Stern alcohol policies fail to improve public health

New research found that the severity of alcohol policy had no association with the number of disability-adjusted life years (DALYs) lost due to alcohol in 30 OECD countries in 2005 nor with alcohol consumption (Table, Figure). DALYs are the most comprehensive measure of public health. DALYs are composites of the number of years lost due to ill-health, disability and premature death. The rate of excise tax on alcoholic beverages was not related to alcohol-related DALYs (25 countries with tax rate data).

Partial correlation coefficients between alcohol policy index, alcohol consumption and number of DALYs lost due to alcohol use in 30 countries, 2005

Partial correlation coefficients between alcohol policy index, alcohol consumption and number of DALYs lost due to alcohol use in 30 countries, 2005

The severity of a country’s alcohol control policies was gauged by Alcohol Policy Index, a composite indicator comprising regulations from five policy domains: availability (restrictions to access) of alcohol, drinking context, alcohol prices, alcohol advertising, and motor vehicle driving restrictions. These policies have been sponsored by the World Health Organization in the hope of decreasing the adverse effects of alcohol.

Alcohol-related DALYs per 100 of population by alcohol policy index score. Country codes correspond to international vehicle stickers

Alcohol-related DALYs per 100 of population by alcohol policy index score. Country codes correspond to international vehicle stickers

The new findings strongly disagree with the prevailing theory, called the total consumption model. The latter forecasts that alcohol-related harms will diminish if the population total alcohol consumption is reduced by increasing the price, lessening availability and controlling the marketing of the beverages. A stern alcohol policy is defined by a high price level and restricted availability of alcohol beverages. Current policy recommendations are often based on this model. Some evidence suggests that stern alcohol policies really can decrease the total alcohol consumption. The weakness of the total consumption model is, however, that all drinkers are assumed to decrease both their consumption and health risks approximately in proportion of the amount of alcohol consumed. Drinking patterns are ignored. In fact, when policy becomes more stringent moderate drinkers are likely to decrease their drinking. Heavy drinkers and alcoholics are less sensitive to severe control measures. Therefore, strict policies may bring about fewer health benefits to moderate drinkers but make little change in the health of alcoholics. In the worst case, stringent alcohol policies can lead to counterproductive effects, such as illicit trade and moon-shining as well as consuming industrial alcohol products that have high toxicity, thus killing more alcoholics. More efforts is called for to reduce alcohol dependence and promote moderate drinking.

Kari Poikolainen

Department of Public Health, University of Helsinki, Finland



The Weakness of Stern Alcohol Control Policies.
Poikolainen K
Alcohol Alcohol. 2015 Jul 6


4 Responses to Stern alcohol policies fail to improve public health

  1. Pia Mäkelä says:

    The study method is very weak. If it wasn’t, Kari Poikolainen could next use similar country comparison to ‘show’ that medications for malaria are a complete waste: the countries where they are used the most do not have a lower level of malaria! If you want to study the effect of a policy, you have to try to exclude the effect of confounding factors, which it the case of alcohol are numerous (drinking patterns, norms relating to behavior while drunk, treatment systems etc. etc.). Proper research designs compare changes in time, when policies change – then the ‘other factors’ usually remain relatively unchanged. Such policy impact studies indicate that e.g. the price of alcohol and the availability of alcohol have an impact not only on total consumption of alcohol but also on the rate of harms from alcohol.

    Pia Mäkelä
    National Institute for Health and Welfare, Helsinki, Finland

  2. Kari Poikolainen says:

    The malaria analogy is false. Malaria medications are given only to those in need, alcohol control policies embrace all members of a national population. It is important to understand that the total consumption model is a national population level theory. The model postulates that increasing control, especially higher price and diminished availability of alcoholic beverages, would improve public health because all drinkers would decrease their consumption, and total consumption determines the alcohol-related ill-health. This is the theory my study was designed to evaluate. No effect was found. If drinking patterns and norms influence alcohol-related ill-health (and I think that they do), this is further evidence against the total consumption model. The model is clearly inadequate, partly because it does not say anything about drinking patterns. Finally, the results from policy impact studies are conflicting. Some studies show positive effects. But some, including those by Dr. Mäkelä, show none or negative effects. None of these policy impact studies have focused on the compehesive indicator of alcohol-related ill-health, disability-adjusted life years, in contrast to my study.

