GENERALLY speaking, if a population drinks more, then there are more heavy drinkers, and more harm from alcohol.
Similarly, if a population drinks less, there will be less harm.
But this link now appears to be unravelling.
One of the core assumptions of public health-focused alcohol research has been the overarching link between levels of alcohol consumption in a population and rates of harm.
This has been demonstrated repeatedly, across a range of settings - when per-capita alcohol consumption goes up, rates of alcohol problems (mortality, morbidity and violence, for instance) go up with them.
Recently, these trends have begun to uncouple in a number of places. In Sweden, per-capita consumption of alcohol has fallen in the last five years; while harm rates have remained fairly stable.
In England, harm rates have increased sharply since 2004 despite a steady decline in per-capita consumption levels.
And a similar pattern is emerging here in Australia.
Over the past decade or so, data from the Australian Bureau of Statistics has shown almost no change in the amount of alcohol consumed per person in Australia.
In 2000-01, it was 10.15 litres of pure alcohol, while in 2010-11 (the most recent year for which data is available), it was 9.99 litres.
In contrast, rates of alcohol-related harm are increasing.
Recent studies in Victoria of both adults and young people have found sharp increases in a range of problems from alcohol.
This includes rates of alcohol-related hospitalisations, presentations at emergency departments due to intoxication, late-night assaults, domestic violence involving alcohol and alcohol treatment.
And a national study of alcohol-related harm between 1995 and 2006 found increases in alcohol-related hospitalisations in all states.
But what can explain these diverging trends?
First, there's the possibility that our data systems or coding practices have changed.
While most measures of harm have increased in recent years, rates of alcohol-related mortality (the most reliably recorded) have not.
So we could be seeing a coordinated shift in how hospital workers, paramedics and police treat alcohol data.
This requires a shift in practice across multiple systems (and multiple states).
While possible, this seems unlikely to explain the full extent of the observed trends.
And if we assume the available data reflects real underlying changes, then something more interesting is going on.
Australian alcohol consumption patterns may actually be fragmenting. Consider this simple example: a large number of light or moderate drinkers may have slightly reduced their alcohol consumption, while a smaller group of heavy drinkers increased theirs.
This would lead to relatively steady per-capita consumption, but the potential for increases in alcohol-related harms, mostly experienced by the heavier drinkers.
A recent study of Swedish youth finds some evidence of this kind of polarisation.
Where does this then leave alcohol policy? Public-health oriented alcohol policy has focused on shifting population consumption, through measures such as taxation or physical availability.
But maybe the important question is not what effect taxation or earlier closing hours have on consumption levels, but rather what effect they have directly on rates of harm.
In Victoria, several recent analyses suggest that the vast increase in the number of places to buy alcohol has had little impact on overall consumption.
But it has directly influenced rates of alcohol-related problems.
Perhaps changes to population level alcohol availability particularly affects risky or marginalised drinkers, those likely to experience harm from their drinking. Studies also show heavy drinkers respond to price changes and that increasing alcohol taxes reduces death and injury.
So it may be that population-level policy solutions still make the most sense, even as population-level consumption and harm rates drift apart.
There are still a lot of questions we need research to address: whose drinking is shifting and why?
Are particular policy changes likely to improve or exacerbate the recent harm increases?
Are there specific demographic or sub-cultural groups that research and policy should target? In any case, increases in alcohol-related harm are of grave concern and it's critical that we reverse this trend.