The unformatted htm text below is from		Copyright © 2004 by Leonard Evans

10 Alcohol

This  html version contains only the text (no figures, tables, equations, or summary and conclusions).  To check printed book appearance see pdf version of Chapter 1 or pdf version of Chapter 16.

Introduction
     Alcohol has figured in human affairs since the beginning of recorded history. Beer making is described in Egyptian hieroglyphics. The use of wine is mentioned early in the Old Testament when Isaac's son "brought him wine, and he drank" (Genesis 27:25). The ancient Greeks had a god of wine, Dionysus, and the Romans had Bacchus. While alcoholic beverages are mainly associated with mood and behavior changes, health benefits have also been recognized. From antiquity alcoholic drinks were known to be free from disease risks posed by other beverages. The Qur'an (2:219) acknowledges positive effects, stating "They question thee about strong drink and games of chance. Say: In both is great harm and utility for men; but the harm of them is greater than their usefulness." Islamic tradition came to forbid alcohol consumption to such an extent that the devout are admonished to avoid medicines and toothpastes containing even a trace of it. Thomas Jefferson's doctor advised him to consume a glass and a half of wine daily for health reasons. He so favored the advice that he doubled the dose and lived to age 83. More recently, sixty studies collectively provide solid evidence that drinking one to two glasses of wine a day reduces the risk of heart attack.
     The problems resulting from consuming large quantities of alcohol far exceed the benefits of moderate consumption. The Old Testament includes "For the drunkard and the glutton shall come to poverty" (Proverbs 23:21). Alcohol plays a major role in a host of social ills in essentially all countries, the only exceptions being a few that have managed to effectively prohibit its availability through strong laws supported by tradition and religion. Foremost among the ills produced by alcohol is its role in traffic crashes.
Alcohol and traffic
     The recognition that drunk driving posed a major danger to the public is as old as motorized traffic. A law passed in 1872 in England specified prison as a possible punishment for being drunk while in charge of a vehicle powered by a steam engine. (p 7) The acronyms DUI (driving under the influence) and DWI (driving while intoxicated) are widely understood by the US public.
     The role of alcohol in traffic safety has produced more activity, literature, passion, and controversy than any other safety topic. In many countries there are advocacy organizations, professional societies, and journals devoted exclusively to the effects of alcohol on traffic safety. A review of literature on just one aspect, namely, how alcohol affects skills related to driving, identified 1,733 titles. A NHTSA review cites 738 recent papers on alcohol as particularly relevant. Scores of additional papers appear each year.
Measurement of alcohol
      Ethyl alcohol, or ethanol, is the active ingredient in beer, wine, and liquor (liquor means distilled spirits of any strength). With chemical composition C2H5OH, ethanol is the second simplest member of a family of compounds chemically classified as alcohols, the simplest being methanol, CH3OH. Ethanol and methanol burn not all that differently from gasoline, and indeed both are used as automotive fuels. In the present context we use alcohol to denote only ethyl alcohol, which is a colorless liquid that generates a homogenous liquid when mixed with water in any proportion. Its specific gravity is 0.79, meaning that the mass (or weight) of a given volume of alcohol is 21% less than the mass of the same volume of water. Thus a solution made by combining equal volumes of alcohol and water will contain 44% alcohol and 56% water by mass, but 50% of each by volume. When indicating the proportion of alcohol in blood or alcoholic beverages it is therefore crucial to specify whether the proportion is of volume or mass.
Measurement of amount consumed
     Although alcoholic beverages come in a wide variety of forms, colors, flavors and bouquets, their chief constituents are water and alcohol. Other ingredients appear to have only minor pharmacological significance, although mold may trigger allergic reactions in some individuals, while other ingredients may add to the severity of hangovers. (p 512) Most US beers contain about 5% alcohol by volume. Light beers tend to be just under 5%, but some can be as low as 3%. Alcohol-free beers still contain about 0.4% alcohol. Alcohol content is not indicated on US beer containers and advertisements in order to preclude marketing based on escalating alcohol content. Alcohol content is printed on beer containers of other countries. Many beers, referred to in the trade as super strong, have over 12% alcohol. One has 17.5% alcohol.
     The percent alcohol by volume in wine is normally printed on the label. In the US the designation table wine without a specified value implies between 8% and 14%. Specified values tend to range from 11.5% to 13.8%, with red wines generally having somewhat higher alcoholic content than white wines. Fortified wines, like sherry and port, tend to be close to 20% alcohol by volume.
     In Europe the alcohol content of all alcoholic beverages is indicated in terms of the percent alcohol by volume. In the US, alcohol content for liquor is given in terms of proof, which, in a strange sort of logic, is simply twice the percent alcohol by volume. Proof originated in ignition tests to confirm alcohol content. Britain formally abandoned the proof measure in 1980 in favor of the simple percent alcohol by volume measure, sometimes referred to in Europe as the Gay-Lussac system, after the French chemist who introduced it. However, proof measures still persist in Britain, with UK proof = 7/8 US proof = 1.75 percent alcohol by volume. Scotch whisky has a minimum 40% alcohol by volume (US 80 proof, UK 70 proof). Gins, whiskeys, and vodkas are typically about 40% alcohol by volume, while liqueurs are less, in some cases much less.
    In the units system used by most of the world, the volume of alcohol and the volume of the drink are understood by the public in the same units (mL), so the volume of alcohol in a drink is readily understood as the volume of the drink times the percent alcohol by volume. In the US and UK small volumes of fluid are usually measured in fluidounces. However, the US and UK fluidounce are defined in entirely different ways. Quantitatively, the difference is inconsequential, with the US fluidounce = 29.6 mL and the UK fluidounce = 28.4 mL. Approximately equal amounts of alcohol, namely 15 mL, are contained in 12 fluidounces of 4.3% beer, 4 fluidounces of 12.5% wine, 2.5 fluidounces of 20% fortified wine, and 1.27 fluidounces of 40% liquor. These are typical sizes for drinks, except that the most common liquor serving in the US is somewhat larger at 1.5 fluidounces. The same amount of alcohol drunk within similar time periods produces fairly similar pharmacological effects, regardless of which alcoholic beverage contained it.
Content in human body -- Blood Alcohol Concentration (BAC)
     The amount of alcohol in the body can be determined by analysis of samples of blood or breath. The alcohol content of blood is commonly measured in terms of the mass of alcohol in a given volume of blood. In the US, laws pertaining to alcohol are commonly based on grams of alcohol per milliliter of blood. If a milliliter of blood had a mass of exactly one gram, this measure would be identical to mass of alcohol per mass of blood. In fact, a milliliter of blood has a mass of 1.05 g. If the small departure from 1.00 g is ignored as inconsequential, then the measure grams of alcohol per milliliter of blood is the same as grams of alcohol per gram of blood. This convenient dimensionless ratio (multiplied by 100) defines the Blood Alcohol Concentration, or BAC, as the percent, by mass, of alcohol in the blood. One part alcohol per 1,000 parts blood (by mass, or weight) gives BAC = 0.1%.
Various other measures appear in the literature. One that has the advantage of providing convenient numbers is milligrams of alcohol per deciliter of blood. A value 80 mg/dL is the same as BAC = 0.08% if the 1.05 factor is ignored. Other units can likewise be converted to BAC by moving the decimal point, and in some cases also ignoring the 1.05 factor.
     The term Blood Alcohol Level (BAL) is sometimes used instead of Blood Alcohol Concentration.
Closely related to the amount of alcohol in blood is the amount of alcohol in breath. This has the advantage that it can be measured by less intrusive means. The earliest practical breath-alcohol measuring instrument, the Drunkometer, was developed in 1938 by Rolla N. Harger. The best known breath-alcohol instrument, the Breathalyzer, was invented in 1954 by Robert Borkenstein. Breath alcohol is closely related to BAC, with BAC = 0.1% being approximately equivalent to 1 gram of alcohol per 2,100 liters of breath.
Continuous variables and ranges
     Science is based on measurement, and BAC provides the foundation of the scientific study of the influence of alcohol. In science there is rarely a need to define ranges of values of a variable even though names such as those listed below are convenient in scientific writing and indispensable in everyday life.
scientific variable non-quantitative useful terms in common use
temperature cold warm hot
age young middle-aged old
BAC sober impaired drunk
     In Chapter 7 technical results on older drivers were presented in terms of the continuous variable age. The terms old, older, etc. were used for descriptive purposes. We found that at ages above 70, various crash risks increased with increasing age. In principle we expect a driver aged 80 years and 25 days to be at greater risk than one aged 80 years and 24 days, even though there is no possibility of measuring such differences empirically.
      The needs of criminal law are quite different from those of science. Laws must be written in terms of thresholds, not continuous variables. US law prohibits people from purchasing alcoholic beverages the day before their 21st birthday. No such restrictions apply one day later. The law makes sense, even though no measurable change in maturity or responsibility can be detected from one day to the next.
     While the above comments might seem too obvious to mention, they seem to be all too often ignored, or even denied, when the subject is alcohol. The law must define offenses such as impaired driving or drunk driving in terms of thresholds. Scientific research can determine how risk depends on BAC, but research cannot reasonably define impairment any more than it can define old or hot.
     The term sober is commonly used to indicate no large observable effects from alcohol. It does not indicate zero alcoholic consumption. In what follows we will tend to use BAC = 0 when common usage might suggest sober. This does not necessarily imply no alcohol. The strict interpretation of BAC = 0 is that it means BAC < 0.005% if measurement precision is two decimal places.
Absorption and elimination of alcohol
      After consumption, alcohol is readily and rapidly absorbed from the stomach, especially from the small intestine, and does not have to be digested before entering the blood. It distributes throughout tissues and fluids of the body in a manner similar to that of water. The bloodstream carries it to the brain, which is where it produces its well-known effects. Alcohol is eliminated from the body mostly through metabolism (enzymatic breakdown). A very small percent of alcohol is excreted unchanged in breath, urine, and sweat. The amount present at a given time in body fluids, organs, and other tissues is determined by rates of absorption, distribution, and elimination. The rate of absorption depends on the quantity drunk, its concentration, and the other contents of the gastrointestinal tract. Food in the tract delays absorption, so that the conventional wisdom that drinking on an empty stomach increases the rate of onset of intoxication is
well founded.
    The greater the concentration of alcohol in a drink, the more rapidly it is absorbed. Thus, alcohol in straight (undiluted) liquor enters the blood stream more rapidly than does the same amount of alcohol contained in a larger volume of wine, or an even larger volume of beer. Such differences have led to an erroneous impression that beer is substantially less likely to cause impairment than liquor, an impression that has in many cases supported less stringent controls on the sale and advertisement of beer (and wine) than on liquor. However, the differences in peak levels of intoxication associated with different levels of concentration when the same amount of alcohol is consumed are minor compared to the main influence, which depends on the amount of alcohol consumed, regardless of concentration.
     Fig. 10-1 shows representative patterns of absorption and elimination of alcohol for a man of average mass, generated from published data. Relationships of this type were first observed in pioneering studies conducted by Erik Widmark in Sweden in the 1920s and 1930s. The approximately constant rate at which alcohol disappears from the blood, referred to as Widmark's Beta, is typically about 0.015% BAC units per hour, say from 0.050% to 0.020% in two hours.  It is found that the peak BAC is approximately proportional to the amount consumed, and inversely proportional to the mass of the person consuming it. For the same consumption and body mass, females reach peak BAC levels about 20% higher than those for men. The reason for this is due mainly to a smaller gastric metabolism in females that leads to more of the alcohol reaching the bloodstream.
     The pattern in Fig. 10-1 is for consuming all the drinks in a short time. If an appreciable period of time elapses between the consumption of drinks, metabolism of the first will be in progress before the absorption of the second is complete. By the time the alcohol from the second drink is producing its peak level in the blood, the effect of the first drink will have diminished, causing the peak from both to be (unlike Fig 10-1) less than twice the peak level due to one drink.
      Peak alcohol levels reported in the literature can be approximated by the following formulae, which do not take into account many details, nor do they reflect variability between individuals:

