I would like to start by complimenting the authors of this paper . It provides an elegant anatomy of drinking patterns and related problems in a developing country with both high rates of abstention and high rates of alcohol use disorders. They have added significantly to the literature on the topic widely referred to as the ‘prevention paradox’ by confirming its relevance in such a population and as well for problems of dependence in addition to social harms. They further confirm the essentially non-paradoxical nature of the wide distribution of alcohol problems across many people who drink, as opposed to these being confined to a small group of extremely heavy drinkers. Much of the ‘paradox’ is explicable in terms of their being a great number of people who occasionally drink to excess but whose overall volume of drinking is within the low-risk drinking guidelines, or at least within the moderate range of drinking by overall volume.
In his much-debated and seminal contribution, Kreitman  first used the prevention paradox as an argument for exhorting all drinkers to drink less as a more appropriate public health message than one which espoused specific low-risk drinking levels. As was subsequently pointed out, Kreitman classified drinkers' risk of harm only in terms of the overall volume of consumption over a week or year, and when drinking pattern was factored in this ‘first-order’ paradox vanished [3–5]. In the present study, a clear majority (75%) of drinkers with social problems were low or moderate drinkers. This number dropped to only 20% when all ‘binge drinkers’ were excluded. For signs of dependence, these figures were 76% and 19%, respectively. Rather than paradoxical, these data indicate the mildly surprising finding that one-fifth of low-volume drinkers who never exceed three drinks (women) or four drinks (men) in 1 day still report some social problems and signs of dependence. Some of this may be explicable in terms of respondents on very low incomes experiencing financial problems when they drink regularly, even at low levels; some of it by underreporting of actual consumption  and some by possible difficulties for some respondents grasping the graduated-frequency questions . Furthermore, it may be possible that questions about ‘problems caused by your drinking’ can be understood in more social than personal terms, i.e. respondents believe that simply participating in a social event where there was drinking caused them to experience a problem. Going to a bar can present a risk of being involved in a violent or impaired driving incident even if you have only one drink. However, even factoring-in all these and other possible explanations these data, along with data on elevated cancer risks at low levels of consumption , remind us that there is no such thing as completely risk-free drinking. They should also remind us that people who choose to drink are best served by public health messages which encourage them to restrict both volume and quantity per occasion, as opposed to a rather vague and ambiguous encouragement to ‘drink less’ at whatever level they happen to be currently drinking.
The prevention paradox has also been employed in the promotion of public health policies to restrict the physical and economic availability of alcohol . A popular counter-argument to such policies is that they punish the many for the sins of the few. The prevention paradox, however, suggests that, on the final day of reckoning, the many commit more sins collectively than the few, and should therefore not be spared the punishment of more expensive and less accessible alcohol. I personally believe the reframing of the paradox in terms of there being many occasional heavy drinkers, even among otherwise low-volume drinkers, to be slightly more helpful in advancing effective alcohol policy than asking people to believe that those who drink a little cause most of the problems. However, this is really simply a nuance, and whatever precise argument is used advocacy of effective broad public health policies such as minimum alcohol pricing and reduced density of liquor outlets is always an uphill battle, especially in cultures in which so many people enjoy and embrace the stuff. Underscoring this point, Macdonald et al.  reported recently that public support for effective alcohol policies in Canada was related inversely to provincial per capita alcohol consumption. Put another way, provinces with most need for effective alcohol policy were also those where it was probably hardest to implement. To the extent that research has a role to play in changing hearts and minds in this difficult policy arena, Caetano et al.'s results help to illustrate that the cosy assumption that most alcohol consumption is without risk does not stand up. In Canada, we tried to put this message into one simple figure: 61.5% of all alcohol consumption was inconsistent with widely used low-risk drinking guidelines .