D

Reducing Alcohol-Impaired Driving: Lessons from a Global Review¹

Adnan A. Hyder, M.D., Ph.D., M.P.H., Lead Consultant;
Andres Vecino, M.D., M.Ec., Ph.D., Co-Author for Part 1
(Thanks to Nukhba Zia, Ph.D., Research Assistant, for help with Part 2)

INTRODUCTION

Road injuries accounted for more than 67 million disability-adjusted life years in 2015, representing a 4.6 percent increase since 1990 (GBD, 2016). Each year, more than 1.3 million people are killed on roadways and another 50 million are injured. Recognizing this serious global public health problem, the World Health Organization (WHO) declared 2011–2020 as the Decade of Action on Road Safety (WHO, 2013).

Drink-driving² is a major cause of global road traffic fatalities. The United Nations General Assembly has expressed its concern that only 7 percent of the world’s population is covered by adequate laws that address all road safety behavioral risk factors including driving under the influence of alcohol. According to WHO’s 2015 Global Status Report on Road Safety, only 34 countries have drink-driving laws in line with best practices, and only 46 countries rated their enforcement of drink-driving laws as “good.” Accordingly, it is critical to support stronger legislation

___________________

¹ This is an abridged version of the paper. A full version including all tables and figures can be found in the public access file for the Committee on Accelerating Progress to Reduce Alcohol-Impaired Driving Fatalities. Available by request from the National Academies of Sciences, Engineering, and Medicine’s Public Access Records Office (PARO@nas.edu)

²Drink-driving is a colloquial term commonly used in countries outside of the United States to refer to driving or operating a motor vehicle while impaired or while one’s blood alcohol concentration is above the permissible limit set by law. This term will be used in lieu of “alcohol-impaired driving” for the purpose of this background paper.

and stricter enforcement to curb drunk driving and mitigate its harmful effect on public health (WHO, 2015).

Approximately 17 percent of all global deaths that are caused by road traffic injuries have been attributed to alcohol consumption (Shield et al., 2012). A strong body of global evidence confirms that alcohol consumption is a risk factor for road traffic injuries and deaths (Aguilera et al., 2015; Bachani et al., 2013; Gururaj, 2004; Peden et al., 2004; WHO, 2013). For that reason, WHO included the prevention of drink-driving as one of the five leading areas to focus on to reduce road injuries and deaths during the Decade of Action on Road Safety (WHO, 2013).

GOALS AND OBJECTIVES

At the direction of the Committee on Accelerating Progress to Reduce Alcohol-Impaired Driving Fatalities (the committee), this commissioned paper was requested to offer insights into the international landscape of drink-driving interventions. The commissioned paper was asked to conduct a purposive review of international drink-driving interventions with a focus on driving policies and countermeasures that target individuals and system changes, and identify how they have been carried out abroad (part 1). The paper was asked to make conclusions on which interventions have worked and why, and in general discuss the potential applicability of those interventions in the United States. In addition, it was requested to focus on countries that are economically and culturally analogous to the United States (part 1). The commission was also asked to conduct a smaller (part 2) scoping review to provide recommendations regarding data collection and application to support the reduction of drink-driving. The paper was also asked to focus the data review and discussion points to highlight new or unique data and tools that the United States could learn from and international data variables that have not been done in the United States, that should be considered, and innovative ways of linking data (part 2). Therefore, this paper responds to the charge from the committee and includes both lessons on interventions (see part 1) and data systems (see part 2).

APPROACH

As requested, this paper provides a purposive review of available knowledge and experiences from global road safety in the prevention and control of alcohol-related harm that may be important for the United States. This last element was critical to the screening of all materials and experiences, and guided the work culminating in this paper. It is to be noted that a general review of alcohol-related interventions and programs

would be more extensive and wide ranging and would include relevance to low- and middle-income countries, which was not the focus of this paper.

Overall, this paper is based on the experience of the authors, interviews with a selected number of global experts, a review of the peer-reviewed literature (since 2000), and a careful screening of key international publications from organizations such as WHO. This paper has been informed by previous discussions in meetings (such as the partner meetings of the United Nations Road Safety Collaboration) and workshops (such as those in the Bloomberg Initiative for Global Road Safety).

The review is presented below in two parts. Part 1 proposes selected intervention domains against drink-driving that may have important lessons for the United States. This is a selection of those interventions considered effective, implemented by other countries (or states), and worthy of consideration in the United States. A deeper discussion of those interventions is described using a systematic approach so the background, details, outcomes, and challenges associated (as reported) are captured for consideration by the committee. References are provided in the event the committee wants more nuanced details.

Part 2 is a review of data around drink-driving and is based on analysis of the best systems currently existing in other high-income countries. Specifically, the section analyzed recommendations from global agencies and a select group of Organisation for Economic Co-operation and Development (OECD) countries, and compares with the Fatality Analysis Reporting System (FARS) data in the United States. Recommendations around both variables and indicators are provided.

The paper ends with a brief conclusion and encourages active discussion on the issues raised here by the National Academies of Sciences, Engineering, and Medicine (the National Academies) committee.

PART 1—INTERVENTIONS AND PROGRAMS

This section proposes a set of domains for interventions against drink-driving for consideration by the National Academies committee. Each proposal is followed by a summary of the evidence, a detailed case study with references, description of some key challenges, and a brief conclusion. Tables 1–4 support the section,³ and the references provide other materials for the committee to obtain more information.

___________________

³ Available by request from the National Academies of Sciences, Engineering, and Medicine’s Public Access Records Office (PARO@nas.edu).

Proposal 1: Enforcement Interventions

The Evidence on Its Effectiveness

During the review of global experiences, enforcement interventions were found among the most effective strategies to reduce mortality, morbidity, and crashes on the road. The evidence found reveals that enforcement interventions, specifically general deterrence strategies, and the performance of random breath-testing stops, have demonstrated improvements in drink-driving outcomes in high-income countries. The characteristics of some key analysis around these interventions can be found in Table 1.⁴

In the case of deterrence strategies, we found that the evidence on effectiveness is heterogeneous; such heterogeneity is likely to be due to differences in the intensity of enforcement and media coverage accompanying such campaigns. Important examples of such varying effectiveness have been observed in Serbia (specifically on enforcement for pedestrians to prevent mortality) (Zivkovic et al., 2016) and Canada (increased enforcement on drivers reducing alcohol-related crashes and casualties) (Mercer, 1985; Vingilis and Salutin, 1980).

The specific case of France is relevant because despite the penalties being increased by 52 percent, drink-driving behavior increased threefold. It has been argued that the legislation change was not properly enforced, leading to increasing drink-driving and highlighting the importance of enforcement accompanying legislation change (Constant et al., 2009).

Random breath-testing (RBT) stops are very effective in reducing serious crashes, with an effectiveness on mortality of up to 35 percent. The key for sustainability of RBT interventions seems to reside in the characteristics of randomness of the stops, the high level of enforcement, and the deployment of mass media campaigns during the RBT intervention (with the purpose of increasing the perception of enforcement and creating the risk of apprehension) (Henstridge et al., 1997; Mercer, 1985; Vingilis and Salutin, 1980).

One example of the relevance of the latter has been revealed in the specific case of Canada, where two campaigns (called blitz) with equal enforcement characteristics were deployed and compared based on the level of media coverage that accompanied the blitz. The study found that the blitz with no mass media coverage campaign failed in reducing drink-driving behaviors when compared to the baseline (Mercer, 1985). Similar findings were found in Australia where RBT interventions (deployed

___________________

⁴ Ibid.

in Tasmania) were found to be less effective, likely owing to the lack of an accompanying mass media campaign (Henstridge et al., 1997), and in the United Kingdom in the 1970s in a study that showed that media combined with random breath testing reduced crashes but the impact faded without renewed media activity (Ross et al., 1970). These findings emphasize the key role of mass media campaigns to effectively deter drink-driving behavior.

Case Study 1: Learning from Random Breath-Testing Activities in Australia

Background Victoria was the first Australian state where RBT campaigns were introduced in 1976 amid public controversy (Henstridge et al., 1997; WHO, 2007). Since then, RBT interventions have been implemented in other states after learning about the successes in Victoria. Most RBT campaigns have usually been implemented in combination with mass media coverage to both increase the perception of enforcement and facilitate the creation of social norms rejecting drink-driving behavior.

One example in Australia highlighting the relevance of concurrent mass media campaigns has been described in Tasmania, where despite some limitations, the general understanding is that the interventions were not as successful as in other states because of the lack of media campaigns associated with the enforcement activities (Henstridge et al., 1997).

Intervention and its characteristics RBT interventions in Australia, and specifically in Victoria, are not implemented as an isolated primary enforcement intervention. Importantly, RBT campaigns are a more general policing approach known in Australia as “problem-oriented policing.” This approach requires police to redirect the focus of enforcement interventions toward prevention. More broadly, this approach allows the police to emphasize solving the problems of safety issues instead of relying on its traditional role of detection and arrest.

Problem-oriented policing is a strategy that requires close collaboration between police departments and state authorities to plan a strategic approach, coordinate safety interventions, and importantly share data on trends on crashes and risk factors. The use of data is extremely important in this approach because it provides police with a surveillance tool to target locations and driver profiles that are found to be at higher risk of drinking and driving.

As part of this policing approach, RBT campaigns make up a key element, and the effectiveness of RBT interventions relies on three main features. First, RBT campaigns are by nature a primary enforcement strategy. Therefore, drivers are pulled over to be tested for drink-driving with no need for another offense to have been committed. This is different from

“selective stops,” when the stop and tests are carried out when another offense has been committed. Random stops are around 50 percent more effective than selective stops in reducing mortality, which reflects the importance of the primary enforcement component in policing drink-driving (Elder et al., 2002; Shults et al., 2001).

The second aspect that has been key to the success of RBT is its actual randomness. This implies that all drivers truly face the same probability of being pulled over and tested. To create a truly random RBT system, the Victoria State Police relies on a computer-generated random assignment of patrols to perform the RBT stops in specific locations and times. The uncertainty around times and locations of the RBT stops effectively increases the perception of risk of detection, improving the effectiveness of the preventive enforcement approach (Henstridge et al., 1997).

The third key aspect for the success of RBT campaigns is their intensity. Experts indicate that the Victoria State Police sets the aim of testing each driver twice per year. However, measures on the optimal intensity of enforcement are not available and need to be the focus of future research (Henstridge et al., 1997).

In Australia, the guidelines around RBT vary across states. For example, in some states alcohol tests are conducted on all drivers stopped, whereas in others this is not necessarily the case. Also, in some states, vehicles are pulled over only for the test to be conducted without a license check (the stop might take less than 30 seconds).

Importantly, a meta-analysis found that the country with the largest crash reduction was Australia, precisely for the use of the blitz type of checkpoints including buses and large media coverage (Erke et al., 2009). One study found that the most effective types of checkpoints are those covering both minor and major roads simultaneously; those that cover at least 20 hours per 100 square kilometers per week; and the level of enforcement is cost-effective when it aims at 1.5 breath tests per year per licensed driver (benefit-cost ratio equal to 2.05) (Cameron, 2013).

Challenges The main challenges faced by Australia around the implementation of RBT are related to resources (Henstridge et al., 1997). The high targets in terms of intensity and randomness requires investment of scarce police resources and the involvement of general duty officers to perform drink-driving checks. The involvement of these police officers increases the coverage of the enforcement activities but at the same time leads to the implementation of low-profile “mobile testing” stations manned by these general duty officers. These mobile stations have low visibility and lack the traditional visible blitz equipment (cones, lights, heavy presence of police and a bus where the tests are taken commonly called the “booze bus”) and other highly visible signs. The low visibility

of the intervention, however, is thought to lead to an overall less effective intervention (Henstridge et al., 1997).

