerving, and overcorrecting. Standard deviation of lane position may be assessed during closed-course and on-road actual driving experiments, but is also well suited to laboratory studies employing computerized driving simulators. Standard deviation of lane position is consistently identified as a particularly sensitive measure of alcohol-related impairment. Significant performance deficits were identified on a lane-change task at a peak BAC of 0.080% by Huemer and Vollrath 32. The impaired performance was primarily attributed to variations in lane keeping (i.e., SDLP), whereas the actual performance during the lane change, that is, deviation from the ideal trajectory was unaffected 32. A number of other researchers have substantiated the relation between alcohol and impairment of tracking ability. For example, Veldstra et al. 33 and Mets et al. 34 conducted independent examinations of the effects of increasing BAC on SDLP with similar results. Veldstra et al. 33 employed BACs of 0.030%, 0.050%, and 0.080% and demonstrated increased SDLP at all concentrations when compared with SDLP following placebo. Similarly, Mets et al. 34 demonstrated concentration-dependent increases in SDLP at 0.050%, 0.080%, and 0.11%. Rakauskas et al. 35 and Fillmore et al. 36 had previously published similar findings. In these studies, subjects with mean BACs ranging from 0.045% to 0.108% demonstrated increased lane position variability accompanied by increased effort as measured by steering variability or steering rate. The impairing effect of alcohol on SDLP was greatest when drivers were challenged by an accompanying distraction condition 35 or increased motivational conflict 36, which provides additional evidence of the importance of task complexity in predicting impairment. Task complexity was evaluated by Rupp et al. 37 in a study that measured lane variability at low BACs under a dual-task condition following a period of extended wakefulness. Impairment was present for all conditions, but performance deterioration was greater during the dual-task condition. Further performance deterioration occurred following an extended period of wakefulness 37. More recently, Harrison and Fillmore 38 examined the interaction between a BAC c. 0.080% and driver distraction on simulated driving performance. In this study, driving precision (SDLP) was impaired both with and without the presence of a secondary, distracting task; however, impairment was greater when subjects were required to simultaneously complete a divided attention task. This effect was measurable as a mean increase in SDLP, relative to placebo, of 5.18 cm for the alcohol condition, compared with a mean increase of 10.97 cm for the combined alcohol/divided attention task condition 38. These studies substantiate a concentration-dependent relation between alcohol and impaired tracking ability at low BACs. Furthermore, these experiments illustrate that impairment is readily apparent when the complexity of the driving task increases. As illustrated by the diversity of these studies, task complexity can be altered by a variety of parameters, for example, motivation, distraction, increased mental workload, extended wakefulness, the latter being a concept described by Ogden and Moskowitz 8 who stated “the effects of alcohol are dependent both on the quantity consumed and the nature of the performance required.”
Other Faculties
Other faculties relevant to driving that have been studied via laboratory experiments and demonstrated impairment at low-to-moderate BACs include: driving speed and acceleration 36, vision and eye effects [e.g., glare recovery time 39], saccadic and smooth pursuit eye movements 40, 41, peripheral attention 42, and vigilance 21, 25, 43-46. Significant impairment due to alcohol has been consistently demonstrated in critical tracking 21, 25, 47-49. The critical tracking task is described as a measure of a “subject's ability to concentrate attention and to integrate visual input with motor skills” 21.
Hazard perception latency, which includes elements of complex reaction time, has also been evaluated 50, 51. For example, Deery and Love 51 assessed the perception and reaction to a hazard in 32 drivers. The authors reported that at BACs ranging from 0.040% to 0.060%, subjects took longer to detect a hazard, and when subjects did react, they reacted more abruptly. Similarly, West et al. 50 demonstrated that BACs between 0.025% and 0.075% produced a detectable increase in traffic hazard latency in a double-blind, placebo-controlled simulator study.
Laboratory research is an invaluable means of identifying alcohol-related changes to the faculties that are integral to the operation of a motor vehicle. Such research has also contributed to elucidating the mechanisms by which the alcohol-related changes occur. It is beyond the scope of the current review to describe these mechanisms in detail; however, other researchers have provided such explanations. For example, Howat et al. 52 described a model for driving that comprised four phases of information processing: (i) seeing the situation develop, (ii) recognizing the stimulus, (iii) making a decision and (iv) executing an appropriate response. A comprehensive explanation of the automatic and controlled processes required for driving was described by Harrison and Fillmore 38: automatic processes were defined as the over-learned aspects of driving that require little conscious effort; controlled aspects of driving were defined as those that demand greater cognitive resources.
Despite the important contributions of laboratory-based research, there are limitations associated with these types of experiments. The operation of a motor vehicle is a complex task requiring the coordination of a variety of interdependent processes that occur nearly simultaneously. Laboratory simulations typically represent only one aspect of actual driving and require extrapolation of the effects as a predictor for other secondary tasks. Because the complex nature of driving cannot be characterized by a single variable, those studies that focus on a single measure of driving behavior may fail to demonstrate statistically significant impairment due to alcohol. Perrine 53 also identified that determining BAC limits for driving on the basis of laboratory experimentation alone risked setting limits that were too high due to insufficient complexity of experimental situations and the unnatural situation that a laboratory study presents. Starmer 54 judged insufficient complexity as the underlying cause of a “credibility gap” between what can be demonstrated experimentally and what has been observed empirically when crash-risk assessments are performed. As excerpted from Starmer 54, “unexpected high demand situations which might be intuitively considered to be more sensitive to the deleterious effects of alcohol are infrequent and difficult to duplicate in the laboratory or on the track.”
