WORLDWIDE DECLINE IN DRINKING AND DRIVING

WORLDWIDE DECLINE IN DRINKING AND DRIVING - 955 -

- 956 -

Worldwide Trends in Drinking and Driving: Has the Progress Continued? B. M. Sweedler Safety and Policy Analysis International, Lafayette, California, USA. Keywords Drinking and Driving, worldwide trends Abstract This paper will summarize the trends in drinking and driving in a number of countries around the world and the nature of the trends. The countries include: Germany, France, The Netherlands, Canada and the United States. The trend appears to be following recent history. Nothing dramatic seems to be occurring. In some countries, such as the United States and The Netherlands there were increases. In France and Germany there were decreases. Introduction In the decade of the 1980s, there were impressive declines in drinking and driving in much of the industrialized world. The declines included about 50% in the U.K., 28 % in Canada and The Netherlands, 32% in Australia, 37% in Germany and 26% in the U.S. These declines did not continue in the early part of the 1990s. In some countries, there were actually increases. Toward the middle and latter part of the decade the increases stabilized and we again began to see some decreases. However, these decreases have been at a slower rate than the dramatic decreases in the 1980s. At the end of the decade, some countries began to again see small increases. The information and data in this paper comes from papers prepared by researchers from Canada, France, The Netherlands, Germany, and the United States for presentation at T2002, the 16 th International Conference on Alcohol, Drugs and Traffic Safety in Montréal, Québec, Canada in August 2002. For additional detail about each country, you are encouraged to review each of the referenced papers. This is the fifth occasion where experts from around the world met to continue discussions began in 1993. The reasons for the changes that occurred were discussed and were published in a special report (1). The results of the continued discussions in 1995, 1997 and 2000 were also published (2,3 and 4). Methods Crash and survey data was analyzed by researchers in the countries listed. Because of different methods used to collect, record and analyze this data in the participating countries, it is not possible to draw comparison between countries. However, the results in each country were compared to historical data in that country to develop trends in drinking and driving. The results of these analyses is summarized. Where it was possible, a number of the researchers also - 957 -

discussed the reasons for what was occurring. The programs in place to combat drinking and driving and their effectiveness is reviewed, where applicable. Results Canada In 1981, 62% of the drivers killed in Canada had been drinking. By 1989, this figure had reached a low of 44%. This represents a 29% reduction in the magnitude of the alcohol-fatal crash problem. Importantly, from 1981 to 1989, the relative change in the number of fatally injured drinking drivers and the change in the percent of fatally injured drivers who had been drinking were similar 31% and 29% decreases, respectively. The downward trend was clearly interrupted in 1991 and 1992 when the percentage of fatally injured drinking drivers increased to 46% and 48%, respectively. But this increase occurred because the number of fatally injured non-drinking drivers declined but the number of fatally injured drinking drivers remained relatively stable. Since 1992, there has been an annual decline in the percentage of fatally injured drivers who tested positive for alcohol i.e., a decrease from 48% in 1992 to 34% in 1999. The level achieved in 1999 was the lowest point reached in the past three decades and this downward trend strongly suggests a resurgence of the declines in the magnitude of the alcohol-fatal crash problem characteristic of the 1980s. It is, however, important to note that the decline in the percent of fatally injured drinking drivers that began in 1993 was again a function of two things a decline in the actual number of drinking-driver fatalities, combined with an increase in the number of non-drinking driver fatalities. This divergence was particularly marked after 1996 and had a salutary effect on the percentage. Nonetheless, from 1992 to 1999, the absolute number of drinking drivers did decrease by 29%, an amount identical the decrease in the percentage of fatally injured drivers who tested positive for alcohol i.e., a 29% reduction. However, it is important to recognize that when progress is measured in terms of changes in the percent of fatally injured drivers who had been drinking, this index can produce spurious effects, if the number of non-drinking driver fatalities increased at the same time. If the number of nondrinking driver fatalities had remained unchanged during the latter part of the 1990s, or had decreased, the decline in the percentage of drinking driver fatalities would have been less. Apart from the caution this demands in reporting and interpreting such data, they reveal as well another important finding that has yet to be explored adequately. At issue is the divergence in trends between the number of fatally injured non-drinking drivers and drinking drivers. Why is the number of alcohol-related driver deaths declining while the number of non-alcohol related driver deaths is increasing? Several alternative explanations are discussed in the paper (5). France In France, alcohol tests are compulsory in cases of injury accidents or when an offence has been committed. Random tests are also conducted. The number of compulsory tests has been fairly static over the past 10 years at about 1,500,000 per year. However, the number of random tests has risen sharply. The number of random tests increased from about 2,881,000 in 1990 to about 7,925,000 in 1999. The number of tests in 2000 declined slightly. The positive rate (over the legal limit) for the random tests was 1.1% in 1990 and 1.3% in 2000. But the numbers since 1996 reflect the lowering of the legal limit from.08 to.05%. The positive rate for those tested after an offence declined slowly from 3.6% in 1990 to 2.9% in 1996. In 2000, that rate was 3.6%, but again reflects the lower limit. More importantly, the number of fatal and injury accidents has declined steadily from 1990 to 2000. The percentages of injury and fatal accidents involving - 958 -

drivers over the legal limit have also declined in that period. In 2000, in 5.5% of injury accidents and 16.2% of fatal accidents the driver was over the legal limit. This progress is attributable to the massive alcohol screening enforcement on the roads. Records show that more than one out of three drivers convicted of DWI with a BAC over.14% need medical care for an alcohol problem. The challenge for the future is to develop more effective interventions for alcohol impaired drivers. Analysis by age stresses the fact that it is important to recognize the road risk problem of the younger driver and the probable health risk problem of the older driver. Since alcohol has been classifies by experts as a hard drug, it is probably a priority to use traffic enforcement as a means of educating drivers about alcohol risks (6). Germany In the years after unification up to 1993 in Germany (East) the road accidents in general and especially alcohol related accidents worsened. But the figures from 1994 to 2000 show a stabilisation and improvement in the number of road accidents in Germany (East) especially with respect to related injuries and fatalities. Up to 2000 a favourable and continuous drop of alcohol related road accidents in all of Germany can be seen. Alcohol-related fatalities and percentage of total fatalities dropped from 1,828 (18.6%) in 1994 to 1,022 (13.6%) in 2000. The BAC limit was reduced from.08 to.05% in 1998, but because of the short time frame, its effects could not be assessed. Most drivers (up to 95 percent) remained below the legal BAC limit of 0.08 % and tend also to remain below the new legal BAC limit of 0.05 % as recent police records since 1998 demonstrate. The frequency distribution of the BAC level of involved car drivers influenced by alcohol indicates that the problem will not be solved only by lowering the legal BAC-limit. From this point of view, it is not only the problem of the low-level-driver, but also of driving alcoholics, especially within the age-group of 25 years and older. In view of the complicated structure of social control, a solution to the problem may be found by concentrating on primary prevention. A denser network of police controls would probably not increase the "yield" of undetected offenders much beyond the present level - according to surveys not higher than 2 percent. However, the establishment of breath tests gives the possibility of a less costly and more efficient policing of drink driving. In further years it has to be determined in which way a more efficient policing policy together with more severe sanctioning of drink driving (fine and suspension of licence for at least one month upwards 0.05 % BAC) could contribute to reduce alcohol-related accidents and thereby improve traffic safety (7). The Netherlands Between the mid-1980s and the early 1990s, DUI in the Netherlands decreased strongly: in weekend nights, the proportion of drivers with an illegal BAC (> 0.5 g/l) dropped from 12% in 1983 to 3.9% in 1991. This favourable development followed the introduction and extension of random breath testing, facilitated by the introduction of electronic screening devices and evidential breath testing. In recent years, however, DUI has not decreased any further, and even tended to increase. This may have been caused by an indecisive government policy towards drink-driving. After a reorganisation of the Dutch police forces in the first half of the 1990s, traffic law enforcement was given a lower priority than before. The introduction, in 1996, of a mandatory rehabilitation program for severe DUI-offenders was not accompanied by a large- - 959 -

scale publicity campaign. The introduction of a 0.2 g/l BAC limit for novice drivers, originally intended to become effective in 2001, was postponed by approx. 3 years. Furthermore, the formation of special traffic enforcement units in all 25 Dutch police regions, which started in 1999 and should have been completed in 2001, did not in the short term result in a higher enforcement level throughout the country. In weekend nights of 2000, 4.6% of Dutch motorists had an illegal BAC. Finally, results of a case-control study, conducted in 2000/2001, raised questions on the effectiveness of police enforcement and rehabilitation programmes in substantially reducing the number of hardcore drinking drivers and the resulting road trauma (8). United States For more than a decade, rates of alcohol-related crashes had declined in the United States. In 2000, however, 40 percent of all fatalities involved alcohol, up from the historic low of 38 percent in 1999. It was the first increase in alcohol-related deaths since 1995. In 2000, 16,653 fatalities were alcohol-related, compared to 15,976 in 1999. In 2001, the rate and number of alcohol-related fatalities remained essentially the same as in 2000. This was very discouraging news, especially in light of the bold new national goal that was established in 1995 to reduce alcohol-related traffic fatalities in America to no more than 11,000 by the year 2005. At that time, 125 recommendations were made to meet the ambitious goal. Despite all this activity, progress in achieving this national public health goal has been slow and for the past two years movement has been in the wrong direction. In addition, the prevalence of drinking drivers on the roadways on Friday and Saturday nights did not change much in the U.S. between 1986 and 1996. An update of a model used to analyze the effectiveness of various impaired driving laws in the U.S. projects that alcohol-related fatalities will still be at about 16,600 in 2005 if present trends continue. This slowness in progress is particularly discouraging because the tools to make significant progress are available. Many effective strategies are well known, but not implemented as widely or as vigorously as possible. Because of the significant progress that has been made in the past, complacency may have set in among policy makers. The level of public awareness and concern seems to have waned, with attention deflected to other issues. Several well established strategies exist that can significantly reduce impaired driving in the United States if they are implemented more broadly and more vigorously. Action should be taken at the national, state, and local level to ensure that we continue to make progress in reducing alcohol-related traffic crashes. These strategies include: administrative license revocation (ALR), lowering the illegal per se blood alcohol concentration (BAC) limit to.08 g/dl, graduated driver licensing, strengthened occupant protection laws, vehicle sanctions, sobriety checkpoints, enforcement of minimum drinking age and zero tolerance laws, improved public information and awareness, and alternatives to drinking and driving (9). Discussion As the new decade began, in The Netherlands, roadside surveys in 2000, showed an increase in drivers with an illegal BAC. In the United States, alcohol-related fatalities increased for the first time in 2000, since 1995. In 2001, the numbers were the same as 2000. On the other hand, in the Federal Republic of Germany, the share of accident-involved persons influenced by alcohol has continued to decrease from 1994 to 2000. In Canada, the downward trend continued. In France, the drinking and driving prevention system appears to have reduced the number of accidents attributed to alcohol. In a number of countries were progress has continued, the results can be - 960 -