Another challenge is that despite the sharing of data by state authorities with the police, police officers did not necessarily keep records on the key features of enforcement activities. These features have been cited as potential factors determining the variability of the effectiveness of RBT across Australian states. However, since such data were not systematically collected, there is some uncertainty as to which aspects of enforcement make it more or less effective. Finally, public opinion was initially unfavorable toward the implementation of RBT stops, but currently it is a widely accepted strategy.

Conclusion

Overall, RBT campaigns demonstrated an “immediate, substantial, and fairly permanent impact” on drink-driving outcomes including mortality and crashes (Henstridge et al., 1997). Therefore, it is worth serious consideration for the United States. However, it needs to be emphasized that there is an understanding in Australia that the successful implementation of the RBT campaigns is not isolated from other institutional changes that led the Australian police to establish closer collaborations with the state authorities and to change its focus to a preventive emphasis rather than a punitive one (Sweedler and Stewart, 2009).

Proposal 2: Reduction in Blood Alcohol Concentration (BAC) Levels

The Evidence on Effectiveness

A policy option that was detected during our review and was consistently mentioned in expert interviews as a potential strategy to reduce drink-driving in the United States was a reduction of the current legal blood alcohol concentration (BAC) level. All of the experts spoken with for this paper believe that the 0.08% BAC per se laws currently in place in most states in the United States are too high. Such views are confirmed in clinical and experimental research that demonstrates that a level of BAC of 0.05% (Kennedy et al., 1993; Tagawa et al., 2000) or even lower (Breitmeier et al., 2007) is enough to impair a driver. In fact, the literature shows clear evidence on the effectiveness of reducing BAC to 0.05% and the characteristics of such evidence is summarized in Table 2.⁵

___________________

⁵ Available by request from the National Academies of Sciences, Engineering, and Medicine’s Public Access Records Office (PARO@nas.edu).

During our review, we found consistent evidence on the effectiveness of reducing BAC levels on drink-driving outcomes. An important example is Australia, in which the reduction of BAC levels from 0.08% to 0.05% led to a reduction of fatal and nonfatal serious injuries of 8 and 7 percent, respectively, in New South Wales, and 18 and 14 percent, respectively, in Queensland (Henstridge et al., 1997). Another example of successful reduction of BAC is Austria, which reduced the BAC levels from 0.08% to 0.05% in 1998, and this was associated with an initial decrease in injuries of about 10 percent (Bartl and Esberger, 2000).

In the Netherlands, the alcohol limit of 0.05% was imposed in 1974, leading to a 10 percent reduction in the prevalence of drivers with positive alcohol levels both over and below the limit (Noordzij, 1994). Such reduction was followed by a 10-year period of stability in the prevalence figures, followed by a further drop over time. Even though the long-term reduction in the prevalence of drink-driving cannot be attributable to the establishment of the 0.05% legal limit alone, it is understood that the downward trend started with this intervention.

In Norway, the reduction of BAC limit from 0.08% to 0.05% did not show statistically significant reductions in nighttime and weekend single-vehicle crashes and deaths (Assum, 2010). It is possible that the lack of effect in this study was caused by measurement issues or the lack of a control group. Reductions lower than 0.05% have also shown positive results (Table 2)⁶ such as in Japan (Desapriya et al., 2007) and Brazil (Andreuccetti et al., 2011). Japan showed a marked decrease in alcohol-related crashes after the reduction of BAC limit from 0.05% to 0.03%; Brazil also had marked reductions in mortality after the BAC limit set by law changed from 0.06% to 0.02%. Studies on changes in BAC limit have limitations as some of them were carried out several decades ago when road safety studies did not often report statistical test results; however, the changes observed in these earlier studies present point estimates of relevant magnitude in large populations.

Case Study 2: Lessons from Europe and National Legislation on BAC

Background In Europe, the BAC limits are defined by national legislation in each country. In 1984, the European Parliament recommended to the European Commission (one of the institutions belonging to the legislative arm of the then European Communities) that it should submit a proposal to establish a single 0.08% limit for all European countries. This was based on the high cross-border traffic in Europe and the fact that one-third of road traffic deaths had drink-driving as a contributory factor (European

___________________

⁶ Ibid.

Commission, 1989a). In 1988 the European Commission proposed a directive to reduce the BAC limits for all countries, but established a different lower limit of 0.05%, considering preliminary evidence of its effectiveness (European Commission, 1989a). This was very progressive and made Europe lead the change in this space for the world.

Legislative change and its characteristics The legislative change in 1988 was aimed at reducing drink-driving. At the time, the Section for Transport and Communications of the European Commission challenged the proposal, arguing that the new 0.05% limit “would seem unjustified in the absence of conclusive evidence that it will enhance road safety.” Furthermore, the committee requested the commission to do the following:

The main issues around the proposal (according to the committee) were the lack of conclusive evidence supporting the proposal, and the potential effect on public opinion, as they requested a transitional period and that research findings should be publicized to generate support (European Commission, 1989b).

Despite this the European Commission’s proposal did not pass the committee. However, 8 out of the 15 member countries in 2003 adopted national legislation to reduce BAC levels between 1988 (when the reduction in BAC levels was first proposed) and 2003 (Albalate, 2008). Even though the adoption of reduced levels in these European countries cannot be solely attributed to the failed proposal, it is understood that this was the first step in a series of institutional changes that led to the reduction of national BAC limits.

In 1997 the European Commission released a 4-year program for “Promoting Road Safety in the European Union: The Programme for 1997−2001.” The European Commission brought back the 1988 proposal and this time included evidence obtained by the same commission in which they estimated reductions in 5 to 40 percent in mortality if a comprehensive intervention (legislation + enforcement + telematics + education) reduced drivers with alcohol levels above 0.05% (European Commission, 1997). This second attempt is thought to have led to changes in

legislation for BAC in five countries in 2 years, including Germany, which had opposed the reduction of BAC limits through common European legislation 10 years before (Albalate, 2008).

In 2001, the European Commission brought back the proposal to change BAC limits in all member countries to 0.05% and added a further reduction of BAC for inexperienced and commercial drivers as well as motorcyclists to 0.02% (European Commission, 2001). For example, Italy reduced the BAC levels to 0.05% in that year as a consequence of the recommendation.

The evidence assessing these changes has shown that the continuous efforts of the European Commission throughout these years, as well as the peer effects generated across country members of the European Union (EU), led to the reduction of BAC levels in Europe (Albalate, 2008). The assessment of the reduction of the BAC levels in Europe was associated with a reduction in mortality of 5.7 percent among males and 11.5 percent among younger road users, especially in urban areas (Albalate, 2008). This trend has continued since the end of the program in 2001.

In 2006 Cyprus joined the 0.05% limit, while Luxemburg reduced its limit to 0.05% in 2007. All but two member states have changed their BAC limits to at least 0.05%, four countries have a zero BAC limit, and half of EU states have implemented 0.02% limits for inexperienced/commercial drivers (Anderson et al., 2012; EU Directorate General for Health and Consumers, 2009).

In addition, it has been shown that awareness about drink-driving among citizens of countries with BAC levels of 0.05% is higher than in the countries in which the limit remains 0.08% (Babor, 2010; EU Directorate General for Health and Consumers, 2009). This adds a social and public perception value for this approach.

Challenges Three main challenges have been reported in the reduction of BAC levels in European countries. The first challenge was the passage of the European law itself, which was not passed as European legislation. The issue appeared to have been that the original proposal was not presented with enough evidence at the time and the fact that the law included neither a transitional period, nor a comprehensive approach including mass media interventions (European Commission, 1989a). Also, other sources have reported heavy lobbying efforts from the alcohol industry to prevent reducing BAC limits, particularly in countries where the BAC limit remains to be 0.08%, specifically Ireland (Hope, 2006) and the United Kingdom (Hawkins et al., 2012). (This is a point we discuss later in the paper.)

The second issue has been the low levels of national enforcement accompanying these legislative changes. Enforcement levels in Europe

are relatively low implying that even though road traffic injuries have been reduced all over Europe as a consequence of the changes in BAC limits, there is a potential to further decrease such deaths if more intensive enforcement, including random breath-testing strategies, would be implemented (Anderson, 2008).

Finally, this experience (like other reviews on BAC laws) found the importance of mass media campaigns accompanying the interventions as critical to maximize effectiveness. BAC laws require media campaigns accompanying the legislation change (Killoran et al., 2010).

Conclusion

Overall, BAC legislation is critical for a national or state-level effort and has been demonstrated to have a substantial effect on drink-driving outcomes. Therefore, such laws and their revisions are worth serious consideration for states in the United States.

Proposal 3: Social Interventions on Alcohol

The Evidence on Effectiveness

Social interventions on alcohol have been shown to have heterogeneous effects on drink-driving outcomes. In Table 3,⁷ we present evidence of different interventions and their effects. These social interventions have varied designs, are applied in diverse settings, and are not always easy to evaluate in a comprehensive way. However, it is relevant to take them into account because the evidence shows that they are necessary complements of other interventions (as stated in this paper), such as reduction of BAC levels or the deployment of random breath-testing checkpoints.

The results of our review reveal that mass media campaigns have a small but significant effect in reducing alcohol-related crashes of around 13 percent (Elder et al., 2004). School-based programs have shown reductions in the prevalence of drunk drivers and riding with drunk drivers (and associated behaviors), which reflect the effect of knowledge and a “social norm effect,” especially on younger people (Elder et al., 2005).

Designated driver programs have shown a much smaller effect, and the evidence available is not very strong (Ditter et al., 2005). In fact, our interviews revealed that these interventions were not favored by our experts. The reasons stated were that these programs tend to be pushed by the alcohol industry as feasible interventions, but the evidence supporting

___________________

⁷ Available by request from the National Academies of Sciences, Engineering, and Medicine’s Public Access Records Office (PARO@nas.edu).

them is weak and that the effects are small. Furthermore, from a policy standpoint it is not sensible to rely solely on individual-level choices (with no need for accountability) rather than interventions that implement a structural or institutional restriction to drink-driving behavior.

During expert interviews, social interventions often considered “ideal” were those outside the specific scope of drink-driving and those that target alcohol consumption more broadly. Even though no direct evaluations are available for the effect of the STAD (Stockholm Prevents Alcohol and Drug Problems) program on road safety, and the focus has been on process outcomes (refusal rates to pseudo-intoxicated patrons) and on alcohol-related violence outcomes, our experts consider that to tackle drink-driving in the United States, alcohol policy needs to be considered more broadly than only targeting individuals drinking while driving. For example, comprehensive interventions need to be made at the point of distribution since in the United States, about half of individuals who get alcohol and then drink and drive get it from licensed stores (Gruenewald et al., 1996). Therefore, the STAD intervention, such as in Stockholm, is worth considering for state and county authorities in the United States.

Case Study 3: A Program for Bartenders in Sweden

Background In the last decade of the 20th century, Stockholm created the STAD program as a comprehensive approach to reduce the consequences of alcohol consumption, primarily to reduce alcohol-related violence (Norström and Trolldal, 2013).

