voila la phrase:
" In stratified analyses, disability risk decreased with light/moderate drinking in a dose-dependent fashion in men and women with good or better self-reported health but not in men or women with fair or worse self-reported health."
"dose-dependent" is a technical term in pharmacology (and by extension, in other fields) indicating that the response to a drug varied with the quantity administered. The usual interpretation is that dose (quantité) and response change in the same direction, which may or may not be the case here: "decreased" is what is giving me trouble; I can't tell whether the reference is to quantity of alcohol drunk or to the risk of side-effects when some other drug is given to moderate drinkers.
In my attempt I made an analogy to the English " X varies as function of Y" and a guess that "dosage" was the technical equivalent of "dose", hence "en fonction de dosage". I'll give it another try if you clarify my points above. If my answer already gives you all you need, I would still like more context, to satisfy my own curiosity.