Epidemiologic Perspectives on Drunk Driving

Transcription

1 35 Epidemioogy and Data Management Epidemioogic Perspectives on Drunk Driving M.W. Perrine Vermont Acoho Research Center, Burington, Vermont Raymond C. Peck Research and Deveopment Office Caifornia Department of Mot& Vehices, Sacramento, Caifornia James C. Fe Nationa Centerfor Statistics andanaysis, Nationa Highway Tmfic Safety A&n&istration, Washington, D. C. What is the one eement found in approximatey haf the U.S. highway fataities? This question has been raised over the ast few decades and the answer is sti the same: Acoho. This answer generates another question: If a singe, identifiabe eement is invoved in such a arge portion of a serious pubic heath and pubic safety probem, shoud it not receive top priority for investigation, in tervention, and prevention? Acoho produces both peasure and pain, euphoria and depression. Acoho aso produces many jobs and biions of doars in tax revenues to the States and to the Nation. Each year, acoho aso produces unintentiona death to thousands and injury to miions. When mixed with driving, acoho is the basis for a major pubic heath and pubic safety probem. In our automotive society, the car is used for amost a facets of socia activity. Therefore, since acoho is invoved in many aspects of socia behavior, driving after drinking is a reativey frequent occurrence. Fortunatey, the vast majority of such driving-after-drinking instances do not resut in crashes. One very important task for researchers is to identify variabes that differentiate between those driving-after-drinking instances that do resut in a crash and those that do not. How do we earn about the contribution of acoho to unintentiona injury and death on the highways? In attempting to do so, we sti find a arge gap between description and expanation that, at this time, can be bridged ony provisionay through inference. Two widey separated research approaches have been used to date as a basis for inferring the contribution of acoho to highway crashes: epidemioogic and experimenta. NOTE: Primary responsibiity forparts of this chapter is as foows: J.C Fe, the subsection entited Acoho fnvovement in Fata Highway Crashes ; RC. Peck, the subsection entited Tharactersitics of Drunk Dfiw-s ; M.W. Perrine, a other sections. The first author is gratefu to Robert B. Voas for eary discussions of this chapter and for materia p&i&d in the sections on Other Roadside Research and Enforcement Checkpoints. Preparation of M.W. Perrine s part of this chapter and production of the manuscript were supported by PHS Research Grants.4,474, AA06926, and AA07876 from the Nationa Institute on Acoho Abuse and Acohoism.

2 36 BACKGROUND PAPERS The epidemioogy of acoho and highway safety can be traced from the first review of the probem presented in 1933 (Mies 1934). Over the years, high bood acoho concentration (BAC) has &en thoroughy impicatedin serious and fata injury highway crashes by post hoc epidemioogic studies. Most evidence for reating-this acoho contribution to highway crashes has been obtained by examination of the distribution of BAC both among drivers invoved in actua crashes (fata and nonfata) and- on the basis of case-contro roadside surveys-among drivers using the highways, but not invoved in crashes at the time. A number of such case-contro studies have demonstrated that acoho is overrepresented among deceased drivers reative to drivers in the popuation-at-risk using the highways at corresponding times and paces (e.g., Borkenstein et a. 1964, 1974; Perrine et a. 1971; for reviews see NRC 1987; NHTSA 1985; Perrine 197%; b). The second approach consists of controed administration of acoho in experiments conducted on isoated variabes that are assumed to be reevant for actua driving. Acoho impairment of rea-word driving performance is then typicay inferred from the mosaic of these bits of behavior examined separatey in the aboratory, in driving simuators, and in instrumented cars driven on cosed courses (NRC 1987). Both research approaches to the study of drunk driving are necessary and have been productive (NRC 1987; Perrine 1976). However, this chapter is imited to epidemioogic aspects; it is organized as foows: A discussion of the scope of the drunk-driving probem from an epidemioogic perspective A brief outine of the major components invoved in studying the probem by means of the avaiabe data sources A review of the most reevant iterature, focusing on acoho invovement in fata as we as nonfata highway crashes and in noncrash drivers; crash risk and acoho; and characteristics of drunk drivers An examination of current issues and probems Scope of the Probem The primary probem ceary consists of those motor vehice crashes that resut in fata injuries. It is now generay we estabished that acoho is invoved in approximatey haf of a such fata crashes. For exampe, the tota number of highway fataities in 1986 was 46,056, of which some 24,000 (52 percent) invoved acoho. More specificay, BACs exceeded the typica ega imit (0.10) in 41 percent of a fata crashes. An estimated 4.8 percent of deaths in the United States during 1980 were directy or indirecty attributabe to acoho (NIAAA 1987). Of these, motor vehice crashes were the argest singe cause of death. Approximatey 26,000 deaths in 1980 were attributed to acoho in motor vehice crashes; these deaths constituted about 27 percent of the tota number of deaths (approximatey 98,000) attributabe to acoho (NIAAA 1987, p.6). The number of acoho-invoved motor vehice deaths is about two times that of the second argest singe cause of acoho-invoved death, namey, homicide (approximatey 12,000 or 12 percent) (NIAA4 1987, p-6). The scope of the acoho and traffic safety probem has recenty been reviewed briefy both from a pubic heath perspective (NIAAA 1982,1985,1987) and from a pubic safety perspective (NHTSA 1985,1987b; NRC 1987). In this chapter, the probem is examined further to provide a more integrated synthesis of the iterature from both these perspectives.

3 EPIDEMIOLOGY AND DATA MANAGEMENT 37 Major Components of the Study Aside from epidemioogic methodoogy considerations, the major components invoved in the present perspective on drunk driving consist of the data sources. The two primary sources of data for this area are offtcia records and surveys of various types. The officia records consist primariy of the foowing: The citation report for driving under the infuence of acoho (DUI) The accident report, if acoho is invoved The prosecutor record Court records Department of Motor Vehice records Treatment/service provider records Probation department records Of specia importance are those reports of accidents in which a fataity resuted, since these data are coected at the State eve and then forwarded to the Fata Accident Reporting System (FARS) of the Nationa Highway Traffic Safety Administration (NHTSA). As an exampe of using such data, anayses of DUI processing from the point of the arrest citation through the other officia records, incuding the postconviction countermeasures, have been prepared for the State of Caifornia (Perrine 19&1; Heander 1986; Peck 1987). The other major source of data for epidemioogic studies consists of surveys. The main varieties are: roadside surveys, teephone surveys (in recent years, the random-digit-diaing teephone SuNeY), househod surveys, and specia ocation surveys (bars, jais, etc.). Of these various types, ony the roadside surveys can obtain direct measurements of the major criterion variabe- namey, BAC- from drivers actuay using the roads at the time. A the other survey methods depend on sef-reported information from the respondents, incuding data concerning driving after drinking. Thus, ony the direct measurement of BAC at roadside can be used to provide criterion measures for estimating acoho crash risk and for evauating the impact of countermeasure programs on the motoring pubic. Review of the Most Reevant Literature Acoho Invovement in Fata Highway Crashes In 1986, 46,056 peope were kied in traffic crashes (NHTSA 1988), which are the eading cause of death for Americans age 6-34 (Richardson 1985). Traffic fataities in 1986 resuted in 1,425,517 years of potentia ife ost before age 65, an amount greater than deaths from cancer, heart disease, and a other causes. Traffic crashes cost society approximatey $74 biion annuay in terms of damage, insurance costs, injury treatments, ost work, and so forth. (NHTSA 1987c). Since 1900, over 2,600,OOO Americans have died in traffic crashes; that is 1,500,OOO more than the tota number of Americans kied in a the wars in U.S. history.