Intervention and its characteristics The STAD program was implemented in northern Stockholm in 1996 with the aim of reducing episodic heavy (binge) drinking. The STAD intervention was implemented through three different strategies: community mobilization, training for bartenders in responsible beverage service, and strict enforcement (Norström and Trolldal, 2013). The training strategy for bartenders became compulsory for all licensed premises that stayed open until 1 a.m. or later, and was aimed at (1) preventing sales of alcohol to minors; (2) preventing sales of alcohol to intoxicated customers; (3) improving the ability of staff to recognize high-risk situations and intervene appropriately; and (4) helping staff from licensed establishments to develop their own alcohol service guidelines (Wallin et al., 2005). The 2-day training program, which issues a certification, focuses on six main subjects:

1. The effects of alcohol,
2. Alcohol legislation,

3. The extent of alcohol-related violence and how this can be prevented,
4. Service refusal issues and skills,
5. Drug issues, and
6. Conflict management skills (Wallin et al., 2005).

The enforcement component of the program includes fines for the establishment, and fines and even prison for the server if he or she provides alcohol to an intoxicated individual (Wallin et al., 2005).

Five years after starting the intervention, the program had trained 1,318 restaurant employees (Wallin et al., 2005) and initial monitoring using pseudo-intoxicated patrons revealed that the serving refusal rate increased from 5 to 70 percent. These results have been disputed since control areas also increased refusal rates in similar magnitude, but this is thought to be attributable to contamination effects of the intervention, which was widely publicized in the media (Rehnman et al., 2005). Further evaluations of this intervention concluded that the intervention reduced violent crimes between 21 and 32 percent (Norström and Trolldal, 2013; Wallin et al., 2003), which proved the effectiveness of the intervention in reducing problematic alcohol consumption. Therefore, it is not difficult to infer that an effect on drink-driving is likely; furthermore, a study considering only the direct costs of the intervention found it to be cost-effective in terms of cost per quality-adjusted life years gained (Mansdotter et al., 2007). The success of the STAD interventions has led the European Commission to expand the program and tailor it to seven European countries (Ames et al., 2016).

Challenges One obvious challenge is the ability to measure the effect of this, and other social interventions, as the target population and measurement strategies are difficult to define. Also, context matters in this type of intervention, and a successful strategy in one setting will not always be replicable in another.

It is important to note that there are no reports of any types of “resistance” to this intervention, and actually other areas (the “control” areas) freely adopted the intervention, making it difficult to measure its effect. This in turn provides a strength to this approach.

Conclusion

Even though the evidence for this intervention is programmatic, such types of interventions combine legal aspects, with strict enforcement, and training in high-risk establishments. Therefore, we suggest that this be considered for potential implementation in the United States.

Proposal 4: Alcohol Ignition Interlock Devices

The Evidence on Effectiveness

Expert interviews highlighted the role of alcohol ignition interlock programs in Australia as exemplary for their role in reducing recidivism and contributing to rehabilitation of drink-driving offenders. For that reason, this type of intervention is considered here and the evidence summarized in Table 4.⁸

Studies in the United States demonstrated a reduction of 15 percent in mortality of alcohol-involved road traffic fatalities among states that implemented ignition devices (Kaufman and Wiebe, 2016); this is attributable to its effect on recidivism among repeat offenders. A Cochrane review on the effectiveness of interlock devices showed a consistent average reduction in recidivism of 64 percent (Willis et al., 2004). One caveat is that this reduction was only during the time in which the devices were installed, so sustainability of the measure needs to be further assessed. A study in Canada also showed a (lower) reduction in recidivism of around 6 percent (Voas et al., 1999); this was attributed to the selection effect of the devices since they are not consistently applied to all offenders.

Case Study 4: Alcohol Ignition Interlock Devices in Australia

Background Several states in Australia demonstrated interest in implementing alcohol ignition interlock devices for convicted drink drivers. With this aim, road authorities from all jurisdictions were consulted to create national standards in 1995. The purpose of these standards was to make sure the devices and the implementation were technically, economically, administratively, and logistically harmonized across Australian states. Such guidelines implied the following:

• The devices would be legally recognized and enforced across all states;
• At the discretion of courts they must be restricted to participants in interlock programs who must be repeat offenders or high BAC offenders and they can appeal;
• The devices should be installed for at least 3 years with a minimum service period of 1–3 months;
• Some minimum period of license cancellation and conditional license categories need to be set with its use;

___________________

⁸ Available by request from the National Academies of Sciences, Engineering, and Medicine’s Public Access Records Office (PARO@nas.edu).

• Costs must be contained for both the state and the user; and
• The program needs to be evaluated (Christie et al., 1995).

As a result, some states in Australia had to delay implementation to address the conditions stated above. For example, alcohol ignition interlocks were actually not introduced until 2001, first in Queensland, with the specific aim of reducing repeated drink-driving behavior. This case study summarizes lessons to help inform implementation of this strategy in the United States in order to reduce repeated drink-driving behaviors (Schonfeld and Sheehan, 2004).

Intervention and its characteristics The implementation of alcohol ignition interlocks by the justice system in Australian states was significantly delayed since the establishment of the national guidelines in 1995 until its final implementation in Queensland in 2001. Multiple reasons have been suggested for this delay, and these comprise lessons for other countries attempting similar interventions. First, some doubts about the effectiveness and compliance of the devices were raised and more evidence on this was required before moving on with the actual implementation of the devices. Second, the costs of the devices for the offenders were too high, which was against specified national standards. Third, there was a marked unwillingness to modify the license suspension provisions and to complement such suspension sentences with the use of the devices as part of the penalty. Together, these had to be addressed before the program was implemented at scale in the country.

Postimplementation analysis has revealed that certain characteristics differentiate a set of different models of interlock programs, all of them with pros and cons (Bailey et al., 2013; Fitzharris et al., 2015; Klipp, 2013; Schonfeld and Sheehan, 2004). For example:

• Some states established compulsory interlock device programs; this increases the coverage of the program but tends to more heavily affect lower-income individuals.
• There are differences across states on the use of the device data to measure compliance, and to use those data to identify individuals with alcohol dependence who may need specific (or further) treatment for their condition.
• There has been variable use of interlock device data for monitoring improvement and increasing awareness about drink-driving behavior.
• Given the issue that losing a license can generate among convicted drivers, including problems in their recovery programs and the risk of driving unlicensed, there is a general tendency

• to understand the period of “suspended license” as a protective rather than a punitive measure. Therefore, the program in most states encourages the early adoption of interlock devices to allow the offender to drive as soon as he or she has finished the rehabilitation program. This makes the interlock device a protective and punitive measure that is taken against the driver for even longer than the length of an alternative suspended license period, but allows them to recover and get back to normal life as soon as possible.
• Interlock device programs are pragmatically used for both offenders who were, and were not, licensed at the time of the offense, recognizing the high risk for the latter to attempt to drive unlicensed again as a protective measure.

Evidence on the economics of alcohol ignition interlocks found that the cost-benefit ratio of these devices ranges between 1 and 3.4 depending on the assumptions of the model—this is for all devices with effectiveness in terms of “noncircumvention rates” (the rate at which the device can be tampered) higher than 50 percent (Lahausse and Fildes, 2009).

Challenges The main issues reported are costs associated with the ignition device. Nearly 70 percent of offenders were deemed unable to use the device during the first year of the program in Queensland because of high costs (Terer and Brown, 2014). This was despite the argument that since the larger society will benefit from ignition locks, it ought to be the society that should pay for it.

Another significant issue was the difficulty for states to adopt the proposed standard legislation (Schonfeld and Sheehan, 2004). This was deemed key when the national standards were established in Australia since drivers should not face different legal frameworks while driving across the country. This was also deemed to be important for consistency in the legislation regarding administrative or judicial sentences and because administrative sentences cannot provide discretionary powers for specific conditions that might make the adoption of the program exceptionally difficult. A final issue is the sustainability of the effect of the device, which appears to disappear when the device is uninstalled. It is for this reason that rehabilitation programs must be combined with ignition interlock programs (Terer and Brown, 2014).

Conclusion

Alcohol interlock devices are effective and offer a safe way to recovery for repeat offenders, allowing them to return to their usual lives sooner.

Particularly important are the high costs of the devices, which are usually faced by the offender. It is important to take this into consideration in terms of the overall social benefits of having those offenders driving with interlock devices in the United States.

Proposal 5: Alcohol Is the New Tobacco

Introduction

One of the reasons for the lack of effective drink-driving policies in many countries is the alcohol industry’s opposition to meaningful regulations, including effective drunk-driving policies, alcohol taxation, and restrictions on the sale of alcohol. Instead, the alcohol industry has been promoting weak interventions to control drink-driving; a 2016 study showed that less than 1 percent of the industry’s actions to reduce drink-driving aligned with evidence-based recommendations (Esser et al., 2016). Under the umbrella of corporate social responsibility, the industry has not only continued to market its products but has also increased its involvement in policy making and scientific research (Babor and Robaina, 2013) while continuing their lobbying efforts to prevent legislation that saves lives (Hawkins et al., 2012; Hope, 2006).

The globalization of the alcohol industry and growth of multinational beverage alcohol corporations is a potential barrier to the implementation and enforcement of effective drink-driving policies (Caetano and Laranjeira, 2006; Casswell, 2011a,b; Jernigan, 2009; WHO, 2014). Furthermore, alcohol industry-sponsored alcohol programs are rarely evaluated around the world (Mosher, 2012), and when they are evaluated, have shown little or no effect (Ditter et al., 2005).

These concerns have been raised and are becoming more acute globally. For example, the International Network on Brief Interventions for Alcohol & Other Drugs (INEBRIA) released a strong position statement dissuading their members from engaging with the alcohol industry altogether and highlighting that any form of industry involvement in evidence-based interventions should be of concern (INEBRIA, 2015).

Case Study 5: Understanding the Alcohol Industry

This case study will demonstrate examples to show the ease with which the alcohol industry is able to gain access to high levels of the United Nations (UN) and national policy processes, and obtain valuable brand association. Alcohol companies and their funded organizations such as the International Center for Alcohol Policies (ICAP) and the International Alliance for Responsible Drinking have a history of promoting

ineffective drink-driving strategies such as education and awareness, and undermining effective approaches such as lower BAC limits. It is important for the United States to recognize, understand, and then develop a strategy to address this potential challenge. The examples listed below provide a glimpse into the types of issues raised by the previous and current working activities of the alcohol industry.

Example: Coalition formation In 2015, the alcohol industry sponsored road safety events involving the International Telecommunications Union (ITC), the United Nations Economic Commission for Europe (UNECE), and the UN Headquarters. The ITC and UNECE symposium on the Future Networked Car was held in Geneva on March 5–6, 2014, and gave prominent and equal branding to the world’s largest beverage company AB InBev alongside UN logos. On November 13, 2014, the UN played a major role in the launch of a private-sector road safety initiative led by AB InBev called Together for Safer Roads. The event was held in the UN, broadcast on UN Web TV, and involved the participation of Mrs. Amina Mohammed, Special Advisor on Post 2015 Development Planning. On March 23, 2015, the UNECE, UNECA, and the newly formed alcohol industry-sponsored International Alliance for Responsible Drinking launched an e-book titled Preventing Drink-Driving in Africa⁹ on the UN premises in Geneva, Switzerland.

Given that the ITC and the UNECE have a direct involvement in setting both global and European regulatory standards and, therefore, there is room for conflicts of interest, it is disturbing to see an alcohol company with obvious commercial interests obtaining such privileged access possibly in exchange for an undisclosed sponsorship fee. Because WHO has led responsibility for road safety within the UN system, it is important to note that it was reportedly not asked for their advice before approval was given for the events stated above and were not officially present at the event.