These formulae provided the illustrative values in Table 10-1, which are for all drinks consumed in a short period (T = 0). The numbers in bold type are violations of the 0.08% BAC legal limit in most US States. The table shows that a 165 pound (75 kg) man consuming 4 drinks in a short time will achieve a peak BAC of 0.091%, but if consumption takes 2 hours, Eqn 10-1 calculates that the peak BAC drops to 0.061%. The equations are available on an Internet link which calculates peak BAC for any body weight and any number of drinks consumed over any time period. Many tables with information presented in a format like Table 10-1 are widely available. Types and details of presentation vary, but the BAC values given are similar, but usually not identical. Devices to measure BAC by chemical means are also available, though there may be disadvantages in intoxicated drivers measuring their own BAC levels.

How alcohol affects humans
     Three earlier chapters (Chapters 6 on survivability, Chapter 8 on performance, and Chapter 9 on behavior) were devoted to human characteristics central to traffic safety. Alcohol is important in all three, through its effect on:
1. Survivability.
2. Performance.
3. Behavior.
The effect of alcohol on survivability from the same physical impact was unknown until a few decades ago. As shown in Chapter 6, a driver or passenger with BAC = 0.08% is 73% more likely to die from the same crash experience than one with BAC = 0.
      Performance and behavior changes related to alcohol were likely observed shortly after alcohol first appeared. The Bible and other ancient writings mention people feeling happy, staggering, and falling asleep under the influence of alcohol. Shakespeare notes that, in respect to lechery, "it provokes the desire, but it takes away the performance." (Macbeth: Act 2, Scene 3). A large number of effects have been associated with increasing BAC, some examples of which are shown in Table 10-2. Responses vary greatly between individuals, and for each individual, changes are gradual. Many changes following alcohol consumption have performance and behavior aspects. Increased reaction time is purely performance, while increased aggressiveness is behavior. The earlier finding of that far more research was conducted on performance than on behavior applies particularly to the effects of alcohol.

Table 10-2. Performance and behavior characteristics that have been associated with increasing BAC levels. Bold type indicates behavior (but not performance) changes likely to have greatest impact on traffic safety.

