Increased population drinking is not always associated with increased number of drink driving convictions

Abstract

Norström & Rossow 1 found a significant association between population alcohol consumption estimated at different time points and the concurrent number of convictions for driving while under the influence of alcohol (DWI) for Norway and Sweden in the period 1957–89. The study was based on total alcohol sales and official statistics on DWI convictions in Norway, and for Sweden the number of police-reported traffic accidents with personal injuries where the driver was under the influence of alcohol. Based on these data, they suggested that when alcohol consumption rises by 1% the number of DWI cases increases by 2%, on average, in Norway and somewhat less in Sweden. Alcohol consumption in Norway has increased substantially since 1989 and we have studied similar data on DWI for the period 1990–2010 to determine whether or not the same association can be found for later years. In addition, we have looked at the results of studies on the prevalence of DWI in roadside surveys over the period 1981/2 to 2008/9. Our task was complicated by the fact that there was a change in the definition of a DWI-related conviction in 1997, according to Statistics Norway. This has caused the official number reported by Statistics Norway to drop significantly from that year onwards. Therefore we studied the number of arrested drivers with blood alcohol concentration (BAC) ≥ 0.5 g/l or the equivalent concentration in breath (0.25 mg/l air) by using data obtained from the National Mobile Police Service (NMPS; Stavern, Norway) and the Norwegian Institute of Public Health. As shown in Fig. 1 the alcohol consumption per adult increased by 35% from 1990 to 2010, whereas the number of DWI cases with BAC ≥ 0.5 g/l decreased by 42%. Alcohol consumption per capita (litres 100% alcohol per inhabitant aged 15 years and above) and number of drivers apprehended with blood alcohol concentration (BAC) ≥ 0.5 g/l per 100 000 inhabitants aged 15–69 years We have also compared data from the two national roadside surveys of DWI that have been performed in Norway. The first study, which was performed in 1981–82, found that 0.27% of drivers had BACs above the legal limit of 0.5 g/l 2. The second study, which was performed in 2008–09, found that 0.2% of 9410 drivers had BACs above 0.2 g/l and about 0.1% had BACs above 0.5 g/l 3. The first study was carried out as random police controls. If an initial breath alcohol test was positive, a blood sample was taken for determination of BAC (in Norway, the police can take a blood sample for alcohol testing by force, so the participation rate was 100%). The second study was a voluntary study that was also performed in collaboration with the police, but data from the breath testing that the police performed was not included in the study, only results from saliva testing. The refusal rate was about 6%. However, the police found through their initial breath alcohol controls that six of the drivers who refused to participate in the project had detectable breath alcohol levels and were therefore brought to the police station for evidential breath testing or blood sampling. Those drivers corresponded to 0.06% of the drivers. Data from random breath testing performed by the police for control purposes in 2009 also indicated that 0.2% of the drivers in routine controls had BACs above the legal limit of 0.2 g/l 4. Therefore, in total, there seems not to have been any increase in the prevalence of DWI, but rather a reduction in the prevalence of BAC above 0.5 g/l between 1981–02 and 2008–09 despite the fact that the total alcohol sales per adult increased by 42% during this period. The number of DWI convictions is related to the percentage of drunk drivers, the number of motor vehicles on the road and the probability of arrest. These factors may change over time for many reasons, e.g. increased number of cars, increased alcohol consumption, changes in drivers' attitudes to DWI and changes in DWI control activities of the police. Norström & Rossow studied the number of DWI convictions per 100 000 population. However, it might have been more relevant to study the number of DWI convictions per 100 000 motor vehicles (as a surrogate for the number of active drivers), because the proportion of the population who had access to a motor vehicle increased significantly during the study period; in 1957 there were 362 630 registered motor vehicles in Norway, whereas in 1989 the number had increased to 2 343 204 5, i.e. a six-fold increase. During this period the number of DWI convictions per 100 000 motor vehicles decreased from 507 in 1957 to 282 in 1989 5, 6. The prevalence of drunk driving has become low in Norway, possibly at least partly because drunk driving has increasingly become socially unaccepted over the years 7. The decreased prevalence of DWI from 1982 to 2009 as revealed in the roadside studies might also, to some extent, have been related to the reduction of the BAC limit from 0.5 to 0.2 g/l in 2001, which may have affected public opinion and thus had an additional deterrent effect on DWI. Public transportation has improved, and it has become common to select a designated driver who does not drink alcohol when going to a party. This may also have contributed to a reduction in drunk driving. The probability of DWI arrest may also have changed during this period. Drunk drivers are arrested for a number of reasons, mainly: (i) police controls and surveillance; (ii) drunk drivers reported to the police by peers, family or other observers; and (iii) involvement in road traffic accidents according to the NMPS. The number of blood or evidential breath samples taken by the police decreased by 19% from 1990 to 2010 (data not shown). Differences between other countries also indicate that the prevalence of DWI in random traffic is not only related to annual alcohol consumption. The average recorded adult per-capita alcohol consumption for 2003–05 in Belgium, the Netherlands, Denmark and Portugal was 9.8, 9.6, 11.4 and 12.5 l, respectively 8. Including estimated unrecorded consumption, the total consumption was 12.8, 13.8, 14.4 and 17.9 l, respectively. The prevalence of driving with BAC above 0.5 g/l in 2007–09 among drivers in normal traffic in those countries was 2.2, 0.6, 0.5 and 1.2% 9. This confirms that other factors than per-capita alcohol consumption are important. Our findings call into question the link between population alcohol consumption and DWI and suggest both that other factors that aggregate consumption may play an important role in DWI and that the link between aggregate consumption and DWI previously observed may not be causal. The data also shows that it is possible to reduce the prevalence of DWI in the driving population even in periods when the total consumption of alcohol is increasing. None.