4 38 BACKGROUND PAPERS It is we known that acoho is a eading factor in trafic crashes. It was invoved in over haf of thetraffic fataities in 1986, resuting in cose to 24,000 deaths (NHTSA 1988). Each year, neary 560,000 additiona peope suffer injuries in acohoi-reated crashes - an average of one person injured every minute of the day. About 43,000 of these injuries are serious (NHTSA 1988). During the period, approximatey, 119,000 peope ost their ives in acohoreated traffic crashes-an average of one acoho-reated fataity every 22 minutes over the past 5 years. About two out of five Americans wi be invoved in an acoho-reated crash in their ifetime (NHTSA 1987u). Approximatey 1800,000 drivers were arrested in 1986 for DUI - an arrest rate of about 1 out of every 90 icensed drivers in the United States (Greenfied 1988). The probem is especiay devastating for young peope. In 1986, more than 40 percent of a teenage deaths resuted from motor vehice crashes. Over haf of these were acoho reated, making acoho-reated trafic crashes the eading cause of death for teenagers. For traffic crash victims age 20-24, cose to 70 percent of the 8,000 who died in 1986 were in acoho-reated crashes (NHTSA 1988). The probabiity that a given death is due to a traffic fataity is 55 times as great for a u-year-od mae as for a 65year-od mae; the corresponding ratio for femaes is 43 (Evans 1987). The average BAC of drinking drivers invoved in fata crashes was 0.15 in 1986 (NI-ITSA 1985). The ega intoxication imit in most States is In a recent survey of drivers jaied for drunk driving offenses, over a quarter of the drivers had consumed at east 20 beers or 13 mixed drinks within 3-4 hours before they were arrested (Greenfied 1988). Research has shown that a driver with a BAC of 0.15 has a 26times greater probabiity of beii invoved in a crash than a sober driver (NHTSA 1985). FARS indicated that 41 percent of the traffic fataities in 1986 invoved either a driver or a pedestrian with a BAC of 0.10 or greater. This percentage transated to 18,890 fataities. An additiona 11 percent (5,100 fataities) invoved a driver or pedestrian with some acoho (BAC = ). Ony 48 percent of the fataities invoved a drivers and pedestrians with zero acoho. Acoho invovement did vary by time of day, day of week, and type of crash (tabe 1). Seventy-seven percent of the fata crashes that occurred between 8 p.m. and 4 a.m. on any night of the week invoved acoho. Acoho was aso much more prevaent in singe-vehice crashes than mutipe-vehice crashes. Amost haf the coisions resuting Tabe 1. Acoho invovement in fata crashes: 1988 Crashes Fata Daytime (4 a.m. - 8 p.m.) Nighttime (8 p.m. - 4 a.m.) Weekday Weekend (8 a.m. Fri - 4 a.m. Mon) Singe vehice Mutivehice Nonoccupant (pedestrian/bicycist) BAC and higher N (pzznt) (percent) (percent) 41, , , , , , , ,

5 EPIDEMIOLOGY AND DATA MANAGEMENT 39 Tabe 2. Drivers and nonoccupsnts (pedestrians/bicycists) invoved in fsu crsshes: BAC and higher N (percent) (percent) (percent) A drivers 60, Driver fataities 26, surviving drivers 33, Nonoccupant fataities 7, Mae drivers 46, Femae drivers G in a nonoccupant (pedestrian or pedacycist) death invoved acoho, mosty on the part of the pedestrian. When examining data for a drivers invoved in fata crashes, keep in mind that in mutipe-vehice crashes, at east two drivers are invoved in one crash. In 1986,60,297 drivers were invoved in the 41,067 fata crashes. Twenty-six percent of these drivers were egay intoxicated (BAC greater than or equa to 0.10) at the time of their crashes (tabe 2). Of the 26,613 drivers who were kied in their crashes, 39 percent were egay intoxicated compared with ony 15 percent of the drivers who survived fata crashes. Mae drivers were amost twice as ikey to have over 0.10 BAC at the time of their crashes as femae drivers (28 percent versus I.5 percent). Acoho invovement did vary substantiay by driver age in 1986 (tabe 3). Whie 21 percent of teenage drivers were egay intoxicated at the time of the crash, an additiona 13 percent had aso been drinking. Drivers years od had the highest acoho invovement rate: 47 percent. In contrast, ony 7 percent of drivers age 65 and oder were egay intoxicated at the time of their crash. Examining certain combinations reveaed that whie amost two-thirds of the fata Tabe 3. Acoho invovement by driver age, 1986 Driver s age 0.00 (percent) Driver s BAC O.O and higher (percent) (percent) N* , , , , , and oder ,881 A ages ,297 N = Number of drivers in age group where age was known.

