< Previous | Contents | Next >

EFFECTIVE AND COST-EFFECTIVE MEASURES TO REDUCE ALCOHOL MISUSE IN SCOTLAND: A LITERATURE REVIEW

CHAPTER FIVE ENFORCEMENT

Introduction

5.1 The previous chapter included evidence relating to legislative interventions. The impact of legislation also depends on the action taken to enforce the measures and the visibility of the enforcement process. This chapter covers evidence relating to specific enforcement actions in the areas of:

• legal drinking age; and
• drink driving.

The studies reviewed are summarised briefly in tables 5.2 (effectiveness) and 5.3 (cost-effectiveness) at the end of the chapter.

Legal drinking age

Types of intervention

5.2 Enforcement of under-age sales laws frequently takes the form of ‘decoy sales’, in which under-age buyers attempt to purchase alcohol under observation of the enforcement agency. Enforcement campaigns may be accompanied by media publicity and training schemes aimed at sales outlets.

Effectiveness

5.3 Communities Mobilizing for Change on Alcohol was an initiative in which communities were encouraged to develop their own specific interventions to reduce under-age access to alcohol (NIAAA 2000). Interventions, which varied across communities, included activities to enforce restrictions on sales of alcohol to minors. After 2 ¹/₂ years, age checks at outlets were more frequent and the likelihood of sales to minors reduced.

5.4 The Community Trials Project had 5 interacting components aimed at reducing alcohol related injuries and deaths. Results were mixed; for example, voluntary server training was not effective at the community level but outlets in intervention sites were half as likely to sell alcohol to an apparent minor.

Quality and relevance of evidence

5.5 Only a small number of studies have been carried out and the results may be specific to the communities in which they were developed. Multi-agency interventions in the UK have not been effectively evaluated.

Driving related

Types of intervention

5.5 A range of interventions has been employed to enforce legislation relating to drinking and driving. The main initiatives that have been subject to evaluation are:

ignition interlock devices. These require the driver to provide a breath sample before starting a vehicle. If the sample corresponds to a blood alcohol content above a pre-specified level the ignition is locked, preventing the vehicle being driven;

‘administrative per se’ licence suspension. This allows for the immediate suspension of a driving licence subsequent to failing a sobriety test and without requiring conviction through the judicial system; and

random alcohol screening. This involves stopping vehicles without prior suspicion of alcohol use. In Australia, all drivers stopped have a breath test administered whereas in the US breath tests are administered only if the use of alcohol is suspected after the driver has been stopped. Random alcohol screening works through both deterrence and detection.

Sanctions against convicted drink drivers may include a range of interventions combining license penalties and treatment (Wells-Parker et al 1995).

Effectiveness

5.6 There is some evidence that ignition interlock devices may be effective in preventing re-offending during the time they are installed (Coben and Larkin 1999). Although there are some design problems with the studies, the findings are confirmed by the one RCT in this area. The relative risk of re-offending was 0.36 for the interlock group (95% CI; 0.20,0.61).

5.7 The evidence relating to the effectiveness of administrative per se licence suspension is mixed (McArthur and Kraus 1999). The intervention was effective in 3 states for at least one year after initial licence suspension but did not work in two other states.

5.8 There is evidence to support the effectiveness of random alcohol screening (Peek-Asa 1999). All studies except one showed reductions in fatalities, injuries and crashes and the exception could also be reanalysed to show a decrease. Decreases in alcohol related fatalities and injuries were greater than total decreases. The results were generally the same for studies that measured alcohol use directly and those that used proxy measures. Decreases were generally greater in Australia than in the US. (One excluded study from Sweden did not find a significant decrease).

Table 5.1 Range of reductions in traffic fatalities, injuries and crashes.

Intensity of intervention.

5.9 With respect to random alcohol screening, two studies examined the effects of the level of enforcement. One found greater reductions with increased enforcement and one found no effect.

Quality and relevance of evidence

5.10 For evaluations of ignition interlock devices, there was only one RCT and the randomisation process was inadequately described. All studies relied on re-arrest rates as outcome measures and this is a relatively rare event. There was potential for allocation of lower risk offenders to ignition interlock programmes in non-randomised studies. There was a lack of description of other components of repeat offender programmes or other sanctions applied.