traced to the effectiveness of the prevention programs and the degree of attention placed on the issue. In those countries were progress has stopped or even reversed, the reasons can be traced to a lack of action, commitment or shifting priorities. References 1. Sweedler, BM (Ed). The Nature of and the Reasons for the Worldwide Decline in Drinking and Driving. Transportation Research Board Circular No. 422, Washington, DC, April 1994. 2. Sweedler, BM. The Worldwide Decline in Drinking and Driving. In C.N. Kloeden and A.J. McLean (Eds). Alcohol, Drugs and Traffic Safety T 95, Adelaide, Australia, NHMRC Roas Accident Research Unit, University of Adelaide, pp 493-497, August 1995. 3. Sweedler, BM. The Worldwide Decline in Drinking and Driving Where are we Now, In C. Mercier- Guyon (Ed). Alcohol, Drugs and Traffic Safety T 97, Annecy, France, CERMT, Centre d Etudes et de Recherches en Medecine du Trafic, pp 1205-1210. 4. Sweedler, BM. The worldwide decline in drinking and driving: has it continued. In Laurell (Ed) Alcohol, Drugs and Traffic Safety - T2000, Stockholm, Sweden, May 2000. 5. Mayhew, DR., Beirness, DJ. and Simpson, HM. Are The Declines In Drinking Driving Fatalities In Canada Being Overestimated? Paper prepared for presentation at T2002, the 16 th International Conference on Alcohol, Drugs and Traffic Safety, Montréal, Québec, Canada, August 2002. 6. Biecheler-Fretel, MB., Facy, F., Peytavin, JF. Drinking and Driving in France in 1999-2000 : Changes in The Decade and New Perspectives, Paper prepared for presentation at T2002, the 16 th International Conference on Alcohol, Drugs and Traffic Safety, Montréal, Québec, Canada, August 2002. 7. Kroj, G. and Lerner, M. Alcohol-related road accidents in the Federal Republic of Germany Status till 2000. Paper prepared for presentation at T2002, the 16 th International Conference on Alcohol, Drugs and Traffic Safety, Montréal, Québec, Canada, August 2002. 8. Mathijssen, MPM. Indecisive Drink-Driving Policy Allows for Increase of DUI in the Netherlands, Paper prepared for presentation at T2002, the 16 th International Conference on Alcohol, Drugs and Traffic Safety, Montréal, Québec, Canada, August 2002. 9. Stewart, K., Fell, J. Trends in Impaired Driving in the United States: Complacency or Backsliding? Paper prepared for presentation at T2002, the 16 th International Conference on Alcohol, Drugs and Traffic Safety, Montréal, Québec, Canada, August 2002. - 961 -

- 962 -

Alcohol Related Road Accidents in the Federal Republic of Germany Status till 2000 G. Kroj and M. Lerner Federal Highway Research Institute, Bergisch Gladbach, Germany Keywords Alcohol, accidents, traffic safety Abstract This article describes the development and actual status of alcohol related accidents in Germany and discusses the effectiveness of police controls, legislative and educational measures. Alcohol and road traffic in Germany Drink-drive limits In 1973, the critical BAC limit of 0.08 % was introduced as an offence which by statutory definition carries a fine ( 24a Road Traffic Act, StVG). After the unification of the Federal Republic of Germany and the German Democratic Republic from 1989 to 1993 there were two different legal BAC limits in the two former German states (West: 0.08 %; East: 0.00 %). In January 1993 in all parts of Germany the same BAC limit of 0.08 % became a legal requirement. In 1998 a second BAC-limit of 0.05 % was introduced, which indicated an offence without suspension of licence. The stricter limit was accompanied by a newly introduced limit for breath alcohol concentration of 0.25 mg/l. Instead of taking blood samples the police can carry out breath tests using special measuring instruments. This new and less costly testing procedure is admissible evidence in a court of law. Since April 2001 the limit of 0.05 % has replaced the limit of 0.08 % as an offence, which carries suspension of licence for at least one month. Alcohol related injury accidents - Data from the Federal Republic of Germany Alcohol related road traffic accidents are accidents in which at least one involved person has an alcohol concentration exceeding 0.03 %. In 1975 in West Germany 86 of 1 000 injury-accident involved car drivers and 63 of 1 000 pedestrians were influenced by alcohol. Since this time the figure decreased up to the year 2000 to 35 per 1 000 involved car drivers and increased on the other hand up to the year 1993 to 73 per 1.000 pedestrians (East: 84/West: 69). Since 1993 the figure of intoxicated injury accident involved pedestrians also decreased to 56 per 1 000 pedestrians in 2000. Development till the unification of the two German states In the German Democratic Republic up to 1989 the trend of the development of alcohol related accidents was similarly favourable: compared with 1980, 14 % less alcohol related accidents with - 963 -

injuries were counted in 1989 and 25 % less fatalities in alcohol related accidents, whereas the number of injury accidents not involving alcohol even rose by 3 % in that time. The number of fatalities also decreased by 8 %, which was clearly lower than in alcohol related accidents [1]. In the period from 1975 to 1990 in the Federal Republic of Germany (West) alcohol related injuries and fatalities in road traffic accidents decreased continuously: The alcohol related injury accidents decreased significantly (-32 %). On the contrary the non alcohol related injury accidents increased between 1975 and 1990 (+6 %). The alcohol related casualties also decreased dramatically in this period (-37 %), meanwhile the non alcohol related casualties rose (+2.5 %). The alcohol related fatalities in road accidents decreased from 1975 to 1990 by 57 %, whereas the non alcohol related fatalities in road accidents accounted only for -44 %. Three years before unification (1986 to 1989), the share of alcohol related injury accidents in the German Democratic Republic was approximately the same as in Federal Republic of Germany (10 %). The share of alcohol related fatalities in road accidents was also similar (FRG 18 %; GDR 17 %). But after the unification the road accidents increased dramatically in the former GDR. Especially the alcohol related injury accidents as well as the alcohol related casualties increased substantially from 1990 to 1993, while in Western Germany these figures continued to decrease. Since 1994 in the western part as well as in the eastern part of Germany all figures of alcohol related injury accidents have decreased. Although, the share of alcohol related accidents in the eastern federal states is higher, the figures of western and eastern countries become more and more equal. Alcohol related injury accidents in Germany since 1994 In the period from 1994 to 2000 all injury accidents decreased slightly (-2.5 %) while there was a continuous decrease concerning alcohol related injury accidents (-31 %). Especially the alcohol related fatalities (-44 %) as well as fatalities in general (-24 %) decreased substantially till 2000. 27 749 accidents involving alcohol occurred in the year 2000. That means a reduction of nearly one third compared to 1994. These accidents accounted for 36 764 casualties, 1 022 of them died. Both the share of alcohol related accidents and casualties have decreased since 1994: The share of alcohol related injury accidents decreased from 1994 to 2000 from 10.2 to 7.2 % [see table 1]. The share of alcohol related casualties decreased from 1994 to 2000 from 10.3 to 7.6 % [see table 2] while the share of alcohol related fatalities decreased even from 18.6 to 13.6 % [see table 3]. The share of accident-involved persons influenced by alcohol decreased between 1994 and 2000 from 5.3 to 3.7 %. - 964 -

Table 1: Time series of injury accidents in total and alcohol related injury accidents in the period from 1994 2000 in Germany 1994 1995 1996 1997 1998 1999 2000 Injury accidents in total Alcohol related injury accidents Share of alcohol related injury accidents 392 745 388 003 373 082 380 835 377 257 395 689 382 949 39 892 36 966 34 468 32 884 28 736 28 350 27 375 10.2 % 9.5 % 9.2 % 8.6 % 7.6 % 7.2 % 7.2 % Table 2: Time series of casualties in road accidents in total and alcohol related casualties in the period from 1994 2000 in Germany 1994 1995 1996 1997 1998 1999 2000 Casualties in road accidents Alcohol related Casualties Share of alcohol related casualties 526 229 521 595 501 916 509 643 505 111 528 899 511 577 55 093 51 346 47 348 45 020 38 483 38 110 36 764 10.5 % 9.8 % 9.4 % 8.8 % 7.6 % 7.2 % 7.2 % Table 3: Time series of fatalities in road accidents in total and alcohol related fatalities in the period from 1994 2000 in Germany 1994 1995 1996 1997 1998 1999 2000 Fatalities in road accidents Alcohol related fatalities Share of alcohol related fatalities 9 814 9 454 8 758 8 549 7 792 7 772 7 503 1 828 1 716 1 472 1 447 1 114 1 114 1 022 18.6 % 18.2 % 16.8 % 16.9 % 14.3 % 14.3 % 13.6 % Most of the persons involved influenced by alcohol were car drivers (63 %). Only a minority of them were women (11 %) whereas compared with 1994 (8 %) the share of female impaired car drivers rose slightly. The problem of alcohol-impaired driving clearly shows age- and sexspecific features. Most of the male drivers involved are aged between 21 and 25 whereas the share of female drink-drivers involved raises continuously till the age of 44. - 965 -