Example: Platforms for marketing In mid-2016, the beverage giant Heineken struck an agreement with Formula One (F1), which allowed the company to become one of the main sponsors of F1 races, rendering Heineken a global platform for aggressive alcohol marketing. Already, F1 was predominantly sponsored by the alcohol industry; a 2015 report showed that audiences of F1 are exposed to alcohol brands every 5 seconds. In fact, the UN Special Ambassador for Road Safety was involved

___________________

⁹ See http://www.unece.org/trans/roadsafe/ece-eca-icap_ebook.html (accessed January 2, 2018).

in the ceremonies related to this event. This is alarming as it sends mixed messages about alcohol and driving.

Example: Obscuring independent research The alcohol industry’s ICAP, for example, funded six countries (China, Colombia, Mexico, Nigeria, Russia, and Vietnam) to conduct reviews of available data on drink-driving outcomes and policies (Stewart et al., 2012). Between 2000 and 2006, the Worldwide Brewing Alliance showed that 23 grants were given to university-based alcohol researchers in 13 countries. Such partnerships may interfere with the public health agenda, as the alcohol industry’s involvement in drink-driving initiatives may serve to promote corporate interests (Jernigan, 2009). Concerns about scientific publications funded by the alcohol industry have been expressed, including their potential to jeopardize the integrity of science and reliable health information, their potential to obscure the evidence on the effectiveness of alcohol policies, and at times their promotion of findings that favor the alcohol industry and undercut those of public health (Babor, 2009; Caetano, 2008).

The public and policy makers may not be aware of the alcohol industry’s involvement in this line of research, especially without statements of funding sources, leaving them vulnerable to uncritically accept studies as research from public health professionals (Babor, 2009). Many drink-driving studies published in scientific journals (Johnson, 2012; Li et al., 2012; Ngoc et al., 2012; Stewart et al., 2012) fail to include a statement disclosing that they are funded by the alcohol industry or even an acknowledgment to support from the alcohol industry.

Stenius and Babor (2010) have argued for the need to sever financial ties with the alcohol industry in order to uphold the integrity of scientific research and prevent potential conflicts of interest. As stated by the Director-General of WHO, “In the view of WHO, the alcohol industry has no role in the formulation of alcohol policies, which must be protected from distortion by commercial or vested interests” (Chan, 2013). Countries may face pressure, as the alcohol industry typically emphasizes the importance of partnerships between themselves and public health professionals (Babor, 2009; Caetano, 2008; Jernigan, 2012). Furthermore, several studies have found that the alcohol industry goals (including ICAP initiatives) are at odds with those of public health (Babor, 2009; Hawkins and McCambridge, 2014; Jernigan, 2012; McCambridge et al., 2013, 2014; Miller et al., 2011).

Conclusion

Given this context and the ongoing debate about alcohol industry funding, this paper makes the case to the alcohol and road safety

communities in the United States, as well as policy researchers, to take these issues seriously. It is important to generate the data needed to take action against stakeholder marketing and other alcohol industry-sponsored activities that undermine effective, evidence-based, alcohol control policies and encroach on the integrity of scientific research. It is critical for the health and scientific community to hold the alcohol industry to high standards for the promotion and protection of public health, especially that of vulnerable and younger communities. It is also imperative that public health scientists work to prevent any efforts that seek to undermine evidence-based best practices and that policy makers keep a critical view of industry-sponsored studies, as has been the case with studies sponsored by pharmaceutical and nutritional companies in recent years.

Recommendations for Programs

The proposals above, including the case studies, provide both evidence and stories of how intervention programs have been implemented in other high-income countries. They provide guidance on what lessons global experiences provide to inform the recommendations of the National Academies committee, which are being formulated separately. This paper suggests serious consideration of the following:

• Enhance enforcement, including random breath testing and creation of an environment where effective and data-based policing becomes the norm.
• Enhance national (and state) legislation around BAC to become consistent with other high-income countries for all drivers (≤0.05%) and focusing on high-risk groups (~0.02%).
• Implement specific, evidence-based social interventions that target high-risk situations and include not only awareness creation but legal and individual consequences.
• Promote alcohol ignition interlocks for repeat offenders together with rehabilitation programs.
• Control the unrestricted marketing and engagement of the alcohol industry through a unified multifaceted approach across sectors, and maybe this can begin by interdepartmental dialogues at the national level.

These remain strategic recommendations for the National Academies committee to compare with its own analysis of the U.S. situation and consider where global lessons enhance its own advice. These are not meant to stand alone in the absence of focused U.S.-based analysis of the current climate of alcohol-based interventions and their effect. Of course, if picked

up by the committee and the National Highway Traffic Safety Administration (NHTSA), then more detailed work will need to be done in each area, to understand the specifics of how such a change or introduction of new programs can be done within the U.S. (or state) context.

Finally, it is important to caution against the use of facile comparisons between countries—each context is specific and the relevant complexities need to be understood. To the extent that despite these realities lessons can be transferred, cross-national learning can be a useful strategy for global road safety.

PART 2—DATA PROPOSALS

This section provides suggestions for the United States on data variables and indicators that are either recommended by international organizations or used by other countries for drink-driving. The section provides details on the approach used, descriptions of data systems in selected countries, and recommendations for consideration by the committee.

Drink-Driving Data and Variables

Introduction

The main objective of this section of the paper is to understand the architecture of drink-driving data in the United States and compare it with other high-income countries that have well-established data systems. Furthermore, innovative ways in which databases in selected high-income countries link drink-driving data with other data, such as hospital data and insurance data, are also explored. This section provides a review of drink-driving databases in selected high-income countries and compares them to the United States to provide recommendations that can improve such data and data collection processes.

Methods

Our approach is based on the following:

• Assessing drink-driving data in selected high-income countries by understanding the type of variables collected, frequency of data collection, reporting, and dissemination;
• Comparing U.S. data with those of selected countries to identify variables that are in the databases of the selected high-income countries and not in the U.S. database;

• Understanding innovative ways of linking different data; and
• Providing recommendations for improvement in the U.S. drink-driving database.

The information in this section was primarily collected by searching websites of road traffic crash and injury databases, their manuals (when available), reports related to the databases, peer-reviewed literature, and suggestions from global road safety experts (see Annex 1).¹⁰ This is further described below.

Recommended Drink-Driving Variables

Drink-driving evidence is crucial to the understanding of development, implementation, and evaluation of drink-driving laws. Efforts have been made by WHO to guide countries toward establishing standardized data systems. In this regard, two manuals—Drinking and Driving: A Road Safety Manual for Decision-Makers and Practitioners (2007) and Data Systems: A Road Safety Manual for Decision-Makers and Practitioners (2010)—list variables that should be part of any drink-driving database (WHO, 2007, 2010). Table 5 lists the WHO-recommended variables and compares them to the U.S. fatality database, FARS.¹¹

The recommended road traffic injury variables are categorized into person, crash, road, vehicle characteristics, and cost. Person characteristics include age, sex, road user type, seating position, safety equipment, and so on. Crash characteristics include date and time of crash, weather, location, type, effect of crash, and lighting conditions. Road characteristics include road type, surface, obstacles, and speed limit (WHO, 2007, 2010). Variables specific to drink-driving include suspected use of alcohol, status of alcohol test, type of testing conducted, and test results. The suspected use of alcohol is reported by police and is a binary (yes/no) variable. The responses for alcohol test status include test not given, test refused, test given, and unknown if tested. Test type includes blood, breath, urine, other, or unknown; test results report value, pending value, or results unknown (WHO, 2007, 2010).

Drink-driving needs involvement of multiple stakeholders, which means that data from other organizations, such as insurance and hospital data, are also needed for a holistic understanding (WHO, 2007). Reporting of data through standardized indicators provides a common language for data presentation. It is useful for understanding time trends, thus

___________________

¹⁰ Available by request from the National Academies of Sciences, Engineering, and Medicine’s Public Access Records Office (PARO@nas.edu).

¹¹ Ibid.

enabling stakeholders to assess and evaluate the effectiveness of interventions. Table 6 compares recommended drink-driving-related indicators with the FARS.¹²

The Fatality Analysis Reporting System (FARS)

The FARS is a data system established in the United States in 1975 by NHTSA to collect data on fatal road traffic crashes that resulted in the death of a road user within 30 days of the crash event. The data are collected from the 50 U.S. states, the District of Columbia, and Puerto Rico. It includes more than 150 variables that relate to person, vehicle, crash, and road characteristics from various data sources including police reports, emergency medical services reports, hospital records, state driver and vehicle registration records, death certificates, coroner reports, and state highway department data (FARS, 2014). The FARS data can be used based on the requirements of researchers through a data query system that allows selections of variables based on research objectives and questions¹³ (FARS, 2014). Specific variables in the FARS related to drink-driving include police-reported alcohol involvement, method of alcohol determination by police, alcohol test status, test type, and test results.

However, NHTSA has found that there are significant data missing for BAC (FARS, 2014). This vacancy is handled using multiple imputation methods (FARS, 2017; Jewett et al., 2015). NHTSA, with support from the National Center for Statistics and Analysis, publishes reports that provide national and state-level statistics related to road traffic crashes and fatalities, including drink-driving data (FARS, 2014).

Data Linkage

Linking drink-driving and road traffic crash and injury data is important for understanding the full spectrum of issues associated with drink-driving. Several data sources including police data and hospital data can be linked for this purpose (Amorima et al., 2014; Derriks and Mak, 2007). However, there are challenges associated with linking various databases.

The International Traffic Safety Data and Analysis Group reports that one of the main reasons for lack of data linkage is protection of individual-level data as no personal identifiers are part of such databases. The majority OECD countries deidentify the data to protect individual identity, which has resulted in development of mechanisms that can link databases based on other variables such as date, time, location of crash, age, and sex

___________________

¹² Ibid.

¹³ See https://www-fars.nhtsa.dot.gov/Main/index.aspx (accessed January 10, 2018).

(Derriks and Mak, 2007). Exploring innovative mechanisms employed by other high-income countries is crucial to holistically addressing risk factors of drink-driving.

Selection Criteria

For this section of the paper, our approach was to present case studies from countries that have databases that collect drink-driving data. Two main criteria for selection of countries for case studies were high-income countries with similar levels of economic development to the United States, the presence of data systems (or databases) related to road traffic injuries that also collect alcohol-related data, and the demonstrated use of such data for interventions or programs against drink-driving.

Based on the above selection criteria, drink-driving-related information for 35 OECD countries was collected from the 2015 Global Road Safety Status report. This included data related to BAC limits for general population and young drivers, self-reported enforcement of drink-driving law, estimated fatality rates caused by road traffic crashes, reported number of road traffic deaths, and percentage of road traffic deaths caused by alcohol. In addition, information related to data sources for drink-driving percentages was also recorded (WHO, 2015).

The 2017 report OECD Health Statistics for injuries in road traffic crashes was also reviewed for additional databases for road traffic crashes and injuries (OECD, 2017). All the 35 OECD countries were found to have road traffic databases that collected data related to drink-driving. However, for this section of the paper, a sample of the 35 OECD countries were selected as case studies: Australia, France, New Zealand, Sweden, and Switzerland. These countries were selected because each has a comprehensive road traffic database with drink-driving variables. These databases have been shown to generate data that has led to the development and implementation of law enforcement and social media campaigns to address the issue of drink-driving in these countries.