Recent laboratory studies have attempted to simulate the complexity of actual on-road driving situations by incorporating distractions and multiple task requirements in their simulated driving experiments 17, 35, 38. Nevertheless, the experimental context of the laboratory study remains a limitation of this type of work. Even the most sophisticated driving simulators can only approximate actual driving conditions and suffer from lack of car dynamics and inadequacy of the visual scene. Furthermore, although a driving simulator may better model complex driving scenarios when compared with laboratory psychomotor tests, simulated driving is subject to both over- and underestimation of driving performance. Underestimation of performance, as noted by Fillmore et al. 36, can arise because simulated driving does not engender the same degree of driver motivation as actual driving, given that there are no salient consequences, such as risk of personal injury. Conversely, overestimation of performance can arise from the absence of distractions typically encountered in real-life driving 36, 38. As observed by Harrison and Fillmore 38, technology-based distractions such as mobile telephones, global-positioning systems, and entertainment systems overburden the capacity of a driver's attention. The removal of such elements from simulated driving studies is critical to ensuring experimental control, which is the primary benefit of laboratory-based research, but may reduce the relevance of the study to actual on-road driving.
While it is important to identify the limitations of laboratory-based studies, such analysis should not detract from the overall contribution of the research. Through the resulting scientific literature, the effects of alcohol on driving have been described and the importance of BAC and task complexity is buttressed further. The limitations of laboratory research illustrate why these studies alone should not provide the basis for an opinion concerning the effects of alcohol on driving. Rather, such opinions should consider the results of a variety of types of investigation: laboratory, closed-course, on-road, and epidemiological, to increase the scientific confidence associated with the opinion.
Closed-Course and On-Road Driving Studies
Studies assessing the impairing effects of alcohol in subjects during the actual operation of a motor vehicle have high external validity; results from these studies generalize readily to actual driving behavior. For safety reasons, these studies are typically conducted on a closed course. Attempts to simulate realistic driving conditions have been made through the introduction of unpredictable or emergency traffic situations. For example, Laurell 55 incorporated a “surprise situation” on a closed course comprising the sudden appearance of a human-shaped obstacle blocking the roadway. Half the participants in the alcohol group (BACs <0.050%) collided with the obstacle, whereas only one participant in the control group was unable to stop safely. Subjects were also tested in an emergency situation, which required an evasive maneuver to be elicited in response to the appearance of two red brake lights. Significant increases in number of pylons hit and stopping distance were observed, including for 10 subjects with BACs ranging from 0.024% to 0.042%. Thus, the results of this study provide direct evidence of alcohol impairment on driving performance in emergency situations at a BAC <0.050% 55.
Closed-course studies often assess the effects of alcohol on the integrated task of driving and define impairment based on an aggregate degradation in the driver's skill. A review of closed-course studies on alcohol and driving impairment conducted between 1950 and 1973 by Huntley 56 noted most studies relied on human observation and judgment and performance measures such as number of pylons upset, elapsed time on course, or smoothness of operation. It was reported that alcohol changes driving behavior “on almost all dimensions observed,” including increased time to complete tasks, increased number of pylons upset, decreased stopping accuracy, decreased driving smoothness, and decreased cornering ability. Nevertheless, Huntley 56 concluded that the effects of alcohol on psychological components of driving required further study. Alcohol-related impairment of intellectual function, that is, decision-making related to driving, has received comparatively less attention in closed-course and on-road studies. A notable exception is that of Cohen et al. 57 who evaluated the effects of alcohol on risk-taking and judgment. Subjects were experienced bus drivers whose skill had been recognized previously with awards for safe driving. Three groups of drivers (control, low dose, moderate dose) were required to predict their success in driving their bus through gaps of various widths. While all subjects, including the control group, attempted to drive their bus through gaps narrower than the width of their bus, as drivers consumed more alcohol, they were prepared to drive through increasingly narrower gaps. Thus, while there was underestimation of the risk associated with a driving task, driving ability was overestimated and even in elite drivers judgment was deemed impaired at BACs below 0.050%. The closed-course driving performance of professional drivers at low-to-moderate BACs was also assessed by Lovibond and Bird 58. In this study, a composite score representing general driving performance and using measures such as braking reaction distance was calculated. Impaired performance was established for both “ordinary drivers” as well as for 16 drivers with racing and rally competition experience at BACs of 0.050%, 0.080%, and 0.100%. The authors concluded their results are “consistent with the conclusion derived from laboratory studies that any measurable quantity of alcohol in the blood impairs, in some measure, the skills important for driving” 58.
A broad approach, examining the effects of alcohol on overall driving performance, was employed by Robbe 59 who conducted an on-road experiment on city streets open to other traffic. Studies of this type are rare, due to safety concerns, but have a high degree of external validity given that they encompass the high degree of complexity that is associated with the typical driving environment, for example, traffic, pedestrians, traffic signals. It was concluded that even a modest dose of alcohol, yielding a BAC of c. 0.040%, produced significant impairment in 16 drivers when compared with their alcohol-free performance. Specific areas of deficiency were identified as vehicle handling and traffic maneuvers.
Other closed-course and on-road driving studies demonstrating impairment have focussed on car-following tasks as a dependent variable 49, 60-62. Although a stopping task was found to be insufficiently sensitive to identify intoxicated drivers, Attwood et al. 60 reported that the car-following task was highly sensitive to any amount of alcohol, with all subjects classified as impaired at an average (±SD) BAC of 0.033% (±0.008).