The temporal distribution of alcohol related accidents shows the connection of alcohol impaired driving and leisure time. Nearly half (49 %) of alcohol impaired involved car drivers in 2000 had an accident at weekend (Saturday and Sunday), more than a quarter (27 %) in the nights of Friday and Saturday between 8 PM and 4 AM. With regard to the frequency distribution of the BAC level of involved car drivers, it must be taken into account that driving with lower BAC-level occurs more often than with higher levels. On the other hand the risk of an accident rises with the BAC-level. In combination of both factors the frequency distribution shows a peak between 0.14 and 0.17 % BAC. The mean BAC figure was - depending on the age - between 0.12 % (18-21 and above 75 years) and 0.19 % (40-50 years). Despite the mostly positive development of the alcohol related accident statistics the results of alcohol related accidents are more serious than the results of accidents in general. The severity of drink-drive accidents is - expressed as fatalities per 1 000 injury accidents nearly twice as high as that of accidents in general. From 46.5 in 1994 this figure decreased to 37.3 in 2000. Discussion Until 2000 a favourable and continuous drop of alcohol related road accidents in Germany can be seen. In the years after unification till 1993 in Germany (East) the road accident development in general and especially concerning alcohol related accidents worsened. But the figures from 1994 to 2000 show a stabilisation and improvement of the road accident development in Germany (East) especially with respect to related injuries and fatalities. Because of the short period of analysis one could not assess clearly the effects of the lower legal BAC-limit in 1998 and 2001 on the alcohol related accident statistics. But the frequency distribution of the BAC level of involved car drivers influenced by alcohol indicates that the problem is not to be solved only by lowering the legal BAC-limit. From this point of view, it is not only the problem of low-level-driver, but also of driving alcoholics, especially within the age-group of 25 years and older. Outlook In view of the complicated structure of social control, a solution to the problem may be found by concentrating on primary prevention. Most drivers (up to 95 percent) remained below the legal BAC limit of 0.08 % [2] [3] and tend also to remain below the new legal BAC limit of 0.05 % as recent police records since 1998 demonstrate. A denser network of police controls would probably not increase the "yield" of undetected offenders much beyond the present level - according to surveys not higher than 2 percent [4]. However, the establishment of breath tests gives the possibility of a less costly and more efficiently policing of drink driving. In further years it has to be examined in which way a more efficient policing together with a more severe sanctioning of drink driving (fine and suspension of licence for at least one month upwards 0.05 % BAC) could contribute to reduce alcohol related accidents and so to improve traffic safety. Present efforts in the Federal Republic of Germany towards a more systematic integration of driver improvement programs into the existing legal and administrative framework seem to - 966 -

provide a more promising route to traffic safety in the medium term than do demands for radical changes in sanctions [2] [5] [6] [7]. Driver improvement programs can already be considered as making an important and unique contribution to road safety in the Federal Republic of Germany, but they cannot be seen as a substitute for legal sanctions and measures. References 1. Kretschmer-Bäumel E. Drinking and Driving in Germany: Behavioural Patterns and Influencing Factors-- a temporal and cross-cultural comparison. J Proceedings "Alcohol, Drugs and Traffic Safety-T'92", Vol. II, Verlag TÜV Rheinland, Köln, 1993, 1011-16. 2. Stephan E. Wirksamkeit der Nachschulungskurse bei erstmals alkoholauffälligen Kraftfahrern. Forschungsberichte der BASt, Bereich Unfallforschung, H 170, Bergisch Gladbach, 1988. 3. Krüger HP. Das Unfallrisiko unter Alkohol, Gustav Fischer Verlag, Stuttgart, Jena, New York, 1995. 4. Kerner HJ. Gesetzgebung, polizeiliche Überwachung und Strafgerichtsbarkeit in der Bundesrepublik Deutschland. Forschungsberichte der BASt, Bereich Unfallforschung, H. 115, Bergisch Gladbach, 1985. 5. Winkler W, Jakobshagen W, Nickel WR. Wirksamkeit von Kursen für wiederholt alkoholauffällige Kraftfahrer. Forschungsberichte der BASt, H 224, Bergisch Gladbach, 1991. 6. Kroj G. Rehabilitation of drunken drivers in the Federal Republic of Germany. J Proceedings "Alcohol, Drugs and Traffic Safety - T'92", Vol. I, Verlag TÜV Rheinland, Köln, 1993, 378-84. 7. Kroj G (ed.). Psychologisches Gutachten Kraftfahreignung. Deutscher Psychologen Verlag, Bonn, 1995. 8. Statistisches Bundesamt. Alkoholunfälle im Straßenverkehr. Auszug aus Fachserie 8, Reihe 7 "Verkehrsunfälle", Wiesbaden, Ausgabe 1993 und 1998. 9. Statistisches Bundesamt. Fachserie 8. Reihe 7 Verkehrsunfälle 2000, Wiesbaden, 2001. - 967 -

- 968 -

Are The Declines In Drinking Driving Fatalities In Canada Being Overestimated? D.R. Mayhew, H.M. Simpson and D.J. Beirness Traffic Injury Research Foundation 171 Nepean St. Suite 200 Ottawa, ON Canada K2P 0B4 Keywords Trends, drinking-driving, alcohol-impaired driving, driver fatalities, alcohol-fatal crash problem, driver characteristics Abstract Previous research has shown that the magnitude of the alcohol-fatal crash problem in Canada declined between 1992 and 1997, when it reached the lowest point in the past three decades. This paper examines trends in the alcohol-fatal crash problem to determine if reductions in the magnitude of the problem have continued. The analyses revealed continued progress from 1997 to 1999 as the percentage of fatally injured drivers who were positive for alcohol, as well as the number of drinking driver fatalities dropped even further. By contrast, the number of nondrinking driver fatalities actually increased over this period. The implications of these findings for estimating changes in the magnitude of the alcohol-fatal crash problem are discussed. Introduction In Canada during the 1980s, consistent and significant declines were reported in the alcohol-fatal crash problem (1, 2, 3). However, these declines halted rather abruptly and significantly for two years in 1991 and 1992, when increases were recorded (4). The following year, the downward trend re-emerged with decreases occurring each year. As a result, by 1997, the magnitude of the alcohol-fatal crash problem reached the lowest point recorded in the previous three decades (5). This paper examines more recent trends to determine if reductions in the magnitude of the problem have continued. In so doing, it seeks to identify reasons for both the historical and recent trends, some of which are related to how the problem itself is measured. In this context, a measure that is a widely accepted in many countries as a valid and reliable measure of changes in the magnitude of the alcohol-crash problem (6, 7, 8, 9) is the annual percentage of fatally injured drivers who test positive for alcohol. Measuring the magnitude of the problem in this way can, however, create interpretive difficulties because the index is sensitive to annual changes in both the numbers of drinking and nondrinking driver fatalities. For example, if both the number of non-drinking driver fatalities and the number of drinking driver fatalities decreased at the same rate each year, the percent of fatally - 969 -

injured drivers who were positive for alcohol would not change. On the one hand, this could be interpreted as no improvement in the drinking-driving problem, since the percent of fatalities that were alcohol-related did not decline; on the other hand, such results could be interpreted in a positive light, since the absolute number of alcohol-related fatalities did decline each year. Accordingly, this paper also considers the extent to which changes in the percent of fatally injured drivers testing positive for alcohol, faithfully reflect changes in the magnitude of the problem. Data Sources Since 1973, the Traffic Injury Research Foundation (TIRF) has collected and maintained a database containing the results of tests for the presence and amount of alcohol performed on fatally injured drivers in seven provinces -- British Columbia, Alberta, Saskatchewan, Manitoba, Ontario, New Brunswick, and Prince Edward Island. In 1987, the Fatality Database was expanded to include all ten provinces and two territories. Rates of testing for alcohol vary somewhat among jurisdictions but, on average each year, about 80% of drivers of highway vehicles who died within six hours of the crash are tested for the presence of alcohol. Information on the presence of alcohol in fatally injured drivers is used in this paper as an index of the alcohol fatal-crash problem in Canada. Consistent with previous papers on this issue, trends in the problem are examined using an indicator derived from the fatality database i.e., the percentage of fatally injured drivers who tested positive for alcohol. To understand the basis for the changes in this indicator, annual changes in the number of non-drinking driver fatalities and the number of drinking driver fatalities are also examined. Results The numbers of fatally injured drivers who tested negative (non-drinking) for alcohol and the number who tested positive (drinking), as well as the percent of tested driver fatalities with positive BACs are shown in Figures 1, 2, and 3 for the 1970s, 1980s, and 1990s, respectively. During the 1970s, the percentage of fatally injured drinking drivers remained relatively stable ranging from 58% to 60% (see Figure 1). This trend occurred because the number of fatally injured non-drinking drivers mirrored annual changes in the number of fatally injured drinking drivers i.e., their numbers rose and fell together. Importantly, however, from 1973 to 1979, even though the percentage of drinking driver fatalities remained basically unchanged, the absolute number of drinking driver fatalities actually decreased by 11%. Figure 1 Numbers of Non-Drinking and Drinking Fatally Injured Drivers and Percent with Positive BACs: Canada*, 1973-1979 100 1000 90 800 Percent 80 70 60 50 58 57 59 60 59 59 58 Pct Positive Non-Drinking Drinking 600 400 200 Number 40 73 75 77 79 * Seven provinces Year - 970-0

During the 1980s, the trend was decidedly different (see Figure 2). The percentage of fatally injured drivers who were drinking increased slightly in 1980 and again in 1981 and then declined to the end of the decade. In 1981, 62% of the drivers killed in Canada had been drinking. By 1989, this figure had reached a low of 44%. This represents a 29% reduction in the magnitude of the alcohol-fatal crash problem. The downward trend in the 1980s occurred not only because the actual number of fatally injured drinking drivers declined but also because the number of non-drinking drivers increased. The divergence in these trends was underscored in 1987, the last year in which the majority of fatally injured drivers had been drinking. Importantly, from 1981 to 1989, the relative change in the number of fatally injured drinking drivers and the change in the percent of fatally injured drivers who had been drinking were similar 31% and 29% decreases, respectively. Figure 2 Numbers of Non-Drinking and Drinking Fatally Injured Drivers and Percent with Positive BACs: Canada*, 1980-1989 100 1000 Percent 90 80 70 60 62 60 61 58 56 50 52 51 51 Pct Positive Non-Drinking Drinking 48 40 80 82 84 86 88 Year * Seven provinces 44 800 600 400 200 0 Number As shown in Figure 3, the downward trend was clearly interrupted in 1991 and 1992 when the percentage of fatally injured drinking drivers increased to 46% and 48%, respectively. But this increase occurred because the number of fatally injured non-drinking drivers declined but the number of fatally injured drinking drivers remained relatively stable. Since 1992, there has been an annual decline in the percentage of fatally injured drivers who tested positive for alcohol i.e., a decrease from 48% in 1992 to 34% in 1999. The level achieved in 1999 was the lowest point reached in the past three decades and this downward trend strongly suggests a resurgence of the declines in the magnitude of the alcohol-fatal crash problem characteristic of the 1980s. In fact, both the decades of the 1980s and 1990s witnessed an initial increase in the magnitude of the problem followed by a consistent and comparable drop reductions of about 30% in both of these decades. It is, however, important to note that the decline in the percent of fatally injured drinking drivers that began in 1993 was again a function of two things a decline in the actual number of drinking-driver fatalities, combined with an increase in the number of non-drinking driver fatalities. This divergence was particularly marked after 1996 and had a salutary effect on the percentage. Nonetheless, from 1992 to 1999, the absolute number of drinking drivers did - 971 -