Alcohol and performance
      From the previously mentioned literature review3 of 1,733 studies on the effects of alcohol on driving skills, 285 articles satisfying strict criteria were selected to examine the influence of measured levels of BAC on a host of performance measures. These included cognitive tasks, critical flicker fusion, divided attention, driving on simulators, drowsiness, perception, psychomotor skills, reaction time, tracking, vigilance, and various visual functions. Strong evidence showed performance declines for some driving-related skills at any measured BAC > 0. Performance declines were reported by the majority of studies for BAC ³ 0.05%, and for 95% of studies for BAC ³ 0.08%.
     It is unfortunate that the legal term impairment is used outside legal contexts. All the reviewed experiments explore a dose (the amount of alcohol) versus response (the performance measure) relationship. Such relationships are nearly always one of two types. Either there is a threshold dose below which there is no response, or else the response is a continuous monotonically increasing function of the dose. None of the studies reviewed suggested a threshold. Thus a plausible assumption is that performance always declines as dose increases. Whether the performance reduction can be measured depends only on the precision of measurement.
     Impairment has been defined as a statistically significant decrease in performance under alcohol treatment from the performance level exhibited under placebo treatment. This definition unambiguously identifies impairment as a property of the measuring technique, not of the phenomenon being measured. It is unhelpful to say that a certain level of alcohol does not lead to impairment today, but will when a study with sufficient precision is performed. Believing that human performance degrades with increasing alcohol consumption does not directly lead to policy conclusions any more than believing that human performance degrades with increasing fatigue, illness, sleep deprivation, aging, etc. has policy implications.
     As discussed on p. 190, NHTSA is providing $5.1 million to the National Advanced Driving Simulator. A main goal is to "determine the degree of impairment associated with a particular blood alcohol concentration." What could possibly be found that would add useful safety-relevant knowledge to what was already documented in the 1,733 studies available in 2000-3 Or for that matter, to the 557 studies reviewed in 1987. There are many important questions relating to alcohol and safety that cry out for serious investigation, yet so much of the scarce resources available are squandered on activity of so little value.
Alcohol and behavior
     Alcohol has major effects on behavior, including reducing inhibitions and caution, and increasing aggressiveness and risk taking. While it seems plausible that such behavior changes would have a large impact on traffic safety, research is lacking. There is little literature on even so central a question as the influence of alcohol on speed choice. Drivers with illegal BAC ³ 0.05% were observed in Adelaide, Australia, to drive about 3 km/h faster than BAC = 0 drivers. Small sample sizes precluded more definitive findings, such as a relationship between BAC and chosen speed. The main barrier to larger sample sizes is limited resources.
     Many obvious studies seem not to have been performed. I have not encountered any paper reporting speeds and BAC levels of drivers stopped for speeding who were additionally administered BAC tests. I believe useful information might be extracted from large data files, even though there are problems to be overcome. Chapters 8 and 9 show that driver behavior has a larger influence on crash risk than driver performance. It therefore seems likely that it is alcohol's influence on behavior that is even more important for traffic safety than its already very well documented effect on performance. By about 1970 the effect of alcohol on performance was sufficiently well known for traffic safety policy purposes, but its much more important effect on behavior remains largely unquantified.
Crash risk and alcohol
     It is not the direct effect of alcohol on driver performance or driver behavior that makes it so important in traffic safety, but the changes in crash risk that flow from these changes in performance and behavior. From the earliest days of the automobile it was well recognized that alcohol consumption sharply increased crash risk. An editorial in 1904 Quarterly Journal of Inebriety mentions 25 fatal crashes in which 19 of the drivers had consumed alcohol within an hour of their crashes. The first case-control study to quantify effects was conducted from 1935 to 1938 in Illinois. In case-control studies, the BAC of a case driver who crashes is compared to the BAC of a matched control driver traveling on a similar road at a similar time. The control driver must be stopped by police, without cause, and then provide an alcohol test result.
The "Grand Rapids" study
      While a number of case-control studies have been performed, the most important is that conducted by breathalyzer inventor Robert Borkenstein and his colleagues in 1962-1963 in Grand Rapids, Michigan. , This study is important for reasons that go beyond even its large sample sizes of 5,985 case drivers and 7,590 controls. In the early 1960s a larger proportion of drivers had BAC > 0 than after the later introduction of additional drunk driving countermeasures. Also, later changes in US law prevented police from stopping drivers at will in the way that the control drivers were stopped and requested to provide a breathalyzer reading voluntarily for research purposes only.
     The question of responsibility for crashing is most easily addressed in single-vehicle crashes, yet even in the large Grand Rapids total samples, only 622 case drivers were involved in single-vehicle crashes. This provided insufficient data for effective analysis. In order to focus on how alcohol affects the risk of crashing, the drivers judged to be responsible in multiple-vehicle crashes were combined with drivers in single-vehicle crashes to produce a sample 3,305 case drivers responsible for their crashes. Of these 21% had BAC > 0, compared to 11% of the 7,590 control drivers.
The effect of alcohol on the risk of being responsible for a crash is plotted in Fig. 10-2. Crash risk increases so steeply with BAC that a logarithmic scale is used. A driver with BAC = 0.17% is 32 times as likely to crash as a BAC = 0 driver. The case-control method does not compare a driver's risk at a given BAC to that same driver's risk at BAC = 0, but to the risk of another BAC = 0 driver. It is logically possible that the high BAC driver would be just as risky when sober. The steeply increasing risk with increasing BAC in Fig. 10-2 is corroborated by other case-control studies.15, -      Additional evidence is provided by a study that combined fatalities recorded in FARS with exposure estimates obtained in the 1996 National Roadside Survey of Drivers to estimate risks for different age and gender groups.Many substances, legal and illegal, affect driver performance, behavior, and crash risk. Often a mix of other substances is detected in conjunction with alcohol in post-crash autopsies. While there is much literature on drugs and safety, it is only for alcohol that pharmacological effects relate closely to the amount measured at the time of measurement. There are no known quantitative relationships like that in Fig. 10-2 for substances other than alcohol.
The Grand Rapids dip. At BAC = 0.025% the nominal indication in Fig. 10-2 is a (12 ± 7)% risk reduction. This so-called Grand Rapids dip, has been, and continues to be, the source of much speculation and controversy. It may be an artifact resulting from a weakness common to all case-control experiments. The case and control subjects may have different risks for unknown reasons. The BAC = 0 group contains many people who never drink. It is implausible to believe that drinkers at BAC = 0 would be identical in any attribute, including crash risk, to those who never drink. While a 17% greater risk by controls would convert a real 3% increase into an apparent 12% decrease, the same bias would convert a real 38 times increase into an observed 32 times increase (the value plotted in Fig. 10-2 for BAC = 0.175%), an important difference but hardly the stuff of controversy. Based on much smaller sample sizes, another case-control study likewise associated small quantities of alcohol with reduced risk.18 A study using a different method also found lower risk at low BAC for one of a number of cases studied.21 However, three studies report an increase in risk for low alcohol levels.15,19,20
     The effects of alcohol on behavior, as distinct from performance, are so complex that the possibility that small quantities of alcohol might reduce risk cannot be dismissed as implausible. Behavior changes at low doses are not simply smaller amounts of the behavior changes at high does, but can be in the opposite direction (more pleasant social behavior at small doses, less pleasant at high). Also, anecdotally one hears of drinkers claiming that they drive more carefully after drinking to reduce the risk of being stopped by the police. If so, this could generate lower crash risk after low levels of alcohol consumption.
Alcohol in fatally injured road users
     Figure 10-3 shows the distribution of BAC for fatally injured drivers with measured BAC > 0 in FARS 2002. A BAC reading (including BAC = 0) was recorded for 64.9% of fatally injured drivers. The probability that BAC is recorded in FARS varies widely over the US, from over 85% in the twelve highest reporting states to under 33% in the five lowest reporting states. As Table 10-3 shows, the probability that BAC is recorded depends also on many factors important in traffic safety, so the recorded cases should not be interpreted as a random samples of all cases. FARS advises "Alcohol Test Results from this database should be interpreted with caution."
     The majority of drivers whose BAC was recorded had BAC = 0. These are not included in Fig. 10-3. The average BAC for fatally injured drivers with any alcohol in their blood was 0.173%.
      The distribution for fatally injured pedestrians whose BAC was recorded shows even higher levels than for drivers, the average being 0.202%. BAC > 0.3% was measured for 14% of the pedestrians, but under 5% of the drivers. This alcohol level is likely to induce unconsciousness or deep sleep (Table 10-2), making vehicle driving unlikely. Many of the pedestrians killed at such alcohol levels may have been asleep on the road. Irish data indicate almost one in ten fatally injured drunk pedestrians appeared to have been lying on the roadway prior to being struck, perhaps attracted to a dry crown of the road in a wet climate.
      Two points are apparent from Table 10-3. First, when any amount of alcohol is present in the body of a road fatality, it is likely to be a large amount, much more than would be attained by a typical alcohol user. Second, the BAC levels of fatally injured drivers are not dramatically different from those of other fatally injured road users, or even drivers who are involved in fatal crashes in which they are not killed. In all cases, average BAC levels of those with any alcohol are higher than the highest legal driving limit in effect anywhere. This supports that the major contribution is from problem drinkers and alcoholics rather than social drinkers. ,5(p 518)
The number of US fatalities involving alcohol
      BAC is not known for 18,633 of the 42,815 people killed on US roads in 2002 (Table 10-3). Because of the need to estimate the role of alcohol in the nation's fatalities, NHTSA developed procedures to impute the missing BAC values based on relationships between factors known to correlate with alcohol use, such as nighttime driving and single-vehicle crashes. Using such procedures NHTSA estimates the percent of all fatalities in crashes in which alcohol was involved, some examples of which are shown in Table 10-4. Their estimates do not indicate the causal role of alcohol, nor how many fatalities would have been prevented if alcohol did not exist. Coffee is involved in most fatalities, yet eliminating coffee would have little effect on safety beyond preventing some drowsy driving crashes. Alcohol increases fatalities only because it increases crash risk. Sober drivers do not have zero risk, and drunk drivers do not have infinite risk. In order to estimate how much alcohol increases fatalities one must relate crash risk to BAC, using relationships such as in Fig. 10-2.
The causal role of alcohol. Suppose 100 drivers with BAC = 0.17% were killed in single-vehicle crashes. The risk of crashing at this alcohol level is 32 times the risk of crashing with BAC = 0, so even if alcohol had been absent, about 3 drivers would still be killed. One can therefore conclude that alcohol caused the death of the other 97. By applying similar calculations to the BAC distributions of drivers and pedestrians involved in different types of crashes, the fraction of traffic fatalities causally attributable to alcohol was estimated for 1987 to be 47%. This estimated value is in bold type in Table 10-4. It is 0.9 times the 52% estimate of the fraction of fatalities in which alcohol was involved. That is, it was found that 90% of crashes in which alcohol was involved were caused by the alcohol. The fraction of fatalities caused by alcohol for the other years (for which the detailed calculation is not available) are obtained by multiplying the percent of all fatalities that involved alcohol by this same 0.9 factor, producing the estimates in Table 10-4.
The calculation indicates that alcohol was causally responsible for 54%
of 1982 traffic deaths, and 38% of 2002 traffic deaths. For every fatality not attributable to alcohol in 1982, there were 1.17 fatalities attributable to alcohol.

Table 10-4. Percent of fatalities involving alcohol estimated by NHTSA, and percent caused by alcohol, estimated as explained in text.