    Kari Poikolainen

  3. Pia Mäkelä says:

    Dr. Poikolainen writes ”This is the theory my study was designed to evaluate” – unfortunately the design was poor, because there are a number of other factors in addition to alcohol policies which systematically vary by the countries in question, and this confounding is ignored by the design. An association is different from causation (that drownings and ice-cream sales are temporally associated is due to summer).

    Poikolainen claims that the acknowledgment of the importance of drinking patterns would be ’further evidence against the total consumption model’. Here, he is clearly mistaken.

    Overall, I do not recollect any key voices in the scientific literature that would have doubted the role of drinking patterns. Quite the contrary – this is something ‘everybody agrees on’. Alcohol-related violence is a good case for illustration. If it was possible to change the drinking patterns and norms regarding drunken comportment of Finns to match those of Italians (drinking only rarely beyond moderation, and a strong norm not to openly show signs of intoxication), it would be self-evident that the rate of alcohol-related violence in Finland would be much lower. Actually, this idea was the basis for the main alcohol policy line in Finland in the 1960’s favouring mild alcoholic drinks over spirits, but it was noticed that drinking patterns are relatively resistant to change: changes in drinking culture happen rather in generations than within a few years. The outcome of the policy was, due to increasing consumption of mild alcoholic beverages, also increased total consumption and increased rate of harms. I would welcome Dr. Poikolainen to present evidence-based interventions that have changed drinking patterns or culture at the population level (rather than e.g. an individual school). These have been much sought for not only in all Nordic countries but in all countries where ‘bad drinking habits’ have been a problem historically – and which are the root cause behind both the relatively high rate of problems and for the more stern alcohol policies.

    It is indeed a well-established fact that in addition to drinking patterns and culture, also the level of total alcohol consumption affects the rate of harms, e.g. alcohol-related violence. This can be seen both in descriptive trend analyses (e.g. Finland from 1960’s to 2000’s) and in scientific time series analyses (a stronger connection in Nordic countries, but a connection also in southern European countries). However, on the basis of this fact (or on the basis of the ‘total consumption model’), nobody would postulate that when comparing Italy and Finland (or a greater number of countries, like in Poikolainen’s analysis), the difference in the rate of alcohol-related violence should directly mirror the difference in total consumption (because other factors are important, too!), and therefore also the type of analysis presented by Poikolainen has very little evidence value for whether given alcohol policies are effective in reducing harms. There are numerous important factors, but some are more easily changed than others. And the existing literature suggests that it is more easy to change the level of total consumption (by e.g. taxation – this has been shown in several meta-analyses) than e.g. the drinking patterns of a country.

  4. Kari Poikolainen says:

    I try to make my response as clear and simple as possible. The total consumption theory claims that stern alcohol policies decrease total consumption, and that total consumption determines alcohol-related ill-health in a population. All these three variables were included in my test of the theory. No association was found between stern alcohol policy and alcohol-related ill-health. Thus, the total consumption theory is incorrect. Drinking patterns are not a factor in the total consumption theory and thus need not to be included in the test. Counfounding by drinking patterns may be important in some other theoretical models but this is a completely different question. For example, in some data-sets, total consumption correlates with some outcomes related to alcohol. This is likely to be due to confounding – alcoholic drinking patterns (alcoholism) influence both total consumption and ill-health, producing a spurious association between the latter two. I do think that drinking patterns are important, as I have said earlier. That total consumption may be easy to change is not a sufficient argument for policy if no reduction in ill-health is gained. The argument just falsely creates the illusion that something good can be done. I rest my case.

Leave a Reply