Closed-course studies have provided complementary data to the numerous laboratory studies that identify SDLP as sensitive to the impairing effects of alcohol 49, 61-64. Gawron and Ranney 64 employed a within-subject, placebo-controlled design to evaluate the performance of six males on a closed-course and a driving simulator. Alcohol effects observed on-road replicated the effects observed in the laboratory: tracking skills, lane-departure frequency, and SDLP were significantly impaired at a BAC of 0.070%. Kuypers et al. 49 documented significant increases in SDLP at a mean (±SD) BAC of c. 0.030% (±0.010) 4 h postalcohol administration. Another example, which confirmed the sensitivity of SDLP as a measure of alcohol impairment and further explored the concentration-dependent nature of the relation, includes Louwerens et al. 63 who observed an exponential increase in SDLP with increasing BAC, with significant impairment at a BAC of 0.060% and greater. The impairing effects of alcohol on SDLP are sufficiently well established that ethanol has been employed as a positive control against which to compare the effects of other drugs on human psychomotor performance 65.
In contrast, the effects of alcohol on driving speed are equivocal. Numerous authors have indicated that despite the clear effect of BAC on the ability to maintain lane position, speed perception, and speed control are far less sensitive to the effects of alcohol. For example, West et al. 50 failed to demonstrate a significant effect of alcohol on the time required to drive a fixed route. Kearney and Guppy 66 also reported a lack of significant effect of alcohol on perception of speed in a closed-course driving study despite an average (±SD) subject BAC of 0.095% (±0.004). Laboratory studies measuring driving speeds have largely concurred with these observations 33, 67, although Mets et al., 34 measured a significant increase in standard deviation of speed once a BAC of 0.080% or higher was attained. A simulated driving study conducted by Veldstra et al. 33 also demonstrated no significant effect of alcohol on driving speed when faced with a highway or a rural road condition; however, an urban driving task did produce significant increases in driving speed at BACs of 0.050% and 0.080%. This finding may reflect increased driving complexity in the urban condition resulting in increased sensitivity of the measure 33.
In a recent study, Harrison and Fillmore 38 concluded that the absence of a significant effect of alcohol on driving speed may itself represent impairment. Forty adults were tested on a driving simulator with and without the presence of a distraction task under alcohol or placebo. Blood alcohol concentrations of c. 0.080–0.090% did not significantly alter the driving speed of participants; however, drivers under placebo slowed their driving speed significantly when faced with the additional burden of the distraction task 38. The authors postulated that the alcohol group may have failed to adopt compensatory strategies to mitigate increased task complexity. This may be the result of an overestimation of driving ability while under the influence of alcohol.
Therefore, while it may be concluded that speed perception and speed control are relatively resistant to impairment at low-to-moderate BACs, caution must be applied to this interpretation. The limitations of closed-course and on-road driving studies may be a factor. Due to ethical and legal considerations, studies of this type are typically conducted at limited speeds [e.g., maximum 50 km/h 66]. Similarly, on-road highway and rural driving studies, and the speeds associated with such driving are rare, resulting in a lack of information regarding the effects of low-to-moderate BACs on driving speed in these environments. Robbe 59 and Ramaekers et al., 62 have conducted on-road research in normal traffic but identify the potential for reliance on the accompanying instructor and redundant vehicle controls as a possible artifact of this type of experiment.
In addition to slow driving speeds, legal and ethical considerations produce another potential experimental confound in this type of research. Task complexity is often limited by the removal of interactions with other vehicles, hazards, and pedestrians 63. Additionally, on-road driving studies in actual traffic may be limited to alcohol concentrations below established per se limits. Finally, for practical reasons and to ensure a relatively stable BAC throughout the test sequence, the duration of driving may be short and may not adequately address the vigilance requirements of real-world driving. Even with this limitation, on-road driving studies may more realistically reflect the increased complexity of actual traffic situations as demonstrated in the results reported by Parks et al. 68, who observed significantly increased SDLP in 60 subjects taking part in an on-road driving study where BACs averaged 0.042% before driving and dropped to an average of 0.031% after driving.
Given these constraints, it is clear that closed-course and on-road driving studies are likely to underestimate the true differences between driver performances under the influence of alcohol versus an alcohol-free state. The foundation for this conclusion includes (i) modification of subject behavior due to awareness of participation in an experiment (“Hawthorne effect”) and (ii) decreased task complexity due to practice effects, as identified by Laurell 55. Milner 69 asserted the following, “it is likely that if a drug is shown to affect driving skills in an experimental situation, its effects are probably even more pronounced in general driving behaviour.”
Epidemiological Studies
Epidemiologic data assess the probability of a driver's involvement in a motor vehicle collision as a function of BAC. These studies have consistently demonstrated an exponential increase in crash risk with increasing BAC.
The Grand Rapids study 70, the most extensive and widely recognized study of this type, compared drivers in 5988 motor vehicle collisions with 7489 control drivers. Borkenstein et al. 70 concluded, “[b]lood alcohol concentrations (BACs) over 0.040% are definitely associated with an increased accident rate.” Specifically, the risk of causing a crash was c. 2, 3, 7, and 25 times greater at BACs of 0.060%, 0.080%, 0.100%, and 0.150%, respectively, when compared with control drivers. More recent epidemiological studies 71, 72 have reproduced the risk function previously described by the Grand Rapids data. For example, Blomberg et al. 72 employed a case–control design, similar to the method employed in the Grand Rapids study, in examining drivers in 4919 motor vehicle collisions compared with 10,066 control drivers and demonstrated a statistically significant increase in crash risk at 0.040% with an exponential increase occurring from 0.100% onward 72.