     If in 2002 there had been 1.17 fatalities attributable to alcohol for every fatality not attributable to alcohol (instead of my estimate of 0.61), then fatalities in 2002 would have been larger by a factor (1 + 1.17)/(1 + 0.61) = 1.35. Thus, if alcohol had played the same role in 2002 that it did in 1982, about 15,000 additional fatalities would have occurred in 2002. By 1982 many anti-drunk driving measures were in place. While no quantitative estimate is available, it seems plausible to assume that perhaps another 15,000 annual deaths were being prevented, leading to a very crude approximation that all the measures in place in 2002 were preventing about 30,000 annual deaths. While this substantial reduction reflects the combined influence of many countermeasures, alcohol still killed more than 16,000 US road users in 2002.
Alcohol use by drivers in FARS. FARS 2002 includes 57,803 drivers involved in fatal crashes. 7,654 have measured BAC > 0.08%. Thus, FARS provides no evidence that 87% of the drivers involved in fatal crashes were in violation of a 0.08% limit. Test results are available for less than half of all drivers. Even though the probability of being tested increases with increasing BAC, 70% of those tested did not exceed 0.08%. For those drivers whose BAC was measured after involvement in a single-vehicle crash in which a pedestrian was killed, 83% had BAC < 0.08%. Enormous though the problem of drunk driving is, one must keep in mind that sober drivers cause far more harm than drunk drivers.
     If all the drivers with illegal BAC > 0.08 became marginally legal drivers with BAC = 0.08, this would reduce US traffic fatalities by 34%, rather than the 38% from eliminating alcohol entirely given in Table 10-4 (based on the calculation in Ref. 28). So, if all violations of drunk-driving laws were eliminated, 66% of US fatalities in 2002, or over 28,000 deaths per year, would remain. Even if alcohol were to disappear entirely, over 26,000 deaths per year would remain.
Alcohol's role in crashes of all severities
     Table 10-5 shows estimates of the monetary costs of crashes in which alcohol was involved using data from the study previously described in Chapter 2. The important feature to note is that the more severe the crash, the more likely it involves alcohol. Under 10% of the cost of minor crashes (mainly property damage, with at most a MAIS = 1 injury) is for crashes involving alcohol, whereas 45.5% of the cost of fatal crashes is for crashes involving alcohol. The increasing role of alcohol with increasing crash severity suggests that alcohol's main influence is changing driver behavior towards accepting higher risks and choosing higher speeds. If the only effect was impaired performance leading to increased driver error, then similar increases in crash risk at all severities might be a more likely outcome. It appears that drivers do things when they are drunk that they would not attempt when sober, rather than merely executing poorly the same things they would do more skillfully when sober.

Table 10-5. Cost of crashes involving alcohol compared to cost to all crashes. Data for 2000.

Drunk driving countermeasures involving criminal sanctions
      The enormous harm that alcohol causes in traffic has spawned a long history of countermeasures. While much progress has been made, alcohol remains a major contributor to traffic deaths in every society in which alcohol is consumed. The earliest countermeasures focused exclusively on using criminal law to punish offenders. Evidence of intoxication was usually provided by a police officer reporting that the accused was unable walk a straight line or speak clearly. Such subjective judgments of performance measures were easily challenged in courts.
Per se laws
      The pioneering work of Widmark9 in the 1930s led to development of instruments to measure alcohol content in the body. This made possible per se laws making it a crime to drive with a BAC exceeding a statutory limit. This probably represents the largest single advance in controlling drunk driving, because the offence could be defined by objective chemical analysis rather than subjectively judged behavior. Per se laws were usually accompanied by implied consent laws. These required the driver to consent to be tested as a condition to hold a license, and agree that a later refusal to be tested would create a presumption of intoxication. Another measure that sometimes accompanies per se laws is administrative license revocation, the immediate removal of the license if the BAC exceeds the proscribed limit.
The first per se law was enacted in Norway in 1936. It criminalized driving with BAC > 0.05%. The other Scandinavian countries, Sweden and Denmark, adopted similar laws. The term Scandinavian approach indicates per se laws enforced by severe punishments. This approach was generally considered successful in reducing drunk driving, although the evidence did not convince all researchers. ,
     The first per se law outside the Nordic countries (the three countries of Scandinavia plus Finland) was included in the British Road Safety Act of 1967, which made it an offense to drive with BAC > 0.08%. Immediately after implementation, fatalities and serious injuries occurring on weekend nights, a surrogate for drunk driving, dropped by 66%. (p 30) Further evidence of the success of the law is provided by a time series analysis that found that total traffic fatalities per unit distance of travel for 1968 dropped 11% below the long-term trend, but returned later to the trend.
The apparent success of the British law led Canada in 1969 to make it illegal to drive with BAC > 0.08%. Most of the world followed by making per se laws the kingpin of their drunk-driving policies. By 1978 all US states had laws making it illegal to drive with BAC > 0.10%. (BAC > 0.08% in Utah and Idaho). In 2000 the US Congress passed legislation providing financial incentives for states to have BAC > 0.08% laws in effect by 2004. All but a few states did.
BAC limits (like speed limits) are usually specified amounts that must not be exceeded. It is generally not an offense to be tested at the limit, but only at a higher value. One encounters comments like "a pedestrian was above the legal alcohol limit". I am not aware of any jurisdiction that has a legal alcohol limit. The limits specified in per se laws apply only to vehicle drivers.
     It was straightforward to examine the immediate effect of the British per se law by comparing casualties just prior and just after it went into effect. As time passes it becomes more difficult to estimate the effect of a law, because even if it were not passed, casualties would still increase or decrease for a whole host of reasons. Despite the difficulties, it is clear that the initial casualty reductions from the British law quickly declined. A major reason why crash rates tend to drift back towards prior levels after the introduction of interventions is that the objective risk of detection is small. The intervention is introduced with much publicity, convincing motorists that if they transgress, they will be subject to well-advertised penalties. Later, drivers become aware by observing or exchanging experiences with others that there is not a police officer at every corner or outside every drinking establishment. The key to sustaining casualty reductions is to maintain the belief that the probability of detection is high. An effective way to do this is to actually make the probability of detection high.
Random breath testing
     Random breath testing was introduced in the Australian state of New South Wales on 17 December 1982; as in the US, traffic law in Australia is largely a matter for the individual states. The program in New South Wales, with its legal limit of 0.05% BAC, gave rise to the slogan "under .05 or under arrest." (This is not strictly correct, as the offense was exceeding 0.05%). About 1.3 million tests were conducted annually on a driving population of 3.2 million; in other words, about a third of all drivers were tested each year, many being tested more than once. Figure 10-4 shows a time series of the number of fatalities per month. A drop of about 19% followed the introduction of random breath testing. This is one of the clearest and largest changes in traffic safety associated with a specific intervention.

Fig. 10-4. Traffic fatalities per month in New South Wales, Australia, three years before and after the introduction of mandatory breath testing on 17 December 1982. (p 7) The value (84 deaths) for December 1982 is not plotted.

     An examination of the fraction of fatal crashes that involved alcohol shows a corresponding drop downwards, from about 28% to 22%.39 (p 21) Such a change implies that in the pre-testing period, there were 28 fatal crashes involving alcohol for every 72 not involving alcohol. In the post-testing period there were 22 involving alcohol for every 78 not involving alcohol, or alternatively, 20.3 crashes involving alcohol for every 72 not involving alcohol, so that due to reductions in the alcohol contribution, fatalities declined by a factor (72+28)/(72+20.3), or 8%. Although this is an approximate calculation resting on uncertain assumptions, the difference between the 8% effect estimated and the 19% decline apparent in Fig. 10-4 suggests strongly that part of the reduction in fatalities is due to factors other than reductions in driving while intoxicated.
     A likely explanation is that the increased fear of interacting with police administering the testing program exercised a controlling influence on other types of driving behavior also likely to lead to fatal traffic crashes, or that driving, especially by high risk groups, was reduced. The primary goal of random breath testing is not to apprehend drunk drivers, but to make the probability of detection sufficiently high to deter drunk driving. Regardless of specific mechanisms, major casualty reductions resulted from the random breath-testing program. The benefits of the program were estimated to exceed its costs by over a factor of 30.
     Random breath testing has been widely adopted in other Australian states with comparable results. It had in fact been first introduced in Victoria in 1976, but in a less abrupt manner making evaluation more difficult. The program there was found to be effective, especially after major changes were introduced in 1989.
Sobriety checkpoints. The Australian approach could not be transferred directly to the US for legal reasons. However, in 1990, the US Supreme Court decided that sobriety checkpoints did not constitute illegal search and seizure if conducted strictly in accord with specified guidelines. At sobriety checkpoints, law enforcement officials evaluate drivers for signs of alcohol or drug impairment at certain points on the roadway. Vehicles are stopped in a specific sequence, such as every other vehicle or every fourth, fifth or sixth vehicle. The frequency with which vehicles are stopped depends on the personnel available to staff the checkpoint and on the traffic conditions. Alcohol is measured only if there is judgmental evidence suggesting impairment. Sobriety checkpoints have been used in most US states and are effective in deterring alcohol-impaired drivers and in reducing crashes.
     Without sobriety checkpoints or similar programs, the probability that a driver with an illegal BAC will be detected on an individual trip is very low. Detection requires other clearly illegal driving to be observed by a police officer before an alcohol test can be administered. The probability that a trip by a drunk driver will lead to an arrest has been estimated at 1 in 2,000, with a higher rate of 1 in 300 reported for high enforcement zones. I suggest in Chapter 16 that one of the most effective drunk-driving countermeasures is to automatically detect illegal speeding and red-light running.
What is an appropriate legal BAC level?
      Different jurisdictions choose different BAC limits for per se laws. In Sweden in the late 1980's there was support to make it an offense to drive with any detectable alcohol in the body. For various practical reasons, a zero BAC law was not accepted, but instead a 0.02% BAC law was passed by the Swedish Parliament, and took effect in July 1990. An average person will exceed this limit after one drink (Table 10-1).
     The trend to lower legal BAC limits is in part propelled by the increasing body of research discussed earlier showing skill reductions at BAC values even lower than 0.02%. Yet the data in Table 10-3 show that the average BAC for any fatally injured driver with any alcohol in the body is 0.173%, far in excess of the level specified in any per se law. The average values in Table 10-3 are typical of those found in Sweden and other countries, and do not appear to be depend much on the BAC limit defining drunk driving.
     The percents of the fatally injured drivers plotted in Fig. 10-3 with BAC exceeding various levels are as follows:

