Alcohol and other drug use and the transition from riding to driving

In this issue, Poulin, Boudreau & Ashbridge provide an intriguing analysis of the problem of adolescents riding with drunk drivers [1]. Their analysis, based on a survey of 13 000 students in grades 7–12 in the four Maritime Provinces of Canada, includes individual, school and province variables. Highlighted are the factors not within the student's direct control that determine the extent of riding with a drunk driver (RDD). They found that rural residence, single parent or no parents, socio-economic status (SES), prevalence of driving under the influence (DUI), lower licensure rate and lower educational attainment in the community were associated with RDD. Of particular interest was their finding that, once the impact of those factors was controlled, being licensed reduced the probability of RDD. Presumably, this is because youths with licenses have less need to ride with others and thus are less likely to ride with a drunk driver. The Poulin et al. results highlight the need to pay more attention to the risk that unlicensed youths will become riders in alcohol-related crashes. Sixteen-year-old drivers have crash rates that are three times greater than 17-year-olds and five times greater than 18-year-olds [2]. The current official response to this problem is GDL, which extends the period of adult-supervised driving and provides for an intermediate period when the novice driver may not carry teenage passengers or drive late at night. This should reduce the opportunity for teenage passengers to be injured in crashes involving drinking novice drivers. There is substantial evidence that GDL programs are effective in reducing the crashes of novice drivers [3–8]. Most recently, Chen, Baker & Li [9] found that the presence of GDL programs in the United States was associated with an 11% decrease in the fatal crash rate involving 16-year-old drivers. However, there have been no studies of the impact of GDL on teenage passengers. Although imposing night-time restrictions on teenage drivers [10] and teenage passengers [11] appears to effectively reduce novice-driver injuries, there is evidence that GDL laws achieve their primary impact through a delay in the licensing of 16- and 17-year-olds. For example, Williams, Ferguson & Wells [12] failed to find a reduction in night-time fatal crashes involving 16-year-old drivers between 1993 and 2003, a period when many states were implementing GDL laws featuring restrictions on night-time driving. They further reported that during those years, the licensing rate of 16-year-olds declined. This suggests that the GDL laws achieve more by reducing the licensing of high-risk 16-year-olds than by lengthening supervision and imposing night-time restrictions. If GDL laws are achieving their impact primarily by limiting the number of licensed 16- and 17-year-olds, then more teenagers in that age group will need rides. It is therefore important to study the limits on mobility (a potential economic burden on some families) and safety (if lack of a license increases RDD) if a teenager is unlicensed. Poulin et al. begin that process by pointing to the increased risk of RDD for unlicensed youths. This suggests that increasing licensing requirements may be a special burden on teenagers from disadvantaged families, as indicated by the association Poulin et al. found between low SES, low educational attainment and living with only one parent. Low SES may indicate the limited availability of a vehicle for learning to drive. Living with a single parent may limit the availability of adult supervision for the extended period of the learner's permit. The relationship of licensing to driving is not clear. The role of licensing in promoting RDD in the Poulin et al. study is somewhat clouded by the failure to have a separate measure of driving frequency. (No driving is confounded with no drinking; see their Table 2.) Without a separate measure of driving, we do not know whether teenagers with licenses are actually driving or the extent to which unlicensed teenagers are driving; consequently, the significance of license status is reduced. This limitation in driving exposure data arises several times in their report. They note, for example, the lack of agreement on the role of gender across studies, which they suggest may be due to ‘. . . the extent to which female adolescents have access to vehicles . . .’; but no measure of vehicle access is available to them. Failure to collect adequate information on driving exposure is typical of most alcohol and drug research instruments. Drinking-and-driving questions usually follow the form used by Poulin et al. (e.g. ‘In the past 12 months, how often have you driven a motor vehicle within an hour of drinking two or more drinks of alcohol?’). Without a denominator that reflects driving/riding exposure (e.g. annual mileage or access to a vehicle and alternative driver), such RDD or DUI frequency questions clearly do not provide a common