Assessment of Selected Countries

To understand the differences between the FARS and databases from selected high-income countries, a set of five questions was selected to analyze these databases:

1. What type of data system is available in the country to collect data on drink-driving?
2. What agencies or organizations are collecting the data?
3. Who collects the data (i.e., police or hospital)?

4. What is the frequency of data dissemination?
5. If available, how is the database linked with other databases?

Table 5 compares the FARS with a recommended list of variables and databases from selected high-income countries.¹⁴ Described below are a series of case studies of data systems based on the above criteria.

Case Study A: Australia

Australia has a population of 24 million with a gross national income per capita of US$54,420 (World Bank, 2017). In 2015, the estimated rate of road traffic fatalities was 7.5 per 100,000 population and the estimated number of road traffic deaths was 1,823 (GBD, 2016). The BAC limit for the general population is 0.049 g/dl, and it is 0.00 for young drivers. The percentage of road traffic deaths involving alcohol is reported to be 28 to 30 percent (OECD, 2016; WHO, 2015). Australia has two main crash datasets that collect data on road traffic fatalities: the Australian Road Deaths Database and the National Crash Database.

Australian Road Deaths Database

The Australian Road Deaths database provides basic information (demographics and crashes) related to road traffic fatalities. The database has been hosted by the Bureau of Infrastructure, Transport and Regional Economics (BITRE) since 1989. The data are collected by police on a monthly basis, reporting both fatalities and crashes, and are linked through a unique ID (BITRE, 2014). BITRE also collects data on economic development, infrastructure, and regional developmental issues including social capital, household wealth, and industry structure. In addition to road data, it also collects data on aviation, maritime, and rail transport incidents (BITRE, 2017a). A steering committee helps improve data collection systems for timely and efficient dissemination of information to stakeholders for planning and implementation of interventions focusing on infrastructure and transportation (BITRE, 2017b).

National Crash Database

This is a national database on fatal and injurious crashes, and it is updated annually (BITRE, 2017c). Crash data in Australia are collected from eight states by police and transport authorities. Uniformity

___________________

¹⁴ Available by request from the National Academies of Sciences, Engineering, and Medicine’s Public Access Records Office (PARO@nas.edu).

in methods of crash data collection has resulted in development of this national crash database, which is maintained by the Department of Infrastructure and Regional Development. It gives fatality counts related to deaths from drink-driving, the number of drivers and motorcycle riders killed with a BAC above the limit set by law, and the number of deaths from crashes involving a driver or motorcycle rider with a BAC above the limit set by law (BITRE, 2017c).

Developing a national data system for serious injuries is a priority in Australia. The effort is directed to have a data system that links road crash data and hospital data. Currently, there is great variation in the data collected for serious road traffic injuries and hence a lack of a national database for serious injuries (OECD, 2016). However, there is a database that provides data related to nonfatal road traffic crashes, including variables related to drink-driving.

Australian Institute of Health and Welfare (AIHW)

AIHW collects data on many health-related issues including expenditure data, hospital data, alcohol, and disability (AIHW, 2017a). It reports injuries from road traffic as the “number of people injured in road traffic crashes per million population.” The data also report on “hospitalized injury that results from road traffic crashes” (OECD, 2017). In addition, the agency collects alcohol-related data from the National Drug Strategy Household Survey, which is conducted every 3 years with the latest one conducted in 2016 (AIHW, 2017b). The data are reported as the percentage of disease burden attributable to alcohol and alcohol-related hospitalizations including road traffic injuries that resulted from the use of alcohol (AIHW, 2017b). AIHW collects data on multiple aspects of health and social issues including maternal child health, youth health, ageing, disability, welfare, burden of disease, noncommunicable diseases, and injuries; it therefore offers an opportunity for linking different databases (AIHW, 2017a).

In Australia, data linkage is possible because AIHW is based within a national agency and has access to national and state-level databases. The example of AIHW is relevant in this case as it collects data on several different aspects of health. It is an accredited data integrating authority in Australia and has access to non-AIHW databases such as the National Hospitals Morbidity Database. It has a dedicated group that helps investigators and researchers by putting together linked data. AIHW facilitates data linkage by developing a customized database on requests from researchers and policy makers. Furthermore, the data linkage requires

approval from the AIHW Ethics Committee (AIHW, 2017c,d; BITRE, 2017b).¹⁵

The data from the Australian Road Deaths Database are disseminated through monthly bulletins on fatal road traffic crashes and quarterly reports on fatal road crashes involving heavy vehicles; however, these do not report data specific to drink-driving. The monthly bulletin on fatal road traffic crashes is available around the 14th of each month and contains information on counts, provides trends of fatalities since the beginning of the year and over the past 5 years for the reporting month, and the percentage change in trend. The monthly bulletin also reports count of fatalities by road user type, age group, gender, crash type, day, and time of day. Annual death rates per 100,000 population for each jurisdiction over the past 5 years for the reporting month are also given (BITRE, 2017d). The quarterly reports on fatal road crashes involving heavy vehicles include count on deaths by jurisdiction, crash type, and road user type over the past year. It also reports the percentage change over the past 1 year and 3 years.

Data from the above data sources are compiled for two annual reports: one related to fatal crashes and the other to fatal crashes involving heavy vehicles. The annual report on road traffic crashes reports the number and standardized rates of fatal crashes, injuries, and fatalities for the previous 10 years. These data are reported based on jurisdiction, road user type, age group, gender, crash type, day and time of day, and hospitalized injuries. The annual report includes a section on historical series covering data on fatalities, crashes, and rates for the past 30 years and exposure series with data in population, age group, and registered vehicles for the past 30 years. The data from this database have been used to follow progress on targets identified in Australia’s National Road Safety Strategy 2011–2020. The targets are assessed with their baseline that was reported in 2008–2010. A specific target related to drink-driving is “Number of deaths from crashes involving a driver or motorcycle rider with a BAC above the legal limit.” It was reported to be 205 at baseline and 134 in 2015, showing a 37.4 percent reduction in drink-driving-related fatalities (BITRE, 2017c,d).

The annual report for heavy vehicle crashes includes data on counts and rates of fatal road crashes involving heavy vehicles, as well as fatalities and injuries requiring hospitalization. The percentage change in these parameters is reported for the last calendar year as well as for the past few years. This report also has historical series and exposure data, such as the annual report on all road traffic crashes. It reports the BAC status of drivers involved in heavy vehicle crashes that resulted in deaths (BITRE,

___________________

¹⁵ A list of approved research projects is available at https://www.aihw.gov.au/our-services/data-linkage (accessed January 10, 2018).

2017d). In addition to national reports, these data are also reported in the Annual International Road Safety Comparisons Report published for OECD nations, allowing for comparisons to be made with other OECD nations (BITRE, 2017d).

The innovation related to data use from these databases is that the process addresses the needs of various stakeholders in multiple jurisdictions at national and international levels including the public, government officials, policy makers, several ministries, and even international road safety organizations. Public availability of these data and reports allows stakeholders to understand gaps in road safety. Specifically, these reports have been used to assess time series trends in road traffic fatalities in Australia and provide evidence related to implementation of laws and their effect on reduction in fatalities. Furthermore, the use of these data to report on progress of the National Road Safety Strategy is an effective way for the country and each jurisdiction to assess improvements in road safety interventions and policies, and to plan for future road safety strategy.

Efforts of the steering committee for timely dissemination of these results to stakeholders are crucial to the effective use of these data. The data are used for developing information sheets for sharing with the public and to raise awareness among the masses. The database is also available for research and publication related to various aspects of road safety, including vulnerable users and assessments of road safety measures in improving outcomes (BITRE, 2017d).

Case Study B: New Zealand

New Zealand has a population of 4 million with a gross national income per capita of US$39,070 (World Bank, 2017). The estimated rate of road traffic fatalities was 8.3 per 100,000 population, and the estimated number of deaths was 378 in 2015 (GBD, 2016). The BAC limit for the general population is ≤0.05 g/dl and 0.00 for young drivers. The estimate for percentage of road traffic deaths involving alcohol is reported to be 31 percent (WHO, 2015). It is reported that for every 100 people who die on roads in New Zealand because of their drink-driving, about 47 of their passengers and 24 road users who were sober also die in the same traffic crash (Ministry of Transport, 2016a).

The Ministry of Transport in New Zealand reports road traffic injuries per 100,000 people using injury and crash data from several resources, including police crash reports, hospitalization data, and the Accident Compensation Corporation (ACC) new claims data from the Motor Vehicle Authority. Road traffic crash data are collected by the police on all fatal crashes. Information collected includes crash location, drivers, vehicles,

road, number of injured, and deaths in the crash. Detailed description of the crash site and a diagram are also included as part of crash reports. Once the data are collected, they are submitted to the New Zealand Transport Agency for coding and data entry into the Crash Analysis System (CAS) (New Zealand Transport Agency, 2017). Data from the CAS are accessible through a secure login system. The process of data collection and data availability takes time; therefore, police develop an initial report that is made available within 24 hours of the crash.

Data related to person characteristics, crash event, and road conditions are collected for each crash along with road user behavior. It also includes data on the number of infringements and offences collected on a monthly basis. For fatal drink-driving crashes, the type of vehicle is also recorded. Specific to drink-driving, the number of deaths in alcohol crashes, proportion of total road traffic deaths, number of drivers tested, and number found to be over the legal alcohol limit is reported using this data (Ministry of Transport, 2016b; OECD, 2016).

In addition to crash data, linking of data is done to calculate social costs of crashes and injuries. The social cost data differ from the ACC, which collects data related to the claims cost incurred by ACC (Ministry of Transport, 2017a). These data have been reported since 2006 and are updated annually after accounting for inflation. The cost is reported for fatal and nonfatal road injuries and includes the total cost of road crashes to the nation (loss of life and life quality, loss of productivity, medical, legal, and vehicle damage costs) and average social cost per injury and per crash (Ministry of Transport, 2017a). It is important to note that the social cost data are based on crash and injury estimates and are calculated by using injury and crash conversion factors (ratio of estimated to reported numbers of injuries or crashes). This gives an estimate of the total number of road traffic incidents and accounts for nonreporting crashes and injuries (Ministry of Transport, 2006; OECD, 2017).

The data are reported on a monthly basis and are available on the website. The monthly overview of crash statistics reports deaths and injuries by age, sex, road user type, outcome of crashes, road user behavior, and social cost-related factors that contribute to crashes. It also gives time trends related to road traffic deaths since 1990. The monthly report also includes the numbers and percentage of drivers killed because of excess alcohol; this has been available since 2005. Drink-driving data for earlier years are not reported because the data are not comparable owing to changes in the crash reporting processes (Ministry of Transport, 2017b). There is also an annual statistical statement, the Motor Vehicle Crashes in New Zealand Report, which collects data from the CAS and reports annual statistics on injuries, crashes, and deaths for different road user types and ethnicities. It also gives historical trends since 1950, and includes data on

population and vehicles. This annual statement also gives international comparisons by road user type and age; this is reported as percentage and rates per 100,000 people. These annual data are available in the form of Excel sheets that are freely available from the website (Ministry of Transport, 2017b).