6 40 BACKGROUND PAPERS Tabe 4. Drivers invoved in fata crashes, 1986 N iac - O.O-.09 (percent) 0.10 and higher (percent) Mae/weekend/night Femae/weekday/day 6, Driver age groups , , and oder W@ crashes invoving a mae driver on a weekend night were acoho reated, ony 10 percent of the crashes invoving a femae driver in a weekday crash in the daytime invoved acoho (tabe 4). Acoho invovement was aso found to vary consideraby by the type of vehice driven (indicating the type of driver, in most cases) (tabe 5). Drivers of motorcyces invoved in fata crashes had by far the highest acoho invovement rate: 54 percent. Ony 3 percent of heavy-truck drivers invoved in fata crashes had BACs over Drivers of oder vehices were more often egay intoxicated than drivers of newer vehices (34 percent versus 22 percent). Intoxicated drivers in fata crashes aso tended not to use safety bets. Of the fatay injured drivers who were at zero acoho, 20 percent were wearing safety bets compared with ony 7 percent of. the fatay injured drunk drivers. Thirty-six percent of the zero-acoho surviving drivers were reported as using bets, in contrast to ony I.5 percent of the intoxicated surviving drivers. Contrary to some popuar misconceptions, the victims of acoho-reated fata crashes are most often the drinking driver or drinking pedestrian. Two-thirds (66 percent) of the 23,990 victims of acoho-reated crashes in 1986 were the drinking driver or driig pedestrian (tabe 6). An additiona 28 percent of the victims were passengers in the Tabe 5. Drivers invoved in fafa crashes, 1986 BAC Drivers of: and higher N (percent) (percent) (percent) Motorcyces 4, I.3 Passenger cars 35, Light trucks and vans 11, Medium trucks Heavy trucks 4, Oder vehices (oder than 1976) 13, Newer vehices ( )

7 EPIDEMIOLOGY AND DATA MANAGEMENT 41 Tabe 6. Acoho-invoved fata crash victims, 1986 N Percent Drinking drivers kied 13, Drinking pedestrians and pedacycists kied 2pJ 11 Passengers in drinking driver s vehice kied 4, Sober drivers kied in crash with drinking driver s vehice 1,680 7 Passengers in sober driver s vehice kied in crash with driiig driver Sober pedestrians/pedacycists kiied by drinking driver Tota =,=J 100 drinking driver s vehice. Additiona anayses reveaed that in fata crashes where BACs were known for drivers and their passengers, 36 percent of the time the driver was egay intoxicated but the passenger was not. Tabe 7 shows the basic trend with regard to the acoho probem in fata crashes over the past 5 years. The percentage of drivers in fata crashes who were intoxicated (BAC = 0.10 or greater) at the time of the crash decreased from 30 percent in 1982 to 26 percent in 1986-a Gpercent reduction, which is substantia. The reduction was especiay great for teenage drivers (tabe 8). Whie 29 percent of the teenage drivers in 1982 were egay intoxicated, this amount dropped to 21 percent in 1986, a 28percent reduction. Whie this teenage driver trend is encouraging, one must sti keep in mind that teenage driver invovement in fata crashes per mie driven is substantiay higher than other driver age groups (Fe 1987).. The nature of this S-year acoho reduction trend was examined in the foowing manner. Specific decreases of certain types of drivers and certain types of crashes were compared with the overa reduction. If these specific reductions were substantiay greater than the overa reduction, then that woud indicate that these drivers or conditions were affected most, Figure 1 summarizes the key fmdings concerning the nature of the reduction. The argest reductions noted were for teenage drivers (28 percent), foowed by teenage pedestrians kied in coisions (26 percent). Aso affected were drivers of vans (23 percent reduction), femae drivers (21 percent), and drivers who survived the fata Tabe 7. BACs for a drivers invoved in fata crashes, (in percents) BAC change and higher N 56,029 54,656 57,512 57,883 60,297

8 42 BACKGROUND PAPERS Tabe 8. BACS for teenage (&19) drivers Invoved in fata crashes (m percents) \ - BAC change % % and higher N 7,467 7,050 7,366 7,151 7,854 crashes (17 percent). The absoute reduction was aso arger in weekday crashes (17 percent) and in mutivehice crashes (16 percent). Drivers age had ony a sight reduction during this S-year period (6 percent). Motorcyce drivers, with the highest percentage of acoho invovement to begin with, experienced no change in the percentage of drivers egay intoxicated during this period. Pedestrians age aso had no reduction in the percentage egay intoxicated between 1982 and Late night crashes and singe-vehice crashes showed ony modest reductions in the percentage of drivers who were at 0.10 BAC or higher (6 percent and 9 percent, respectivey). The average BAC of drinking drivers in fata crashes in States where most of the drivers were tested showed a modest decrease from in 1980 to in Acoho consumption per capita decreased in the United States between 1982 and But if that decrease was a prime factor in the decreased acoho invovement of drivers in fata crashes, then a simiar reduction shoud have occurred in intoxicated adut pedestrians in fata crashes, which was not the case. A drivers Teen drivers Van drivers % I -1 28% 1 123% Femae drivers I t 21% Surviving drivers II 21% Teen pedestrians r 126% Daytime crashes (6 a.m.-6 p.m.) I : 1 17% Weekday crashes Mutivehice crashes I % B 1 16% Figure 1. Nature of acoho reduction in fata crashes, Decrease in percentage drunk (BAC = 0.10 or higher)

9 EpDEMOLOGY AND DATA MANAGEMENT 43 Tabe 9. Nature Of acoho reduction among fatay injured &iv- with known BACs in 15 good reporting states* (percents) BAC of fatay injured drivers change m m m *Test and report BACs on 85 percent of fata drivers. The nature of the reduction of acoho in fata crashes does seem to point to main effects in responsibe socia drinkers, i.e., substantia reductions in daytime crashes, by femae drivers, drivers of vans, and teenagers. However, there is some evidence that the percentage of drivers with very high BACs is aso decreasing, at east in the 15 goodreporting States in FARS. Tabe 9 shows that, in 1980, amost a quarter (24 percent) of the fatay injured drivers in these States had BACs of 0.20 or greater. That portion in 1986 was 18 percent, which was a 25-percent reduction-greater than the reduction for the drivers at BACs between 0.10 and Most researchers woud agree that drivers at 0.20 BAC or greater are most ikey probem drinkers or acohoics. Yet the percentage of drivers at these eves has decreased significanty since 1980 (in that 5-State sampe). Are these probem drinkers finding aternatives to driving? Are they confining their drinking to their homes? Have many of them stopped drinking? More research is necessary to answer these important questions. Acoho in Noncrash Drivers Accurate determination of acoho actuay present in drivers whiie they are using the highways can be estimated ony by obtaining measurements from sampes of these drivers at roadside. (Thus, sef-reported drink&-and-driving data from teephone or househod surveys are not considered here.) Measurement of acoho in noncrash drivers is generay obtained at roadside for four major purposes: 1. To estimate the contribution of acoho to crash risk 2. To provide data for describing a particuar probem by identifying and specifying reevant parameters 3. To provide data for evauating the resuts of any changes in circumstances surrounding the particuar probem, whether they resut from unpanned natura events or from controed countermeasures 4. To foster genera deterrence of drunk driving and to enforce DUI aws Research designed to accompish the first purpose invoves case-contro studies. Activities designed for the fourth purpose are currenty referred to as either enforcement checkpoints or sobriety checkpoints. Studies designed for the second or third purpose have a broader range of objectives. Usefu epidemioogic data can be obtained from activities designed for any of these four purposes, but the most fundamenta question is addressed in investigations of acoho and crash risk by means of case-contro studies.