The vast majority of fatally injured drivers with any alcohol have BAC levels above even the highest of the legal limits. Unless other factors were at work, it would appear to not make all that much difference which level is chosen. For example, if all the drivers formerly obeying a 0.08% law were to obey a stricter new 0.05%, law, this would affect only 6% of the drivers with some alcohol in their bodies. The change in risk from a BAC of 0.08% to 0.05% is
modest compared to the large risks associated with the average BAC's of fatally injured drivers.
     While the casualty change associated with drivers who remain just under the legal limit after it changes may be small, reducing limits may have a more general deterrent influence on drunk driving. The public discussion stimulated by introducing lower limits may in itself contribute to reductions in drunk driving. The move to ever-lower legal limits is also partly inspired by a view that increased risk from alcohol use is more morally reprehensible than increased risk from other sources, such as speeding, driving while fatigued, sleep-deprived, upset, distracted, medicated, or slightly ill. Driving 65 mph when the speed limit is 55 mph increases risk of involvement in a fatal crash by a factor of 2.0, similar to the risk increase associated with driving with BAC = 0.08% compared to driving at BAC = 0.
Alcohol ignition interlock systems
     On-board devices to evaluate fitness to drive before the vehicle's engine can be started have been developed since the late 1960. Early attempts to use tasks in which performance deteriorated with increasing alcoholic consumption were unsuccessful. Even after customizing the task for a particular driver, this driver would still sometimes fail the test when sober and pass it when drunk. (p 198-202)
      Technology that measures alcohol in breath after the driver blows into a tube has led to effective interlock devices. The vehicle cannot be started unless a BAC below a set limit is recorded. Although such devices work reliably, and in questionnaires seem to evoke positive reactions from the public, there has never been a consumer market for them, nor is there likely to be. People think they are great devices to install in other people's vehicles! It is difficult to imagine any set of circumstances that would lead people to want them in their own vehicles, or to vote to have them installed on all vehicles.
     Where alcohol ignition interlock systems have proved successful is in reducing recidivism. A repeat offender may be offered a choice between driving only a vehicle with such a device installed and prison or license revocation. Reductions in recidivism of up to 90% have been found among interlock participants compared to those with suspended licenses. , Five Canadian jurisdictions, 43 US states, and many countries in Europe have legislation that allows the installation of interlock devices in offenders' vehicles.
     Alcohol ignition interlock programs are highly successful, but address only repeat offenders, who are a small minority of all offenders. For example, 2002 data for Connecticut show that 84% of those subject to administrative license revocation were first time offenders of drunk driving laws, 14% were committing their second offense, and 2% had previously committed three or more offenses. An earlier analysis found that preventing all drivers arrested for drunk driving from ever driving drunk again would reduce fatalities by 4.7%, injuries by 3.5%, and property damage by 2.4%.49(p 202)
    Above, and in common usage, offender means someone arrested and convicted. It is likely that as many as 2,000 offenses occur before an offender is arrested.45
Mothers Against Drunk Driving (MADD)
     The criminal law's approach to drunk driving in the US prior to the early 1980s was somewhat ambiguous. Even when laws were stringent, enforcement tended to be lax. A major change occurred in the 1980s, largely stimulated by citizen activist groups representing the families of victims killed by drunk drivers. Mothers Against Drunk Driving (MADD) is the best known of a number of such groups. MADD was founded by Candy Lightner after one of her 13-year-old twin daughters was struck and killed by a drunk driver. The crash occurred in the middle of the day on 3 May 1980 as her daughter was walking on a bicycle path. The driver had prior convictions, and only two days before the fatal crash had been released from jail on bail for another hit-and-run drunk driving crash. The release of this driver by the court focused attention on the judicial system's failure to protect the public from tragedies like this.
     MADD has grown to over 600 chapters. Its central thrust has been to advocate more severe punishments, such as more and longer prison sentences. MADD furthered these goals by court monitoring, in which members observe the court proceedings and encourage the judicial process to take the rights of victims, and potential victims, into account at sentencing. Since its start, more than 2,300 anti-drunk driving laws have been passed.
Citizen activist groups deserve credit for a major portion of the 15,000 traffic deaths per year reduction from the early 1980s to 2002. I believe that their influence was not so much through the specifics of having new laws passed, but rather in using the media to inform the public and change public attitudes. Widespread media coverage, including a full-length television movie on the tragedy that devastated Candy Lightner's family, stimulated many people to reflect more on negative factors associated with drinking. Although the coverage was modest compared to that of television advertisements associating beer with positive characteristics, I believe that the impact was profound. The testimony of bereaved parents makes it harder for society to continue to look upon the drunk as an endearing figure of amusement. Such changing attitudes made drunk driving less acceptable. The 1981 comedy movie Arthur, named for its alcoholic hero, was a box-office hit, grossing over $95 million. A 1988 sequel Arthur 2: On the Rocks flopped, grossing under $15 million. One of the many factors that contributed to the difference was a change in public attitudes, so drunkenness was seen less as a source of humor and more as a likely precursor to killing a child. The social norm regarding what is acceptable behavior has a large influence on how people behave, and MADD made drunk driving more unacceptable.
In its early history, MADD accepted large donations from the beer industry. While it no longer accepts such donations, it still focuses mainly on the individual drunk driver, while ignoring the crucial role of the rich powerful institutions that encourage and profit by his and, less often, her abusive consumption of alcohol.
Availability of alcohol
     Any decline in the overall consumption of alcohol is expected to lead to reductions in drunk driving. Many factors are known to affect how much alcohol is consumed, including the difficulty of obtaining it. The experience of prohibition in the US is sometimes invoked to support the claim that making alcohol more difficult to obtain does not reduce consumption. Prohibition was the period from 1920 to 1933 when the manufacture, sale, or possession of any drink with more than 0.5% alcohol was prohibited. The effects of this unfortunate attempt at social engineering were catastrophic, giving birth to many problems that persist to this day. However, the failure of prohibition does not mean that it did not reduce alcohol consumption. Data on how it affected alcohol consumption are not available, because all consumption was illegal and therefore not formally documented. However, time trend data for cirrhosis, admissions for alcoholic psychosis, and arrests for drunk and disorderly conduct suggest that alcohol consumption declined by more than half during prohibition. (p 195) Traffic fatalities were changing too rapidly to allow any inferences about the effect of alcohol, which in any event could not yet be quantitatively measured.
     While much less dramatic than prohibition, unmistakable links between abrupt interruptions in the availability of alcohol and alcohol-related harm have been shown in a number of studies from the Nordic countries. In September 1978 workers at Norway's state operated Wine and Spirits Monopoly went on strike for nine weeks. The occurrence of various events during this period was compared to their occurrence in the same period in 1977. Drunkenness was down 40%, domestic disturbances down 22%, and acts of violence against the person down 15%. Comparisons between the same two years for non-strike affected periods showed increases of between 3% and 6%. Closing Norway's Wine and Spirits Monopoly outlets on Saturdays in some towns but not in others led to differences in drunkenness. Small sample sizes precluded detecting changes in crash risk.
Minimum drinking age laws
     Perhaps the clearest indication of reductions in traffic deaths following reduced alcohol availability occurred when the US National Minimum Drinking Age Act of 1984 encouraged all states to raise their minimum age for purchase and possession of alcohol to 21. All states complied, replacing prior laws that had generally specified ages 18, 19 or 20. NHTSA reported a 13% reduction in fatal-crash involvements by affected drivers. A review of 241 studies of the effect of minimum drinking age laws provided overwhelming evidence of reduced alcohol consumption and traffic crashes in the affected age group. These laws are estimated to prevent close to 1,000 traffic deaths per year.56 All states have a zero-tolerance policy prohibiting driving with BAC > 0 at age under 21, as any alcohol in the blood implies violation of the law prohibiting the use of alcohol.
     Minimum drinking age laws reduce traffic deaths without depending on police monitoring of drivers. Instead, the laws reduce the probability of intoxication by those under 21 by prohibiting them from purchasing alcohol in bars, restaurants, and retail outlets. The employees of such businesses are trained to card any patron who looks even remotely likely to be under the minimum age, meaning that proof of age must be provided, in almost all cases by showing a driver license. Very strict adherence to this procedure is widespread throughout the US, because the benefit to the business of an easily detected illegal sale is so small compared to the penalties for violating the law.
     Under-age drinkers use many ways to circumvent the minimum drinking age law, including forged documents, and having older associates buy for them. Among some groups of young males the law is more honored in the breach than the observance, and underage drinking remains a major problem. It is unlikely that any law will meet with complete compliance, but what a law does do is increase the cost of alcohol, interpreting cost to mean all the ways that the user pays for it. Illegal under-age drinkers pay more for alcohol in terms of trouble, inconvenience, and obligations to those who assist them in violating the law, as well as risk of prosecution. Increasing the cost of anything reduces its consumption, but almost never to zero. One countermeasure to drunk driving is to increase any of the costs of consuming alcohol, the most obvious way to do so being to increase the purchase cost.
Cost of alcohol
     Economists describe the relationship between price and consumption in terms of price elasticity. An elasticity of -1 means that a (say) 5% increase in price leads to an equal 5% decrease in consumption, whereas an elasticity of -0.4 means a 5% price increase leads to a 2% decrease in consumption.
Price elasticity for different alcoholic beverages has been determined in different studies in many countries. One source summarizes 73 estimates. Simple averages of these, without regard to reliability or country, give the following elasticities: for beer -0.41, for wine -0.76, and for liquor -0.78. The authors of another study report values of -0.3, -1.0, and -1.5, respectively, but emphasize that these represent "best guesses" because of the wide range of estimates in the studies reviewed.
