(Received 27 May 2004; first review notified 26 August 2004; in revised form 2 November 2004; accepted 14 December 2004)

Transcription

1 Alcohol & Alcoholism Vol. 40, No. 2, pp , 2005 Advance Access publication 17 January 2005 doi: /alcalc/agh135 PREDICTING DRUNK DRIVING: CONTRIBUTION OF ALCOHOL USE AND RELATED PROBLEMS, TRAFFIC BEHAVIOUR, PERSONALITY AND PLATELET MONOAMINE OXIDASE (MAO) ACTIVITY DIVA EENSOO 1, MARIKA PAAVER 2, MAARIKE HARRO 1 and JAANUS HARRO 2 * 1 Division of Health Promotion, Department of Public Health, University of Tartu, Ravila 19, Tartu, Estonia and 2 Department of Psychology, University of Tartu, Centre of Behavioural and Health Sciences, Tiigi 78, Tartu, Estonia (Received 27 May 2004; first review notified 26 August 2004; in revised form 2 November 2004; accepted 14 December 2004) Abstract Aims: The aim of the study was to characterize the predictive value of socio-economic data, alcohol consumption measures, smoking, platelet monoamine oxidase (MAO) activity, traffic behaviour habits and impulsivity measures for actual drunk driving. Methods: Data were collected from 203 male drunk driving offenders and 211 control subjects using self-reported questionnaires, and blood samples were obtained from the two groups. Results: We identified the combination of variables, which predicted correctly, ~80% of the subjects belonging to the drunk driving and control groups. Significant independent discriminators in the final model were, among the health-behaviour measures, alcohol-related problems, frequency of using alcohol, the amount of alcohol consumed and smoking. Predictive traffic behaviour measures were seat belt use and paying for parking. Among the impulsivity measures, dysfunctional impulsivity was the best predictor; platelet MAO activity and age also had an independent predictive value. Conclusions: Our results support the notion that drunk driving is the result of a combination of various behavioural, biological and personality-related risk factors. INTRODUCTION Although drunk driving has decreased in many countries during the past 20 years (European Transport Safety Council, 2001), alcohol consumption is still one of the main causative factors in road traffic accidents. Alcohol impairs the reaction time of the drunk drivers and their ability to estimate risks adequately, and drunk driving is considered to be a serious violation of traffic law. The reason why some people engage in this kind of risky behaviour is probably due to a combination of various factors. Quantity (Wilson and Jonah, 1985) or frequency (Grunewald et al., 1996; Baum, 2000) of alcohol consumption or both (Johnson et al., 1998) are related to drunk driving and the probability of alcohol-related injury. Several studies have found that drunk driving is associated with a lower income (Baum, 2000; Golias and Karlaftis, 2002), a lower educational level and is more frequent among blue-collar workers (Baum, 2000). However, some studies have not found any association between drunk driving and socio-economic measures (Wilson and Jonah, 1985; Grunewald, 1996). Drunk drivers have been found to break other traffic rules more frequently, including driving without a driving licence (Baum, 2000; Begg et al., 2003), not using a seat belt, and exceeding the driving speed limits (Golias and Karlaftis, 2002). Personality traits are the underlying factors affecting peoples estimations, attitudes and behavioural decisions. One of the most studied traits in association with risky driving is sensation seeking (Jonah, 1997; Iversen and Rundmo, 2002). Sensation seeking, which is sometimes also referred to as novelty or excitement seeking, is a trait described by the constant need for novel and intense sensations and experiences. Another trait commonly associated with risk taking and rule breaking, as well as alcohol-related problems, is impulsivity (von Knorring and Oreland, 1996). The association between impulsivity, risky driving and motor *Author to whom correspondence should be addressed at: Tel.: ; Fax: ; jaanus.harro@ut.ee vehicle crash injuries has also been shown (Hilakivi et al., 1989; Cherpitel and Tam, 2000). Impulsivity is defined by a range of various tendencies, including rapid and thoughtless action, low self-control, risk-taking and the inability to hold back one s desires. It is known that genetic factors predict about