10 44 BACKGROUND PAPERS Case-Contro Roadside Surveys and Acoho Crash Risk That acoho is found in approximatey 50 percent of fatay mjured drivers tested does not necessariy prove that acoho actuaiy contributed to the occurrence of these crashes. To begin buiding a case for or against the actua contribution of acoho, it is first necessary to determine the extent to which fatay injured drivers with acoho are representative of drivers with simiar ekosure, but not invoved in the crashes. Thus, it is necessary to compare the distribution of BACs obtained from contro or comparison drivers randomy seected whie passing the same pace as the crashes and at equivaent times. By comparing these two sets of data, it is then possibe to determine the simiarities and differences between the two sets of drivers in terms of the percentages of each with no acoho, with detectabe acoho, with medium BACs, with high BACs, and so forth. For exampe, a number of studies have indicated that between 40 and 50 percent of fatay injured drivers examined.had BACs of 0.10 or higher. If we had been abe to examine the other motorists who were actuay driving at the s&e times and paces that these fata crashes occurred, and if we had found that about 45 percent of these noncrash-invoved drivers aso had BACs of 0.10 or higher, then we woud have no basis for concuding anything at a about the contribution of acoho to highway crashes. That is, the percentages of high-bac drivers in the fatay injured sampe woud have been the same as the percentage of high-bac drivers in the comparison sampe from the popuation-at-risk-namey, about 45 percent. Therefore, in this hypothetica instance, high-bac drivers woud have been neither under- nor overrepresented in terms of the percentage of the popuation-at-risk made up of high-bac drivers, namey, about 45 percent. Conversey, if we had found a significant difference between the percentage of high-bac fatay injured driers and the percentage of high-bac contro drivers from the same popuation-at-risk, then we woud be abe to make some strong inferences about the reative contribution of acoho to these fata crashes. Thii ine of reasoning provides the ogica basis for attempting to obtain these BAC data from the popuation-at-risk using the case-contro design with roadside research surveys. The first such study was conducted in Evanston, Iinois, 50 years ago by Hocomb (1938), and severa more studies have been conducted in the United States and abroad since that time. These case-contro studies have been anayzed from a variety of perspectives and summarized in a number of pubications (Hurst 1973, 1985; NHTSA 1985; Perrine 1975u, b; Reed 1981; Zyman 1971). However, the materia that foows in this subsection is taken primariy from the most recent review (NRC 1987). In ai these reviews, a consistent pattern is reveaed by the case-contro studies: crash risk increases sharpy as BAC rises. The reative probabiity of being invoved in a crash is defined as the ratio of the BACs of comparison drivers to those of drivers invoved in crashes. This probabiity remains roughy equivaent for crash-invoved drivers compared with noncrash drivers up to about 0.08 BAC (figure 2). (However, these reative risk curves understate the risk of invovement at ow BACs.) Athough the rate of increased risk varies across studies (m part because some studies examine a crashes and some examine ony fata crashes), the. risk increases after about 0.08 BAC in a cases and increases dramaticay after 0.10 BAC in most studies. The curves depicted in figure 2 are based on groups of drivers of different ages who have varying experience with acoho and with driving. Because of the heterogeneity of contro groups and the ack of perfect comparabiity, the effect of acoho at ow BACs is masked by other variabes. For exampe, the major shortcoming of the Grand Rapids study (among the most cited case-contro studies) is the ack of comparabiity between the drivers invoved in crashes and the contro drivers regarding the frequency of consuming acoho. This ack of comparabiity is the source of the apparent improvement in crash risk at ow BACs in the Grand Rapids data (the much debated Grand Rapids

11 EPIDEMIOLOGY AND DATA MANAGEMENT Grand Rapds data (5,965 tota crashes) Grand Rapds data (300 tota or serous crashes) Evanston data (270 Injury crashes) Toronto data (423 tota crashes) Manhattan data (34 fata crashes) Estimated approxmate extenson of Manhattan trace Vermont data (106 fata crashes) Huntsvie data (615 Injury crashes) Adeaide data (299 Injury crashes) 45 : : : : : : : : : : : O BAC %WN Figure 2. Reative probabiity of crash invovement as a function of BAC. (Hurst 1985.) Reprinted with permission. dip ) and the understatement of the risk of crash invovement at ow BACs. Hurst (1973) noted that the contro group had a higher percentage of drivers who were reguar consumers of acoho. Their apparenty greater toerance for acoho had made them safer drivers at ow BACS than the drivers invoved in crashes at ow BACs, presumaby because the atter had ess experience as drinkers. Hurst recacuated the reative risk of crash invovement in the Grand Rapids data based on the drivers sef-reported frequency of acoho consumption (figure 3). He drew three concusions from the resuts. First, drivers with frequent experience as drinkers are ess ikey to be invoved in crashes than ight and medium drinkers at comparabe BACs. Second, regardess of the toerance for acoho, the risk of crash invovement increases with BAC. Third, the curves greaty underestimate the risk for the average driver at any BAC; they ony demonstrate the reative hazard to drivers who reguary drink and drive. The curviinear reationship between reative risk of crash invovement and BAC is therefore caused in part by the comparison of drivers with varying degrees of experience as drinkers and experience driving under the infuence of acoho. When experience with acoho is controed for, the risk of crash invovement increases with BAC without evidence of a threshod effect. As noted by Perrme (197%) in his review of the iterature, the reative risk of invovement is not the same as evidence of causaity. Given the many interacting factors that may contribute to a crash (and the ack of data on many of them), the roe of atry singe factor is difficut to isoate. Three of the case-contro studies deserve specia attention because they aso estimate the effect of acoho on the probabiity of being responsibe for a crash. The methodoogy for estimating crash responsibiity was first deveoped by McCarroU and Haddon (1%~) in their case-contro study of fata crashes in Manhattan. They