Even greater crash risks have been observed for collisions resulting in fatal injury to the driver, which have been shown to become significant at a BAC of 0.020% 73. At BACs ranging from 0.050% to 0.079%, the relative risk of driver fatality is increased to c. 3.6–17 times that of alcohol-free drivers 73. The increased risk observed in studies of alcohol in fatal collisions likely reflect the decreasing ability to process and to respond to highly complex driving situations, even at low BACs.
Recently, Phillips and Brewer 74 examined injury severity for nearly 1.5 million crashes reported to the Fatality Analysis Reporting System (FARS) from 1994 to 2008. Using the ratio of serious to nonserious injuries as a measure, the authors reported that BACs of 0.010% and greater were associated with increasing collision severity when compared with alcohol-free drivers. The authors further reported that at a BAC of 0.010%, drivers were “significantly more likely to speed, to be improperly restrained and to drive the striking vehicle.” These conditions were associated with more severe motor collisions. Connor et al. 75 also reported increased crash risk at very low BACs. In a case–control study of drivers involved in crashes in which at least one occupant of the vehicle was killed or hospitalized, driving with BACs of 0.003% to 0.050% were associated with more than 10 times the risk of driving with no measurable alcohol. Drivers with a BAC > 0.050% had 40 times the serious injury crash risk 75.
Epidemiological studies can be limited by missing data (e.g., BAC, case history), lack of appropriate controls, and potentially confounding variables such as age, seatbelt use, and time of day. Failure to account for these variables can produce a distorted relative risk probability curve, the most notable example being the “Grand Rapids Dip,” which at face value implies a decrease in relative crash risk at BACs between 0.010% and 0.040%. Further analysis has established the Grand Rapids Dip to be an artifact of single variable analysis 76. Subsequent epidemiologic studies that control for covariates (e.g., gender, age, education, ethnicity, employment status, vehicle type) in their analyses have also negated the implied decrease in risk of collision at low BACs 72, 73, 75.
Analysis of crash statistics following the revision of per se alcohol limits can reveal whether the change has been effective in reducing the number of alcohol-related crashes. Where a reduction in crashes occurs, an inference can be made that higher BACs contribute to a larger fraction of alcohol-related motor vehicle collisions. For example, comparisons among alcohol-related crash statistics in jurisdictions with differing per se limits indicated that a reduction in the permissible BAC from 0.100% to 0.080% (or lower) was effective in reducing alcohol-related crashes 77-79.
Subject Variables
There is variability in the impairing effects of alcohol: Faculties are not impaired equally, and study participants do not demonstrate equal impairment 15, 31, 36. One study reported that while 82% of subjects exhibited “highly significant deterioration in driving performance” after alcohol consumption when compared with their control performance, 9% of the subjects showed no difference, and an additional 9% apparently improved 80. Sources of variability in alcohol-related impairment can be attributed to experimental design, for example, dosing regimen, task complexity, and practice effects. As noted by Flanagan et al. 80, the drivers in their study were not permitted to become too familiar with the course prior to testing and therefore would be expected to improve their performance on their second run, which represents a confound in differentiating between continued learning and the effects of alcohol on driving.
Other sources of variability that must be considered are subject variables such as age, gender, tolerance to alcohol, and driving proficiency. Practically, these variables may be offered in courts of law as factors that reduce the extent of impairment at a measured BAC. Thus, it is essential that the impact of subject variables on alcohol-related impairment be reviewed.
Age
Motor vehicle crash and fatality risks are greatest for young and elderly drivers with these populations exhibiting a higher risk of collision even in the absence of alcohol consumption 72, 81-83. Young drivers in particular have been shown to contribute to a disproportionate number of road and traffic collisions. In 2001, drivers under the age of 25 accounted for c. 25% of all U.S. traffic fatalities 84.
Increased risk of crash involvement for younger and older drivers is not due to a greater impairing effect of alcohol in these groups, but to the presence of additional risk factors. In the aged, the diminution of psychomotor faculties (e.g., vision, attention, perception/reaction time) has been offered as explanation of increased risk of collision 85, 86. Simulated driving studies illustrate the influence of age on driving performance with and without the administration of alcohol. Quillian et al. 87 observed that male drivers over age 60 performed more poorly when compared with drivers between 30 and 50 years. Even in an alcohol-free state, the older drivers drove more slowly, had more variable speed, failed to stop fully when required, had more crashes, and braked inappropriately. More relevant to this review, however, the authors observed that all subjects were impaired by the consumption of alcohol to a similar degree, regardless of age. Similar conclusions were reached by Tupler et al. 88 who demonstrated that subjects aged 59–65 years, performed more poorly than middle-aged and young subjects when alcohol-free. After the ingestion of alcohol, impairment of the magnitude of observed performance decrements was the same across all age groups.
In young drivers, higher crash risk has been attributed to, among other factors, adolescent characteristics such as increased risk-taking behavior 89, 90. Following the consumption of alcohol, young drivers are at even greater risk relative to the risk of alcohol-related collision for mature drivers 73. Blomberg et al. 72 observed a trend toward increased crash risk in drivers under 21, which was not statistically significant. Subsequent reanalysis of these data by Peck et al. 91 concluded the interaction between age and BAC was statistically significant. The interaction was described as a synergistic effect whereby the relation between BAC and crash risk was greater for drivers under 21 years of age when compared with older drivers. Estimated relative risks for drivers in this age group were elevated at all BACs, even those as low as 0.010%. For example, the relative risk of crash involvement for drivers under 21 years of age was 1.13, 2.75, and 16.0 at BACs of 0.010%, 0.050%, and 0.100%, respectively. In comparison, for drivers aged 21 and older, the relative risks were 0.94, 1.07, and 2.43 at the same BACs 91.