half of the variance of personality traits, depending on the variety of environmental impacts. Social maladaptation and personality traits such as impulsivity, sensation seeking and monotony avoidance have been correlated with low levels of monoamine oxidase (MAO) activity in platelets, which is a peripheral marker of serotonergic activity in the central nervous system (von Knorring et al., 1984; Oreland, 2004). Platelet MAO activity is also lower in alcohol-dependent subjects (Cloninger et al., 1988; von Knorring and Oreland, 1996). It has been proposed that risky driving has a common biological basis with sensation seeking (Jonah, 1997). Nevertheless, empirical support for such assumptions is scarce. We have compared a group of police-referred drunk drivers with a group of controls, taking into account self-reported drinking and driving, socio-economic variables, usual alcohol consumption, smoking, other risky traffic behaviour habits, impulsivity measures and platelet MAO activity. METHODS Subjects Driving while impaired by alcohol (DWI) group (n = 203; mean age ± SD: 33 ± 11 years) was composed of men who were caught by the police during the previous 12 months. Control group (n = 211, mean age 36 ± 12 years) was selected by a computerized random choice from the driving licence database of the Estonian Motor Vehicle Registration Centre. Subjects were contacted by telephone and a description of the study was provided. Four hundred and sixteen men (27% of the contacted people) agreed to participate in the study. Two men dropped out of the study at the stage of filling out the 140 Alcohol & Alcoholism Vol. 40, No. 2 Medical Council on Alcohol 2005; all rights reserved

2 DRUNK DRIVING PREDICTORS 141 questionnaires. Eight subjects recruited as controls had an earlier record of drunk driving in the police database, and were moved into the DWI group. The Ethics Committee at the Faculty of Medicine of the University of Tartu approved this study. Socio-economic background, smoking, alcohol use and traffic behaviour Subjects reported their socio-economic status, smoking, alcohol use habits and traffic behaviour in a self-reported questionnaire. Questions about socio-economic background included marital status, education, monthly income, occupation and religiousness. Based on self-reported smoking behaviour, the subjects were divided into non-smokers, ex-smokers and smokers who smoked 10 cigarettes per day, cigarettes per day and 20 cigarettes per day. For evaluating alcohol consumption habits, the questionnaire contained items about the frequency of using strong and light alcoholic drinks during the previous year on a 6-point scale (none, some times during the year, one to three times per month, one to two times per week, three to four times per week, almost every day). The amount of alcohol consumed during the previous week before the study was evaluated on the basis of the amount of different alcoholic drinks and was expressed in grams of pure alcohol. The score of alcohol-related problems was obtained by summing up five questions based on the DSM-IV diagnostic criteria for alcohol abuse, relating to specific life events (having turned aggressive while drunk, having had longer periods of alcohol use, having had conflicts with friends and family, having been absent from work, and having lost one s job ; reported as present or not, total score 0 5). Questions relating to traffic behaviour habits such as the frequency of car driving, using the seat belt, exceeding the speed limits, paying for parking, stopping before a zebra crossing and overtaking during the previous year, as well as the duration of having a driving licence, were included. The subjects were also asked about driving while impaired by alcohol during the previous year. In addition, the knowledge of time limits for safe drinking before driving was checked. Personality measures Impulsivity was measured by a short instrument based on Dickman Impulsivity Inventory (Dickman, 1990) and impulsivity-related subscales of NEO Personality Inventory (NEO-PI, Costa and McCrae, 1989). Altogether, four scales were formed Dysfunctional and Functional Impulsivity based on Dickman Inventory, and Impulsivity and Excitement Seeking based on NEO-PI subscales. Each of the four scales consisted of six items. Measurement of platelet MAO activity Venous blood samples were