12 6ACKGFIOUND PAPERS SELF-REPORTED DRINKING FREQUENCY a. YEiRLY OR LESS MONTHLY -=. - - WEEKLY --- BXNVEEK - DAILY i I 1 I I I I BAC, %W/V Figure 3. Reative probabiity of crash invovement by sef-reported drinking frequency (Hurst 1973). Reprinted with permission from thejoorna/ofsafetyresearch, a joint pubication with the Nationa Safety Counci and Pergamon Press, Ltd. categorized the crashes into five casses, the first three of which were assigned responsibiity: 1. Ony one vehice invoved 2. Two vehices invoved but ony one moving 3. More than one vehice invoved and in motion, with responsibiity assigned based on circumstances of the crash (cases in which there was any doubt were excuded from this category). The Manhattan study is based on a sampe of 43 drivers fatay injured in crashes that occurred between June i950 and June For the 26 drivers in the assigned responsibiity casses, 19 (65 percent) had positive BACs, of which 14 (46 percent) had BACs greater than Of these 14 drivers, 12 had BACs of 0.25 or greater. Of 156 drivers randomy seected as contros at or near the sites of the crashes, 39 (25 percent) had positive BACs, of which ony 8 (5 percent) were at or above The Grand Rapids investigation invoved by far the argest sampe of a the case-contro studies (5,985 crashes of a types) (Borkenstein et a. 1964,1974). By comparison, the 423 cases in the Toronto study constituted the next argest sampe (Lucas et a. 1955; Hurst 1985). Using McCarro and Haddon s method for assigning responsibiity, Borkenstein, Crowther, Shumate, Zi, and Zyman estimated that 3,305 of the invoved drivers were responsibe for the crash that occurred. They used the innocent drivers as contros. The Vermont study was the third to estimate crash responsibiity, based on 106 cases (a fata crashes) (Per&e et a. 1971). These crashes resuted in 113 fataities, and 97 of the drivers were assigned responsibiity, again reying on the method deveoped by McCarro and Haddon. Of the drivers judged responsibe, 60 percent had positive BACs and 46 percent had BACs at or above Perrine, Waer, and Harris (1971) aso cacuated a crash-responsibe curve, but in contrast to the Grand Rapids study, the drivers stopped at roadbocks were used as contros.

13 EPDEtdCL~Y AND DATA MANAGEMENT U GRAND RAPIDS DATA CRASHES. DRIVER ASSUMED RESPONSIBLE - MANHAT AN DATA, 24 FATAL CRASHES, DRIVER ASSUMED RESPONSIBLE o----q VERMONT DA-t-As 75 FATAL CRASHES. DRIVER ASSUMED RESPONSIBLE 1 I I I I I I I I O AC % (WA ) OF DRIVER Figure 4. Reative crash responsibiity for drivers assumed responsibe and those not assumed responsibe as a function of BAC, where 1.O = reative probabiity at zero acoho. (Hurst 1973.) Reprinted with perniission from thejourna/ofsafefyreseafch, a joint pubication with the Nationa Safety Counci and Pergamon Press, Ltd. Hurst (1973) repotted the curves from these three studies on a ogarithmic scae to faciitate comparison (figure 4). Athough risk of crash responsibiity increases as BAC increases in a three studies, severa disadvantages with the underying data shoud be noted. The trend in the Manhattan data is based on a very sma number of crashes: 25 responsibe drivers with positive BACs. In addition, the trend at the higher BACs is greaty understated. For the fata crashes in which the driver had a BAC of 0.25 or higher (about haf those in the driver-responsibe category), no driver in the contro group had an equivaent BAC. Hence, the reative hazard cacuated from the case/contro ratio woud be infinite within the range, were it possibe to graph it (Hurst 1973). One of the shortcomings of the reative risk curve estimated in the Grand Rapids study is the incusion of drivers invoved in singe-vehice crashes in the group of responsibe drivers. Athough the responsibiity of the driver is not in question, because of the nature of the crash, a contro driver is not avaiabe. The pubished data do not provide sufficient detai to aow the curve to be compieteiy recacuated without the singe-vehice crashes to determine the effect of incuding these crashes, but the avaiabe data suggesthat the curve woud shii to the right. It woud sti acceerate after 0.04 BAC and at an exponentia rate, but the curve woud not rise as quicky as shown in figure 4.

14 48 BACKGROUND PAPERS One probem with the Vermont data is the sma number of crashes in the sampe. In the comparison of crash risk as BAC rises, one. or two drivers are responsibe in some of the BAC ranges. Chance occurrence coud distort the resuts when-so few drivers / are the basis of the cacuations. Despite the weaknesses in the case-contro studies, some important concusions can be drawn. In severa of the case-contro studies done in the United States and abroad, a consistent increase in risk of crash invovement has been shoti. When experience with drinking is controed for, this risk increases with BAC without any evidence of a threshod effect (or dip). The three studies that attempted to estimate crash responsibiity showed that the risk of causing a crash increases even more rapidy than the risk of crash invovement as BAC increases (NRC 1987). Another important aspect of the acoho contribution to crash risk is reported by Voas (NHTSA 1985). To emphasize the signiicance of the difference in BAC between drivers assumed to be responsibe versus those assumed not to be responsibe for crashes, Hurst (1974) aso presented an additiona cacuation on the data from the Grand Rapids study. His resuts for the drivers assumed to be responsibe are represented by the center pot in figure 4. However, the probabiity of being innocenty invoved in a crash remains essentiay eve and does not increase with increasing BAC, the pot is basicay fat and woud ie between the reative crash probabiity of 1 and 2 in figure 4. The same resut was found in the Huntsvie/San Diego study by Farris, Maone, and Liiefors (1977). These resuts provide further evidence for the causa roe of acoho in crashes. Other Roadside Research The success and utiity of the case-contro procedures for investigating acoho crash risk stimuated interest in using the roadside survey technique for evauating acoho safety programs by measuring the change in the number of high-bac drivers actuay on the roads. Standardized procedures for conducting roadside surveys were deveoped (Perrine 1971) and appied successfuy in 28 of the 35 Acoho Safety Action Projects (ASAPs) funded by the bepartment of Transportation between 1970 and 1975 (Voas 1972; Carr et a. 1974). In these programs, the roadside surveys were used to evauate project effectiveness (Levy et a. 1978) by serving as a means to coect data used as the primary criterion or dependent variabe (BAC). Roadside surveys were conducted before and after program impementation to measure the change in average driver BAC (ii any) resuting from project activities (Lehman et a. 1975). When used for program evauation, samping was conducted during periods when a high percentage of drinking drivers was on the road (i.e., Friday and Saturday nights) rather than at times and paces at which accidents had occurred. By a return to the same sites, changes over time can be measured. Roadside surveys provide a more direct method of evauating acoho safety countermeasure programs than does the use of accident data, because highway crashes resut from a arge number of factors (weather, roadway construction, economic conditions, etc.) that are unreated to the evauation of enforcement activities. The BAC vaues of drivers serve as an intermediate measure between action programs and the utimate criterion of accident prevention. Whie a reduction in the average BAC of drivers on the road does not guarantee a reduction in crashes, the reationship between driver BAC and risk of crash invovement is cose enough to make this measure a credibe criterion for program effectiveness. During the ASAP period (1970 through 1974), some 77 roadside breath-testing surveys of nighttime drivers were conducted. In addition, a nationa roadside breathtesting survey was conducted in 1973, and a computer archive of these 78 roadside surveys is stored at the University of Michigan Highway Safety Research Institute (Lehman et a. 1975). The tie contains breath-testing resuts, demographic data, and so forth, for some 78,000 randomy seected drivers, as we as 2,700 passengers. Anaysis