A review by Mayhew et al. 89 considered the mechanism for the combined effect of age and alcohol consumption on increased crash risk. The authors concluded there was little support for the hypothesis that higher crash risk in young drinking drivers is due to inexperience with drinking, driving, or a combination thereof. Rather, they proposed that young drivers are more likely to undertake certain risky behavior, and the consumption of alcohol further contributed to the likelihood of risk-taking behavior. This effect may result in an increased likelihood that young drivers will attempt maneuvers beyond their capabilities 89. It has been reported that for drivers aged 15–20 years, a greater percentage of their fatal motor vehicle collisions occur at night, without the use of seat belts, and with positive BACs, relative to other age groups 92. Williams et al. 93 recently reviewed the circumstances of fatal motor vehicle collisions between drivers aged 16 and 17 years and concluded that alcohol-positive drivers were “more likely than those in nonalcohol-related crashes to be male, unbelted, in single vehicles, and speeding, and their collisions were more likely to have occurred on Saturday or Sunday, at night, and when passengers were present” 93. Such analyses demonstrate the risk-taking behavior that is inherent to crash risk in young drivers; behavior that may be exacerbated by the presence of alcohol.
Unfortunately, experimental studies typically fail to provide information on alcohol and driving impairment in the youngest of drivers (i.e., ages 16–18) given the legal and ethical constraints of administering alcoholic beverages to individuals who have not yet reached minimum legal drinking age. This gap in the experimental research prevents full elucidation of the source of elevated crash risk among young drivers. Where research is conducted in areas bound by higher minimum legal drinking ages, (e.g., 21 years of age in the U.S.A.), an even larger gap may exist.
Preventative measures such as legal restrictions on alcohol consumption and deterrents such as zero-tolerance laws for alcohol in novice drivers have been shown to be effective in reducing crashes among young drivers 94. Much like reductions in alcohol-related crashes following the establishment of lower per se limits a reduction in alcohol-related crashes among young people after the introduction of these deterrents suggests that alcohol is a contributing factor to these crashes.
Gender
Systematic investigation of the effect of gender on alcohol-related psychomotor impairment is rare, and the results of the available studies are equivocal. Data from epidemiological studies infer a reduced crash risk for female drivers when compared with male drivers; however, laboratory experiments of the behavioral and cognitive effects of alcohol are inconclusive regarding gender-based differences in impairment. Miller et al., 95 compiled findings from seven previous studies that investigated gender differences and the effects of BACs between 0.067% and 0.088% on a variety of driving-related tasks. While the results of this study demonstrated that impairment was present in both men and women on all tasks at the BACs tested, it was evident that women displayed greater alcohol-related impairment when compared with men on dependent measures such as simulated driving, tracking ability, divided attention, and speed of information processing 95. Given that women exhibited significantly higher SDLP when compared with men at an equivalent BAC, Louwerens et al., 63 also postulated that women were more sensitive to the behavioral effects of alcohol.
In contrast, Friedman et al. 96 found that the male participants in their study displayed greater impairment than a group of women at approximately the same BAC. Low BACs (c. 0.050%) were achieved prior to laboratory testing of cognitive processes in a group of 10 men and compared with the performance in 11 women. All participants demonstrated higher error rates in the alcohol condition relative to placebo; however, the increase in error rate was statistically significant only for the males.
Other studies have revealed no significant contribution of gender to alcohol-related impairment of driving-related skills. A relatively large study of 84 men and 84 women provided no evidence of differential alcohol effects between gender groups. 18. Parks et al. 68 reported no effect of gender on actual driving performance or laboratory tests of divided attention, and Drew et al. 31 reported that the effects of alcohol in 40 volunteers tested on a driving simulator were not affected by differences in age, gender, previous driving experience, or drinking habits. Because the subjects in this study varied widely in their initial driving skill, a within-subject design was used to control individual differences in driving skill 31.
Data from some epidemiological studies, including the Grand Rapids Study, suggest reduced crash risks for female drivers when compared to male drivers at similar BACs 70, 73; however, the Grand Rapids Study revealed “females to be over-represented in accidents when no alcohol was present” resulting in an apparent reduced risk of crash involvement when alcohol was present 70. Peck et al. 91 provided relative-risk curves for males and females that were “almost identical” thus neither males nor females had any greater risk of being involved in a collision when compared to the other group. Connor et al. 75 also indicated that the gender of the driver had little effect on risk of serious injury crash.
Zador 97 observed a significant increase in fatality risk in young drivers and female drivers with positive BACs when compared with male drivers, 25 years of age and older; however, fatality risk increased in all groups when compared to control drivers. Accordingly, the authors concluded that “BAC is the most important factor followed by age and then by the BAC-by-sex interaction” 97.
Driving Proficiency
For many adults, the operation of a motor vehicle is performed on a daily basis and has been described as an over-learned, automatic behavior. Moreover, large interindividual differences in driving proficiency are inevitable in the population. It might be assumed that the ability of a highly experienced driver would be less disrupted by the impairing effects of alcohol. As suggested above, this has been postulated for young drivers, where driving inexperience has been often cited as the possible cause of increased crash risk following the consumption of alcohol.