collected into 4.5 ml vacutainers containing K 3 EDTA as an anticoagulant. MAO activity was analysed in platelet-rich plasma by a radioenzymatic method with [ 14 C]- -phenylethylamine as the substrate, according to the procedure described by Hallman et al. (1987) and Harro et al. (2001). All samples were analysed in duplicate and blindly, and corrected using a reference sample. MAO activity is expressed as nanomoles of substrate oxidized per platelets per minute. Statistical analysis Nominal variables were described using frequency tables. Pearson s Chi-square test with post hoc analysis of individual cell contributions was used. If a significant difference was found with analysis of variance (ANOVA), Fisher s multiple comparison procedure was used. Covariation analysis (ANCOVA) was used for controlling the effect of the third variable if significant effects were revealed in the ANOVA analyses. Kruskal Wallis test was used for comparing the nonparametrically distributed variables. Spearman s correlation analysis was used. Discriminant analysis was performed to identify the combination of quantitative variables that predict drunk driving. RESULTS Self-reported driving after drinking versus police-reported drunk driving In the control group, 81 (38%) men reported driving after drinking (DAD) sometimes or often during the previous year and in the DWI group, 45 (22%) subjects denied DAD. Thus, for obtaining more homogenous groups, we separated DADreporters from the control group and DAD-deniers from the DWI group, obtaining four groups. Hence, we obtained the following groups: Control I (129 subjects) random choice from driving licence database who denied DAD during the previous year; Control II (81 subjects) random choice from driving licence database who reported DAD sometimes or often per year; DWI I (45 subjects) DWI subjects who denied DAD during the previous year; and DWI II (157 subjects) DWI subjects who reported DAD sometimes or often per year. Socio-economic background, smoking and alcohol use habits The four groups did not differ by marital status, monthly income or by occupational categories (data not shown). Statistically significant differences appeared with respect to median age, education, belief in religion, smoking behaviour, frequency of using strong and light alcoholic drinks, the amount of alcohol consumed in the past week before the study and having alcoholrelated problems (Table 1). The median age of the DWI II group s subjects was significantly lower, compared with other groups. The subjects of the DWI II group also had a lower educational level compared with both control groups. In the DWI II group, there were more subjects who had elementary to high school education, and fewer subjects who had a university education compared with the control groups. The Control I group was clearly different from the other groups with respect to religiousness, smoking status and alcohol consumption habits. There were more religious subjects and non-smokers in the Control I group, and the use of strong and light alcohol was also lower among the Control I group when compared with the other groups. While the DWI II group did not differ from the Control II group by the amount of alcohol consumed during the previous week and by the frequency of using light alcoholic drinks, the score of alcohol-related problems and frequency of using strong alcoholic drinks were higher in the DWI II group. Traffic behaviour The four groups did not differ with respect to the frequency of stopping before a zebra crossing to allow the pedestrians to

3 142 D. EENSOO et al. Table 1. Comparison of socio-economic background, smoking habits and alcohol use of the DWI and control groups Control I Control II DWI I DWI II Parameter (n = 129) (n = 81) (n = 45) (n = 157) P Age (years) Median a,b,c Range Education, n (%) Elementary to high school 85 (66) 52 (64) 36 (80) 129 (83) University 44 (34) 29 (36) 9 (20) 27 (17) Religious, n (%) No 99 (77) 75 (93) 38 (88) 135 (86) Yes 30 (23) 6 (7) 5 (12) 22 (14) Smoking, n (%) < Non-smokers 54 (42) 20 (25) 8 (18) 23 (15) Ex-smokers 34 (26) 28 (34) 10 (22) 34 (22) Smoked 1 10 cigarettes per day 21 (16) 12 (15) 9 (20) 30 (19) Smoked cigarettes per day 4 (3) 14 (17) 10 (22) 30 (19) Smoked 20 cigarettes per day 16 (13) 7 (9) 8 (18) 40 (25) Frequency of using strong alcohol, n (%) < None to some times