     While precise quantification is unavailable, there is little doubt that increases in price produce reductions in consumption. Given the difficulty of obtaining overall elasticities, there is scant information on elasticities for population sub-groups. However, there is no basis for thinking that heavy drinkers are exempt from basic economic laws. While alcohol is more intensively desired by heavy drinkers, it consumes a larger portion of their disposable income, so that resource constraints are more relevant than for moderate drinkers. Most alcohol is consumed by heavy drinkers, , so the elasticity values measured reflect mainly consumption changes by heavy drinkers.
US Federal Excise Tax
     The US Federal Government provided the leadership that led to minimum age drinking laws, thereby acknowledging its responsibility to address the national problem of drunk driving. The same Federal Government already exercises a statutory role in influencing the price of alcohol through the Federal Excise Tax, and therefore has at its disposal a potent weapon to reduce drunk driving. Not only has this weapon not been used, but the Federal Excise Tax has actually declined steeply in real terms from initially small amounts to even smaller amounts. Currently, the tax on the standard drinks defined earlier is 5¢ on beer, 4¢ on wine, and 13¢ on liquor. Adjusted for inflation, these amounts represent an 86% reduction from 1951 to 2003. (There are also state taxes on alcohol).
     It is the consumption of beer, the beverage of choice of young males, which causes most drunk driving. Even if the Federal Excise Tax had kept step with inflation since 1951 and risen to 19¢ compared to its present 5¢, it would still add a small percent to the cost of a beer. While non-alcoholic beverages have increased in price relative to the consumer price index, alcoholic beverages have decreased in price.61(p 8)
     A strange irony of US alcohol policy is that the beverage that is responsible for the most harm is treated is if it were the least harmful. A substantial increase in the tax on beer would have an important impact on drunk driving. At an absolute minimum, the tax on beer should not be less than that on liquor.
     Seemingly, the death of 16,000 people from alcohol in traffic is not a political problem, but increasing the Federal Excise Tax on beer is. Such a tax increase could be politically acceptable if it was made clear that the purpose was to save lives. Its political acceptability would be assured if it were rendered revenue neutral by reducing another tax that did not save lives. The small minority who consume most of the alcohol would pay far more in taxes each year, notwithstanding their reduced consumption. The total tax paid by moderate drinkers would be less if a revenue-neutral change were enacted. Drunk driving and other social ills can be substantially reduced by increasing the excise tax on alcohol, particularly on beer.
Alcohol sales
     Beer accounts for more than half of alcoholic beverage retail sales in the US (Table 10-6). The heaviest 5% of drinkers, who, on average, consume more than four drinks per day, consume 42% of the alcohol sold, while the heaviest 2.5% of drinkers, who consume more than six drinks per day, account for more than a quarter of alcohol sales.62 29% of the population is teetotal (consume no alcohol). , Young people (not necessarily underage) who consume hazardous quantities of beer are the alcohol industry's most important customers. Hazardous drinking, defined as 5 drinks or more per day, accounts for more than half the alcohol industry's market and 76 percent of the beer market. Underage and adult excessive drinking account for half of the alcohol industry's sales.
Alcohol advertising
      The facts of alcohol consumption are in stark contrast to the belief that the alcohol industry has skillfully fostered, through massive advertising, that drinking is universal, glamorous, and largely devoid of negative consequences. In 2002, $1.9 billion was spent on alcohol advertising in measured media (television, radio, print, outdoor, major newspapers, and Sunday supplements). The largest portion of this was on television advertisements for beer, most of which are placed in sports programs. Budweiser and Bud Light spent more than 87% of their combined television advertising expenditures on sports programming in 2001 and 2002. Working from alcohol company documents, the Federal Trade Commission estimated that, in 1999, the alcohol industry's total expenditure to promote sales (including through sponsorship, Internet advertising, point-of-sale materials, product placement, items with brand logos, and other means) was three or more times its expenditure for measured media advertising. This would mean that the alcohol industry spent a total of $5.7 billion or more on advertising and promotion in 2002. About 65% of the expenditure was for marketing beer. The American Medical Association estimates that young people are bombarded with $4 billion of alcohol marketing each year.61(p 6)
Beer is most advertised, causes most harm
     Beer, which is 55% of alcoholic consumption, but a larger proportion of problem consumption, is advertised so widely on television as to constitute an important portion of television advertising revenues. In contrast, there is a voluntary ban on television advertising of liquor, the alcoholic beverage that accounts for 30% of alcohol consumption, and a yet smaller proportion of problem alcoholic consumption. Two percent of television advertising revenue in 2002 was from alcohol advertisements, largely for beer. The combination of massive advertising expenditures, and a television industry too timid to allow mention of obvious truths that would adversely affect business with a major customer, has led to uncritical acceptance of patently false claims.
      The alcohol industry claims that advertising does not increase consumption. They allege that its sole purpose is to persuade customers to choose one brand over another without changing the total number of customers. The industry's actions show that they do not believe anything so foolish. If the industry believed that advertising did not increase consumption, then they would be expected to support (perhaps quietly) a universal ban on all advertising so that they could pocket the billions they pay in order to play what they are alleging is a zero-sum game with each other. Two companies, Anheuser-Busch and Philip Morris (owner of Miller Brewing Company), account for two-thirds of all beer sales. It is hard to believe that each symmetrically believes that the other would enjoy a sharp increase in market share at their expense if television advertising were discontinued. Their opposition to voluntary or statutory limits on advertising makes sense only if they believe that advertising increases beer consumption.
     It is difficult to take seriously any claim that the large alcohol-advertising billboards that dominate the American urban landscape are there solely to persuade the generally poor inhabitants to switch brands. The advertising of just about any product increases consumption of it, as well as encouraging switching to the advertised brand. Even if the only effect of the advertising were to persuade some to switch brands from less advertised non-alcoholic drinks to alcoholic drinks, this would still increase alcohol consumption.
Advertising to under-age drinkers. The alcohol industry claims that it does not advertise to under-age drinkers. It would be unusual for any industry to not want to acquire new customers, and to acquire them at as early an age as possible. The industry behaves in accord with this economic law while denying it does so. The beer industry is a major sponsor of television sports with mainly young male viewers, a large portion known to be under 21.
Problem drinkers - core customers of alcohol industry
The alcohol industry claims that it wants to eliminate problem drinking and sell only to responsible drinkers. Successful businesses owe their success to their best customers, not to those who do not buy their products, or buy them sparingly. The highest consuming 5% of the population, those who consume four or more drinks per day, account for 42% of alcohol sales.62 If all these individuals were to suddenly become responsible drinkers, drunk driving would largely disappear. The nearly 42% reduction in sales would transform profits into deep losses, forcing the alcohol industry to undergo major restructuring.
Tax and advertising policies that would save lives
     Two simple "laws" apply to alcohol's role in traffic deaths.
Law 1. Decreasing national alcohol consumption leads to fewer traffic deaths.
As is usual with any law, we assume other things remain unchanged. Decreased consumption might not reduce traffic deaths if the lower consumption was more concentrated among fewer people, or more peaked by time of day, or if other successful anti-drunk-driving policies were discontinued. Law 1 is so compellingly obvious that the onus is on anyone who does not accept it to provide specific evidence or convincing reasons why it is not so. This law does not have any immediate policy implications, nor does the similarly valid law that lower speeds reduce fatalities. There are benefits in higher speed and in consuming alcohol. Such laws illuminate policy decisions, but do not define policy. One profound difference between speed and alcohol is that there is no large politically powerful industry whose earnings depend directly and primarily on higher speeds.
Law 2. Alcohol consumption is decreased by:
· Decreased Advertising.
· Increased Price.
· Decreased Availability.
It is hard to imagine any set of circumstances, even of a hypothetical nature, in which any component of Law 2 would not apply. Claims by the alcohol industry that aspects of Law 2 do not apply are about as convincing as their claims that the only purpose of advertising is to move customers from one brand to another.
Reasonable approaches to harmful substances
      While the ways to reduce drunk driving are clear, they involve a clash of interests and a US political tradition of foolish policy when it comes to substances that cause harm. Alcohol was banned entirely from 1920 to 1933 with catastrophic consequences. Efforts to reduce the 16,000 traffic deaths caused by alcohol are often countered by charges of prohibition. Prohibiting television and billboard advertising, and increasing the Federal Excise Tax are no more prohibition than are present prohibitions of beer vending machines in public places, selling to under-age drinkers, or the existence of the Federal Excise Tax. The alcohol industry opposes new restraints by invoking grand principles that apply just as strongly to previously passed well-accepted regulations that they also often opposed before implementation.
      The US seems intent on not learning from the experience of prohibition in its present war on drugs. The political process has classified a number of harmful substances as illegal, just as alcohol was classified as illegal between 1920 to 1933. Because alcohol is now legal, the alcohol industry claims that it should be no more constrained than the manufacturer of any other legal product. At the same time, mere possession of another substance can lead to a prison life sentence. The distinction between legal and illegal substances is determined by the political process, not by how much harm is caused. It is a distinction that has served the US poorly. Legalizing any illegal substance will inevitably increase its use (legalizing alcohol in 1933 approximately doubled its use), but the increased harm must be balanced against the seemingly unbounded costs of making any widely demanded product illegal.
     The distinction should be between products that cause large amounts of harm, particularly harm to people who do not use them, and the vast majority of products that do not cause appreciable harm. Products that cause major harm should be subject to regulation aimed at reducing the harm they cause, with regulation being more forceful if substantial harm is caused to non-users. Many teetotalers and light drinkers are killed by drunk drivers. The consumption of alcohol causes more harm in the US than the consumption of any other legal or illegal substance, with the possible exception of tobacco. Unlike alcohol, the victims of tobacco are overwhelmingly those who use it.
     Each harmful substance should be evaluated in a similar manner. A reasonable analysis would rarely conclude that an absolute ban supported by severe penalties was the best policy. Nor would a reasonable analysis conclude that manufacturers should be permitted to increase consumption by using such potent means of persuasion as television advertising.