decrease by 29%, an amount identical the decrease in the percentage of fatally injured drivers who tested positive for alcohol i.e., a 29% reduction. Figure 3 Numbers of Non-Drinking and Drinking Fatally Injured Drivers and Percent with Positive BACs: Canada*, 1990-1999 80 700 Percent 70 60 50 40 43 48 46 46 44 43 41 39 39 30 Pct Positive Non-Drinking Drinking 20 90 92 94 96 98 Year * Seven provinces 34 600 500 400 300 200 100 0 Number Discussion The above findings demonstrate that the magnitude of the alcohol-fatal crash problem in Canada declined in the 1980s and again in the 1990s. In fact, both the percentage and number of drinking driver fatalities decreased at a comparable rate in the 1980s as well as the 1990s about a 30% reduction for both measures in each of these decades. However, it is important to recognize that when progress is measured in terms of changes in the percent of fatally injured drivers who had been drinking, this index can produce spurious effects, if the number of non-drinking driver fatalities increased at the same time. If the number of non-drinking driver fatalities had remained unchanged during the latter part of the 1990s, or had decreased, the decline in the percentage of drinking driver fatalities would have been less. Apart from the caution this demands in reporting and interpreting such data, they reveal as well another important finding that has yet to be explored adequately. At issue is the divergence in trends between the number of fatally injured non-drinking drivers and drinking drivers. Why is the number of alcohol-related driver deaths declining while the number of non-alcohol related driver deaths is increasing? Several alternative explanations are possible. One hypothesis is that the prevailing trend in the 90s was actually for an increase in driver deaths. However, drinking-driving initiatives may have exerted a powerful countervailing influence on factors that were leading to an increase in the number of fatally injured drivers. The myriad of safety measures targeting drinking drivers in the 1980s and 1990s resisted these prevailing influences and the alcohol-fatal crash problem decreased rather than increased. If so, the declines in drinking driver fatalities, as measured by the percentage of fatally injured drivers who tested positive for alcohol, is actually not at all spurious or an overestimate of the magnitude of the reductions. In fact, it would be an appropriate index of the downward trend because it takes into account the increase that would have occurred in the absence of the effective initiatives i.e., the number of drinking driver fatalities would have increased at the same rate as did the number of non-drinking driver fatalities. - 972 -

Alternatively, drinking-driving initiatives may have fostered a climate in which driving after drinking has become socially unacceptable. Drivers who would have previously driven after drinking refrained from doing so, but they crashed anyway. If this were the case, one would have to conclude that in such crashes alcohol was not a primary causal factor; rather, other factors, such as driving inexperience, overconfidence, poor skills, aggressive driving, and fatigue played a principal role in the increased number of non-drinking driver fatalities. This speaks to the fact that the presence of alcohol in a collision is not synonymous with alcohol being a cause in the collision. If this hypothesis were supported, it would suggest that although drinking-driving initiatives may have contributed to reductions in alcohol-related fatal crashes, they have not positively influenced the overall road crash problem. It is also possible that drinking-driving initiatives have played a negligible role in determining the divergent trends that have emerged in the past two decades. Perhaps powerful secular forces have exerted the major influence on these trends. Demographic changes, economic conditions, shifts in attitudes about health and alcohol, and a myriad of other factors largely define the context in which drinking and driving and road crashes takes place and, consequently, could be the driving force behind the divergent trends in the alcohol-and non-alcohol-fatal crash problems. Such explanations are speculative and provocative. But they underscore the need to reconsider the meaning of changes in indicators of the magnitude of the alcohol-crash problem, and perhaps more importantly, the need for research to understand why the number of alcohol-related fatal crashes has declined but the number of non-alcohol-related crashes has increased. Conclusion The meaning and continued value of indicators typically used to examine trends in the magnitude of the alcohol-fatal crash problem need to be examined. Apparent reductions in the problem as measured by declines in the percentage of fatally injured drinking drivers can result simply from increases in the number of non-drinking fatally injured drivers and this has been happening in recent years. Even so, it is important to underscore that during the 1980s and 1990s both the percentage and number of drinking driver fatalities declined at about the same rate in Canada. The implication of these divergent trends for understanding the impact of drinking-driving countermeasures, however, remains unclear. This is especially the case if powerful secular forces and not safety measures have exerted the major influence on these trends. Basically, the explanation for these trends remains open to speculation. The factors responsible for them have not been adequately documented. Further research beyond the descriptive level is needed to understand what precipitated the changes in the magnitude of the problem observed in the 1980s and 1990s. Acknowledgement The Fatality Database is funded by the Canadian Council of Motor Transport Administrators (CCMTA) and Transport Canada. - 973 -

References 1. Beirness DJ, Simpson HM, Mayhew DR, Wilson RJ. Canadian trends in drinking driver fatalities. In H. Utzelman, G. Berghaus and G. Kroj (Eds) Alcohol Drugs and Traffic Safety. Cologne: Verlag TUV Rheinland, 1993, pp 1062-1067. 2. Simpson HM, Beirness DJ, Mayhew DR. Decline in drinking and driving crashes, fatalities and injuries in Canada. In B. Sweedler (Ed.) The Nature of and Reasons for the Worldwide Decline in Drinking & Driving. Proceedings of a Workshop of the TRB Committee on Alcohol, Other Drugs and Transportation. Washington: (DC) 1994, TRB Circular No. 422. 3. Simpson HM, Mayhew DR, Beirness DJ. The decline in drinking-driving fatalities in Canada: A decade of progress comes to an end? In C.N. Kloeden and A.J. McLean (Eds) Alcohol Drugs and Traffic Safety T 95, Adelaide, Australia: NHMRC Road Accident Research Unit, University of Adelaide, Volume I, 1995, pp 508-512. 4. Simpson HM, Mayhew DR, Beirness, DJ. The decline in drinking-driving fatalities in Canada: Progress re-emerges? In C. Mercier-Guyon (Ed) Alcohol, Drugs and Traffic Safety T 97, Annecy, France: Centre d Études et de Recherches en Médecine du Trafic, Volume 3, 1997, pp 1219-1225. 5. Mayhew DR, Beirness, DJ, and Simpson, HM. Trends in drinking-driving fatalities in Canada Progress continues. In: Alcohol, Drugs and Traffic Safety T 2000, Stockholm, Sweden, May 2000. 6. Sweedler, B.M. (ed) The Nature of and the Reasons for the Worldwide Decline in Drinking and Driving. Transportation Research Board Circular No. 422, Washington, D.C. April 1994. 7. Sweedler, B.M. The worldwide decline in drinking and driving. In C.N. Kloeden and A.J. McLean (Eds) Alcohol Drugs and Traffic Safety T 95, Adelaide, Australia: NHMRC Road Accident Research Unit, University of Adelaide, Volume I, 1995, pp 493-497. 8. Sweedler, B.M. The worldwide decline in drinking and driving Where are we now. In C. Mercier-Guyon (Ed) Alcohol, Drugs and Traffic Safety T 97, Annecy, France: Centre d Études et de Recherches en Médecine du Trafic, Volume 3, 1997, pp 1205-1210. 9. Sweedler, B.M. The worldwide decline in drinking and driving : Has it continued. In: Alcohol, Drugs and Traffic Safety T 2000, Stockholm, Sweden, May 2000. - 974 -

Indecisive Drink-Driving Policy Allows for Increase of DUI in the Netherlands M.P.M. Mathijssen SWOV Institute for Road Safety Research, Leidschendam, The Netherlands. Keywords Alcohol, drugs, policy, risk Abstract Between the mid-1980s and the early 1990s, DUI in the Netherlands decreased strongly: in weekend nights, the proportion of drivers with an illegal BAC (> 0.5 g/l) dropped from 12% in 1983 to 3.9% in 1991. This favourable development followed the introduction and extension of random breath testing, facilitated by the introduction of electronic screening devices and evidential breath testing. In recent years, however, DUI has not decreased any further, and even tended to increase. This may have been caused by an indecisive government policy towards drink-driving. After a reorganisation of the Dutch police forces in the first half of the 1990s, traffic law enforcement was given a lower priority than before. The introduction, in 1996, of a mandatory rehabilitation program for severe DUI-offenders was not accompanied by a large-scale publicity campaign. The introduction of a 0.2 g/l BAC limit for novice drivers, originally intended to become effective in 2001, was postponed by approx. 3 years. Furthermore, the formation of special traffic enforcement units in all 25 Dutch police regions, which started in 1999 and should have been completed in 2001, did not in the short term result in a higher enforcement level throughout the country. In weekend nights of 2000, 4.6% of Dutch motorists had an illegal BAC. Finally, results of a case-control study, conducted in 2000/2001, raised questions on the effectiveness of police enforment and rehabilitation programmes in substantially reducing the number of hardcore drinking drivers and the resulting road trauma. Introduction Between 1970 and 1999, SWOV carried out periodic roadside surveys into the alcohol consumption of Dutch motorists during autumn weekend nights (10 pm-4 am). Since 2000, the surveys are being conducted by the Transport Research Centre (AVV) of the Ministry of Transport. The objective of these surveys is to obtain an insight into the patterns of drink-driving and into the effects of countermeasures. - 975 -

In 2000/2001, SWOV, in collaboration with Utrecht University, conducted a case-control study to determine the relative injury risk of psychoactive substance use by motorists. Until then, only rough estimates of the DUI road toll could be made, based on incomplete official statistics. Methods Roadside surveys Since 1991, roadside surveys have been conducted on a yearly basis. In each of the twelve Dutch provinces a varying number of survey areas is selected, dependent on population size of the province. The sample of survey areas is geographically spread over the province and stratified by degree of urbanisation. In each survey area, a police team, instructed and accompanied by a researcher, is performing RBT-activities at four to six consecutive locations, situated along main roads inside built-up area. The frequent change of location is intended to minimize the predictability of the combined survey and enforcement activities with respect to time and place. Motorists are taken at random from moving traffic and breath-tested by means of a Dräger Alcotest 7410 Plus screening device. Since random breath testing by the police is legally admitted in the Netherlands, non-response is virtually non-existent. Police survey teams are equally distributed over Friday and Saturday nights, and random breath testing is performed between 10 pm and 4 am. Each test result, as well as sex and age of the motorist, is entered on a registration form with preprinted date, time and location. In the end, provincial samples are put together, forming one nationwide sample. Since the distribution of observations over the various provinces is not equal to the distribution of the population (as an indicator of traffic volume), the BAC-distribution of the sample is weighted for provincial population size. The sample size has grown from about 3,000 tested motorists in the 1970s to nearly 25,000 in recent years. Data analysis is performed with the log-linear Weighted Poisson Model (1,2). In 1997 and 1998, a random sub-sample of 893 motorists was also urine-tested for a number of licit and illicit drugs, i.e.: (meth)amphetamines, cannabis, cocaine, opiates, methadone, benzodiazepines, barbiturates, and tricyclic antidepressants (3). In addition to the roadside survey, accident data is analyzed, and police co-ordinators of the police teams are interviewed on developments in enforcement levels and tactics. Furthermore, data on publicity campaigns, rehabilitation programmes, medical examinations, and accident data is collected and analyzed. Case-control study From May 2000 until August 2001, a prospective case-control study was conducted in the town of Tilburg and surroundings, covering a population of approximately 350,000 inhabitants in the south of the Netherlands. Cases consisted of seriously injured motorists who were admitted to the emergency department of the Tilburg St. Elisabeth Hospital. Controls consisted of motorists who were taken at random from moving traffic in the Tilburg police district, which covers the catchment area of the St. Elisabeth Hospital. - 976 -