per year 73 (58) 12 (14) 14 (31) 25 (16) One to three times per month 46 (37) 50 (62) 25 (56) 93 (60) One time per week or more often 6 (5) 19 (24) 6 (13) 38 (24) Frequency of using light alcohol drinks, n (%) < None to three times per month 80 (64) 23 (29) 26 (58) 46 (29) One to two times per week 33 (26) 30 (38) 14 (31) 61 (39) Three times per week or more often 12 (10) 27 (34) 5 (11) 50 (32) Alcohol consumption in the previous week (g) < Median a 34 a,b 52 a,c Range Mean±SD 40±64 118±152 95± ±22 Alcohol-related problems < Median 0 1 a 1 a a, b, c 2 Range Mean ± SD 0.5 ± ± ± ± 1.5 Control I, control subjects who denied driving after drinking; Control II, control subjects who reported driving after drinking; DWI I, police-referred drunk drivers who denied driving after drinking; DWI II, police-referred drunk drivers who reported driving after drinking. a P < 0.05 for comparison with Control I group; b P < 0.05 for comparison with Control II group; c P < 0.05 for comparison with DWI I group. Significantly different proportions in distribution (by post hoc cell contribution analysis) are bold typed. Mean ± SD data are presented for illustrative purposes. cross the road and by the frequency of overtaking (data not shown). Statistically significant differences between the groups appeared in the frequency of car driving, duration of having a driving licence, frequency of using a seat belt, exceeding the driving speed limits, paying for parking and a knowledge about time limits for safe drinking before driving (Table 2). Most of the subjects were in the habit of driving almost every day. A few men in the DWI groups reported that they had stopped driving or that were in the habit of driving without possessing a driving licence. In the DWI II group, the number of subjects having a driving licence for 2 years was three or more times higher compared with other groups. In their frequency of using the seat belt, the subjects of the DWI II group answered more often mostly not and the Control I group more often almost always compared with the other groups. In the frequency of exceeding the driving speed limits, the subjects of the Control I and also the DWI I groups answered more often mostly not and the subjects of the DWI II group answered more often occasionally. In the frequency of paying for parking, the subjects of the Control I group answered more often almost always and the subjects of the DWI II group more often occasionally. Altogether, two thirds of subjects did not know about safe time limits for drinking before driving. The difference between the groups was minimal, but knowledge was a little better among DWI II group subjects. Personality measures The groups did not differ significantly in functional impulsivity (mean score ± SD: 19.6 ± 4.8, 19.7 ± 4.7, 20.4 ± 3.6, 20.3 ± 4.2 for Control I, Control II, DWI I, DWI II, respectively), which showed a tendency to be higher in the DWI groups compared with control groups. Statistically significant differences between the groups appeared with respect to the measures of dysfunctional impulsivity (13.9 ± 4.6, 15.4 ± 4.5, 15.5 ± 5.6, 16.8 ± 4.9, respectively; F 3,408 = 8.55, P < ), NEO-PI Impulsivity (16.3 ± 4.6, 17.8 ± 4.0, 17.9 ± 4.9, 18.8 ± 4.0, respectively; F 3,408 = 8.37, P < ) and excitement seeking (20.0 ± 5.5, 20.6 ± 5.5, 18.7 ± 5.7, 21.4 ± 5.2, respectively; F 3,408 = 3.69, P = 0.012). The scores of dysfunctional impulsivity, NEO-PI impulsivity and excitement seeking were higher in the DWI II group compared with other groups. DWI II group subjects were younger compared with the other groups and it is known that younger people, in general, score slightly higher in

4 impulsivity (Eysenck et al., 1985). Age had a significant covariate effect (F = 19.29, P < ) on NEO-PI impulsivity, but the group effect remained significant (F = 6.41, P < ). The significant difference between the groups in excitement seeking disappeared after taking age into account (the covariate effect of age on excitement seeking was F = 77.95, P < , and the group effect was not significant F = 1.35, P = 0.3). DRUNK DRIVING PREDICTORS 143 Table 2. Traffic behaviour of the driving while impaired (DWI) and control groups Control I Control II DWI I DWI II Parameter (n = 129) (n = 81) (n = 45) (n = 157) P Frequency of car driving, n (%) Some times per year to three times per month 17 (13) 4 (5) 2 (4) 12 (8) 1 4 times per week 15 (12) 8 (10) 4 (9) 19 (12) Almost every day 97 (75) 69 (85) 36 (80) 123 (79) Does not drive any longer (7) 1 (1) Duration of having the driving licence, n (%) years or less 7 (6) 6 (7) 3 (7) 21 (13) 3 5 years 26 (20) 13 (16) 8 (18) 29 (19) 6 15 years 43 (33) 29 (36) 14 (31) 58 (37) years 30 (23) 26 (32) 7 (16) 26 (17) 26 years or more 23 (18) 7 (9) 11 (24) 15 (10) Without driving licence (4) 6 (4) Frequency of using the seat belt, n (%) Mostly not 7 (6) 10 (12) 2 (5) 27 (17) Occasionally 21 (16) 21 (26) 10 (22) 43 (27) Almost always 101 (78) 50 (62) 33 (73) 87 (56) Frequency of exceeding the limited driving speed, n (%) Mostly not 63 (49) 26 (32) 26 (58) 57 (36) Occasionally 47 (36) 36 (44) 17 (38) 82 (52) Almost always 19 (15) 19 (24) 2 (4) 18 (12) Frequency of paying for parking, n (%) 0.03 Mostly not 8 (6) 11 (14) 1 (2) 16 (10) Occasionally 9 (7) 9 (11) 4 (9) 26 (17) Almost always 112 (87) 61 (75) 40 (89) 114 (73) Knowledge of safe drinking before driving, n (%) Yes 34 (26) 28 (35) 10 (22) 62 (39) No 95 (74) 53 (65) 35 (78) 95 (61) Control I, control subjects who denied driving after drinking; Control II, control subjects who reported driving after drinking; DWI I, police-referred drunk drivers who denied driving after drinking; DWI II, police-referred drunk drivers who reported driving after drinking. Significantly different proportions in distribution are bold typed (by post hoc cell contribution analysis). Platelet MAO activity Mean values of platelet MAO activity were 8.3 ± 4.1(SD), 7.7 ± 4.1, 6.2 ± 2.2 and 6.7 ± 4.5 nmol/10 10 platelets/min for Control I, Control II, DWI I, and DWI II groups, respectively. After logarithmic transformation of MAO activity, ANOVA and Fisher s multiple comparison procedure was performed. The DWI I and DWI II groups mean MAO activities were both significantly lower compared with the Control I and Control II groups (F 3,399 = 7.51, P < ; Fig. 1). Platelet MAO activity was significantly negatively correlated with alcoholrelated problems (r = 0.20, P < ) but not with any measures of alcohol consumption. Predicting DWI To identify the combination of quantitative variables, which differentiate the most distinct groups (Control I group and DWI II group), discriminant analysis was carried out. Alcohol-related problems, frequency of using strong and light alcoholic drinks during the previous year, seat belt use, Fig. 1. Platelet MAO activity in the drunk drivers and control groups. Control I, control subjects who denied drunk driving; Control II, control subjects who reported drunk driving; DWI I, police-referred drunk drivers who denied drunk driving; DWI II, police-referred drunk drivers who reported drunk driving. ***Difference from Control I group P < 0.001; # different from Control II group P < smoking, paying for parking, dysfunctional impulsivity, platelet MAO activity, age and the amount of alcohol consumed in the past 7 days had salient importance in the discriminant function (Table 3). Canonical discriminant

5 144 D. EENSOO et al. Table 3. Standardized canonical coefficients for predicting subjects belonging to the DWI II and Control I groups analysis yielded a Wilks s lambda of 0.57, F 9,268 = 22.8, P < Number of observations and the proportion of cases classified correctly into the first class (Control I group) was 105 (83%), and 123 (81%) for the second class (DWI II group). Using this discriminant function for the groups Control II and DWI I, 60 (76%) and 36 (80%) observations, respectively, were classified correctly. DISCUSSION Canonical coefficients Predictor** Alcohol-related problems 0.41 Frequency of drinking 0.27 Seat belt use 0.25 Smoking 0.15 Paying for parking 0.11 Dysfunctional impulsivity 0.03 MAO activity 0.03 Age 0.02 Alcohol consumption in the previous week Class means Control I 0.96 DWI II 0.79 Control I, control subjects who denied driving after drinking; DWI II, police-referred drunk drivers who reported driving after drinking. **P < 0.01 for all predictors. We used both the police database and the self-reports to identify people who had been driving drunk during the previous year. It is well known that in the case of violating rules, people might want to show themselves in a socially more desirable way, thus, the research based only on selfreported questionnaires for behaviour in traffic might not be fully