metric across all respondents. Unfortunately, the most widely used alcohol and drug surveys are markedly incurious about access to a vehicle, even though investigators frequently use the information to assess national trends in impaired driving [13] and as a marker for Diagnostic and Statistical Manual version IV (DSM-IV) [14] alcohol abuse [15]. For example, although the Monitoring the Future survey [16] contains a question on miles driven, the two principal national college surveys—the College Alcohol Study [17] and the Core Alcohol Survey [18]—as well as the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) [19] do not. Without an exposure measure, the utility of self-reported drinking and driving is compromised. Hasin et al. [15], for example, reported that a diagnosis of DSM-IV alcohol abuse for half the individuals in their study was based on only one symptom: driving after drinking too much. Clearly, the frequent use of drinking and driving as a marker for abuse is because it is the most readily available situation in which to measure risky behavior [e.g. abuse related to alcohol and other drug (AOD) use]. Accepting it as a psychiatric disorder obviously has an ironic effect, suggesting that selling the car or moving from Los Angeles where a car is a necessity to New York City where it is a burden is a ‘treatment’ for alcohol abuse. To clarify the relationship of impaired driving to a diagnosis of abuse, Hasin et al.[15] compared individuals who met DSM-IV abuse criteria based only on driving after drinking too much with respondents who qualified based on other measures. They found that, although the latter differed on all the variables studied from individuals not classifiable as abusers, the drinking drivers shared some characteristics with the non-abusers. A measure of driving exposure could have helped to clarify that relationship. Clearly, some alcohol abusers fail to report drinking and driving because they do not drive. With a question on vehicle use, non-drivers (which include some abusers) could be separated from the sample rather than classifying them all as non-abusers by default. In estimating the percentage of abusers, it might be appropriate to weight reports of drinking and driving based on reported number of miles driven. In addition, based on the demonstration by Poulin et al. that drinking is associated with RDD, a question on RDD might identify abusive drinkers who do not drive. Other examples of the utility of information on driving can be found in prevalence studies such as that of Chou et al. [13]. They calculated population-based percentages of respondents who reported driving after drinking and compared age, gender and ethnic groups based on the NESARC, which does not have a measure of driving exposure. Chou et al.[13] appropriately used percentage-based measures to compare groups. Because the measures are population-based rather than miles driven, a number of the comparisons could be influenced significantly by vehicle-access differences rather than by impaired-driving differences. It is well established that, when comparing age groups, underage youths and older people are less likely to have access to a vehicle and therefore drive less. This restricts the percentage of those age groups who have a basis for reporting driving after drinking. The addition to the NESARC of an extent-of-driving measure could provide a basis for making impaired-driving estimates based on driving population, rather than total population, as a measure that would be more useful in the development of interventions. Because a substantial fraction of the US population drives little or not at all, a miles-driven-based denominator for measuring drinking and driving is more relevant than is a population-based measure. Further, the addition of another question on crash involvement would offer the possibility of deriving a relative risk measure for those respondents who report drinking and driving compared to those who do not, providing a clearer picture of national trends in crash risk resulting from drinking and driving. The Poulin et al. paper challenges us to increase our knowledge about RDD as a signal for alcohol problems. It suggests a need to examine more closely the trade-off between discouraging early licensing versus reducing RDD by making licensing more available. It also suggests more research on programs that discourage RDD that go beyond current educational efforts. Along with the Poulin et al. results, Yu & Shacket's [20] finding of a reciprocal relationship of DUI and RDD suggests that we might target impaired drivers by increasing the scope of passenger endangerment laws to include adults as well as children. Alternatively, we might provide sanctions for the passenger who chooses RDD as well as the impaired driver. Poulin et al.'s paper also illustrates the need to expand information on vehicle access and miles driven in AOD surveys where driving is used as a signal for abuse.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.