15 EPIDEMIOLOGY AND DATA MANAGEMENT 49 of these aggregated data show the foowing percentages of drivers with BACs at or exwe&mg 0.10: 1 percent of weekday-eary drivers, 3 percent of weekend-eary drivers, and 6 percent of weekend-ate and weekday-ate drivers. Significant reductions in the percentages of drivers above the ega imit (0.10 BAC) were demonstrated for those jurisdictions that used this evauation method (Levy et a. 1978). Based on the utiity of this roadside BAC measure in the ASAP program, it was appied again in a 4-year study of a specia DUI enforcement effort in Stockton, Caifornia (Voas and Hause 1987). In this study, survey procedures were modified to permit ow cost and ow profre surveys that were conducted every weekend for 3 k years (Hause et a. 1982). Drivers with a BAC of 0.10 or greater on Friday and Saturday nights decreased from 88 per thousand before the Stockton project to 50 per thousand during the third year. The roadside survey technique aso permits (through an appication of Bayes Theorem) estimation of the probabiity that a driver at a given BAC wi be arrested by the poice. This procedure was first appied by Beite, Sharp, and Gauz (1975) to the ASAP survey data in Kansas City. Hause, Voas, and Chavez (1982) used the same procedure in Stockton. These studies provided roughy simiar resuts indicating that the chances of being arrested at a BAC of 0.15 is roughy 1 in 100, whie the chance of arrest at 0.10 BAC is haf that amount, about 1 in 200. Since both these studies invoved intensive enforcement programs, they provide a reasonabe indication of the maximum arrest rate that can be achieved with traditiona patro methods. The ASAP experience with roadside research surveys and with the success of the manua for conducting and evauating them (Perrine 1971) provided the basis for subsequent internationa activity. An invitationa internationa workshop was conducted in Paris in an attempt to coordinate the methodoogy for roadside research surveys to be impemented in other countries in order to maximize the comparabiity of the obtained data. The workshop resuted in a usefu manua (Carr et a. 1974) and compemented parae activities being conducted under the auspices of the Organization of Economic and Cooperative Deveopment. As a resut of these activities, use of roadside surveys for internationa comparisons of countermeasure programs was stimuated in Canada, the Netherands, Norway, Sweden, and Finand. The resuting data permitted an internationa comparison of driver BACs (Voas 1982), which indicated that approximatey 12 percent of drivers on weekend nights were at or above 0.05 BAC in Canada, the Netherands, and the United States, whereas ess than 2 percent of drivers were at this eve in Scandinavian countries. Athough use of roadside research surveys has diminished in the United States since the end of the ASAP activities in the mid-197os, the technique continues to be used effectivey in other nations, for exampe, Canada, Austraia, Denmark, and Norway. Nevertheess, a few studies of the popuation-at-risk in the United States either have been conducted recenty or are currenty being conducted. In the spring of 1986, U.S. Nationa Roadside Breathtesting Survey II (Wofe 1986) was conducted in a representative sampe of 32 ocaities, 18 of which had participated in the 1973 U.S. Nationa Roadside Breathtesting Survey I (Wofe 1974). Statisticay significant reductions were found in the percentage of medium and high BAC drivers samped at high-risk times (Friday and Saturday nights from 10 p.m. to 3 a.m.). Drivers at or above the iega BAC of 0.10 decreased from 5.0 percent in 1973 to 3.1 percent in 1986, drivers at or above a BAC of 0.05 decreased from 13.5 percent in 1973 to 8.3 percent in It shoud be noted that the breathtest competion rates were 86 percent in 1973 and 92 percent in 986. In Vermont, a arge-scae roadside research study invoving a projected 42,000 nocturna drivers samped at high-risk times (Friday and Saturday nights from 10 p.m. to 3 a.m.) is currenty being conducted. It is funded by the Nationa Institute on Acoho Abuse and Acohoism (NIAAA, Grant AAO7876). This 5-year fied study is primariy

16 50 BACKGROUND PAPERS designed to determine the prevaence of drivers with high acoho toerance, and to determine their saient and differentiating characteristics. Since the high acoho toerant driver is apparenty rare, a arge number of motorists (42,ooO) driving at high-risk times must be stopped and screened for BAC to identity a sufficient number of such peope (40 to 60) to be abe to conduct a meaningfu study. In the process of conducting this fied study, a arge number of peope wi be breathtested using both the new passive acoho sensor and the more traditiona hand-hed evidentiary devices. Approximatey 4,000 of these motorists, samped across the fu distribution of BACs, wi participate in extensive.persona interviews concerning sef-reported background data; drinking, driving, drinking-and~driving, drugs-and-driving information and attitudes; and seected personaity characteristics. Data wi aso be gathered on these motorists driver records, performance on the most vaid fied sobriety tests (gaze nystagmus, wak-and-turn, and standing steadiness), and ratings on cinica signs of intoxication. With a test competion rate of % percent, the resuts from the first 650 drivers indicate that 3.7 percent had a BAC of 0.10 or higher, whereas 10.2 percent had a BAC of 0.05 or higher. Athough the sampe size issti reativey sma, these 1988 data show a decrease in distribution of BAC when compared with data obtained in a 1974Vermont study (Perrine 1976) of 1,663 drivers at high-risk nocturna times (Thursday, Friday, and Saturday nights between 1030 p.m. and 3 a.m.): 4.6 percent had a BAC of 0.10 or higher and 14.7 percent had a BAC of 0.05 or higher. Thus, these roadside studies of the high-risk popuation woud seem to show that some progress is being made in the war on drunk driving, if motorists with BACs in excess of the ega standard (0.10) are taken as the criterion. Enforcement Checkpoints Poice officers conduct sobriety checkpoints at which they stop motorists at random and test them for breath acoho. Such activities are conducted primariy for enforcement purposes, athough they aso serve as genera deterrence. Athough usefu data for epidemioogic purposes are avaiabe from these enforcement checkpoints, few systematic studies have been conducted to anayze such data. If they were anayzed carefuy and propery, these data coud provide a vauabe source of reativey ow-cost information concerning the popuation-at-risk. In a recent Charottesvie, Virginia study, Voas, Rhodenizer, and Lynn (1985) evauated the effectiveness of such sobriety checkpoints, especiay in comparison with the drivers arrested for DUI by traditiona roving patros. In addition, thii study found evidence of poice biases in those arrested, whereby both young drivers and women were underrepresented among arrested drivers, whie minority and very high BAC drivers were overrepresented. Thus, such studies ceary demonstrate that researchers can avai themseves of enforcement checkpoints as an opportunity to coect vauabe data for epidemioogic purposes. Characteristics of Drunk Drivers During the past 20 years, numerous statistica and cinica studies have been pubished on various aspects of the drinking driving probem. However, surprisingy itte rigorous research has been pubished on characteristics of convicted DUI offenders, particuary when contrasted with the vast iterature on probem drinkers/acohoics, and on characteristics of drivers invoved in fata accidents. As suggested by Zyman (1974) and by Moskowitz, Waker, and Gomberg (1979), the popuation arrested and convicted of DUI offenses is not typica of impaired drivers in genera or of drivers invoved in acoho-reated accidents. The mean BAC of convicted DUI offenders in Caifornia during 1984 was 0.18-a concentration far in excess of the State s 0.10 imit per se, and we beyond the eve at which impairment occurs. In one of the few forma statistica studies of differences between acoho-invoved fata accident drivers and convicted DUI offenders, Fridund and Hagen (1977) used discriminant function anaysis in comparing 146 DUI offenders in Los Angees County with a sampe of 191 acoho fataities. The DUI conviction group had significanty more prior DUIs, more prior reckess-driving

