Neither the characteristic odor of ethanol nor various markers of alcohol exposure (eg, ethyl glucuronide, ethyl sulfate, carbohydrate deficient transferrin, and so on) are useful in estimating BAC or degree of intoxication. The ability of an individual to estimate his or her own BAC accurately is poor.30,40 A lack of accuracy in estimating BAC is also observed in trained medical providers and police officers.3,12 A large confounding factor in these observations is tolerance. Greater tolerance, as occurs in heavy drinkers and alcoholics, imparts greater difficulty in detecting the clinical effects associated with alcohol intoxication.
Tolerance involves a central adaptation to the intoxicating effects of alcohol. As such, persons with significant tolerance to the effects of ethanol may not manifest signs of intoxication despite having a high BAC. Multiple case reports and case series in the medical literature describe alcoholic individuals with very high BACs but muted or absent clinical effects.14,22,26,32,45 As a result of tolerance in the alcoholic population, it may not be possible to detect clinical intoxication even though an individual has a high BAC. This circumstance does not, however, suggest that even in the clinically sober alcoholic that driving abilities are unaffected and the individual is necessarily capable of safely operating a motor vehicle.3 Furthermore, tolerance has no bearing on the potential prosecution of a DUI case on a per se BAC basis.
The circumstance is slightly more straight forward in the social drinker. The intensity of the effects of alcohol on the central nervous system (CNS) is generally proportional to the concentration of alcohol in the blood. Dubowski has tabulated the stages and effects of acute alcohol intoxication, and these tables are used in educational programs and some legal proceedings. In general, the greater the BAC, the greater the clinical effect observed.
However, while useful as a pedagogic tool for explaining the continuum of alcohol intoxication, the table must be used with care as the effects are defined only over a population, thereby making assignment of a specific effect or degree of effect in an individual impossible. The inherent individual variability in the table is apparent in the fact that the BAC ranges for the various stages of alcoholic influence overlap. Furthermore, the population and methods used to compile the table are not described, and the table itself has not been subjected to peer review.
The CNS effects of alcohol intoxication are typically more pronounced on the ascending portion of the blood alcohol kinetic curve than on the descending side due to acute tolerance.19 In other words, the clinical effect of intoxication is greater during the absorptive arm of the kinetic curve than on the elimination arm, even though the same blood alcohol concentration is measured in both kinetic phases. This principle is known as acute tolerance, or the Mellanby effect.
In a dram shop case, the ultimate question often becomes this—what is the typical BAC at which it is more likely than not that the average nontolerant individual will exhibit signs of intoxication (eg, the odor of alcohol on the breath, clumsiness, difficulty walking or maintaining balance, slurred speech, inappropriate behavior) that are apparent to a bystander or third party? A 1986 report by the Council on Scientific Affairs of the AMA examined a review of seven studies spanning 50 years that included more than 6500 participants for identification of BACs at which individuals appeared “drunk” (ie, clinically intoxicated).1 BACs were stratified by increments of 50 mg/dL, beginning at 0.0 to 50 mg/dL and extending to 401 mg/dL. In the lowest BAC group (0.0–50 mg/dL), observer perceptions of subject drunkenness ranged from 0% to 10%, with an average of 4%. The percentage of individuals determined by observers to be drunk increased steadily from a mean of 32% (range, 14%–68%) at a BAC of 51 to 100 mg/dL to a mean of 62% (range, 47%–93%) at a BAC of 101 to 150 mg/dL. In this latter range, four of the seven studies indicated a perception of drunkenness by less than or equal to 50% of observers. However, in the next BAC increment of 151 to 200 mg/dL, observers judged a mean of 89% (range, 83%–97%) of the drinkers to be drunk. In each of the subsequent higher increments, the mean percentage of persons classified as drunk ranged from 95% to 100% with no individual study value less than 90%. Therefore, it can be reasonably concluded that, in a nontolerant individual, a BAC of 151 to 200 mg/dL will more likely than not result in observable signs of drunkenness.
The validity of a police officer detecting intoxication in drivers involved in motor vehicle crashes also increases with increasing BAC. A 1996 report examined a total of 1336 subjects over age 15 who were admitted or died at a level 1 trauma center in Seattle during a 5 year period from 1986 to 1993 and in whom both a recorded BAC and a police assessment of sobriety were conducted.21 The blood alcohol measurement was conducted in the hospital, and it is not expressly stated if the analytical matrix used was whole blood, serum, or plasma. Four categories of sobriety assessment were used by police: (a) had not been drinking, (b) had been drinking–not impaired, (c) had been drinking–sobriety unknown, and (d) had been drinking–impaired. Officers used a battery of specific criteria to judge whether a driver was intoxicated, including odor on the breath, slurred speech, chemosis, poor coordination of motor function, and the ability to simultaneously perform multiple tasks. The greatest number of drivers were in the two extreme categories: had not been drinking (n = 746) and had been drinking–impaired (n = 568). A direct correlation between measured BAC and police officer assessment of sobriety was observed. The mean BAC associated with the “had been drinking–impaired” group was 190 mg/dL with a 95% confidence interval (CI) of 180 to 200 mg/dL. Those in the “had been drinking–sobriety unknown” group had a mean BAC of 130 mg/dL (95% CI, 110–150). Among all drivers, police field assessment of sobriety had a positive predictive value of 85% with sensitivity and specificity of 91% and 90%, respectively, demonstrating recognition of drunk driving by police with a high degree of accuracy, especially in the group with the highest BAC.