reliable; at the same time, not all drunk drivers are caught by the police. Thus, we have separated controls and drunk drivers according to whether they reported or denied drunk driving. Socio-economic background of the groups was quite similar, except that a lower educational level was overrepresented in the DWI-groups and religiousness in the Control I group. Previous studies have described different (Baum, 2000) as well as a similar (Wilson and Jonah, 1985) socio-economic background of the drunk drivers and controls. In our study, the variables describing the socio-economic background did not appear as independent predictors of drunk driving in discriminant analysis, which suggests that their significance is closely related to other factors. DWI subjects, who self-reported driving after drinking (DWI II), appeared to be the youngest group. At present, it remains unclear whether older drivers lie more about their drunk driving or whether they were caught more often in a hangover, thus, not acknowledging their drinking and driving problem. Investigations of drunk drivers who had caused traffic accidents, as well as population studies, show that younger drivers tend to drive after drinking, more often (Wilson and Jonah, 1985; Johnson et al., 1998; Golias and Karlaftis, 2002). As expected, the DWI subjects who reported drunk driving consumed more alcohol in the previous 7 days in comparison with the other groups. Out of alcohol-related variables, the best discriminating indicators appeared to be alcohol-related problems and the frequency of using alcohol. It has been found that drunk driving incidents are more frequent among drivers having alcohol-related problems (Del Rio et al., 2001). Johnson et al. (1998) described both the frequency of drinking and the amount of consumed alcohol as predictors of drunk driving. With respect to alcohol consumption measures, DWI subjects who denied drunk driving were similar to the subjects of the Control I group by the frequency of using light alcoholic drinks; in the use of strong alcoholic drinks, they were placed in between controls and drunk driving reporters. They also reported consuming less alcohol compared with the groups who reported drunk driving. It is possible that they were caught in a hangover without being aware of their blood alcohol levels. Alternatively, they may have underreported their alcohol consumption. We do not know whether this reflects their bigger reporting bias in general. Subjects who were caught driving while impaired, and reported this in the questionnaire (DWI II group), reported more often their generally careless traffic behaviour and violation of other traffic rules. They reported using a seat belt and paying for parking more rarely, and exceeding speed limits more often. This group also included more subjects who had been driving without having a valid driving licence, and novice drivers. In the investigations of drunk drivers, similar differences in low seat belt use, speeding (Golias and Karlaftis, 2002), driving without a driving licence (Baum, 2000; Begg et al., 2003) and novice driving (Cooper et al., 1995) have been described. In our study, drunk drivers and controls were discriminated significantly by seat belt use and paying for parking. Knowledge about safe time limits for drinking before driving was quite low, which suggests that such information should be included in the DWI prevention programmes. It has been shown that knowledge about drunk driving improves after participating in a DWI prevention programme (Kayser et al., 1995). DWI I group subjects had a lower level of knowledge about time limits for safe drinking before driving, compared with other groups, which might be the reason for underreporting their drunk driving. On the contrary, among the DWI II group, subjects knowledge about time limits for safe drinking before driving was a little better in our study compared with the other groups. This might be explained by the fact that there were more subjects having a driving licence for 2 years in the DWI II group and who hence better remember the time limits for safe drinking before driving from their studies. However, this knowledge does not reflect in their actual driving habits, as other variables such as their higher level of impulsivity and alcohol use habits have a stronger impact. Drunk drivers had a higher dysfunctional impulsivity and NEO-PI impulsivity. Dysfunctional impulsivity, which appeared to be the best predictor of drinking and driving, is a feature of cognitive style, supporting quick and thoughtless action in complex situations, where this kind of information processing leads to errors and inaccuracy (Dickman, 1990). This information-processing style might bring along their inadequate estimation of the risk, when deciding to drive after drinking. Functional impulsivity, which is defined as adaptive