Few studies have specifically included driving skill as a consideration in the experimental design. Where this variable has been considered, there is little support for the hypothesis that driving proficiency can overcome alcohol-related impairment. Harrison and Fillmore 98 documented wide variations in the baseline performance of subjects on a simulated driving task. Following the administration of alcohol sufficient to achieve a target BAC of 0.080%, participants demonstrating the greatest decrease in driving precision, as measured by within-lane deviation, were those with the highest deviation scores while alcohol-free 98. Although only two measures of driving skill were assessed (speed and within-lane deviation), the authors suggested that poor baseline driving skill can intensify alcohol-related impairment on a driving-related task. Nevertheless, it should be noted there was a significant increase in within-lane deviation for the alcohol group relative to their pretreatment score in general, illustrating that alcohol produced impairment in this measure regardless of driving proficiency.
A review of the literature conducted by Mayhew et al. 89 concluded that alcohol decreases driving performance in both skilled and less-skilled drivers to the same extent. Lovibond and Bird 58 also demonstrated that increasing BACs resulted in progressive impairment of driving performance even among a group of skilled race and rally car drivers. Although the overall driving performance of the experienced, competitive drivers was superior relative to the noncompetition drivers, both groups showed a marked deterioration in driving performance at a BAC of 0.050%.
It is important to note studies such as these have shown that while the relative proficiency of the drivers may be retained pre- and postalcohol consumption, impairment is still demonstrable in both groups following the administration of alcohol. It follows that an alcohol-impaired “skilled” driver may perform better than an alcohol-impaired “lesser-skilled” driver. The most important consideration is that alcohol impairment is still present relative to the individual's ability at a zero BAC.
When assessing the likelihood of impairment in an individual, it is apparent that subject variables such as age, gender, and driving proficiency are of limited value. The most reliable objective determinant of alcohol-related psychomotor impairment is BAC. It is also important when assessing the impairment of an individual to compare their performance to their own alcohol-free performance and not to the performance of other drivers.
Functional Tolerance
Under controlled experimental conditions, it has been demonstrated that repeated performance of certain tasks while under the influence of alcohol may make a subject less sensitive to the effects of alcohol, referred to as drug state-dependent learning 99 or functional tolerance.
Repeated consumption of alcohol may increase the functional tolerance of an individual resulting in a decreased response to the drug. Behavioral research concerning functional tolerance has demonstrated a relation between the expectation of drug effects, compensatory responses, and the consequence of the compensatory responses. For example, an experienced drinker can predict with assurance the effect that alcohol will produce (expectation), and if the consequence of the compensatory response is desirable, it is more likely to occur, for example, successfully operating a motor vehicle 99. While this theory may, in part, explain observations that individuals can display a high degree of tolerance to some effects of alcohol and no tolerance to other effects, the literature is equivocal. Criticisms include reliance upon anecdotal reports, variable definitions for heavy drinking, experimental confounds (e.g., different degree of novelty of test settings), and poor external validity (e.g., research performed using dogs or rats).
One study reported that heavy drinkers demonstrated less impairment of driving performance when compared with light drinkers at similar BACs; however, as the authors observed the driving task was well practiced, the attitude of the subjects was similar to that of a competitor in a game, and speed never exceeded 24 km/h 100. The authors concluded “that under actual driving conditions … the effect of alcohol on driving performance would be more pronounced than that observed in these experiments.”
Therefore, while impairment from alcohol may not be evident during simple repetitive tasks, where the outcome is predictable and desirable, it is highly unlikely that functional tolerance to alcohol exists in tasks that require a rapid response in unexpected situations. King and Byars 101, in their laboratory study of performance effects of low and moderate BACs in heavy episodic and light social drinkers, noted that impairment was generally comparable between the two groups. Where exceptions were noted, the authors speculated that heavy drinkers may have gained tolerance at a given BAC to tasks of a stereotyped and repetitive nature, but did not acquire tolerance to more complex tasks, such as those requiring the processing and execution of new strategies. The authors concluded that an “assumed lack of performance impairment … in regular heavy alcohol consumers is not valid” 101.
In a review of the literature concerning functional tolerance to alcohol and driving, DeLuca 102 concludes “the most that can be said is that at moderate BACs (50–70 mg/100 mL [0.050–0.070%]) more frequent drinkers appear to be less effected [sic] and at a lower risk of accident involvement. At higher BACs, the risk of an accident and the impairment of driving ability were both significantly increased in all drinking frequency groups.”
There have been reported differences in a drinker's ability to accurately self-assess their degree of impairment dependent upon their classification as a social or heavy drinker. Brumback et al. 103 demonstrated that habitual binge social drinkers show comparable alcohol-induced behavioral impairment, but less self-rated perception of impairment when compared with light social drinkers. The authors concluded that experienced habitual binge drinkers might have false beliefs in their functional tolerance to the psychomotor impairing effects of alcohol. Other researchers have studied similar populations with similar findings. In one study, binge drinkers and nonbinge drinkers were found to be impaired in multiple aspects of driving performance; however, the binge drinkers overestimated their driving ability following alcohol consumption 104. In this study, lack of awareness to the decreased ability to perform the driving task was attributed to a decrease in self-reported sedation among the binge drinkers; a potential cue to impaired psychomotor skills.
In a simulated driving study designed to determine the influence of binge-drinking frequency on driving behavior and subjective ratings of impairment, Bernosky-Smith et al. 105 observed a significant increase in the number of collisions and number of center-line crossings at a BAC of c. 0.080% regardless of binge-drinking history. High-frequency and low-frequency binge drinkers were observed to make a similar number of errors on a test of simulated driving following an alcohol binge that produced a BAC of c. 0.080%.