17 EPIDEMIOLOGY AND DATA MANAGEMENT 51 convictions, and more prior moving t&tic convictions. The differences on the incidence of DUIs and reckess convictions was arge, with the DUI group having about three times as many entries during the prior 3-year period. Moskowitz, Waker, and Gomberg (1979) conducted a detaied review of the iterature on DUI offender characteristics. This subsection reies heaviy on their monograph for the pre-1979 iterature, but concentrates on recent studies and studies not incuded in the 1979 review for its primary source references. However, a few pre-1979 studies of specia importance are reviewed here as primary references even though they are aso incuded in Moskowitx et a. (1979); Moskowitz, Waker, and Gomberg organized their review by type of offender characteristic (prior driver record, age, etc.), and reached the foowing concusions with respect to each domain: Marita status: DUI con&tees are much more ikey to be divorced, separated, or widowed than are non-dui contro popuations. Some studies have reported five- to sixfod differences in rates compared with contro popuations. Empoyment history: Convicted DUI offenders are more ikey to be unempoyed, with rate differentias ranging from two- to fourfod higher across various studies. Occupation: Convicted DUI offenders are more ikey to have ower status occupations. DUI offenders in bue coar jobs averaged 65 percent across studies compared with-5 percent for contro sampes. Income: Convicted DUI offenders tend to have ower incomes- about 18 percent ower than contros across the reviewed studies. BAC: The mean BACs for the offenders averaged from 0.18 to (Statewide Caifornia figures have consistenty averaged 0.18.) Drinking behavior: DUI offenders drink more often and consume more acoho per sitting than do non-dui popuations. Beer is the preferred beverage of DUI offenders. The findings of the Southern Caifornia study by Poack (1%9) are typica. Poack reported that 18 percent of DUI drivers drank every day compared with 11.5 percent for a contro sampe. In terms of drinks per sitting, 35.2 percent of the DUI sampe typicay consumed five or more drinks compared with 5.7 percent of the contro sampe. Reason for drinking: Convicted DUIs (and acohoics) are more ikey to drink to reease tension and to cope with stress. Probems caused by drinking: Convicted DUIs are much more ikey than contros to exhibit poor heath, famiy disorganization, financia probems, and poor job performance. Prior acoho treatment history: Convicted DUIs are more ikey than contros to have previousy entered some form of acoho treatment program. The median across 20 studies was 6.0 percent, with a maximum of 42.5 percent. (Thii characteristic is highy dependent on the institution and deivery systems of a particuar region, and woud be expected to vary across jurisdictions and over time.) Probem drinking status: Studies using the Michigan Acohoism Screening Test (MAST) indicate that percent of convicted DUIs fa in the piobemdrinking and acohoic range. Studies using the Mortimer-Fikins test produce sighty ower prevaence figures. Driving after drinking: Convicted DUIs are much more ikey than contros to drive after drinking. Poack (1%9), for exampe, reported that 49 percent of DUI offenders admitted to driving at east once a week after two drinks compared with 12 percent of a contro sampe.

18 52 BACKGROUND PAPERS a Tota prior- arrests: Convicted DUI offenders are more ikey to have prior arrests for both acoho and nonacoho offenses. Driving history: Convicted DUI offenders have substantiay more driving record entries of a types than do contros (more DUIs, more acoho and tota accidents, more moving traffic vioations, and more icense actions). The rate increases across studies and variabes range from 100 percent to 500 percent. The driving record histories of convickd DUIs are aso substantiay worse those of medicay diagbosed acohoics. Personaity traits: Convicted DUIs have a significanty higher prevaence of personaity trait disorders. They are more ikey than contros to exhibit neuroticism, depression, paranoid ideation, ow sef-esteem, and to have a ower sense of persona responsibiity and contro and greater. feeings of aggression/hostiity. Stress: Convicted DUI offenders are more ikey to report experiencing.i stress from famiy, financia, and job probems. Education: Convicted DUI offenders are more ikey to be high schoo dropouts and have fewer years of education. \ Age: Convicted DUI offenders tend to be sighty oder than non-dui contros, with the highest disproportionate concentration in age interva 304. Race: Most convicted DUI offenders are white, but minority groups (hispanics and backs) are overrepresented compared with their representation in the popuation. Sex (not covered by Moskowitz): The great majority of convicted DUI offenders are mae. The range for femaes is from 5 to 20 percent, depending on the characteristics of the specific DUI popuation (first versus repeat offenders), region, and so forth. A arge statewide samping in Caifornia indicates that 13 percent of those convicted for a DUI offense in 1982 were femae (Tashima and Peck 1986). Among fust and second offenders, femaes accounted for 17 percent and 10 percent, respectivey. Drinking Status It is cear from the above summary that persons convicted of drunk driving offenses deviate greaty from the genera driving popuation on a wide variety of characteristics. That DUI offenders contain a disproportionate number of probem drinkers woud be expected, of course, since the offense of drunk driving is, per se, a probem associated with the consumption of acoho. The number of drinks required to produce manifesty detectabe impairment in driving and the BACs typicay attained by DUI offenders impies a eve of acoho consumption that is statisticay deviant. Given the very ow probabiity that any given incident of impaired driving wi resut in detection (arrest or accident), the percentage of DUI offenders who were simpy unucky, in the sense of getting caught in a rare instance of impaired driving, woud be reativey sma. One woud therefore expect most DUI offenders to be heavy consumers of acoho. Most of the empirica iterature, and most authorities in the area, agree with this concusion, athough controversy has deveoped over the percentage of DUI offenders who are acohoics in the cinica disease context. This controversy stems more from semantic and epistemoogica compexities than from disagreements over data, and is not pursued here. Lest the impression be created that opinion and data are unanimous on the drinking status of DUI offenders, the resuts of a recent Caifornia study of first offenders wi be summarized in detai. This study was carried out by the Pacific Institute for Research