The ministry also publishes Crash Fact Sheets on an annual basis using the data from the CAS. It highlights key findings of drink-driving data from the previous year and reports drink-driving-related crashes based on road type, age group, gender, license status, day and time of crash, and other risk factors like use of seat belts. Time series data on fatal crashes, deaths, and injuries have been reported since 1990, and time series based on age groups and gender have been reported since 1995. The Crash Fact Sheets also summarize the history of legislation related to drink-driving in New Zealand (Ministry of Transport, 2017b; RoadSafetyBC, 2016). In addition, ad hoc reporting is also done; for example in 2012, the Ministry of Transport published a special report, High-Risk Drivers Statistics, which reported statistics from data available to the ministry. It reported on two drink-driving-related high-risk behaviors: drivers with a BAC of at least 50 percent over the adult legal limit and repeat alcohol offenders, specifically drivers in alcohol-related crashes who have at least one prior alcohol conviction in the previous 5 years (Ministry of Transport, 2012).

Road safety is a high-priority area identified by the government of New Zealand. These data are used for planning safe road and transport systems and for assessing programs that are planned and implemented to achieve goals based on New Zealand’s road safety strategy 2010–2020 (Ministry of Transport, 2017c). In this regard, the data have been used by stakeholders involved in planning and legislation related to road safety. In addition, the data are used for developing targeted campaigns for the public. They also serve to track road safety trends since 2010 and have helped to address gaps and identify opportunities that can drive work at the national level to make roads safer in New Zealand.

Specific drink-driving-related goals that have been achieved since 2010 include implementation of an alcohol interlock program, implementation of zero BAC levels for drivers under 20 years of age, and lowered BAC limit to 0.05% for those above 20 years of age. In addition, efforts to increase community responsibility for reducing drink-driving were also implemented. These efforts resulted in a reduction in road crashes caused by drink-driving (National Road Safety Committee, 2016). Currently, collaborative efforts between the Ministry of Transport and the police are under way to enforce, educate, and use social media to reduce drink-driving.

Like Australia, in New Zealand multiple stakeholders work in collaboration and have taken a data-driven approach to address the issue of

drink-driving. This approach is innovative in the context of New Zealand because it considers road safety a systems-level issue and is addressing it using a systems approach that accounts for infrastructure, roads users, and road behaviors, and uses a multiprong method to address these components of road systems to make them safer (National Road Safety Committee, 2016).

Case Study C: Sweden

The population of Sweden is 9 million with a gross national income per capita of US$54,630 (World Bank, 2017). In 2016, the estimated rate of road traffic fatalities was 4.6 per 100,000 people, and the estimated number of fatalities was 447, respectively (GBD, 2016; OECD, 2016; WHO, 2015). The BAC limit for the general population and young drivers is 0.02 g/dl. The estimate for the percentage of road traffic deaths involving alcohol is reported to be 19 to 24 percent (OECD, 2016; WHO, 2015).

The database related to road traffic injuries in Sweden is called the Swedish Traffic Accident Data Acquisition (STRADA). This national-level database was established in 2003 and collects data from the police and hospitals. Police reporting is mandatory and no additional funding is given to the police, while hospital reporting is voluntary and each participating hospital receives fixed funding each year to cover costs related to meetings, training, and materials (Howard and Linder, 2014). The police data provide details related to the crash (date, location weather, road conditions, type of crash), traffic elements (road user type, injury severity), and person (age). In terms of drink-driving data, only information related to the driver being under the influence of alcohol is collected. No further information on drink-driving is reported by the police (Howard and Linder, 2014). More than 60 out of 80 Swedish hospitals share data with STRADA (Howard and Linder, 2014). Together, these two sources of data facilitate the understanding of circumstances of road traffic crashes and their severity in the country (Howard and Linder, 2014; OECD, 2016).

Although the STRADA data system is a comprehensive source of data on serious crashes, it does not include data on individuals, crashes that were less severe, injured persons not seen in an emergency department (ED), or crashes not reported to the police. In cases where the police have information related to a minor crash or individual with minor injury, data are sent to STRADA (OECD, 2016; Transport Styrelsen, 2015). In addition, the STRADA is linked with national-level registries that collect data related to driver’s licenses and vehicles registered in the country (Howard and Linder, 2014). The STRADA data are available for use by the Swedish Transport Agency in the form of a Microsoft Access database and

Web-based system. The agency also provides 1 day of training to those interested in using the data (Howard and Linder, 2014).

Even though STRADA collects data from the police and hospitals, these two sources of data are not always linked. Only 40 percent of police data are linked with hospital data, and 30 percent of the hospital data are found to be linked to police data. The data are linked based on predefined matching criteria that develop an algorithm with the STRADA. The criteria for matching include a patient identification number, crash timing within 24 hours, and crash location within a 1,000-meter radius (person, time, and place matching). The linked data are then given a unique crash number. This unique identification number is stored in the database along with a quality estimate which is a value between 0 and 100 and is based on differences in time and location (Howard and Linder, 2014). Several reasons have been identified for the lack of 100 percent matching; police may not have registered crash and person information related to vulnerable road users (cyclists, pedestrians) or the information needed for matching is not available or it is incorrect. Hospital data may not match because individuals involved in the crash may not have gone to a hospital that is part of the STRADA system (Howard and Linder, 2014). Duplicate cases are checked through monthly quality checks performed by the Swedish Transport Agency and by collecting hospital data from the first hospital where the crash victim was seen. No data from hospitals to where the patient are later transferred are collected (Howard and Linder, 2014).

Data are disseminated through monthly and annual reports (Howard and Linder, 2014; Transport Styrelsen, 2015, 2016, 2017). Sweden has implemented Vision Zero to address the issue of road safety in the country and targets zero road traffic-related deaths by the year 2050. The innovation for the use of data in Sweden is to make road systems safe for its users so that there is no crash, injury, or fatality where the road system is a contributing factor; thus the focus is more on engineering rather than enforcement of laws.

The data from the STRADA are used for understanding changes in road traffic crashes and fatalities because of road interventions. The data have been used for making improvements in the existing system through collaborations with stakeholders (engineers, transport agencies, and ministries) as well as government. The data identify areas of improvement and facilitate development and implementation of new road safety interventions in the country. These data were also presented at the Vision Zero conference held in Sweden annually for the past 9 years. One of the major steps taken in this regard was the relaunch of Vision Zero by the Swedish government in 2016 with the key focus on stakeholder engagement to achieve the zero target (Swedish Transport Administration, 2017a,b).

The hospital component of the data is used for different research projects. These include analysis of vulnerable road users, such as pedestrians and bicycle users; accuracy of data in terms of response rate; and classification of injury. Some work is more regional while the rest is at the national level (Howard and Linder, 2014). The data are also used for assessing interventions implemented for achieving Vision Zero in Sweden, such as interlock systems (Transport Styrelsen, 2015).

Case Study D: France

The population of France is 66 million with a gross national income per capita of US$38,950 (World Bank, 2017). In 2016, the estimated rate of road traffic fatalities was 7.9 per 100,000 people, and the estimated number of road traffic deaths was 5,179 (GBD, 2016; WHO, 2015). The BAC limit is ≤0.05 g/dl for both the general population and young drivers. The percentage of road traffic deaths involving alcohol is reported to be 19 to 29 percent (OECD, 2016; WHO, 2015).

In France, road traffic crashes that result in an injury are recorded by the police. Variables collected include information related to vehicle occupants and classification of injury severity into four categories: noninjured, slightly injured (outpatients or inpatients with less than a 6-day hospital stay), seriously injured (those requiring a longer hospital stay), or killed (individuals who die within 6 days of the crash) (Amoros et al., 2008). These data are then collected centrally, which has led to the development of a national road traffic crash database. The data management is done by the French Road Safety Observatory (ONISR, Observatoire National Interministériel de la Sécurité Routière) within the Ministry of Interior (Ministry of Road Safety, 2015a, 2017).

Data from the national database are accurate for fatal and serious crashes; however, crashes resulting in minor injuries are underreported owing to variation across the country in reporting such events (OECD, 2016). The database has data related to person, road, vehicle, and crash characteristics. Drink-driving-related data include variables related to BAC levels; the ONISR reports data on fatal crashes that involve at least one driver with a known BAC level (Ministry of Road Safety, 2015a).

ONISR also assesses road user behavior using three data sources that include observations of road user behavior; these observations are conducted by ONISR. In addition, it also collects data on offenses and permits that are gathered by law enforcement agencies. Data on convictions associated with road traffic offenses are collected from the Ministry of Justice (Ministry of Road Safety, 2015b). There is a lack of information related to how these data sources can be linked to provide a complete picture related to road safety risk factors.

There are no county-level data in France except for Rhone County, where police data are compared with the Rhone trauma registry that collects medical care data from health facilities. The facilities include primary care clinics, EDs, and rehabilitation units. The data from police and health care facilities are linked based on date and time of crash, crash location, year, month, and gender of crash victims (Amoros et al., 2008).

ONISR has multiple stakeholders, including enforcement agencies and road safety departments. These data are shared with them on a regular basis to help with the enforcement of road safety and drink-driving laws in the country (Ministry of Road Safety, 2015a, 2017). These data are disseminated via monthly reports that provide the number of injuries, casualties, and hospitalizations. It reports trends from the current year and the previous year and helps identify areas of road safety that need immediate action (Ministry of Road Safety, 2017). The annual report gives fatality trends since 2010. It also gives comparisons of crashes, fatalities, injured people, and hospitalization from the current year and previous year and presents the percentage change in these data. Data also include road type, day of the week, road user type, age group, and sex. It also gives a comparative analysis of data on vulnerable road users in France and the rest of Europe. Specific to drink-driving, data on percentage of crashes caused by the use of alcohol are reported for the overall population and for different age groups (Ministry of Road Safety, 2017).

In the year 2014, ONISR reported local performance indicators for France and its counties. These indicators were developed based on main road safety issues in France and included drivers driving under the influence of alcohol (Ministry of Road Safety, 2016). Data from ONISR are also made available for the IRTAD Annual Report to provide a comparison with OECD nations by providing annual trends in road safety based on road user type, age group, sex, and road safety risk factors including drink-driving (Ministry of Road Safety, 2017).

Case Study E: Switzerland

Switzerland has a population of 8 million with a gross national income per capita of US$81,240 (World Bank, 2017). In 2015, the estimated rate of road traffic fatalities was 4.7 per 100,000 people, and the reported road traffic deaths were 390 in 2015 (GBD, 2016). The BAC limit for the general population is ≤0.05 g/dl and ≤0.01 g/dl for young drivers. The percentage of road traffic deaths involving alcohol is reported to be 14 to 16 percent (OECD, 2016; WHO, 2015).

The Federal Roads Office (FEDRO) in Switzerland is responsible for the collection of road traffic crash data in the country. These data are collected by police. It is hosted by the Federal Department of the Environment, Transport, Energy and Communications (DETEC) (FEDRO, 2017).

Like other high-income countries, Switzerland also collects data on person, crash, vehicle, and road characteristics. The data have been available since 2002. They are accessible via Excel spreadsheets available free of cost at the FEDRO website (FEDRO, 2017).

The crash data reported by police are compared with insurance claim data related to road traffic crashes. This comparison is done to make an accurate estimation of road traffic injuries. It also accounts for underreporting based on road user type and age group. Recently, efforts were made to link police and hospital data, and data are reported by age group and road user type (FEDRO, 2016; OECD, 2016). A pilot program was conducted that showed promising results and led to the implementation of data linkage procedures in the national database (OECD, 2016).

