Alcohol policy impact studies
There is a very substantial literature on the effects of alcohol control policy changes on drinking amounts, patterns and problems. Data used in these analyses has primarily been of two types: social and health statistics, such as alcohol sales data, police statistics and mortality and hospital discharge data; and before-and-after surveys, mostly cross-sectional but in a few cases longitudinal. Some studies have included control sites, and one or two notable studies have included a random assignment to intervention or control condition (e.g. Norström and Skog, 2003).
Alcohol policy impact studies have primarily been carried out in a limited range of countries, generally excluding both the developing world (Room et al., 2002) and Southern European wine cultures. Even between somewhat similar societies, there are substantial variations in the research emphasis on particular topics (Room, 2004).
There is an imperfect fit between what those involved in liquor licensing decisions may want to know and what is available in the literature on alcohol controls. This gap between the content of alcohol control legislation and the research literature has been documented in the USA (Wagenaar and Toomey, 2000), but exists also elsewhere — particularly in countries where the tradition of alcohol policy impact studies has not been strong. The studies are sometimes done because a change was controversial in a particular jurisdiction, and funding an evaluation was a way of defusing the controversy.
Other studies have been opportunistic, where a researcher seizes the chance to do a
‘natural experiment’ study (‘natural’ here means that the researcher did not have a voice in the circumstances of the change, so that the study’s design is often constrained).
Often studies have made use of available data, such as per-capita consumption data or mortality registers. Since research is usually a national government responsibility, its topical focus is not necessarily attuned to the concerns of local jurisdictions.
Nevertheless, the growth of the literature evaluating the effects of alcohol controls has been a substantial achievement involving a number of national traditions, and lessons from it can be applied, with suitable caution, across jurisdictions, and drawn on in thinking about cannabis policy. Reviews are now available (e.g. Babor et al., 2003;
Room et al., 2002) which summarise the findings and implications of the literature. A new step forward, as part of the WHO-CHOICE programme (‘Choosing interventions which are cost effective’, available at: www3.who.int/whosis/menu.cfm?path=evidenc e,cea&language=english), has been the estimation of the relative cost-effectiveness of different strategies and combinations of strategies to prevent alcohol-related problems (Chisholm et al., 2004), in terms of dollars per saved DALY (disability-adjusted life
Chapter 8
123 year). Table 1 shows some of the results from these analyses for the ‘Europe-A’ WHO subregion, which is roughly coextensive with the European Union. Since evidence was lacking for any effectiveness of mass media persuasion of school-based education, these strategies were excluded from the analysis as having no apparent cost-effectiveness. In terms of cost-effectiveness per DALY saved in developed European countries, then, the policies tested ranked as follows (most cost-effective first): taxes (even without counting the revenues from taxes); advertising ban; closing times (specifically, Saturday closing for off-sales); random traffic breath tests; screening and brief medical advice; and (with no cost-effectiveness) mass media persuasion and school education.
Tobacco policy impact studies
There is also a substantial literature of tobacco policy impact studies. As for alcohol, there are several synthetic reviews of the literature (e.g. Jha and Chaloupka, 1999;
Rabin and Sugarman, 2001). Whereas the alcohol policy impact literature aims primarily at assessing the impact of specific interventions, the equivalent tobacco literature is often aimed at assessing the impact of anti-smoking policy packages as a whole (e.g. Siegel and Biener, 1997; Pierce et al., 1998). This partly reflects the reality that policy changes in the tobacco area have often involved the simultaneous application of multiple strategies. It also reflects the different circumstances of the substances in the countries where the main policy impact studies have been done. For alcohol the status quo ante has often been a detailed system of controls on availability and on places and times of use, with the literature often studying what happens when one or more of the controls is removed or relaxed. For tobacco the status quo ante has been very little control on availability, and the literature is primarily studying the effect of initiating measures such as anti-smoking persuasion campaigns, controls on places of use and on age of purchase, and raised prices, which have been increasingly put forward as a coordinated package.
Table 1:
Comparative cost-effectiveness of alcohol interventions in ‘Europe-A’(Chisholm et al., 2004)
DALYs saved/million
population Average cost-effectiveness ratio ($/DALY)
Brief medical advice 1 889 2 351
Tax: current + 50 % 1 764 258
Tax: current + 25 % 1 576 289
Tax: current 1 365 333
Advertising ban 459 594
Saturday closing for off-sales 251 1 087
Random traffic breath tests 247 2 467
In thinking about cannabis policy, what can be learned from alcohol and tobacco?
124
Comparing the alcohol policy and tobacco control literatures, one can find clear differences in emphasis. Taxes loom even larger as a strategy for tobacco than they do for alcohol (see Chaloupka et al., 2001). Although a much greater proportion of the total harms from alcohol than from tobacco are to others, the aim of reducing harm from ‘second-hand smoke’ has proved politically potent for tobacco control in a way that has only been true for drink-driving in alcohol policy. Accordingly, a strong emphasis in the tobacco literature has been put on environmental prohibitions — bans on smoking at work and in public places — which are already, to a considerable degree, taken for granted with respect to alcohol.
In this connection, Hauge (1999) has argued that the modern emphasis on health harm to the drinker has been a policy mistake in the alcohol field. The two policy impact literatures have also reached substantially different conclusions about the effects of counter-advertising campaigns. This probably primarily reflects the differences in the aims and content of the campaigns, as well as differences in the social politics of the substances. The anti-smoking campaigns which have proved effective (Pechman and Reibling, 2000; Sly et al., 2001, 2002; Wakefield et al., 2003) have often involved frontal attacks financed by governmental agencies on the bona fides of the tobacco industry. This is an unusual enough occurrence in a capitalist society to have impressed teenagers, at least in the short run — although the campaigns have often proved politically unsustainable in the longer run (Givel and Glantz, 2000). Also, more available in the nicotine field, though underutilised, has been the option of harm reduction through changing the mode of use of the psychoactive substance (Shiffman et al., 1997).
As for alcohol, the WHO-CHOICE programme has calculated estimated cost-effectiveness ration for specific interventions, and for combinations of interventions (Shibuya et al., 2003). Results for ‘Europe-A’ are shown in Table 2. Again, the cost-effectiveness calculations exclude the government revenue gained from the tax from the calculations. A comparison of the results suggests that somewhat more conservative assumptions were used in the alcohol calculations than in the tobacco calculations.
Instead of impact studies: ‘expert knowledge’
As will be apparent from the discussion above, there is great variability in the availability of published evidence on the effects of policies governing the availability and use of psychoactive substances, both licit and illicit. It should be noted, however, that the lack of a formal academic literature does not mean a lack of practical knowledge of the effects of policies. As Valverde (2003) has documented for the alcohol control system in Ontario, those staffing regulatory systems typically build up a job-based stock of knowledge, often mixing ‘facts’ and values, which guide their everyday actions. On the other hand, there is ample experience from medicine and other professions with
Chapter 8
125 such practice knowledge that its conclusions about effects are often mistaken, when subjected to the harsh test of well-designed outcome and impact studies. It would be advantageous, with respect to cannabis policy, and for that matter policy on all psychoactive substances, to move to an ‘evidence-based’ standard of policymaking. This requires a substantial investment in developing the evidence on which the policymaking can be based.