5.11 The evaluations of administrative per se licence suspension included no randomised or non-randomised controlled trials. Results are based on studies comparing events before and after introduction of laws. Whilst this may be an appropriate study design, it can be difficult to attribute effects to the intervention, particularly when other initiatives may be taken at the same time.

5.12 The evaluation of random alcohol screening also relies on before and after study designs. Studies generally had short follow up periods and lacked direct measures of the number of impaired drivers involved in crashes. Proxy measures used included single vehicle night-time crashes. There was a lack of multivariate analysis to control for other deterrence programs, resulting in a problem of attribution of effects as above.

5.13 A review of the methodology relating to drink driving studies found that most studies had design flaws (Wagenaar et al 1995). In addition, there was insufficient reporting of data for purposes of meta-analysis and most studies had relatively short follow up periods.

Cost-effectiveness

Alcohol server laws

5.14 Levy and Miller (1995) undertook a cost-benefit analysis to investigate a pilot scheme to forbid the serving of intoxicated drinkers. The analysis is based on a case study in Washtenaw, Michigan, of a server intervention programme (SIP). Data on the incidence of adverse events were taken from police files from before and after the programme enforcing alcohol server laws. Costs of alcohol related adverse consequences were scaled down to represent the Washtenaw County population. The population for the analysis was patrons drinking at controlled establishments and the outcomes of interest were fatal and non-fatal injuries as a consequence of drinking. Programme costs were based on local police costs ($48,400) and publicising the programme and training costs ($3,000). The total benefits of the scheme were derived from the total US costs as a result of alcohol-related accidents and medical care. Medical cost savings were an estimated $0.47 million, monetary benefits $3.7 million and total comprehensive cost savings were $10.1 million.

5.15 Levy and Miller present benefits of the programme as much greater than costs. Applying before and after data for the SIP programme and then calculating the change in costs provides a significant positive cost saving to society. However, it is very difficult to see exactly how the figures were calculated and other assumptions may yield different results, e.g. alternative sources of alcohol rather than exclusively licensed premises. The policy relevance is limited outside the study area, as costs of incidents are very different, as are licensing laws.

Community sobriety checkpoint

5.16 Miller et al (1998) investigated the costs and benefits of a community sobriety checkpoint program to reduce the incidence of drink driving and associated accidents in the USA. The analysis was constructed using a hypothetical population of 100,000 licensed drivers. Benefits were estimated using data from 1993 alcohol-involved crash incidence. Costs per crash were taken from published studies. Various estimates of the reduction in accidents were used to estimate the benefits, with estimates taken from previously published studies documenting crash reductions following similar programmes. Miller et al used estimates of 5%, 10%, 15% and 20% reductions, and a rate of 250 driver stops per 4 hour checkpoint.

5.17 The study was based on a hypothetical community of 100,000 drivers, with outcomes based on averted fatalities, averted non-fatal injuries and averted property damage. The costs of the programme were police wages, checkpoint equipment (trailer and generator, breath test equipment), and a mobile video system at a total cost of $39,000 (£26,500 1990 prices). The best estimate pointed to a reduction in alcohol related crashes by 15%. Benefits of the programme included $7.9 million savings in alcohol-related crashes and $0.3 million in averted property damage. Annual operating costs were $1,181,000. Other costs were travel delays ($44,000) and criminal justice costs ($172,000). Each of 156 checkpoints cost society $7570, imposed $2175 in costs to drivers and yielded $50,000 benefits. Every $1 invested in the programme saved society an estimated $6.

Further research

5.18 There is a lack of UK evidence relating to enforcement actions. Random testing for drink drivers emerges as the intervention most likely to have an effect but studies are needed to address the questions of how long the effects of enforcement actions are sustained and the level of enforcement that is optimal, in terms of costs and benefits. UK studies of enforcement actions relating to under-age drinking are required.

Table 5.2 Summary of Enforcement Effectiveness Studies Reviewed

< Previous | Contents | Next >