The data are disseminated via monthly and annual reports and shared with national and international organization partners to plan interventions and assess their effect (FEDRO, 2016). FEDRO uses these data to plan changes in infrastructure and then assesses the changes to see their effect on reduction in road traffic crashes (FEDRO, 2016). Table 7 compares databases from the above countries.¹⁶

Recommendations for Data Systems

Based on the analysis above on a sample of high-income countries similar to the United States, key points include the following:

• All data systems collect basic data related to road traffic crashes, fatalities, and injuries and present them based on road user type, sex, age groups, day, time of day, and in some cases road type. Most of these data are available in the form of time series to assess trends in road traffic mortality and injuries.
• In all of these countries there is a national-level data system that collects detailed data on road traffic crashes. These data are collected by police and managed at a central level for coding and cleaning. The cleaned data are available in most cases to the public (except for New Zealand, where special training is needed to use the data).
• Data from these countries are reported on a monthly and annual basis and used by local and national governments to inform decision making about road safety measures that are identified within national road safety strategies in these countries (e.g., New Zealand and Sweden).

___________________

¹⁶ Available by request from the National Academies of Sciences, Engineering, and Medicine’s Public Access Records Office (PARO@nas.edu).

• Data are also available for use by researchers and international road safety agencies to generate evidence that is available for policy makers to develop and implement road safety policies within countries based on gaps identified in the data.
• Data systems in New Zealand and Switzerland are linked with other data systems, such as insurance data; in Australia fatal and nonfatal data are linked by AIHW because it has access to several sources of health-related data.

Some considerations in applying findings from the above case studies to the FARS are as follows:

• The FARS has a comprehensive set of variables on fatal crashes including person, vehicle, road, and crash characteristics. There are a few variables, such as passenger position, injury position (arm, head), activity of road user at the time of crash (reversing), and purpose of journey, that are not in the FARS and could be included for better understanding of risk factors associated with drink-driving in the United States. However, it is important to note that the FARS collects all WHO-recommended variables for drink-driving.
• While the FARS collects data on fatal injuries from the participating states, there is a need to establish a similar system that collects nonfatal data at a central level. This requires engagement with various stakeholders, including hospitals, emergency services, police, and state authorities.
• There is no combined database for fatal and nonfatal injuries at the national level in the United States. Several states in the United States have databases that combine fatal and nonfatal data on road traffic crashes and injuries; however, the true burden of road injuries and crashes attributed to alcohol requires analysis of both fatal and nonfatal databases at the national level. Such a system at the national level will help in cost estimations and planning, implementation, and evaluation of interventions geared toward reducing road traffic crashes and injuries caused by drink-driving (Amorima et al., 2014; Derriks and Mak, 2007).

Finally, this paper recommends further study and analysis of U.S.-based data to explore the quality of data collected and associated analytic limitations around alcohol-related data in the FARS and relevant state databases. More work is needed in this area to help efforts to curb drink-driving.

REFERENCES

BITRE (Bureau of Infrastructure, Transport and Regional Economics). 2014. Australian road deaths database. https://bitre.gov.au/statistics/safety/fatal_road_crash_database.aspx (accessed October 31, 2017).

BITRE. 2017a. Regional databases. https://bitre.gov.au/databases/regional.aspx (accessed October 31, 2017).

BITRE. 2017b. National infrastructure data collection and dissemination plan. https://bitre.gov.au/data_dissemination/index.aspx (accessed October 31, 2017).

BITRE. 2017c. Road trauma Australia 2016 statistical summary. https://bitre.gov.au/publications/ongoing/files/Road_Trauma_Australia_2016_Web.pdf (accessed October 31, 2017).

BITRE. 2017d. Safety statistics—road safety statistics. https://bitre.gov.au/statistics/safety (accessed October 31, 2017).

Breitmeier, D., I. Seeland-Schulze, H. Hecker, and U. Schneider. 2007. The influence of blood alcohol concentrations of around 0.03% on neuropsychological functions—a double-blind, placebo-controlled investigation. Addiction Biology 12(2):183–189.

Caetano, R. 2008. About smoke and mirrors: The alcohol industry and the promotion of science. Addiction 103:175–178.

Caetano, R., and R. Laranjeira. 2006. A “perfect storm” in developing countries: Economic growth and the alcohol industry. Addiction 101(2):149–152.

Cameron, M. 2013. Best practice in random breath testing and cost-effective intensity levels. 20th International Conference on Alcohol, Drugs and Traffic Safety (T2013), Brisbane, Queensland, Australia.

Casswell, S. 2011a. Alcohol harm—the urgent need for a global response. Addiction 106(7): 1205–1207.

Casswell, S. 2011b. Alcohol marketing research agenda—let us look at how the industry maintains its hegemony. Addiction 106(3):476–477.

Chan, M. 2013. Re: Doctors and the alcohol industry: An unhealthy mix? BMJ 346:f2647.

Christie, R., D. Carseldine, and S. Brown. 1995. National guidelines for alcohol ignition interlock programs for drink driving offenders. Sydney, Australia: Austroroads.

Constant, A., L. R. Salmi, S. Lafont, M. Chiron, and E. Lagarde. 2009. Road casualties and changes in risky driving behavior in France between 2001 and 2004 among participants in the gazel cohort. American Journal of Public Health 99(7):1247–1253.

Derriks, H. M., and P. M. Mak. 2007. IRTAD special report—underreporting of road traffic casualties. The Netherlands: OECD and International Transport Forum.

Desapriya, E., S. Shimizu, I. Pike, S. Subzwari, and G. Scime. 2007. Impact of lowering the legal blood alcohol concentration limit to 0.03 on male, female and teenage drivers involved alcohol-related crashes in Japan. International Journal of Injury Control and Safety Promotion 14(3):181–187.

Ditter, S. M., R. W. Elder, R. A. Shults, D. A. Sleet, R. Compton, J. L. Nichols, and Community Preventive Services Task Force. 2005. Effectiveness of designated driver programs for reducing alcohol-impaired driving: A systematic review. American Journal of Preventive Medicine 28(5 Suppl):280–287.

Elder, R. W., R. A. Shults, D. A. Sleet, J. L. Nichols, S. Zaza, and R. S. Thompson. 2002. Effectiveness of sobriety checkpoints for reducing alcohol-involved crashes. Traffic Injury Prevention 3(4):266–274.

Elder, R. W., R. A. Shults, D. A. Sleet, J. L. Nichols, R. S. Thompson, W. Rajab, and Community Preventive Services Task Force. 2004. Effectiveness of mass media campaigns for reducing drinking and driving and alcohol-involved crashes: A systematic review. American Journal of Preventive Medicine 27(1):57–65.

Elder, R. W., J. L. Nichols, R. A. Shults, D. A. Sleet, L. C. Barrios, R. Compton, and Community Preventive Services Task Force. 2005. Effectiveness of school-based programs for reducing drinking and driving and riding with drinking drivers: A systematic review. American Journal of Preventive Medicine 28(5 Suppl):288–304.

Erke, A., C. Goldenbeld, and T. Vaa. 2009. Good practice in the selected key areas of speeding, drink driving and seat belt wearing; results from meta-analysis. Police Enforcement Policy and Programmes on European Roads. Contract No. 019744.

Esser, M. B., J. Bao, D. H. Jernigan, and A. A. Hyder. 2016. Evaluation of the evidence base for the alcohol industry’s actions to reduce drink driving globally. American Journal of Public Health 106(4):707–713.

EU (European Union) Directorate General for Health and Consumers. 2009. First progress report on the implementation of the EU alcohol strategy. Brussels, Belgium: European Union Directorate General for Health and Consumers.

European Commission. 1989a. Proposal for a council directive relating to the maximum permitted blood alcohol concentration for vehicle drivers. Official Journal of the European Communities 32:9.

European Commission. 1989b. Opinion on the proposal for a council directive relating to the maximum permitted blood alcohol concentration for vehicle drivers. Official Journal of the European Communities 32:54–55.

European Commission. 1997. Promoting road safety in the EU: The Programme for 1997−2001. http://webcache.googleusercontent.com/search?q=cache:ZXSkku-AdDcJ:aei.pitt.edu/4729/1/4729.pdf+&cd=1&hl=en&ct=clnk&gl=us (accessed December 12, 2017).

European Commission. 2001. Recommendation of 17 January 2001 on the maximum permitted blood alcohol content (BAC) for drivers of motorised vehicles. Official Journal of the European Communities 44.

FARS (Fatal Analysis Reporting System). 2014. Fatality Analysis Reporting System. https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/811992 (accessed October 31, 2017).

FARS. 2017. FARS alcohol data. https://www-fars.nhtsa.dot.gov/QueryTool/Common/alcohol_warning.html (accessed October 31, 2017).

FEDRO (Federal Roads Office). 2016. Developments, facts and figures: Roads and traffic. https://www.astra.admin.ch/astra/en/home/documentation/facts-and-figures/archive.html (accessed October 31, 2017).

FEDRO. 2017. About FEDRO. https://www.astra.admin.ch/astra/en/home.html (accessed October 31, 2017).

Fitzharris, M., S. Liu, S. Peiris, A. Devlin, K. Young, M. Lenne, D. Bowman, and J. Gatof. 2015. Options to extend coverage of alcohol interlock programs. Austroads, September 30.

GBD (Global Burden of Disease). 2016. Global burden of disease study. https://vizhub.health-data.org/gbd-compare (accessed October 31, 2017).

Gruenewald, P. J., P. R. Mitchell, and A. J. Treno. 1996. Drinking and driving: Drinking patterns and drinking problems. Addiction 91(11):1637–1649.

Gururaj, G. 2004. Alcohol and road traffic injuries in south Asia: Challenges for prevention. Journal of the College of Physicians and Surgeons Pakistan 14(12):713–718.

Hawkins, B., and J. McCambridge. 2014. Industry actors, think tanks, and alcohol policy in the United Kingdom. American Journal of Public Health 104(8):1363–1369.

Hawkins, B., C. Holden, and J. McCambridge. 2012. Alcohol industry influence on UK alcohol policy: A new research agenda for public health. Critical Public Health 22(3):297–305.

Henstridge, J., R. Homel, and P. Mackay. 1997. The long-term effects of random breath testing in four Australian states: A time series analysis. Canberra: Department of Transport and Regional Development, Federal Office of Road Safety.

Hope, A. 2006. The influence of the alcohol industry on alcohol policy in Ireland. Nordic Studies on Alcohol and Drugs 23(6):467–481.

Howard, C., and A. Linder. 2014. Review of Swedish experiences concerning analysis of people injured in traffic accidents. Brussels, Belgium: Belgian Road Safety Institute.

INEBRIA (International Network on Brief Interventions for Alcohol & Other Drugs). 2015. INEBRIA position statement on the alcohol industry. http://inebria.net/wp-content/uploads/2016/02/position_statement_on_the_alcohol_industry.pdf (accessed January 10, 2018).

Jernigan, D. H. 2009. The global alcohol industry: An overview. Addiction 104(Suppl 1):6–12.

Jernigan, D. H. 2012. Global alcohol producers, science, and policy: The case of the International Center for Alcohol Policies. American Journal of Public Health 102(1):80–89.

Jewett, A., R. A. Shults, T. Banerjee, and G. Bergen. 2015. Alcohol–impaired driving among adults—United States, 2012. Morbidity and Mortality Weekly 64(30):814–817.

Johnson, M. 2012. Assessing a country’s drink driving situation: An overview of the method used in 6 low- and middle-income countries. Traffic Injury Prevention 13(2):96–100.

Kaufman, E. J., and D. J. Wiebe. 2016. Impact of state ignition interlock laws on alcohol-involved crash deaths in the United States. American Journal of Public Health 106(5):865–871.

