METHOD

The study was based on the Dunedin Multidisciplinary Health and Development Study (DMHDS) which is a longitudinal study of the health, development, attitudes, and behaviour of a cohort (n=1037) born at the only obstetric hospital in Dunedin, New Zealand between 1 April 1972 and 31 March 1973. Members of this cohort have been followed up at two yearly intervals from age 3 to 15 years, and at ages 18, 21, and 26 years. Further details about the cohort and the DMHDS have been published elsewhere.²²

The present study was part of a follow up conducted when the study members were aged 26 years. In a personal semistructured interview study members were asked about one particular event (that is the “critical incident”) that must have occurred at a venue where alcohol was available and there was also the opportunity to travel by motor vehicle. A pilot study had been conducted to develop the procedure to be followed in this interview.²³ The criteria for selecting the “critical incident” are given in fig 1. To assist with recall and limit selection bias the most recent and not necessarily the most risky occasion was deemed to be the critical incident. Information about the critical incident was recorded verbatim during the interview, and coded later.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Procedure for selection of critical incident.

Following guidelines provided by the Alcohol Advisory Council (www.alcohol.org.nz) the incidents were categorised as drink drive or sober drive. Where the study member was the driver, it was classified a sober drive incident if they had four or less standard drinks in the first hour and no more than one standard drink for each subsequent hour, up to a maximum of eight standard drinks (one New Zealand standard drink equals 10 g of alcohol) otherwise it was a drink drive incident. Where the study member was a passenger, it was classified a sober drive incident if the driver had four or less drinks, undetermined if driver had 5–8 standard drinks, otherwise it was a drink drive incident. Cases were excluded where the sobriety of the driver was undetermined, or if the incident did not take place in New Zealand.

The background and demographic measures were obtained in other interviews at the age 26 assessment of the DMHDS. Socioeconomic status was measured using the Elley-Irving scale, which is calculated using occupation and education.²⁴ Measures of alcohol and marijuana dependence were obtained at age 21 years using a modified version of the Diagnostic Interview Schedule.^25–27

χ² Statistical tests were used to examine differences in background characteristics of those involved in sober drive and drink drive incidents.²⁸ For the critical incidents, some quotations from the interview are included to provide additional contextual information.

RESULTS

Of the original 1037 members of the DMHDS cohort 969 (93%) completed the road safety interview at age 26 years. Of these, 750 reported a “critical incident”, 663 (88%) a sober drive incident, and 87 (12%) a drink drive incident. The remaining 219 had no critical incident. Of the exclusions, 18 were known to have died, 11 were excluded for various reasons (for example three were severely handicapped), and 39 (4% of the original cohort) were refusals or missing.

A comparison of the personal characteristics of those who reported a drink drive or sober drive incident are shown in table 1. Being male, unmarried, of low socioeconomic status, no school qualifications, and alcohol or marijuana dependence at age 21, were all significantly associated with being involved in a drink drive incident.

Table 2 shows that compared with sober drive incidents, drink drive incidents were more likely to have been associated with drinking at a bar (51% v 36%), and after-work drinks (24% v 12%), in a provincial town (66% v 81%), and on a Friday night (43% v 25%).

Events resulting in drink drive incidents were less often planned in advance, more likely to have been driven to alone, and more likely to involve drinking before going to the event (see table 3).

Some characteristics of the event are shown in table 4 with the commonest influence on drinking behaviour for those reporting drink drive incidents being the conviviality or atmosphere at the event (36%). For sober drive incidents it was the need to drive (28%), as in the quote “I wanted to drink but public transport is no good so I had to curtail my drinking so I could drive”. In 24% of the drink drive incidents, marijuana and/or LSD had also been used.

Some had predetermined the amount they would drink “I was driving, I had to go before I drank too much”, but for some such an intention was sometimes affected by unexpected factors: “different people who turned up, I had to stay and talk to them”.

Table 5 shows that passengers were often of a similar state of sobriety to the driver, and that those involved in drink drive incidents often went on to another drinking location (43% v 19%). Drinking drivers were more likely to be travelling alone than sober drivers (47% v 30%).

Alternative transport was available for around three quarters of the incidents (table 5) but one of the most common reasons for drink drivers not using it was not wanting to leave their car: “I wanted to go home and didn't want to leave the car”, “I had my own car and it would be a hassle to get the next day”, “It was too expensive and a hassle if I'd left the car”.

Of those in drink drive incidents, 54% acknowledged that it was risky (table 5) with reasons usually related to being caught by the police. Also, of the drink drivers who had planned ahead but changed their plans, the decision to drive, or be driven, appeared to be impulsive and reflected impaired judgment: “I thought I had a place to stay and they were deciding on who was sleeping where and I thought I'm not mucking around wasting time so I just left”, “I only ran into him at the end of the night so I wasn't quite sure how much he'd been drinking and I was a bit drunk too”.

Only two drink drivers (3%) reported that other people put pressure on them to drive. Also, for drink driving incidents where further driving took place, approximately 20% were return trips, and often the reason for the trip was to go out and purchase more alcohol.

Further analyses were conducted to examine in more detail some areas of interest, such as the relationship between driving to and driving from the event. Over half (n=423) of the study members had driven themselves to the event, 92% (n=391) of whom also drove themselves from the event. Of these, 336 were sober drivers and 55 were drink drivers. Of the sober drivers, 86% had specifically planned to remain sober, for the drink drivers the comparable figure was 55% (n=23). Of these 23 drink drivers, 11 drank more than they had intended and acknowledged that driving had been risky, while nine drank the amount they had intended to drink but did not consider that it was risky for them to drive.

