Drinking moderate amounts of alcohol may not have the health benefits of long life that have been previously asserted, due to flaws in research methods.

The latest findings should caution researchers and the public to reconsider the existing evidence and the popular notion that moderate drinking is good for general health.

Several previous studies have demonstrated that drinking moderate amounts of alcohol seems to reduce mortality.

However, the authors point out that those studies have a methodological flaw: they generally compare drinkers with nondrinkers, but fail to distinguish between those who never drank alcohol and those who quit drinking for health reasons.

Lumping nondrinkers with ill health into the group that abstains from alcohol entirely may produce findings of an elevated mortality risk seemingly associated with not drinking.

The investigators questioned more than 22,000 men aged 40 to 69 in northern Japan on their drinking frequency in 1990, then tracked the men until 1997.

By separating the two groups of nondrinkers, investigators found that ex-drinkers did have a higher risk of mortality than nondrinkers and that moderate drinkers did not actually have decreased mortality when compared with those who had never been drinkers.

The result shows that, if an inadequate comparison group (never drinkers combined with ex-drinkers) is used, the protective effect of moderate drinking is seriously exaggerated. Thus, moderate drinkers spuriously showed a decreased risk when compared with nondrinkers, which erroneously included both ex-drinkers and never drinkers.

However, the protective affect of moderate drinking on reducing coronary heart disease has been well-established and was also noted in the study. But because the rate of heart disease in Japan is relatively low, this benefit has less of an effect on reducing overall mortality than it might in the United States, where heart disease is more prevalent.

JAMA September 12, 2001;286:1177-1178