Kennedy, R. S., J. J. Turnage, R. L. Wilkes, and W. P. Dunlap. 1993. Effects of graded dosages of alcohol on nine computerized repeated-measures tests. Ergonomics 36(10):1195–1222.

Killoran, A., U. Canning, N. Doyle, and L. Sheppard. 2010. Review of effectiveness of laws limiting blood alcohol concentration levels to reduce alcohol-related road injuries and deaths. London: National Institute for Health and Clinical Excellence.

Klipp, S. 2013. Towards a national alcohol interlock field trial in Germany. Achievements and challenges. Bundesanstalt für Straßenwesen (BASt). https://kuldoc.com/download/towards-a-national-alcohol-interlock-field-trial-in-germany-achievements-and-challenges-_5948bd121723dd36bf5a1655_pdf (accessed December 19, 2017).

Lahausse, J. A., and B. N. Fildes. 2009. Cost-benefit analysis of an alcohol ignition interlock for installation in all newly registered vehicles. Traffic Injury Prevention 10(6):528–537.

Li, Y., D. Xie, G. Nie, and J. Zhang. 2012. The drink driving situation in China. Traffic Injury Prevention 13(2):101–108.

Mansdotter, A. M., M. K. Rydberg, E. Wallin, L. A. Lindholm, and S. Andreasson. 2007. A cost-effectiveness analysis of alcohol prevention targeting licensed premises. European Journal of Public Health 17(6):618–623.

McCambridge, J., B. Hawkins, and C. Holden. 2013. Industry use of evidence to influence alcohol policy: A case study of submissions to the 2008 Scottish government consultation. PLoS Medicine 10(4):e1001431.

McCambridge, J., K. Kypri, P. Miller, B. Hawkins, and G. Hastings. 2014. Be aware of Drinkaware. Addiction 109(4):519–524.

Mercer, G. W. 1985. The relationships among driving while impaired charges, police drinking-driving roadcheck activity, media coverage and alcohol-related casualty traffic accidents. Accident Analysis & Prevention 17(6):467–474.

Miller, P. G., F. de Groot, S. McKenzie, and N. Droste. 2011. Vested interests in addiction research and policy. Alcohol industry use of social aspect public relations organizations against preventative health measures. Addiction 106(9):1560–1567.

Ministry of Road Safety. 2015a. Statistical series. http://www.securite-routiere.gouv.fr/la-securite-routiere/l-observatoire-national-interministeriel-de-la-securite-routiere/series-statistiques (accessed October 31, 2017).

Ministry of Road Safety. 2015b. User behaviors. http://www.securite-routiere.gouv.fr/la-securite-routiere/l-observatoire-national-interministeriel-de-la-securite-routiere/comportements-des-usagers (accessed October 31, 2017).

Ministry of Road Safety. 2016. Local performance indicators 2014. http://www.securite-routiere.gouv.fr/la-securite-routiere/l-observatoire-national-interministeriel-de-la-securite-routiere/english-version/local-performance-indicators (accessed October 31, 2017).

Ministry of Road Safety. 2017. Road accidents. http://www.securite-routiere.gouv.fr/la-securite-routiere/l-observatoire-national-interministeriel-de-la-securite-routiere/english-version (accessed October 31, 2017).

Ministry of Transport. 2006. The social cost of road crashes and injuries: June 2006 update. Wellington: New Zealand Ministry of Transport.

Ministry of Transport. 2012. High-risk drivers in fatal and serious crashes: 2006–2010. http://www.transport.govt.nz/assets/Import/Documents/High-risk-drivers-in-fatal-and-serious-crashes-2006-2010.pdf (accessed October 31, 2017).

Ministry of Transport. 2016a. Motor vehicle crashes in New Zealand 2015. http://www.transport.govt.nz/research/roadcrashstatistics/motorvehiclecrashesinnewzealand/motor-vehicle-crashes-in-new-zealand-2015 (accessed October 31, 2017).

Ministry of Transport. 2016b. Alcohol and drugs. http://www.transport.govt.nz/research/crashfacts/alcohol-and-drugs (accessed October 31, 2017).

Ministry of Transport. 2017a. The social cost of road crashes and injuries. http://www.transport.govt.nz/research/roadcrashstatistics/thesocialcostofroadcrashesandinjuries (accessed October 31, 2017).

Ministry of Transport. 2017b. Road crash statistics. 2017. http://www.transport.govt.nz/research/roadcrashstatistics (accessed October 31, 2017).

Ministry of Transport. 2017c. Our work. http://www.transport.govt.nz/ourwork (accessed October 31, 2017).

Mosher, J. F. 2012. Joe Camel in a bottle: Diageo, the Smirnoff brand, and the transformation of the youth alcohol market. American Journal of Public Health 102(1):56–63.

National Road Safety Committee. 2016. Safer journeys: Action plan 20l6–2020. http://www.saferjourneys.govt.nz/assets/Safer-journeys-files/Safer-Journey-Action-Plan-2016-2020.pdf (accessed October 31, 2017).

New Zealand Transport Agency. 2017. Crash Analysis System (CAS). https://www.nzta.govt.nz/safety/safety-resources/crash-analysis-system (accessed October 31, 2017).

Ngoc, L. B., N. T. Thieng, and N. L. Huong. 2012. The drink driving situation in Vietnam. Traffic Injury Prevention 13(2):109–114.

Noordzij, P. C. 1994. Decline in drinking and driving in the Netherlands. Paper presented at Transportation Research Circular, The Hague, The Netherlands.

Norström, T., and B. Trolldal. 2013. Was the STAD programme really that successful? Nordic Studies of Alcohol and Drugs 30(3):171–178.

OECD (Organisation for Economic Co-operation and Development). 2016. Road safety annual report 2016. Paris, France: OECD/International Transport Forum.

OECD. 2017. Injuries in road traffic accidents—OECD health statistics. Definitions, sources and methods. Paris, France: Organisation for Economic Co-operation and Development.

Peden, M., R. Scurfield, D. Sleet, D. Mohan, A. A. Hyder, E. Jarawan, and C. Mathers. 2004. World report on road traffic injury prevention. Geneva, Switzerland: World Health Organization.

Rehnman, C., J. Larsson, and S. Andreasson. 2005. The beer campaign in Stockholm—attempting to restrict the availability of alcohol to young people. Alcohol 37(2):65–71.

RoadSafetyBC. 2016. Immediate roadside prohibition fact sheet. Victoria, British Columbia: Ministry of Public Safety and Solicitor General.

Ross, H. L., D. T. Campbell, and G. V. Glass. 1970. Determining the social effects of a legal reform: The British “Breathalyser” crackdown of 1967. American Behavioral Scientist 13(4):493–509.

Schonfeld, C., and M. Sheehan. 2004. Critical overview of alcohol ignition interlock programs in Australia. Carseldine, Australia: Centre for Accident Research and Road Safety-Queensland.

Shield, K. D., G. Gmel, J. Patra, and J. Rehm. 2012. Global burden of injuries attributable to alcohol consumption in 2004: A novel way of calculating the burden of injuries attributable to alcohol consumption. Population Health Metrics 10(1):9.

Shults, R. A., R. W. Elder, D. A. Sleet, J. L. Nichols, M. O. Alao, V. G. Carande-Kulis, S. Zaza, D. M. Sosin, R. S. Thompson, and Community Preventive Services Task Force. 2001. Reviews of evidence regarding interventions to reduce alcohol-impaired driving. American Journal of Preventive Medicine 21(4 Suppl):66–88.

Stenius, K., and T. F. Babor. 2010. The alcohol industry and public interest science. Addiction 105(2):191–198.

Stewart, K., D. Silcock, and F. Wegman. 2012. Reducing drink driving in low- and middle-income countries: Challenges and opportunities. Traffic Injury Prevention 13(2):93–95.

Swedish Transport Administration. 2017a. The number of fatalities on Swedish roads is not decreasing. https://www.trafikverket.se/en/visionzero/vision-zero-conference-2017/news/2017/2017-05/the-number-of-fatalities-on-swedish-roads-is-not-decreasing (accessed October 31, 2017).

Swedish Transport Administration. 2017b. Vision zero and ways to work. https://www.trafikverket.se/en/startpage/operations/Operations-road/vision-zero-academy/Vision-Zero-and-ways-to-work (accessed October 31, 2017).

Sweedler, B. M., and K. Stewart. 2009. Worldwide trends in alcohol and drug impaired driving. In Drugs, driving and traffic safety, edited by J. C. Verster, S. R. Pandi-Perumal, J. G. Ramaekers, and J. J. de Gier. Basel, Switzerland: Birkhäuser Basel. Pp. 23–41.

Tagawa, M., M. Kano, N. Okamura, M. Itoh, E. Sakurai, T. Watanabe, and K. Yanai. 2000. Relationship between effects of alcohol on psychomotor performances and blood alcohol concentrations. Japanese Journal of Pharmacology 83(3):253–260.

Terer, K., and R. Brown. 2014. Effective drink driving prevention and enforcement strategies: Approaches to improving practice. Trends & Issues in Crime and Criminal Justice. Canberra: Australian Institute of Criminology 472.

Transport Styrelsen. 2015. Swedish traffic accident data acquisition (STRADA). http://www.transportstyrelsen.se/sv/vagtrafik/statistik-och-strada/STRADA-informationssystem-for-olyckor-skador (accessed October 31, 2017).

Transport Styrelsen. 2016. Alkola lock after steering wheel. http://www.transportstyrelsen.se/sv/Publikationer/Vag (accessed October 31, 2017).

Transport Styrelsen. 2017. Accident statistics. http://www.transportstyrelsen.se/sv/vagtrafik/statistik-och-strada/Vag/Olycksstatistik (accessed October 31, 2017).

Vingilis, E., and L. Salutin. 1980. A prevention programme for drinking driving. Accident Analysis & Prevention 12(4):267–274.

Voas, R. B., P. R. Marques, A. S. Tippetts, and D. J. Beirness. 1999. The Alberta interlock program: The evaluation of a province-wide program on DUI recidivism. Addiction 94(12):1849–1859.

Wallin, E., T. Norstrom, and S. Andreasson. 2003. Alcohol prevention targeting licensed premises: A study of effects on violence. Journal of Studies of Alcohol 64(2):270–277.

Wallin, E., J. Gripenberg, and S. Andreasson. 2005. Overserving at licensed premises in Stockholm: Effects of a community action program. Journal of Studies of Alcohol 66(6):806–814.

WHO (World Health Organization). 2007. Drinking and driving: A road safety manual for decision-makers and practitioners. Geneva, Switzerland: World Health Organization and Global Road Safety Partnership.

WHO. 2010. Data systems: A road safety manual for decision-makers and practitioners. Geneva, Switzerland: World Health Organization.

WHO. 2013. Global plan for the decade of action for road safety 2011-2020. Geneva, Switzerland: World Health Organization.

WHO. 2014. Global status report on alcohol and health. Geneva, Switzerland: World Health Organization.

WHO. 2015. Global status report on road safety. Geneva, Switzerland: World Health Organization.

Willis, C., S. Lybrand, and N. Bellamy. 2004. Alcohol ignition interlock programmes for reducing drink driving recidivism. Cochrane Database Systematic Review (4):CD004168.

World Bank. 2017. World Bank open data 2016. https://data.worldbank.org (accessed October 31, 2017).

Zivkovic, V., V. Lukic, and S. Nikolic. 2016. The influence of alcohol on pedestrians: A different approach to the effectiveness of the new traffic safety law. Traffic Injury Prevention 17(3):233–237.

This page intentionally left blank.