Body fluids (urine or blood) of both cases and controls were tested for the presence of alcohol and the above-mentioned licit and illicit drugs. The relative risk of psychoactive substances was determined by comparing their prevalence in cases with their prevalence in controls. Odds ratios were computed by relating subjects who had been tested positive for a substance or a combination of various substances, to subjects who had been tested negative for all substances. A 5% probability level (p < 0.05) was used for significance. Results Development of drink-driving Between 1970 and 2000, the proportion of drivers with a BAC >0.5 g/l dropped from 15% to 4.5%. Significant reductions of drink-driving in weekend nights could be observed after police enforcement was intensified and/or the perceived risk of apprehension was enhanced (4): - Immediately after the introduction of the legal BAC-limit of 0.5 g/l, in 1974, the proportion of motorists with an illegal BAC dropped sharply, due to a perceived high risk of apprehension. Although this initial effect diminished quickly, a significant and stable 25% reduction of drink-driving could be observed until the mid-1980s. - From 1985 on, drink-driving again decreased rapidly,coinciding with expanding possibilities for random breath testing (RBT) by the police. RBT was facilitated by the introduction of electronic screening devices and the subsequent introduction of evidential breath testing. The share of motorists with an illegal BAC dropped to 3.9% in 1991. A temporarily strong reduction of the enforcement level after a reorganization of Dutch police forces, which came into effect in 1992, resulted in an increase of drink-driving. In weekend nights of 1994, 4.9% of motorists had an illegal BAC. Since then, the pre-reorganization enforcement level was more or less restored, and the proportion of illegal BACs stabilized at about 4.5% between 1995 and 2000 (see Figure 1). Figure 1: Development of the proportion of motorists with an illegal BAC, in the Netherlands, in weekend nights, in the period 1970-2000 Drink-driving in the Netherlands in weekend nights, 1970-2000 % 16 14 12 10 8 6 4 2 0 1970-74 1975-79 1980-84 1985-89 1990-94 1995-00 - 977 -

High-risk groups Based on roadside survey and accident data, relating to weekend nights of 1996 and 1997, SWOV (4) determined the relative accident risk of the various age categories of male motorists (Table 1). Table 1: Relative accident risk of drink-driving by males of various age categories Age BAC-ditribution in traffic BAC-distribution in injury-accidents <0.5 g/l >0.5 g/l <0.5 g/l >0.5 g/l Relative risk (odds ratios) 18-24 yrs 96.5% 3.5% 82.0% 18.0% 6.1 25-34 yrs 94.3% 5.7% 79.0% 21.0% 4.3 35-49 yrs 93.8% 6.2% 81.6% 18.4% 3.4 > 50 yrs 95.6% 4.4% 89.4% 10.6% 2.6 all > 18 yrs 94.8% 5.2% 82.2% 17.8% 3.9 The high relative accident risk of drinking young males is reflected by their involvement in alcohol-related accidents. While forming less than 5% of the Dutch population, males of 18-24 years form nearly a quarter of all alcohol-intoxicated drivers who are involved in serious injuryaccidents. In view of their high relative accident risk, the development of drink-driving by young male motorists is rather disturbing. In the period 1991-1993, an average of 3.1% exceeded the legal BAC-limit; in the period 1994-1996, 3.5%; and in 1997-1999, 4.0%. Survey data of 2000 (5) showed a further increase of drink-driving by young males, to 4.6% (see Figure 2). Figure 2: Development of illegal BACs among young motorists (18-24 years) in the Netherlands, in weekend nights, in the period 1991-2000 5 4 Illegal BACs among young Dutch motorists, in weekend nights, 1991-2000 male female % 3 2 1 0 1991-93 1994-96 1997-99 2000-978 -

Another category of motorists with an extremely high injury risk was revealed by a case-control study (6) that SWOV conducted in 2000/2001, in collaboration with the Institute for Pharmaceutical Sciences of Utrecht University. The study was aimed at determining the relationship between the use of various psychoactive substances and road trauma. Results showed a high relative risk rate, of nearly 50, for drivers with a BAC >1.3 g/l, who had been tested negative for other licit or illicit drugs. Their prevalence in moving traffic was 0.3%, against 10.0% in serious injuries. An even much higher risk rate, of more than 400, was found for drivers who had combined a high BAC with the use of illegal drugs; their prevalence in moving traffic was less than 0.1%, against 11.8% in injuries. For BACs between 0.5 en 1.3 g/l, a moderately enhanced relative risk of about 5.5 was found. No enhanced risk could (yet) be assessed for BAC s between 0.2 and 0.5 g/l. The size of the case and control group was too small for age and sex differentiations. The above figures indicate that, between 1970 and 2000, serious injuries caused by drink-driving in the Netherlands, have decreased less than the overall proportion of drivers with an illegal BAC. Furthermore, they indicate that drivers with a BAC >1.3 g/l, while forming only 20% of all drivers with an illegal BAC, are responsible for about 90% of serious injuries caused by drinkdriving. Drink-driving government policy In 1996, administrative sanctions against severe offenders were sharpened, comprising the introduction of a compulsary three-day rehabilitation programme, to be followed at the offenders own expense. Since then, about 8,000 offenders per annum had to follow the programme. An evaluation study, conducted in 1999-2000, could not determine a significant positive effect on repeat drink-driving (7). General deterrence as a result of the rehabiliatation program was prevented by an almost complete lack of publicity. In 1999, special traffic police squads were (re)introduced in 7 of the 25 Dutch police regions. Within three years such squads should have been created in all police regions. In the 2000 roadside survey, no positive effects on drink-driving throughout the Netherlands could yet be established. The proportion of illegal BACs increased slightly, from 4.3% in 1999 to 4.6% in 2000. Among young males the proportion increased even stronger, from 4.1% to 4.6% (5,8). In some selected police regions with a newly formed traffic police squad, however, a 10% reduction of drink-driving was observed; in these regions, enforcement activities had clearly been intensified after the introduction of the new squads. Also in 1999, the Dutch government decided to lower the legal BAC-limit for novice drivers to 0.2 g/l, in order to reduce the alcohol-related road toll caused by young (male) drivers. Positive experiences, especially in Austria (9), contributed to the government s decision. On request of the Ministry of Transport, SWOV made an estimate of the effects that could be expected. SWOV estimated that the new limit would result in a reduction of 12 fatalities and 100 serious injuries per annum (10), without any change of the, rather low, enforcement level. The new limit for novice drivers was intended to become effective in 2001, but was delayed by discussions in parliament and society on a 0.2 g/l BAC-limit for all drivers, like in Sweden. Eventually, in 2002, the government reconfirmed its original decision, but introduction is now foreseen not earlier than for 2004. - 979 -

Discussion In the period 1970-2000, the proportion of alcohol-related serious injuries in the Netherlands seems to have decreased much less than the proportion of drivers with an illegal BAC. Since about 90% of alcohol-related serious injuries is caused by a very small proportion of drivers with a very high BAC, hardcore drinking drivers deserve special attention in drink-driving policy. The effect of the existing 3-day rehabilitation programme seems to be too limited for bringing about a substantial reduction of this high-risk group (7). Based on the effects of alcolock programmes on repeat drink-driving in the U.S.A. and Canada (11), the Dutch Ministry of Transport is interested in implementing an alcolock programme in the Netherlands (12). A change of the Road Traffic Act is needed, however. Preparations for a change of law have started in 2002; the moment of actual introduction cannot yet be predicted. Another explanation for the still relatively high proportion of alcohol-related serious injuries, is the ever growing number of young male drivers with an illegal BAC. A quicker introduction of a 0.2 g/l BAC-limit for novice drivers might have stopped this unfavourable development. The (re)introduction of special traffic police squads did, nationwide, in 2000 not result in a substantially higher enforcement level. Based on the number of mouthpieces for breathtest screening devices that were sold to the police, it is estimated that about 600,000 drivers were tested at random. This means that only 1 test in 25 inhabitants (or in 15 license holders) was performed. At this low enforcement level, the risk of apprehension for hardcore drinking drivers, who are forming only 0.3% of all drivers, is almost negligible. Consequently, general deterrence effects on hardcore drinking drivers can hardly be expected. This would even be the case, if the RBT-level would be doubled, unless the additional enforcement capacity would focus on highrisk days and hours, especially weekend nights around the closing times of pubs and bars. In that case, doubling the RBT-level would most likely be highly cost-effective, especially if it would be combined with the introduction of a lower legal limit for novice drivers, and of an alcolock programme for hardcore drinking drivers. But so far, Dutch politicians, and public opinion for that matter, seem to weigh the road toll less heavy than the loss of human life by other causes, e.g. illness, disaster or crime. As long as this mentality does not change, enforcement of drink-driving laws will not get the priority it deserves, when it comes to dividing scarce police capacity. References 1. De Leeuw J and Oppe S. Analyse van kruistabellen: loglineaire poisson modellen voor gewogen aantallen. R-97-41. SWOV, Voorburg, 1976. 2. Vogelesang AW. The analysis of weighted poisson data. D-96-12. SWOV, Leidschendam, 1996. 3. Mathijssen MPM. Drug and alcohol use by motorists in the Netherlands, 1997/1998. R-99-5. SWOV, Leidschendam, 1999. - 980 -

4. Mathijssen MPM. Dutch drink-driving decreases after new policy. In: Mercier-Guyon C (ed.). Alcohol, Drugs and Traffic Safety T97. Proceedings of the 14th International Conference on Alcohol, Drugs and Traffic Safety, Annecy, France. CERMT, Annecy, 1997, pp 1251-1258. 5. AVV. Drink-driving in the Netherlands, 1997/1999-2000 (in Dutch with English summary). Ministry of Transport and Waterways, The Hague, 2001. 6. Mathijssen MPM. The effect of the use of alcohol, illicit drugs and psychoactive medicines on the injury risk of motorists (in Dutch with English summary). SWOV, Leidschendam, 2002. 7. Nägele R and Vissers J. Gedragseffecten van de EMA. TT00-119. Traffic Test bv, Veenendaal, 2000. 8. Mathijssen MPM. Drink-driving and police enforcement in the Netherlands (in Dutch with English summary). R-2001-8. SWOV, Leidschendam, 2001. 9. Bartl G and Esberger R. Effects of lowering the legal BAC-limit in Austria. In: Laurell & Schlyter (eds.). Alcohol, Drugs and Traffic Safety - T2000. Proceedings of the 15th International Conference on Alcohol, Drugs and Traffic Safety, Stockholm. 10. Mathijssen MPM. Schatting van de effecten van verlaging van de wettelijke limiet voor alcoholgebruik in het verkeer. R-99-11. SWOV, Leidschendam. 11. Beirness DJ. Best Practices for Alcohol Interlock Programs. Traffic Injury Research Foundation, Ottawa, 2001. 12. Bax C (ed.). Alcohol Interlock Implementation in the European Union; Feasibility Study. SWOV, Leidschendam, 2001. - 981 -