The existing scientific evidence supports the concept of functional tolerance, albeit only at low-to-moderate BACs, and in simple, predictable driving tasks. Additionally, because functional tolerance is learned through practice, it can be inferred its acquisition would require the repeated operation of a motor vehicle at the same BAC, and repeatedly facing driving tasks with predictable complexity. Functional tolerance cannot mitigate the impairing effects of alcohol when the driving task is complex and unpredictable.
Acute Tolerance
Acute tolerance (i.e., the Mellanby effect) is defined as reduced sensitivity to the effects of alcohol on the descending limb of the BAC curve relative to the ascending limb of the BAC curve. Acute recovery results in reduced subjective ratings of intoxication and/or reduced physical signs of intoxication (e.g., slurred speech, ataxia, motor incoordination) when the BAC is declining 106, 107. Consequently, it has been postulated that psychomotor impairment may also be mitigated if a subject is in the elimination phase of the BAC curve. Supporting evidence for this hypothesis would have implications in the application of laboratory and driving studies to forensic casework. Studies performed in test subjects on the descending limb of the BAC curve may underestimate the alcohol-related impairment potential of a given measure. Conversely, because the vast majority of drinking drivers have been demonstrated to be in the plateau or postabsorptive phase of the BAC curve at the time of driving 108, impairment studies confined to the ascending limb of the BAC curve may overestimate impairment at the time of arrest/incident. Some researchers have controlled this variable by ensuring that study participants are in the postabsorptive phase at the time of testing 4, 49, 50, 55, 62. Parks et al. 68 documented significant increases in the time required to respond to a peripheral stimulus during a laboratory-based divided attention task and similarly observed significant increases in SDLP in an on-road driving study comprising 60 subjects despite a decline in the mean BAC from 0.042% to 0.031% throughout the exercise, that is, in the presence of acute tolerance. In such experiments, an observed decrease in ability to perform a task will likely underestimate alcohol-related impairment when the BAC is rising.
The relevant scientific literature provides little support for acute tolerance ameliorating psychomotor impairment on the descending limb of the BAC curve. As is the case for the functional tolerance characteristic among chronic, heavy users of alcohol, motor-based skills appear to be most subject to acute tolerance effects. As such, acute tolerance to the impairing effects of alcohol on the complex faculties and cognitive processes inherent to the operation of a motor vehicle has not been described 28. Fogarty and Vogel-Sprott 109 observed impairment of a motor skill task was subject to acute tolerance, but that an information processing task was equally impaired at equivalent BACs on both the ascending and descending limbs of BAC curve. Schweizer et al. 110 found that errors during an information processing task remained in the postabsorptive phase despite a return in the speed of information processing to alcohol-free levels.
In an examination of how acute tolerance affects the ability to self-evaluate intoxication, Cromer et al. 111 demonstrated that although individuals report lower levels of subjective intoxication on the descending limb of the BAC curve compared with equivalent BACs on the ascending portion of the curve, executive functions (i.e., error monitoring and spatial working memory) were equally impaired. Thus, while perceived level of intoxication may be subject to the development of acute tolerance, there is continued impairment in the ability to perform complex decision-making and problem-solving tasks. The authors concluded that “subjective perception of intoxication is a poor indicator of sobriety and the ability to operate a motor vehicle” 111. Weafer and Fillmore 107 corroborated this work in a driving simulator study that demonstrated diminished driving precision in response to alcohol (demonstrated by increased within-lane deviation and steering rate) on both the ascending and descending limbs of the BAC curve. Acute tolerance to alcohol on driving performance was not observed; however, motor coordination and perceived intoxication ratings showed marked acute tolerance. Consequently, willingness to drive was greater when subjects were on the descending limb of the BAC curve compared with the ascending limb.
Marczinski and Fillmore 112 similarly reported that although acute tolerance to the subjective effects of alcohol occurred among binge drinkers, this was not accompanied by acute tolerance to the impairing effects of alcohol. Indeed, driving errors such as within-lane deviation, number of center-line crossings and number of excursions to the driving shoulder actually increased on the descending limb of the BAC curve relative to the ascending limb. The authors concluded that while the “intensity of subjective intoxication diminished under dose … the intensity of the impairing effect of alcohol on driving performance increased.”
The most relevant aspect of acute and functional tolerance in drinking drivers may be the accompanying reduction in self-rated perception of intoxication. Psychomotor effects such as motor incoordination, clumsiness, and sedation are highly salient cues that may be influential in the decision whether to drive following the consumption of alcohol. Thus, while acute and functional tolerance has negligible impact on mitigating performance decrements, there may be effects on self-perception of intoxication and willingness to operate a motor vehicle.
Self-rated perception of intoxication may also be affected by the concurrent administration of stimulant drugs. A number of researchers have examined the effects of caffeinated “energy” drinks mixed with alcohol on subjective reports of intoxication and objective measures of impairment (e.g., information processing, motor coordination, visual reaction time) 113-115. The dominant finding of this research is that, subjectively, coadministration of an energy drink reduces mental fatigue and diminishes feelings of intoxication, but there is no amelioration of alcohol-caused deficits in the objective measures. Others have performed similar experiments where the combined effects of ethanol and the stimulant drug methylenedioxymethamphetamine (MDMA, “Ecstasy”) were examined 33, 49. Again, the coadministration of the stimulant drug was found to counteract the subjective ratings of impairment due to alcohol, but was insufficient to overcome the psychomotor impairment objectively determined by measures such as a car-following task. Participants in the study by Veldstra et al. 33, for example, rated their driving performance as only “slightly worse” than normal, when in fact, the quality of driving had “seriously deteriorated” with significant increases in SDLP at each target BAC (0.030%, 0.050%, and 0.080%) compared with placebo. Similarly, at a BAC of 0.050%, 29 subjects who took part in an on-road driving test performed significantly worse than they did while alcohol-free in spite of reporting no change in alertness before driving and no change in driving quality or mental effort after driving 116.