Summary and conclusions (see printed text)

References for Chapter 10 - Numbers in [ ] refer to superscript references in book that do not correctly

 show in this html version.  To see how they appear in book see pdf version of Chapter 1 or pdf version of Chapter 16.

[1] Goldberg IJ, Mosca L, Piano MR, Fisher EA. Wine and your heart. Circulation. 2001; 103: 472.

[2] Beirness DJ, Simpson HM. The safety impact of lowering the BAC limit for drivers in Canada. Ottawa, Ontario, Canada: Traffic Injury Research Foundation; May 2002.

http://www. trafficinjuryresearch.com/publications/PDF_publications/BAC_Limits.pdf

[3] Moskowitz H, Fiorentino D. A review of the literature on the effects of low doses of alcohol on driving-related skills. Report DOT HS 809 028. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration; April 2000.

http://www. nhtsa.dot.gov/people/injury/research/pub/Hs809028/DocPage.htm

[4] Jones RK, Lacey JH. Alcohol and highway safety 2001: A review of the state of knowledge. Report DOT HS 809 383. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration; November 2001.

[5] Waller JA. Injury Control -- A Guide to the Causes and Prevention of Trauma. Lexington, MA: Lexington Books; 1985.

[6] Borkenstein RF. Historical perspective: North American traditional and experimental response. J Studies Alcohol, Supplement. 1985; 10: 3-12.

[7] Jones AW. Enforcement of drink-driving laws by use of ‘per se’ legal alcohol limits: Blood and/or breath concentration as evidence of impairment. Alcohol, Drugs, and Driving 1988; 4: 99-112.[8] Wilkinson PK, Sedman AJ, Sakmar E, Erhart RH, Weidler DJ, Wagner JG. Pharmacokinetics of alcohol after oral administration in the fasting state. J Pharmacokinet Biopharm. 1977; 5: 207-224.

[9] Andréasson R. Widmark’s Micromethod and Swedish Legislation on Alcohol and Traffic. Stockholm, Sweden: The Information Center for Traffic Safety; 1986.[10] Baraona E, Abittan CS, Dohmen K, Moretti M, Pozzato G, Chayes ZW, Schaefer C, Lieber CS. Gender differences in pharmacokinetics of alcohol. Alcohol Clin Exp Res. 2001; 4: 502-507.

[11] BAC after quickly consuming a number of drinks as a function of gender and age. Calculator at:

http://www.ScienceServingSociety.com/AlcoholCalculator.htm

[12] Johnson MB, Voas RB. Potential risks of providing drinking drivers with BAC information. Traf Inj Prev. 2004; 5: 42-49.

[13] NADS awarded $2.9 million for alcohol and driver performance research. University of Iowa press release, 31 December 2002.

http://www.uiowa.edu/~ournews/2002/december/1231nads.html

[14] Moskowitz H, Robinson C. Driving-related skills impairment at low blood alcohol levels. In: Noordzij P, Roszbach R, editors. Alcohol, Drugs and Traffic Safety - T86. Amsterdam, Netherlands: Excerpta Medical Elsevier Science Publisher; 1987, p. 79-86.

[15] Holcomb RL. Alcohol in relation to traffic accidents. J Am Medical Assoc. 1938; 111: 1076-1085.

[16] Borkenstein RF, Crowther RF, Shumate RP, Ziel WB, Zylman, R. The role of the drinking driver in traffic accidents. Department of Police Administration, Indiana University; 1964.

[17] Borkenstein RF, Crowther RF, Shumate RP, Ziel WB, Zylman, R. The role of the drinking driver in traffic accidents. Blutalkohol 11 (supplement 1); 1974.

[18] McCarroll JR, Haddon W Jr. A controlled study of fatal automobile accidents in New York City. J Chronic Diseases. 1962; 15: 811-826.

[19] Perrine MW, Waller JA, Harris LS. Alcohol and highway safety: Behavioral and medical aspects. Report DOT HS 800 599. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration; 1971.

[20] Farris R, Malone TB, Lilliefors H. A comparison of alcohol involvement in exposed and injured drivers. Phases I and II. Report DOT HS 801 826. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration; 1976.

[21] Zador PL, Krawchuk SA, Voas RB. Relative risk of fatal crash involvement by BAC, age, and gender. Report DOT HS 809 050. Washington, DC: National Highway Traffic Safety Administration; April 2000.

[22] National Commission Against Drunk Driving. Drinking and driving in the United States: The 1996 National Roadside Survey. http://www.ncadd.com/112.cfm

[23] Corfitsen MT. Tiredness! A natural explanation to The Grand Rapid “DIP”. Accid Anal Prev. 2003; 35: 401-406.[24] Fatality Analysis Reporting System (FARS) Web-Based Encyclopedia. Create a query, and choose “Alcohol Test Result” as a variable – caution regarding use of BAC appears:

http://www-fars.nhtsa.dot.gov[25] An Foras Forbartha. Road Accident Facts Ireland, 1984. Dublin, Ireland: National Institute for Physical Planning and Construction Research; 1985.

[26] Vingilis, E. Drinking drivers and alcoholics -- are they from the same population? In: Smart RG, Glasser FB, Israel Y, Kalant H., Popham R, Schmidt W, editors. Research Advances in Alcohol and Drug Problems, Vol. 7. New York, NY: Plenum Press; 1983, p. 299-342.