19 EPIDEMIOLOGY AND DATA MANAGEMENT and Evauation (PIRE); it was commissioned by the State of Caifornia pursuant to Assemby Bi 3405 (Stewart et a. 1987). The major objective of,the PIRE study was to deveop a mode curricuum and rehabiitation program for use by courts in sentencing fmt-offender DUI cases. HOWever, this review wi ony consider that component of the study pertaining to offender characteristics (natura variation component). Detaied biographica, drinking habit, and arrest-incident information &as coected by questionnaire on 5,052 respondents from 26 frst-offender treatment programs throughout Caifornia. The authors reported the foowing statistics from an anaysis Of the questionnaire responses: Median number of drii on day of arrest: 6 Median BAC upon arrest: 0.16 Median number of days in past year with four or more drinks: 60 0 Median number of days in past year with eight or more drii 4 Percentage who did not fee intoxicated upon arrest: 35 percent 0 Median number of previous days in past year driven whie impaired: 1 0 Percentage with prior DUI arrests: 20 percent The authors categorized the drinking pattern responses into two typoogies for comparison with a statewide genera popuation survey. The more compex of the typoogics was a ir-point continuum: abstainers, infrequent drinkers, ight monthy, ight weeky, moderate weeky, and frequent heavy. Using a probit anaysis to adjust for p~~puation differences, the authors found no significant differences between the drmking frequency of the two groups after abstainers were removed from the genera popuation sampe. The difference was significant, however, in the frequency of heavy drinking (five or more drinks at east once a week), with substantiay more of the DUI offenders faing into that category. Nevertheess, fewer than 30 percent of the first offenders were paced in this category. In commenting on these findings, the authors concuded: Whie defining a typica pattern of drinking for a subjects is difficut, the median frequencies of use conditiona upon the eve of use at arrest is reveaing: 40 percent of the subjects reported having had 8 or more drii on the day of their arrest. The median frequency of use at this eve among these subjects was 14 occasions over the previous year and ony one occasion in the preceding 30 days. A pattern of use incuding drinking 8 or more drinks at east weeky over the previous year was reported by ony 20 percent of these subjects. Thus, for the majority of subjects, drinking at the eve of use at which they were arrested is reativey infrequent (pp )... first offenders are not unike the genera popuation of drinkers in Caifornia in terms of the typica frequency of use. However, it appears that the incidence of heavier drinking is greater among first offenders (p. 32) Thc authors aso incuded a measure of acoho dependency in their study. Subjects c a Z-item Acoho Dependency Scae (Skinner and Aen 1982), and the c0rcs werc compared with those of cinicay diagnosed acohoics. Ninty-two percent If hc subjects pioduced scores ~&&~g a ow eve of acoho dependency. The aut hors went on to concude the dramatic differences in these distributions suggesthat d pcndcnct symptoms among first offenders are quite ow, as compared to acoho rcjmcnt groups;) Thc resu*& and concusions of the PIRE study are at odds with prevaiing opinion id most Prior studies in this area. If the findings are accepted at face vaue, the great

20 54 BACKGROUND PAPERS majority of first offenders fa within the bounds of socia drinking Even the percentage characterized as heavy does not seem extreme. \ 1 There are severa possibe expanations. Fist, the PIRE study was imited to first offenders, and it is known that the percentage of probem drinkers among first offenders is ower than among repeat offenders. Second, probem drinkers and DUI offenders are known to understate their drinking, sometimes dramaticay. Stewart, Epstein, Greenewad, Laurence, and Roth (1987) acknowedge this possibiity and recommend caution in interpreting the study f&ings. Athough previous studies are aso subject to reporting biases, many have incuded cinician interviews, psychometric instruments empoying ie scaes, and a variety of pubic agency data. No evidence is presented in the PIRE study to indicate that procedura contros were used to minimize the tendency of peope to fake good or empoy various forms of sef-denia..s Third, the first-offender survey ony invoved offenders who were sentenced to an acoho program and who agreed to cooperate by competing the questionnaire. In Caifornia, 23 percent of first offenders are not assigned to programs. Approximatey I.5 percent of the sampe did not return a questionnaire. There are aso inconsistencies in some of the vaues derived from the sef-report. For exampe, it woud have taken more than a median of six drinks to produce a median BAC The fact that 35 percent of the subjects did not fee intoxicated when arrested and that ony 14 percent acknowedged being definitey intoxicated is cause for further suspicion. Finay, it is difficut to accept the median estimate of ony one incident of driving whie impaired in the previous 12 months. The suggestion is that many of the subjects were not being candid in their responses. The PIRE report aso contains a description of first offender biographica and socioeconomic characteristics. The statistics of interest are summarized beow: Mae: 81 percent Singe: 46 percent Divorced, widowed, or separated: 21 percent White: 68 percent Hispanic: 22 percent Back: 3 percent High schoo dropouts: 22 percent Median age: 30 Unempoyed or empoyed part-time: 30 percent Median income: $16,500 These demographic characteristics are reasonaby consistent with the portraya from the Moskowitz review, particuary when aowance is made for differences in time and region. The percentages for ethnic minorities are somewhat ower than woud be expected based on Caifornia ethnicity composition and prior evidence showing that some minorities (e.g., Hispanics) are overrepresented in DUI popuations. It is important to recognize that the PIRE sampe is imited to offenders entering first-offender programs and, within this subset, to those who returned the questionnaire. These factors coud ater the representativeness of the sampe. Prior Driving Record It is aso cear that DUI offenders have statisticay deviant driver records before their