6 DRUNK DRIVING PREDICTORS 145 and situation-specific, was not associated with drinking and driving. The difference between the groups in NEO-PI impulsivity, characterized by problems with controlling one s wishes and impulses, decreased after taking age into account. The difference between the groups in excitement seeking appeared to be mostly due to the younger average age of the group of drunk drivers. Earlier studies have shown the association of platelet MAO activity with risky behavioural tendencies (Fowler et al., 1980), social maladaptation (Oreland, 2004) and alcohol dependence (von Knorring and Oreland, 1996). Platelet MAO activity is a peripheral marker for the activity of the central serotonin system due to a postulated common genetic control (Oreland, 2004) that could occur via common gene promoter sequences and co-regulation of platelet MAO and monoamine transmitter genes. Recently, the ability of the platelet MAO activity to predict central serotonergic activity was further confirmed by the finding that there is a strong correlation between cerebrospinal fluid levels of the serotonin metabolite 5-HIAA and platelet MAO activity (Fahlke et al., 2002). These authors also found that rhesus macaques with low platelet MAO activity were less competent socially, and consumed alcohol to excess. In our study, platelet MAO activity was negatively correlated with alcohol-related problems but not with any measures of alcohol consumption. Studies with alcohol-dependent subjects have found that low MAO activity in platelets is associated with type 2 alcoholism (Cloninger et al., 1988; von Knorring and Oreland, 1996), which is characterized by several social complications and an early debut of the abuse (Oreland, 2004). Our results also lead to an assumption that MAO activity is more strongly related to the tendency for an alcohol-related problem behaviour than to the tendency to consume more alcohol. In both DWI groups, platelet MAO activity was significantly lower compared with both control groups. Drunk driving deniers and reporters in the DWI group did not differ with respect to MAO activity. The present discriminant analysis identified smoking and platelet MAO activity as separate significant factors in predicting drinking and driving. However, platelet MAO activity is directly reduced by tobacco smoking, and this has been suggested to underlie the low MAO activity in alcohol abusers (Whitfield et al., 2000). In a preliminary analysis, we compared the police-referred drunk drivers with controls, with respect to the intensity of smoking and found that the platelet MAO activity of non-smokers among DWI subjects was lower in comparison with the non-smoking controls (Eensoo et al., 2004), suggesting that smoking is not a confounding factor. This result is also in accordance with the primate study by Fahlke et al. (2002). In conclusion, significant differences between the drunk drivers and controls in the pattern of alcohol-related problems, frequency of using strong and light alcoholic drinks during the previous year, safety-belt use, smoking, paying for parking, dysfunctional impulsivity, platelet MAO activity, age and alcohol consumption, suggest that drunk driving is a result of a combination of various behavioural, biological and personality-related risk factors. Acknowledgements This study was supported by the Estonian Science Foundation (grant 5450), the Estonian Ministry of Education and Science (grants 2643 and 0820), and the Estonian Road Administration (grant LSOPH01030). We are grateful to Dr Toomas Ernits and his colleagues at the Estonian Road Administration for help in conducting this study. Marika Paaver was a holder of the scholarship of the World Federation of Scientists during the period of the study. REFERENCES Baum, S. (2000) Drink driving as a social problem: comparing the attitudes and