[27] Subramanian R. Transitioning to multiple imputation – a new method to estimate missing blood alcohol concentration (BAC) values in FARS. Report DOT HS 809 403. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration; January 2002 (Revised October 2002).

http://www-nrd.nhtsa.dot.gov/pdf/nrd-30/NCSA/Rpts/2002/809-403.pdf[28] Evans L. The fraction of traffic fatalities attributable to alcohol. Accid Anal Prev. 1990; 22: 587‑602.[29] Subramanian R. Alcohol involvement in fatal crashes 2001. Report DOT HS 809 579. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration; April 2003.

http://www-nrd.nhtsa.dot.gov/pdf/nrd-30/NCSA/Rpts/2003/809-579.pdf

[30] US Department of Transportation. DOT releases preliminary estimates of 2002 highway fatalities. Press release, NHTSA 13-03; 23 April 2003

http://www.emcommunityconnection.com/nhtsa_13.htm

[31] Blincoe LJ, Seay AG, Zaloshnja E, Miller TR, Romano EO, Luchter S, Spicer RS. The economic impact of motor vehicle crashes, 2000. Report DOT HS 809 446. Washington DC: US Department of Transportation, National Highway Traffic Safety Administration; May 2002. http://www.nhtsa.dot.gov/people/economic/EconImpact2000

[32] Glad A. After 50 years with a per se law -- the drinking and driving problem in Norway. In: Noordzij P, Roszbach R, editors. Alcohol, Drugs and Traffic Safety - T86. Amsterdam, Netherlands: Excerpta Medical Elsevier Science Publisher; 1987, p. 241-244.

[33] Snortum JR. Deterrence of alcohol-impaired drivers. In: Laurence MD, Snortum JR, Zimring FE, editors. Social Control of the Drinking Driver. Chicago, IL: University of Chicago Press; 1988, p. 189-226.

[34] Ross HL. The Scandinavian myth: The effectiveness of drinking-and-driving legislation in Sweden and Norway. J Legal Studies. 1975; 2: 1-78.

[35] Ross HL. Deterrence-based policies in Britain, Canada, and Australia. In: Laurence MD, Snortum JR, Zimring FE, editors. Social Control of the Drinking Driver. Chicago, IL: University of Chicago Press; 1988, p. 64-78.

[36] Ross HL. Deterring the Drinking Driver. Lexington, MA: Lexington Books; 1984.

[37] Broughton J. Predictive models of road accident fatalities. Traf Eng Control. 1988; 29: 296-300.

[38] Job RFS. The application of learning theory to driving confidence: The effect of age and the impact of random breath testing. Accid Anal Prev. 1990; 22: 97-107.

[39] Roads and Traffic Authority. Road traffic accidents in New South Wales 1988. Rosebury, NSW, Australia: Road Safety Bureau; May 1989.

[40] Arthurson RM. Evaluation of random breath testing. Research note RN 10/85. Rosebury, NSW, Australia: Roads and Traffic Authority, Road Safety Bureau; 1985.

[41] Henstridge J, Homel R, Mackay P. The long-term effects of random breath testing in four Australian states: A time series analysis. Report CR162. Canberra, Australia: Federal Office of Road Safety; 1997.

[42] Cameron M, Cavallo A, Sullivan G. Evaluation of the random breath testing initiative in Victoria 1989-1991. Multivariate time series approach. Report #38. Victoria, Australia: Monash University Accident Research Centre; 1992.

[43] National Highway Traffic Safety Administration. Saturation patrols & sobriety checkpoints: A how-to guide for planning and publicizing impaired driving enforcement efforts. Report DOT HS 809 063. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration; Revised October 2002.

http://www.nhtsa.dot.gov/people/injury/alcohol/saturation_patrols/SatPats2002.pdf

[44] Insurance Institute for Highway Safety. Sobriety checkpoints work but they aren’t used often. Status Report. Volume 36 (6); 30 June 2001. http://www.hwysafety.org/srpdfs/sr3606.pdf[45] Borkenstein RF. Problems of enforcement, adjudication and sanctioning. In: Israelstam S, Lambert S, editors. Proceedings of the 6th International Conference on Alcohol, Drugs And Traffic Safety, p. 655–662. Toronto, Ontario; 8-13 September 1974.[46] Voas RB, Hause JM. Deterring the drinking driver: The Stockton experience. Accid Anal Prev. 1987; 19: 81-90.

[47] Sjögren H, Björnstig U, Eriksson A, Öhman U, Solarz A. Drug and alcohol use among injured motor vehicle drivers in Sweden: Prevalence, driver, crash, and injury characteristics. Alcohol Clin Exp Res. 1997: 21: 968-73.

[48] Voas, R.B. Emerging technologies for controlling the drunk driver. In: Laurence MD, Snortum JR, Zimring FE, editors. Social Control of the Drinking Driver. Chicago, IL: University of Chicago Press; 1988, p. 321-370.

[49] Evans L. Traffic Safety and the Driver. New York, NY: Van Nostrand Reinhold; 1991.

[50] Cobden JH, Larkin GL. Effectiveness of ignition interlock devices in reducing drunk driving recidivism. Am. J. Prev. Med. 1999; 16: 81–87.[51] Voas RB, Marques PR, Tippetts AS, Beirness DJ. The Alberta interlock program: The evaluation of a province-wide program on DUI recidivism. Addiction. 1999; 94: 1849–1859.[52] Beirness DJ, Simpson HM, Robertson RD. Commentary: International symposium on enhancing the effectiveness of alcohol ignition interlock programs. Traf Inj Prev. 2003, 4: 179–182.[53] Fazzalaro JJ. Miscellaneous questions regarding drunk driving offenders. OLR Research Report 2003-R-0729. October 2003. http://www.cga.state.ct.us/2003/olrdata/tra/rpt/2003-R-0729.htm

[54] Gerstein DR. Alcohol use and consequences. In: Moore MH, Gerstein DR, editors. Alcohol and Public Policy -- Beyond the Shadow of Prohibition. Washington, DC: National Academy Press; 1981, p. 182-224.

[55] Hauge R. The effects of changes in availability of alcoholic beverages. In: Laurence MD, Snortum JR, Zimring FE, editors. Social Control of the Drinking Driver. Chicago, IL: University of Chicago Press; 1988, p. 169-187.

[56] National Highway Traffic Safety Administration. Fact sheet: Minimum drinking age laws. Community how to guide on public policy, appendix 7. http://www.

nhtsa.dot.gov/people/injury/alcohol/Community%20Guides%20HTML/PDFs/Public_App7.pdf

[57] National Highway Traffic Safety Administration. The impact of minimum drinking age laws on fatal crash involvements: An update of the NHTSA analysis. Report DOT HS-807-349. Washington, DC; January 1989.

[58] Wagenaar AC, Toomey TL. Effects of minimum drinking age laws: Review and analyses of the literature from 1960 to 2000. J. Stud. Alcohol. 2002: Supplement No. 14; 206-225.

[59] Raistrick D, Hodgson R, Ritson, B. Tackling Alcohol Together: The Evidence Base for UK Alcohol Policy. London, UK: Free Association Books; 1999.

http://www.scotland.gov.uk/health/alcoholproblems/docs/lire-05.asp

[60] Leung SF, Phelps, CE. “My kingdom for a drink?” A review of estimates of the price sensitivity of demand for alcoholic beverages. In: Hilton ME, Bloss G, editors. Economics and the Prevention of Alcohol-Related Problems, p. 1-32. Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism, NIAAA Research Monograph No. 25, NIH Pub. No. 93–3513; 1993.[61] American Medical Association. Policy briefing paper: Partner or foe? The alcohol industry, youth alcohol problems, and alcohol policy strategies. Chicago, IL; 2002.

http://www.ama-assn.org/ama1/pub/upload/mm/388/partner_foe_brief.pdf

[62] Greenfield T, Rogers J. Who drinks most of the alcohol in the US? The policy implications. J Studies Alcohol. 1999; 60: 78-89.

[63] Brandes R. Beverage dynamics identifies the fastest-growing brands of wine, beer and spirits in the beverage alcohol industry. Beverage Dynamics; March/April 2002.

http://www.beveragenet.net/bd/2002/0204/0204gb.asp

[64] Global Alcohol Policy Alliance. Partner or foe? (Paper produced by the American Medical Association), p. 18. http://www.ias.org.uk/publications/theglobe/02issue2/globe02issue2.pdf

[65] Rogers J, Greenfield T. Beer drinking accounts for most of the hazardous alcohol consumption reported in the United States. J Studies Alcohol. 1999; 60: 732-39.

[66] Foster SE, Vaughan RD, Foster WH, Califano JA Jr. Alcohol consumption and expenditures for underage drinking and adult excessive drinking. J Am Medical Assoc. 2003; 289: 989-95.

[67] The Center for Alcohol Marketing and Youth. Alcohol advertising and youth. 2 January 2004.

http://camy.org/factsheets/index.php?FactsheetID=1

[68] The Center for Alcohol Marketing and Youth. Alcohol advertising on sports television 2001 and 2002. http://camy.org /factsheets/pdf/AlcoholAdvertisingSportsTelevision2001-2002.pdf

[69] Federal Trade Commission. Self-regulation in the alcohol industry: A review of industry efforts to avoid promoting alcohol to underage consumers. Appendix B: Alcohol advertising expenditures, page iii; September 1999. http://www.ftc.gov/reports/alcohol/appendixb.htm