- 982 -

Trends in Impaired Driving in the United States: Complacency and Backsliding? K. Stewart 1 and J. Fell 2 1 Safety and Policy Analysis, International, 3798 Mosswood Drive, Lafayette, CA 94549, USA; 2 Pacific Institute for Research and Evaluation, 11710 Beltsville Drive, Suite 300, Calverton, MD 20705, USA Keywords Alcohol, trends, impaired driving, prevention Abstract After years of decline, alcohol-related crash rates have stalled and actually increased. This paper describes recent trends and suggests a variety of strategies to reduce impaired driving. Introduction For more than a decade, rates of alcohol-related crashes have declined in the United States. In 2000, however, 40 percent of all fatalities involved alcohol, up from the historic low of 38 percent in 1999. It was the first increase in alcohol-related deaths since 1995. In 2000, 16,653 fatalities were alcohol-related, compared to 15,976 in 1999 (Table 1). This was very discouraging news, especially in light of the bold new national goal that was established in 1995 to reduce alcohol-related traffic fatalities in America to no more than 11,000 by the year 2005 (1). At that time, 125 recommendations were made to meet the ambitious goal. Despite all this activity, progress in achieving this national public health goal has been slow and for the last year movement has been in the wrong direction. In addition, the prevalence of drinking drivers on the roadways on Friday and Saturday nights did not change much in the U.S. between 1986 and 1996 (Figure 1) (2). An update of a model used to analyze the effectiveness of various impaired driving laws in the U.S. (3) projects that alcohol-related fatalities will still be at about 16,600 in 2005 if present trends continue. This slowness in progress is particularly discouraging because the tools to make significant progress are available. Many effective strategies are well known, but not implemented as widely or as vigorously as possible. Because of the significant progress that has been made in the past, complacency may have set in among policy makers. The level of public awareness and concern seems to have waned, with attention deflected to other issues. This paper will discuss some of the major strategies that can contribute to further reductions in impaired driving in the U.S. Many of these strategies have already helped reduce alcohol-related traffic fatalities in this country. Reinvigorated efforts can help regain lost ground and move us towards our goal. - 983 -

- 984 -

Strategies to reduce impaired driving The key strategies for reducing impaired driving include the legislative framework, enforcement efforts, public awareness and alternative transportation strategies. Each of these areas will be discussed below. Legislation Effective impaired driving laws form the foundation for any impaired driving program. These laws send a message to the public that impaired driving is serious and will not be tolerated. In order to have maximum effect, impaired driving laws must be well publicized and well enforced. General Deterrence Laws that send a message to all potential drinking drivers are general deterrent laws. These laws make it clear that offenders, when caught, will receive serious consequences. Some of the most important laws include: Administrative License Revocation (ALR) A total of 40 States and the District of Columbia have adopted such laws, which allow for the immediate suspension of the driving license of anyone arrested with a blood alcohol concentration over the legal limit. If the remaining ten States adopt ALR laws, an estimated 300-350 lives could be saved each year (3). Lowering the illegal per se blood alcohol concentration (BAC) limit to.08 g/dl. Most States previously adopted.10g/dl or.10 BAC as the illegal limit. However, in recent years, a number of States have lowered the illegal limit to.08 BAC. Studies show that States that have lowered the limit to.08 BAC have experienced an average 8 percent reduction in fatal crashes involving drivers with BACs exceeding.10 BAC. The majority of the public are in favor of lowering the per se limit to.08 BAC (4). At present, 30 States plus the District of Columbia (DC) have lowered their illegal BAC limit to.08. If all States adopt the.08 standard, an estimated 590 lives could be saved annually (3). Graduated Driver Licensing About half the States have adopted some form of graduated licensing system for young novice drivers. The system involves a series of distinct stages of licensing to allow young drivers to gain experience in driving under relatively low risk conditions before they graduate to the next level. The licensing system should include certain key driving restrictions such as a nighttime driving curfew and a limit on the number of young passengers in the vehicle when a novice is driving. Studies show that graduated licensing systems adopted by States result in a 5-8 percent reduction in fatal crash involvement by young drivers (5). Occupant Protection Laws While not a law that prevents impaired driving, primary safety belt usage laws have the potential to save almost 700 alcohol-related traffic deaths each year if all States adopt them (6). These laws allow police to stop and ticket drivers for failure to wear safety belts. In states without such laws, the driver must be stopped for some other offense in order to be ticketed for not buckling up. Wearing a safety restraint reduces a driver s chance of a fatal injury in a crash by 45 percent (6). Fewer than 20 States have adopted primary safety belt - 985 -

usage laws. They have the potential to reduce alcohol-related fatalities substantially because as the law permits the enforcement of safety belt laws to become more vigorous, proportionately more high risk drivers including drinking drivers - will buckle up. In States that adopt primary laws, safety belt usage goes from about 60-70% to 75-85%. Specific Deterrence Laws that deal specifically with drivers arrested for driving while intoxicated (DWI) are called specific deterrent laws. There are some promising laws that deal with chronic and repeat offenders. Some of the key laws are described below. Vehicle sanctions Many States have passed laws allowing judges to immobilize, impound, confiscate, or forfeit vehicles owned by drivers convicted of a second or multiple impaired driving offense within the past three or five years. The rationale behind these laws is that if these offenders were not deterred by the consequences of the first conviction, then they must be separated from their vehicles. Recent research is showing that these vehicle sanctions are reducing recidivism by about 50 percent (7). Some States confiscate and destroy the license plates of the vehicles as a more economical vehicle action. There is evidence that this is also effective in reducing recidivism (8). Alcohol ignition interlock devices have also been used by several States to eliminate any impaired driving by chronic DWI offenders. These devices require drivers to pass an alcohol breathalyzer test before they can start their vehicle. Interlocks allow repeat offenders to drive their vehicles on a restricted license, but not after they have been drinking. Research shows that while ignition interlocks are on these offender s vehicles, they are very effective in reducing recidivism (9). Offenders pay for the installation and a monthly fee for the monitoring and maintenance of the interlock. These vehicle-based sanctions, while they are specific deterrents to repeat DWI offenders, could have a general deterrent effect if they are highly publicized and are used in some instances on first offenders. Enforcement Highly publicized and frequent enforcement of impaired driving probably has great potential for reducing impaired driving crashes in this country. Two key areas for enforcement are described below. Sobriety checkpoints Numerous studies have shown that when sobriety checkpoints are conducted fairly frequently in a community, are publicized, and are highly visible, significant reductions in impaired driving occurs (10). Recently, a few States have committed to statewide sobriety checkpoint programs with very dramatic results. After North Carolina conducted a series of checkpoint blitzes called Booze it and Lose it, the proportion of drivers on the roadways with BACs at.05 or greater was cut in half. A different statewide program called Checkpoint Tennessee, where highly publicized checkpoints were held each weekend in 2-4 counties for a period of one year, resulted in a 20 percent reduction in impaired driver fatal crashes in Tennessee. This program saved over 100 lives in Tennessee during that year (11). - 986 -

Only 37 States in the U.S. conduct sobriety checkpoints and many of them are only conducted once or twice a year during certain holiday periods. If sobriety checkpoints are conducted at least on a weekly basis, somewhere in the State, are publicized and are visible, dramatic reductions in alcohol-related traffic fatalities would occur in this country. Enforcement of minimum drinking age and zero tolerance laws The minimum legal drinking age in the US is 21. All states also have laws, called zero tolerance laws that make it illegal for drivers under 21 to drive with even small amounts of alcohol in their systems. While minimum drinking age laws and zero tolerance laws for youth have saved thousands of lives over the years, many States do not enforce these laws to any great extent. Zero tolerance laws effectively reduce under age 21 drinking driver fatal crashes by up to 24 percent. If these laws were enforced to any great extent, their effectiveness would be even greater (12). With the population of 15-20 year old drivers continuing to grow in the U.S. and the reported increase in binge drinking for this age group increasing in recent years, a renewed emphasis on youth underage drinking enforcement is overdue. Public information and awareness Impaired driving has become socially unacceptable to most of the American public. There is evidence that public awareness of the impaired driving problem goes up and alcohol-related fatalities go down when the number of newspaper stories and television and radio coverage of this subject increases substantially. There is no question that the media coverage of citizen activists groups like Mothers Against Drunk Driving (MADD) during the 1980s dramatically changed the public s attitude about the issue. MADD put faces on the numbers and made people realize that these drunk driving tragedies happened to real people and could happen to them or a loved one (13). Publicity about impaired driving has waned in recent years and this could be one of the reasons for the stagnation in progress. Impaired driving stories compete with many other social and public health issues for earned media coverage. Renewed efforts to get impaired driving into the news once again can help capture the public s attention. Alternatives to drinking and driving Assuming there is a renewed vigor in the impaired driving problem, and many drivers get the message not to drink and drive, what alternatives can we offer them? A substantial portion of the population would like to consume alcohol in social settings away from home. Alternatives to drinking and driving must be provided in order to reduce impaired driving. Public transportation is not always easily available, but where it is it should be promoted for transportation both to and from drinking settings. Other alternative transportation strategies include designated drivers, free rides or safe rides programs in which volunteers drive drinkers home, and subsidized taxi rides (14) (15). Discussion Several well established strategies exist that can significantly reduce impaired driving in the United States if they are implemented more broadly and more vigorously. Action should be taken at the national, state, and local level to ensure that we continue to make progress in reducing alcohol-related traffic crashes (16). - 987 -