These reports underscore the unreliability of subjective ratings of intoxication as a measure of impairment while supporting BAC as the most appropriate predictor of driving impairment. Subject variables affecting self-perceived intoxication, including functional and acute tolerance, cannot be offered as viable explanations that a person was unimpaired by alcohol at a given BAC. Impairment has been repeatedly demonstrated to exist independent of subjective feelings of intoxication. Moreover, in one closed-course driving study, 76% of the subjects who had shown deterioration after consuming alcohol actually thought that their driving had improved 80.
Discussion
The findings from laboratory, on-road, closed-course, and epidemiologic studies should be used to crystallize the most important concepts regarding the effects of alcohol on driving performance. Laboratory-based studies have played a critical role in isolating mechanisms for impairment; epidemiological studies have demonstrated the impact of these deficits on crash risk; and on-road and closed-course studies have provided empirical evidence as to the effects of alcohol on the integrated task of driving. When the totality of the scientific evidence is assessed the weight of most subject variables is diminished and the relevant variables are reduced to BAC and complexity of the driving task. The effects of alcohol on driving impairment can thus be reduced to the following two points:
- as BAC increases the magnitude of impairment also increases,
- whether impairment is observable depends upon the complexity of the task.
With respect to task complexity, the key issue is not whether the person is impaired from a scientific perspective, but simply whether the impairment is observable. Moreover, given the inherently complex and unpredictable nature of the operation of a motor vehicle, it follows that the single most important predictor of impairment is BAC.
Previous reviews have reached analogous conclusions 11, 53, 56. For example, Perrine 53 identified the equivocal nature of the effects of alcohol “on the highway” as being related to (i) the pharmacologic effects of alcohol and (ii) the variable nature of the driving task (i.e., the wide range of driving task demands).
In addition to elucidating the key determinants in predicting alcohol-related impairment, an objective of this work was to review the effects of low-to-moderate BACs (≤0.100%) on driving ability. The relevant scientific literature provides confidence that BACs at or in excess of 0.050% produce alcohol-related impairment of driving ability.
Scientifically, there is no evidence for a threshold effect for alcohol. Ogden and Moskowitz 8 reported that “some impairment of performance occurs at the lowest levels that can be measured” with no level at which a sudden transition from unimpaired to impaired expected. Whatever the level of BAC examined, at least some skills can be demonstrated to be significantly impaired 8. A review of 177 studies similarly concluded there is no lower limit of BAC at which impairment does not exist 7. Howat et al. 52 focussed on a statutory limit of 0.050% and concluded that there is “sufficient experimental evidence to demonstrate that BACs of 0.050% and higher can produce impairment of the major components of driver performance for most people.” Moskowitz and Fiorentino 117 enumerated the vast scientific literature in this area and determined that 27% of studies reported impairment by 0.039%, 47% by 0.049%; 92% of studies reporting impairment by 0.079%. Similarly, Mercer et al. 118 concluded that “virtually all drivers are substantially impaired” at a BAC of 0.080% in tasks such as braking, steering, lane changing, and judgment, and that these impairments began at BACs as low as 0.020%.
Both the Canadian and American Medical Associations (CMA and AMA, respectively) have responded to the scientific evidence and endorse a legal BAC limit of 0.050% 119, 120. Other organizations have issued statements that align with the recommendations of the CMA and the AMA. For example, it has been the long-standing position of the U.S. National Safety Council Subcommittee on Alcohol and Other Drugs 121 that “every person, regardless of that person's previous experience with alcoholic beverage consumption is impaired in driving performance if that person's alcohol concentration is 0.08 g of alcohol per 210 L of breath (0.080%) or more.” The American College of Emergency Physicians 122, 123 has declared that all drivers are impaired at a BAC of 0.08% and that a BAC of 0.05% is presumptive evidence of impaired driving. More recently, the Centre for Addiction and Mental Health, in Toronto, Canada 124 stated, “it has become clear that the effects of alcohol on performance can begin with the first drink and are measurable at BACs of 20 mg% (0.020%) and lower. Thus there is no question that at 50 mg% (0.050%) driving skills are significantly impaired.” These opinions are fully supported by critical review of the scientific literature.
It is uncontroversial that alcohol-related impairment is concentration dependent; the magnitude of impairment increases with increasing BAC. As demonstrated by the present review, there is convergent evidence, therefore a high degree of scientific confidence, in support of the conclusion that a BAC of 0.050% impairs faculties required in the operation of a motor vehicle. This conclusion notwithstanding, the present review also demonstrates that impairment may occur at BACs <0.050%. Whether impairment is apparent depends upon the complexity of the driving task; however, because driving is an inherently complex behavior it can be asserted scientifically that the faculties required in the operation of a motor vehicle will be impaired at a BAC of 0.050% or greater.
Acknowledgments
The assistance of M. Brenton and the Centre of Forensic Sciences Library staff in retrieving numerous references is gratefully recognized. The efforts of J.G. Wigmore in founding the Centre of Forensic Sciences alcohol literature database are also acknowledged.