knowledge of drink driving offenders and the general community. Accident Analysis and Prevention 32, Begg, D. J., Langley, J. D. and Stephenson, S. (2003) Identifying factors that predict persistent driving after drinking, and driving after using cannabis among young adults. Accident Analysis and Prevention 35, Cherpitel, C. J. and Tam, T. W. (2000) Variables associated with DUI offender status among whites and Mexican Americans. Journal of Studies on Alcohol 61, Cloninger, C. R., Sigvardsson, S. and Bohman M. (1988) Childhood personality predicts alcohol abuse in young adults. Alcoholism: Clinical and Experimental Research 12, Cooper, P. J., Pinili, M. and Chen, W. (1995) An examination of the crash involvement rates of novice drivers aged 16 to 55. Accident Analysis and Prevention 27, Costa, P. T. Jr and McCrae, R. R. (1989) The NEO-PI/NEO-FFI Manual Supplement. Psychological Assessment Resources, Odessa, FL. Del Rio, M. C., Gonzalez-Luque, J. C. and Alvarez, F. J. (2001) Alcohol-related problems and fitness to drive. Alcohol and Alcoholism 36, Dickman, S. J. (1990) Functional and dysfunctional impulsivity: personality and cognitive correlates. Journal of Personality and Social Psychology 58, Eensoo, D., Paaver, M., Pulver, A. et al. (2004) Low platelet MAO activity associated with high dysfunctional impulsivity and antisocial behavior: evidence from drunk drivers. Psychopharmacology 172, European Transport Safety Council (2001) Priorities for EU Motor Vehicle Safety Design. European Transport Safety Council, Brussels. Eysenck, S. B. G., Pearson, P. R., Easting, G. et al. (1985) Age norms for impulsiveness, venturesomenes and empathy in adults. Personality and Individual Differences 6, Fahlke, C., Garpenstrand, H., Oreland, L. et al. (2002) Platelet monoamine oxidase activity in a nonhuman primate model of type 2 excessive alcohol consumption. American Journal of Psychiatry 159, Fowler, C. J., von Knorring, L. and Oreland, L. (1980) Platelet monoamine oxidase activity in sensation seekers. Psychiatry Research 3, Golias, I. and Karlaftis, M. G. (2002) An international comparative study of self-reported driver behavior. Transportation Research Part F 4, Grunewald, P. J., Mitchell, P. R. and Treno, A. J. (1996) Drinking and driving: drinking patterns and drinking problems. Addiction 91, Hallman, J., Oreland, L., Edman, G. et al. (1987) Thrombocyte monoamine oxidase activity and personality traits in women with severe premenstrual syndrome. Acta Psychiatrica Scandinavica 76, Harro, M., Eensoo, D., Kiive, E. et al. (2001) Platelet monoamine oxidase in healthy 9- and 15-years old children: the effect of gender, smoking and puberty. Progress in Neuro-Psychopharmacology and Biological Psychiatry 25, Hilakivi, I., Veilahti, J., Asplund, P. et al. (1989) A sixteen-factor personality test for predicting automobile driving accidents of young drivers. Accident Analysis and Prevention 21, Iversen, H. and Rundmo, T. (2002) Personality, risky driving and accident involvement among Norwegian drivers. Personality and Individual Differences 33, Johnson, F. W., Gruenewald, P. J. and Treno, A. J. (1998) Age-related differences in risks of drinking and driving in gender and ethnic groups. Alcoholism: Clinical and Experimental Research 22, Jonah, B. A. (1997) Sensation seeking and risky driving: a review and synthesis of the literature. Accident Analysis and Prevention 29,

7 146 D. EENSOO et al. Kayser, R. E., Schippers, G. M. and Van Der Staak, C. P. (1995) Evaluation of a Dutch educational Driving While Intoxicated (DWI) prevention program for driving schools. Journal of Drug Education 25, Oreland, L. (2004) Platelet monoamine oxidase, personality and alcoholism: the rise, fall and resurrection. NeuroToxicology 25, von Knorring, L., Oreland, L. and Winblad, B. (1984) Personality traits related to monoamine oxidase activity in platelets. Psychiatry Research 12, von Knorring, L. and Oreland, L. (1996) Platelet MAO activity in type1/type2 alcoholics. Alcoholism: Clinical and Experimental Research 20, 224A 230A. Whitfield, J. B., Pang, D., Bucholz, K. K. et al. (2000) Monoamine oxidase: associations with alcohol dependence, smoking and other measures of psychopathology. Psychological Medicine 30, Wilson, R. J. and Jonah, B. A. (1985) Identifying impaired drivers among the general driving population. Journal of Studies on Alcohol 46,