Re: Gmel et al (2000) Revising the preventive paradox: the Swiss case.
Once more a general population survey has been used to replicate Kreitman's preventive paradox that drinkers with a moderate volume of drinking collectively contribute more social problems from drinking than do high volume drinkers (Kreitman, 1986; Gmel et al, 2000; Skšg, 1999, Stockwell et al, 1996). Once more, when drinkers are instead classified according to whether they ever drink a significant amount on one occasion (here defined as at least 50g), the 'paradox' vanishes: the majority of social problems are then reported by people who are, at least occasionally, 50g+ drinkers.
Kreitman's finding has become something of an icon in alcohol research and will doubtless always be referred to as the "preventive paradox" though I wholeheartedly agree with Gmel et al (Gmel et al, 2000) that there is nothing at all paradoxical about the observed pattern of results. Gmel et al also attempted to examine Skog's suggestion that a 'second order paradox' lurked beneath these data i.e. that moderate volume drinkers collectively contribute more episodes of intoxication than do high volume drinkers (Skog, 1999). They confirmed this to be the case but, again, rightly suggest this is not a very paradoxical finding either as occasional 'bingeing' is so common. (In passing, I suggest that Gmel et al have not been able to yet lay to rest the possibility that the problems experienced by the high-volume binge-drinkers are nonetheless the most severe and costly Ð measures of individual problem severity and frequency are needed to settle that score).
Regardless of whether a second, third or even fourth order paradox can be uncovered, all parties seem in agreement on these basic points:
(i) 'binge' drinking, almost however defined, is the main predictor of social or acute harms from drinking;
(ii) 'binge' drinking in the populations studied is sufficiently common as to justify general population approaches to prevention.
In a recent analysis of types of alcohol-caused deaths in Australia, Chikritzhs et al (2000) found that acute or mixed acute, chronic causes accounted for 58% of alcohol-caused deaths and 66% of potential years of life lost. Given the high prevalence of occasional high risk drinking these findings are arguably a stronger justification for population-based approaches to prevention than is an argument which, even to a policy adviser, must seem paradoxical.