The type of occasion (table 2) was also examined to compare “after-works drinks” drink drive incidents (n=21) with drink drive incidents after other events (n=66), and the quantity of alcohol consumed by the driver. This showed that for “after-work drinks” only one incident involved a driver who had consumed over 12 standard drinks, whereas for the other incidents over half (n=36) had consumed that amount.

DISCUSSION

While previous research has considered the personal and situational characteristics of drink drive incidents, this is the first time of which we are aware that these have been compared with sober drive incidents. In this study we were also able to examine personal characteristics, such as alcohol dependence, which is not often possible for a community sample. There are several other factors, however, that should be kept in mind when interpreting the results from this study. First, because of the hierarchical method we used to select the “critical incident” the participants were more likely to report incidents where they were a driver, rather than a passenger. This was because it was considered that the study member would be more likely to be know the details of an event (for example, how much the driver had to drink) if they were the driver, rather than if they were a passenger. Also, we wished to compare the background and personal characteristics of those involved in these events, and these data were only available for the members of the DMHDS cohort. The results reported here, therefore, cannot be construed as representative of the prevalence of these different scenarios among the cohort, or any wider population.

Another factor to be considered is the method used to classify the incident as drink drive or sober drive. We estimated whether drivers were sober or drunk based on self report, and the accuracy of these estimates will have varied between individuals. There is evidence that self reported drinking collected in the manner used in this study is, if anything, under-reported as quantities increase.²⁹ Also, acknowledging the imprecise nature of these estimates we used a uniform definition of standard drink consumption for males and females. While this may have resulted in a few females failing to be classified as drink drivers, when perhaps they should have been, this is unlikely to have had much influence on the findings.

The first aim of this study was to identify personal and background characteristics that differed between those involved in drink drive compared with sober drive incidents. The most important finding was the highly significant association between alcohol and cannabis dependence at age 21 and being involved in a drink drive incident at age 26. Similar findings have been previously reported,^19–21 and it is an important point because those who are dependent on alcohol have extreme difficulty modifying their drinking behaviour. They can, however, make decisions about driving, and interventions specifically designed to encourage this group not to drive must be considered. Also, it may be possible to identify those who are hazardous drinkers or alcohol dependent by using a screening tool such as the AUDIT,³⁰ among drivers convicted of drink driving offences, and targeting them with an intervention specifically designed to take account of this condition.

Other personal factors associated with drink drive incidents were being a male, of lower socioeconomic status, having lower educational attainment, and being unmarried. The first three, in particular, have previously been demonstrated to be correlated with drink driving in New Zealand.⁴

The second aim of this study was to compare the situational factors associated with drink driving incidents with sober driving incidents. In doing this, we found it useful to consider the situational factors as pre-event, event, and post-event phases. It has been suggested that decisions made well before the decision to drive are very important in understanding drink driving decision making.^7,17 These decisions include: which events to attend, where they are, who will drive, whom to go with, and what to do once there.⁷ Nelson et al, in 1999, found that making plans in advance to avoid drink driving was highly correlated with actually avoiding drink driving.¹⁷ The current study supports this, with the sober drive incidents being more likely to have been planned in advance than the drink drive incidents.

The location of where the drinking took place also differed. If drinking events took place in a bar, or in a provincial town, it was more common that it would result in a drink driving incident. These results are similar to those from previous research on drink drivers or alcohol related crashes,^8–11,31 and suggest that enforcement around bars may help to deter this behaviour. Bars may also be a useful location to target for educational campaigns.

The type of event at which people drink and drive has received little previous attention. Overall, almost one quarter of the drink drive incidents were classified as after-work drinks, and further cautionary messages may be appropriate.

In concordance with other studies,¹⁷ this study showed that, if persons travelled to an event alone, it was more likely to result in a drink drive incident, than if they had travelled with other people, particularly family members. Also, bar patrons with high blood alcohol concentrations who arrived alone were more likely to drive from the bar.¹⁸ However, this may not apply to younger drink drivers who have been shown to be more likely to have passengers.¹²

Not surprisingly, those who drove to an event were highly likely to drive from the event, irrespective of how much alcohol they had drunk. This shows that people must be dissuaded from driving if they know they are going out drinking. This is arguably the single most important, and modifiable, aspect of these events that could lead to a decrease in drink driving. This is illustrated by those who planned to stay sober because they wanted to drive, but after drinking more than they planned they still drove from the event, despite being aware it was risky. In rural areas, where driving is often a necessity and leaving the vehicle at home is not an option, other approaches may be needed such as encouraging bars to provide food, and promoting the drinking of low alcohol or non-alcohol drinks.

It has been shown that the difference between someone who is impaired (loss of ability but still legally able to drive) and someone who is intoxicated (legally unable to drive) is difficult to distinguish, particularly in large social groups.³² In this study, we found that most of the passengers were in a similar state of sobriety (or drunkenness) to the driver. This meant that, for the drink drive incidents, the passengers were possibly not concerned about the driver's state of impairment, and would not have been able to exert a “sobering” influence.

IMPLICATIONS FOR PREVENTION

By comparing drink drive incidents with sober drive incidents, this study has identified differences between such occasions and the people reporting them. The young adults who were dependent on alcohol were significantly more likely to report drink driving incidents, and these people may be particularly important targets for interventions. There were obvious differences in the situational factors, particularly with regard to decisions made prior to the event itself. The most important intervention would be to encourage people not to drive to drinking events. If driving to the event is necessary, they should either have a plan for limiting drinking, or have prearranged alternative travel. Finally, if neither of the above occurs, or if the drinking plan fails (that is, too much has been consumed to drive safely), then other interventions such as host responsibility need to be employed.