This paper accumulates evidence from several studies and performs very complex statistical models, but does not lead to anything clear to say about the increase (or decrease) in the risk of developing health problems associated with different levels of alcohol consumption in different age groups. In the paper they only consider relative risks, thus failing to comply with The Lancet's own guidelines, which state that absolute risks should always be reported, which are essential to understand the real magnitude of a risk (in this case the risk of developing different health problems associated with alcohol consumption).

When covering this information, one should be wary of expressions such as "alcohol consumption is a risk for young people" or "it may be of benefit to some adults". There are two problems with these kinds of statements. The first is that they use causal language ("is a risk" and "may have") rather than correlational ("is associated with increased risk" and "may have"), which sends the wrong message about what kind of evidence there is. 

Second, and more importantly, the paper does not include the *absolute risks of developing alcohol-associated health problems in young and older adults, so there is no way to substantiate and explain those claims.  

Furthermore, they make recommendations based on the analysis, again ignoring the principle that policies should always be based on absolute risks.

*Absolute risk in people not exposed to the risk factor (or baseline risk): proportion of non-drinkers expected to develop the health problems studied.