References 1. Partners in Progress: An Impaired Driving Guide for Action, National Highway Traffic Safety Administration, 1997. 2. Voas, RB, Wells, J, Lestina, D, Williams, A, and Greene, M. Drinking and driving in the United States: the 1996 National Roadside Survey. J Accid Anal and Prev 1998; 30:2, 267-275. 3. Voas, RB, Tippetts, AS, and Fell, JC. The relationship of alcohol safety laws to drinking drivers in fatal crashes. Acc Anal and Prev 1999; 32; 483-492. 4. National Safety Council and National Highway Traffic Safety Administration. Setting limits, saving lives, the case for.08 BAC laws. National Safety Council and National Highway Safety Administration, Washington (DC) 1999, NSC# 82353-0000, DOT HS 808 524. 5. National Highway Traffic Safety Administration and National Safety Council. Saving teenage lives: the case for graduated driver licensing. National Highway Traffic Safety Administration, Washington (DC) 1998, DOT HS 808 801. 6. National Highway Traffic Safety Administration. Traffic safety facts 1998: occupant protection. National Center for Statistics and Analysis, Washington (DC) 1999, DOT HS 808 954. 7. Voas, RB, Tippetts, AS, and Taylor, E. Effectiveness of the Ohio vehicle action and administrative license suspension laws. National Highway Traffic Safety Administration, Washington (DC) 1999; DOT US 809 000. 8. Rodgers, A., Effect of Minnesota s License Plate Impoundment Law on Recidivism of Multiple DWI Violators, Alcohol, Drugs, and Driving, Vol. 10, pp.127-134, 1994. 9. Beck, KH, Rauch, WJ, Baker, EA, and Williams, AF. Effects of ignition license restrictions on drivers with multiple alcohol offenses: a randomized trial in Maryland. Am J Pub Health 1999; 89(11). 10. National Highway Traffic Safety Administration. Alcohol and highway safety 1984: a review of the state of knowledge. National Highway Traffic Safety Administration, Washington (DC) 1985, DOT HS 806 569. 11. Lacey, JH, Jones, RK, and Smith, RG. An evaluation of Checkpoint Tennessee: Tennessee s statewide sobriety checkpoint program. National Highway Traffic Safety Administration, Washington (DC) 1999, DOT HS 808 841. 12. Ferguson, SA, Fields, M and Voas, RB. Enforcement of zero tolerance laws in the United States. Presented at the Insurance Institute for Highway Safety s 15 th International Conference on Alcohol, Drugs and Traffic Safety, Stockholm, Sweden, 2000. - 988 -

13. McCarthy, J. A media framing context: it s shape, newspaper coverage outcomes and impact upon the citizens movement against drunk driving. Paper presented at Workshop on Social Movements, Counterforces and Bystanders, Berlin, Germany, 1990. 14. Winsten, J. Promoting designated drivers: the Harvard Alcohol Project. Amer J Prev Med 1994; 10 (suppl 1):11-4. 15. Fell, J, Voas, RB, and Lange, JE. Designated driver concept: extent of use in the USA. J Traff Med 1997; 25(3-4). 16. Fell, J. Keeping us on track: a national program to reduce impaired driving in the United States. J Sub Use 2001; 6: 258-268. - 989 -

- 990 -

Drinking and Driving in France in 1999-2000 : Changes in The Decade and New Perspectives M. B. Biecheler-Fretel 1, F. Facy 2, J. F. Peytavin 1 1 INRETS-DERA, 2, avenue du Général Malleret-Joinville, F-94 114 Arcueil cedex 2 INSERM, U 302, 44, Chemin de Ronde, F-78 116 Le Vésinet cedex Keywords Alcohol levels, road accidents, convicted drivers, age, health risk Abstract Several sources of data in France are able to describe the different aspects of drinking and driving behavior, including: consumption, preventive tests and offences, medical consulting, and accidents. Analysis of these data over the last decade raises new questions in France about the strategies to prevent road risk and/or health risk. The main data to be analyzed are from: 1) roadside checks (1990-2000), 2) populations and sub populations of drivers involved in accidents with alcohol (1990-2000), 3) populations of drivers convicted for driving under the influence, their punishment by judges (1995-2000), and 4) surveys about medical care needed by offenders (1998-1999). We compare the profiles of offenders and involved drivers, focusing on an analysis by age. The drinking-driving prevention system in the period between 1990 and 2000 appears to have reduced the number of accidents that can be attributable to alcohol. In 1999-2000, alcohol is as much present among young drivers (18-24 yrs) as among the older (25-40 yrs ): from 40% to 50% illegal BAC in fatal accidents on week-end nights. The national testing system results in around 100,000 convictions each year and a third of the cases need medical care.the roads (and the courts) have a massive potential for detecting the alcoholism of older drivers (above 30 years and moreover above 40 years) and the problem consumption of younger drivers. The possibility of using traffic enforcement as a means of educating drivers about alcohol risks will be discussed. Introduction The blood limit 0.8 g/l, as well as its equivalent breath limit 0.4 mg/l, became an indictable offence in 1983. On the same year, in 1983, was created the French interministerial national observatory of road safety (ONISR), the role of which was to produce annual statistics and trends in road traffic safety. Concerning alcohol the observatory publishes data resulting from alcohol testing by police forces (accidents, offences and random tests), it publishes also the main statistical features of the judicial treatment of offending. In September 1995 the legal blood limit went further down to 0.5g/l (0.5 pro mille). The decree introduced a distinction between two levels : 0.5 g/l (blood) for a minor offence and 0.8 g/l (blood) for an indictable offence. Most part of the epidemiological knowledge in the field of drink driving behaviour in relation to accidents has been produced by INRETS (1) (2). Two other major contributions are to be mentioned : Got et al (3) studying specifically drivers presumed to be responsible of a fatal accident brought on - 991 -

this point a major contribution ; in the survey by the national institute on medical research (INSERM) and the high committee of studies and information on alcoholism (HCEIA) the distinction between chronic and occasional consumption for drivers involved in accidents had been investigated : 36% of men and 10% of women presented signs of chronic alcoholization (4). Since 1996 the only data available come from the ONISR (5), but these do not include a characterization of alcohol-related accidents. A previous work (6) has shown that the dissuasive prevention system to drink-driving based on legislation and enforcement tended over 1990-1998 to reduce the number of accidents that can be attributable to alcohol consumption. In this paper we will firstly control that this trend proves to be stable in 1999 and 2000. We then will present some results helping to discuss to which extent the system for preventing and punishing drinkdriving constitutes a framework for detecting alcohol risks, road risk as well as health risk. Methods and data Evaluation of trends at the national level Since there has been no scientific survey of alcohol levels on roads during the nineties, available data come from official sources based on enforcement roadside checks. They were obtained from the drink-driving monitoring activities of police officers between 1990 and 2000 (5). Data about subsequent convictions are provided by the Ministry of Justice (7). Alcohol levels of drivers involved in injury accidents from 1990 to 1996 have been obtained from the INRETS report file, where are collected one out of fifty accident records each year since 1987 (8). The specific exploitation of this 1/50 e file on the alcohol factor was interrupted in 1996. So in order to investigate the more recent distributions of alcohol levels for drivers involved in injury accidents we used the French national accident file 1999-2000 (the BAAC -Bulletin d Analyse d Accident Corporel de la circulation file). To describe synthetically the prevalence of alcohol in accidents, among the criteria found to be linked to the presence of alcohol in epidemiological studies we selected : age, time and day on which the accident occurred, and seriousness. A preliminary examination of the data led us to select three periods in day :7h-13h, 14-20h and 21h- 6h. An analysis through age of road versus heath risk We used age as the pertinent criteria to evaluate the road and health risks of the population. The seven age categories selected correspond to a common basis through the different sources (< 18, 18-19, 20-24, 25-29, 30-39, 40-59 and >60). Two sources of data, one national and one local, allow to estimate what proportion of drivers arrested with illegal blood alcohol level (BAC) need a medical care. Drivers tested positive when driving are generally invited by courts to attend an alcohologic consultation in a specific health center, to be informed or oriented The national association for prevention of alcoholism (ANPA) register the drivers consulting and is able to distinguish two groups : those who have a problematic consumption and those who need information on road and/or health risk (9). The survey conducted by INRETS in the French department of Val d Oise, allows to estimate what proportion of drivers arrested with illegal BAC need a medical care (10). In this experiment, the drivers were invited, before their judgment to perform biological tests and then to attend a alcohologic consultation. For each driver the doctor make a diagnosis (occasional consumption or alcohol problem). The results provide an estimation of the presence an alcohol related health problem among illegally drink-drivers. - 992 -

Results 1 Enforcement and alcohol related to accident trends (1990-2000) Roadside breath tests : enforcement trends (1990-2000) Under the current legal framework alcohol tests are compulsory in cases of injury accidents or when an offence has been committed. Random tests are carried even when no accident or offence has taken place. A feature of random breath tests is that they cover the entire population of drivers on a non-selective basis. Table 1 shows the way the number of breath tests (both random and those administered after an offence) have changed over the decade. The number of random breath tests has risen steadily over the decade, from 2,900,000 tests in 1990 to 7,900,000 in 1999. For the first time in 2000, we observe a slight decrease : this mean that probably the preventive system has reached its maximum capacity. On the other hand, he number of tests after offences (about 1.5 million) remains more or less the same throughout the period 1990-2000. The percentage of positive random tests has risen after 1995 : from 1% or 1.1% on 1990-1995 to 1.3% or 1.4% on 1997-2000. This is in part be attributable to the lowering of the legal limit to 0.5 g/l in September 1995 which means that levels of between 0.5 and 0.8 now count as positive (6). Table 1: Random breath tests and breath tests after an offence (1990-2000) Breath tests performed Year Random tests Tests after an offence 1990 2,881,232 1,619,531 1991 3,812,442 2,205,045 1992 4,614,079 1,662,413 1993 5,542,319 1,579,987 1994 6,270,045 1,664,788 1995 6,650,690 1,595,707 1996 6,286,419 1,532,461 1997 6,677,808 1,522,785 1998 6,908,932 1,491,951 1999 7,925,604 1,332,022 2000 7,804,267 1,409,560 Percentage of positive tests Random tests Tests after an offence 1.1% 3.6% 1.0% 2.9% 1.1% 3,0% 1.0% 3,8% 1.0% 2.9% 1.0% 2.9% 1.2% 2.9% 1.3% 3.0% 1.4% 3.4% 1.4% 3.8% 1.3% 3.6% From: ONISR(5) Drink-driving offences and sentences (1995-1999) In 1998, 20% of drink-driving offences came under the minor offence category and 79% were indictable offences. The percentage of minor offences came up to 23% in 1999 and 24% in 2000 while the percentage of indictable offences came down to 77% and 76%. Drink-driving offences account for around 23% of the indictable offences dealt by the courts (23% in 1998 and in 1999). They represent throughout the 1990s, over 100 000 convictions each year (101 636 in 1998 and 103 088 in 1999). This is the judicial response to the intensive screening testing by police forces that has been growing since the last ten years. The recidivism rate is increasing each year (national estimation of 10% in 1999 (5)). The punishment of drink-driving is slowly shifting from imprisonment to fines and measures - 993 -

