The message that low amounts of alcohol consumption can have beneficial effects on health status has been built on weak scientific evidence. This article addresses the multiple biases of observational cohort studies in the association between alcohol consumption and mortality, highlighting the use of non-drinkers as a reference category, which usually includes occasional drinkers and ex-drinkers.

It incorporates, with respect to previous works, a classification according to the control or not of multiple biases, differentiating according to the adjustment of each one of them, and generally separating them into low and high quality studies. The conclusion, following the review of 107 longitudinal studies, is the enormous variation in the results underpinning this message, noting that when selecting young cohorts and separating ex-drinkers and occasional drinkers from abstainers, the risk of mortality in consumers of low amounts of alcohol was very similar.

The methodological approach of the study is appropriate to its objectives. Among its strengths is that it is a recent systematic review of published cohort studies. Among its limitations, it might have been interesting to differentiate an analysis group for population-based cohort studies, which have less selection bias than cohorts based on convenience sampling.