Another study designed to determine the ability of police officers, who have more training than the average lay person, to make assessments about alcohol intoxication in drinking target subjects without the aid of special testing normally available to them (eg, standardized field sobriety tests or the ability to smell the odor of alcohol on the subject’s breath).7 This study involved 39 police officers who viewed a series of videotaped interviews with six volunteer moderate drinkers having targeted BACs in three concentration ranges: low (80–90 mg/dL), medium (110–130 mg/dL), and high (150–160 mg/dL). Based on their observations of the taped interviews, the officers answered three questions:
Has the person been drinking?
Was it OK to serve that person one additional drink?
Was the person able to drive a car?
Each of the three questions could be answered “yes,” “no,” or “not sure.” A fourth question assessed the officers’ confidence in their answers as “not sure,” “little uncertain,” or “positive.” Question 2 was to be answered from the perspective of a social host or bartender, not a police officer. None of the police officers had any formal training in the management or service of alcohol to intoxicated people, such as that offered by commercial seller/server training programs such as Techniques in Alcohol Management or Training for Intervention Procedures. With respect to their answers, the police officers were fairly certain that the subjects had been drinking only in the target groups with the highest BAC (150–160 mg/dL). Officers also answered in the affirmative to question 2 (ie, that it was OK to serve the target subject another drink) the majority of the time in the low- and medium-BAC target groups but not in the high BAC target group. The percentage of officers answering affirmatively to this question was 55%, 75%, and 41% for the low, medium, and high target groups, respectively.
Multiple studies have examined the likelihood of on-premise (bars and restaurants) and off-premise (liquor stores, grocery stores, and convenience stores) as well as outdoor events like festivals to sell alcohol to obviously intoxicated persons (typically, paid professional actors pretending to be drunk).17,31,47, 48, and 49 The results are fairly uniform in that the majority of the time the alcohol was sold or served to the individual. Additionally, it was noted that male servers/clerks who appeared younger than age 31 were more likely to make such sales and the sales were more likely to occur in off-premise establishments.17
In a typical study, the authors employed 19 actors who were specifically hired based on their ability to feign intoxication. All actors were male and ranged in age from 31 to 59 years, with a mean of 42 years.48 The actors used a standardized script to attempt to purchase either a single vodka drink after asking what beers were available on tap (on-premise) or a six-pack of beer (off-premise). Prior to entering the establishment, the actor received specific instructions to demonstrate multiple signs of intoxication, including disheveled hair and clothing, smelling of alcohol, lack of coordination, stumbling, fumbling with money, slurring words, repeating questions, appearing forgetful, and laughing inappropriately. If the actor was asked if he was driving, he responded “no” and if he was asked if he had been drinking, he responded, “I’ve had a few beers.” Attempts at alcohol purchase were made at 223 on-premise and 132 off-premise establishments.
Of the 355 attempts, actors were able to successfully make the alcohol purchase in 280 instances (79%). On-premise establishments served the actors 76% of the time, whereas off-premise establishments allowed the sale 83% of the time. Actors and a nondrinking observer also watched for clues that the server/clerk indicated an awareness or suspicion that the buyer was obviously intoxicated. Verbal indications, including asking the buyer to leave, suggesting a nonalcoholic drink, offering to call a cab, and so on, as well as nonverbal indications, such as staring or rolling of eyes, were recorded. A similar series of observations were made of security staff, other staff/bartenders/cashiers, and other customers. In 51% of the attempted purchases, there was an indication from the server of the recognition of the intoxication of the buyer. Even within the group that recognized signs of apparent intoxication, the alcohol was sold 61% of the time. When the server made no indication of the apparent intoxication, alcohol sales were completed 97% of the time. In 45% of purchase attempts, another staff member or customer made an indication that they believed the buyer was intoxicated. Still, alcohol purchases were completed in 66% of these cases. When no other staff member or customer indicating an awareness of the buyer’s behavior, the actor was served 89% of the time. The authors concluded that sales of alcohol to obviously intoxicated individuals in some US communities is very high and recommend additional study to identify effective training and tools to mitigate such illegal sales.