concerning driving license. In 1999 we observe as first sentences : custodial imprisonment 6%, suspended imprisonment 54%, fines 21%, driving disqualification 19% ; when the first sentence is prison or fine, most part of time a second sentence is the disqualification from driving. In case of suspended sentence, the use of probation with compulsory treatment is not largely used in France but it is slowly increasing : 21% of suspended sentences in 1999 vs 14% in 1995. Overall, the suspended imprisonment with probation is applied in 1999 to 11% of the offenders. Drivers involved with alcohol in accidents: general trends (1990-2000) In the period 1990-2000 the number of injury accidents fell continuously (from 162 573 in 1990 to 121 233 in 2000).The general trend has been a 25% fall.table 2 gives the proportion of drivers, involved in those accidents, with an alcohol level over the legal limit. In 1999, 214 156 drivers have been involved in injury accidents : 14% of their alcohol levels were not determined and of the known cases 5.5% were over the legal limit 0.5g/l. In 2000, 207 554 drivers have been involved in injury accidents : 12 % of the alcohol levels were not determined and of the known cases 5.5 % were over the legal limit 0.5g/l. In 1999 (resp 2000) 11855 drivers (resp 1154 drivers) have been involved in fatal accidents, the alcohol levels were unknown in 26% (resp 24%) of the cases, and of known cases 16,4% (resp 16,2%) were over the legal limit. Since 1993 the BAAC information on alcohol has improved and the most probable is that information about alcohol from treatment of INRETS PV and BAAC figures are very similar now. For instance the1996 percentage of unknown levels in the 1/50 e PV file was 14% for injury accidents and 21% for fatal accidents, to be compared to 12% and 24% and in the national file BAAC 2000. It appears that on the decade the weight of the alcohol factor in accidents do not increase and even would tend to decline. Thus it is possible to extrapolate to the entire decade the conclusions drawn two years ago from a analysis of specific accident series 1990-1998 : there has been a greater reduction for the categories of accidents where alcohol is most frequently present than for those where it is generally rare. Table 2: Accidents and drivers involved in accidents : percentages of illegal alcohol levels (the percentages are calculated on known cases) Year Number of Injury accidents of which Fatal Accidents Accident involved drivers Accident involved drivers % results over the legal limit Injury accidents 1990 162 573 9 128 7% 15% 1991 148 890 8 509 9% 16% 1992 143 362 8 114 9% 22% 1993 137 500 8 005 8% 19% 1994 132 760 7 609 8% 19% 1995 132 949 7 453 7% 24% 1996 125 406 7 178 8% 24% 1999 124 387 7 185 214 156 5.5% 16.4% 2000 121 223 6 811 207 554 5.5 % 16.2% * Before 1996 the legal level was 0,8 (blood), since it is 0,5 (blood) From : PV-INRETS Fontaine et Gourlet (8) and BAAC file exploited by Peytavin in 2002 Fatal accidents - 994 -

2 Structure of drink-driving by age (1999) Comparison of driving while impaired and accident involved drivers Table 3 gives the distribution by age of the drivers found with alcohol levels over the limit when involved in accidents on the one hand and when arrested on roads on the other hand. The figures are given for the year 1999 at the national level. The last column gives the ratio of over road-risk by age. Young drivers 18-19 and from 20 to 24 years old are over-represented in alcohol-related accidents compared with the population of the same age driving under influence (ratios 2.6 and 1.5) ; drivers in the age class 24-29 years are also over-represented in accidents but in a less extent (ratio 1.2). A switch occur beyond 30 years : the 30-39 years old drivers are over represented among DUI than among alcohol-related accidents involved drivers (ratio 0.9) ; this observation becomes sharper after 40 years : the ratio is 0.7 for the 40-59 years class. Above 60 years, drivers are equally represented among crash involved with illegal BAC and drink-driving offenders. Frequency of alcohol problems among convicted drivers The Val d Oise survey (8) reveals that a large part of drivers convicted for DUI need medical care for a alcohol problem. Around 60% (62%) of those drivers with an alcohol level over 1.4 g/l (blood) when arrested have an health problem with alcohol. This frequency relatively unfrequent before 25 years (12%) is growing fastly, reaching 41% in the range 20-29 years, 64% in the range 30-39 and culminating at 72% in the range 40-59 years. Adjusting the results to be representative of all offenders of Val d Oise (extropolating this frequency 62% to unknown data and supposing that levels under 1.4 represent occasionnal consumption) lead to estimate to 42% the overall part of drivers with need medical care. The estimation from the national ANPA file is lower: 34% of the drivers consulting after road arrest were orientated towards medical circuit. Anyway more than one out of three drivers alcohol offenders present an health risk (9). Table 3: The distributions by age category of drivers involved in accidents and drivers convicted for DUI (year 1999) Age % illegal BAC of drivers involved 183 887 Involved drivers with illegal BAC 10 056 Offenders (indictable offences) 103 088 Ratio % illegal BAC/ % offenders <18 0.8 % 0.6 % 0.1 % 6 18-19 yrs 4.2 % 4.1 % 1.6 % 2.6 20-24 yrs 6.3 % 18.1 % 11.7 % 1.5 25-29 yrs 6.7 % 18.1 % 14.8 % 1.2 30-39 yrs 6.5 % 26.9 % 29.1 % 0.9 40-59 yrs 5.4 % 28% 38.3 % 0.7 Over 60 yrs 2.5 % 4.3% 4.4 % 1 Total 5.5% 100% 100% 1 From: BAAC file exploited by Peytavin in 2002, Annual judicial statistics (7) - 995 -

Prevalence of illegal alcoholisation in accidents (1999-2000) Table 4 shows the proportions of drivers involved in accidents with an alcohol level over the legal limit. The results confirm previous features of the phenomena, with the high presence of alcohol on nights, on week-ends and especially on week-end nights whatever the age. Table 4: Drivers involved in injury and fatal accidents in 1999 : % illegal alcohol levels by category of day, period of day and age (the percentages are calculated on known cases) Weekday 7h-13h Weekday 14h-20h Weekday 21h-6h Week-end 7h-13h Week-end 14h-20h Week-end 21h-6h Injury Fatal Injury Fatal Injury Fatal Injury Fatal Injury Fatal Injury Fatal <18 0.1 0 0.4 0 2.3 6.2 0.6 0 0.5 3.8 7.6 18.8 18-19 yrs 0.6 5.9 1 6.2 8.4 23.5 3.4 13.3 1.9 8.6 18 42.1 20-24 yrs 0.7 5.2 1.8 5.7 12.7 32 6.4 22.4 3.8 12.6 24 49.6 25-29 yrs 0.6 1.9 2.7 8.2 15.9 32.8 5.3 14 5.6 17.6 27.2 52.6 30-39 yrs 09 3.4 4 124 16.9 34.7 3.4 6.3 8.2 21.6 27 43.7 40-59 yrs 1.1 3.8 5 15.1 13.9 20.6 2.5 6.3 8.8 18 27 27.6 >60 yrs 0.8 3.4 2.5 8.2 8.1 15.1 1.1 0.8 4 13.2 18.4 3.0 Total 0.8 3.6 3.3 10.9 13.8 28.5 3.4 8.7 6.1 16.6 23 42.8 From : BAAC file exploited by Peytavin in 2002 Discussion Alcohol investigation practices have been improving during the last ten years as well for injury accidents (fatal or not) as for fatal accidents. The results of investigation show that the prevalence of illegal alcohol levels in road accidents does not increase and even would tend to diminish, especially for fatal accidents. This progress is attributable to the massive alcohol screening enforcement on roads. To day more than 9 billions of screening tests a year are performed, of which around 150 000 are positive leading to 100 000 convictions (103 088). In front of this preventive-repressive system, 10 000 drivers involved in injury accidents involved have an alcohol level over the blood limit 0.5 g/l representing 5.5% of the drivers involved. Alcohol, whatever the age is linked with specific accident profiles : fatal accidents on nights, week-ends and especially on week-end nights. The system for preventing and punishing drink-driving, based in France on intensive roadside screening tests, leads to detect among the 100 000 drivers convicted each year, of whom more than one out of three need medical care for an alcohol problem. When two types of risk are associated with drink-driving, the critical age threshold, where health risk becomes predominant on road risk is 30 years and more clearly 40 years. The number of suspended imprisonment with probation (11%) is not in proportion with the convicted drivers with alcohol problems (more than 33%), so that the important challenge is to develop more effective interventions for alcohol offending drivers. The analysis by age stresses the fact that it is important to recognize the road risk problem of the younger and the frequent health risk problem of the older. As alcohol has been classified by experts as an «hard drug» (9), it becomes a priority to use traffic enforcement system as a means of educating drivers about alcohol risks - 996 -

References 1. Biecheler-Fretel M B et al. Alcool, conduite et insécurité routière. Synthèse des travaux de l ONSER. Cahier d études ONSER n 65, avril 1985. 2. Biecheler-Fretel M B, Filou C, Fontaine H. Drink driving and alcohol related accidents,trends an appraisal 1985-1995. In : Synthèse INRETS n 35, pp 57-127. 3. Got C, Faverjon G, Thomas C. Alcool et accidents mortels de la circulation. In Bulletin d Information du HCEIA, 1984 ; 1: 38-60. 4. Haut comité d études et d information sur l alcoolisme (HCEIA). Alcool et Accidents. Ouvrage collectif, série documents, ed : La Documentation Française, 1985, pp 1-154. 5. ONISR (Observatoire National Interministériel de Sécurité Routière). La Sécurité Routière en France. Bilan de l Année 2000. Ed : La documentation Française, Paris 2001. 6. Biecheler-Fretel M B, Filou C. Drinking and driving in France in the nineties, can evaluation help prevention? In : Proceedings of the 15 th International Conference, Stockholm 2000, vol 2 : 375-381. 7. Annuaire de la justice. Édition 2001. Séries 1995-1999 ; Ed: La Documentation Française. 8. Fontaine H, Gourlet Y. Mobilité et Accidents, année 1996. Rapport sur convention DSCR/INRETS, juillet 1998, pp 1-67. 9. Facy F et al. Dépistage de l alcoolémie sur route, délits de conduite en état alcoolique et consultations en centre d alcoologie. To be published in Recherche-Transports-Sécurité (accepted march 2002). 10. Biecheler M B, Peytavin J F. Preventing drink-driving and reoffending : the experiment conducted in the french département of Val d Oise (1996-2000). In proceedings of the 16 th International conference on alcohol, drugs and traffic safety 2002. 11. Roques B P. La dangerosité des drogues. Rapport au secrétariat d état à la santé, Ed Odile Jacob, 1999. - 997 -

- 998 -