One in ten heart deaths in middle-aged men 'prevented' by alcohol, on paper only

Whether moderate drinking protects the heart is one of the longest running arguments in public health. Epidemiological studies have repeatedly found that moderate drinkers have lower coronary heart disease mortality than abstainers, producing a J-shaped risk curve in which risk falls at light intake and climbs again at heavy intake. A relative risk, though, tells a country very little about scale: how many actual deaths would such an effect involve, and in whom? These researchers set out to answer that for Finland, adapting standard methods to estimate how many coronary deaths alcohol 'prevents' and how many it 'causes' each year.

The team adjusted established methods for calculating attributable fractions so they could handle a J-shaped risk function, where both abstention and heavy drinking sit above the lowest-risk point. Point estimates of relative risk were taken from a meta-analysis and pooled by fitting a nonparametric cubic smoothing spline. The distribution of alcohol consumption in Finland was estimated from survey data covering 4,818 people, of whom 2,488 were women. They then estimated how various assumed changes in the drinking distribution would alter coronary mortality, presenting the most detailed analyses for men aged 30 to 69 and summarised results for men and women aged 30 to 79.

Among men aged 30 to 69, the beneficial effects of light to moderate drinking were estimated to 'prevent' some 400 coronary heart disease deaths each year, corresponding to 12 to 14% of the observed deaths. By contrast, only around 20 coronary deaths were 'caused' by consumption exceeding the estimated optimum level. Among men aged 70 to 79, roughly 200 deaths were 'prevented', and among women aged 30 to 79 roughly 100, with only a few deaths 'caused' by alcohol in those groups. The authors concluded that about one-tenth of coronary deaths among middle-aged Finnish men were 'prevented' by alcohol, with a considerably smaller relative effect in older men and in all women.

The quote marks around 'prevented' and 'caused' are the authors' own, and they carry the weight of the paper: these are not observed deaths but counts that would follow if the J-shaped association between drinking and coronary mortality were fully causal. The study extends the epidemiology by translating relative risks into population numbers, which is what public health debate actually needs. What it cannot do is settle whether alcohol itself does the protecting, since moderate drinkers may differ from abstainers in diet, income, health and much else. Whether the association survives stricter methods remains the open question it leaves behind.

If you have heard that moderate drinking is good for the heart, this study shows where such figures come from: statistical models applied to survey data and pooled relative risks, not trials in which anyone's drinking was changed. Even taken entirely at face value, the estimated benefit was concentrated in middle-aged men and was considerably smaller for older men and for women, so the headline version does not apply equally to everyone. It is a reason to treat heart-health claims about alcohol as unresolved associations rather than established fact, and to notice when the researchers themselves reach for quote marks.