Any time we're calling one weakener "better," we have a problem. There should never be two answers that weaken (unless it's a "weaken except" question).
In this case, as others have said, the answer must be B. This is a classic correlation vs. causation argument, and B addresses the point that drinking coffee may be correlated with other behaviors that lower the risk of Alzheimer's.
Answer choice A might seem relevant, since it brings up other causes, but without further information it does nothing at all to weaken the argument. We'd need to know that some of these other causes are correlated with drinking coffee. If they merely exist independently, then there is still a correlation between coffee drinking and lowered risk that we need to explain. It would be like if every time you ate sushi you got a headache, so you concluded that sushi causes headaches. Knowing that other things (illness, blunt trauma, loud music) can also cause headaches wouldn't matter. Sushi could still cause headaches, too, and we have evidence linking the two. However, if every time you had sushi, you were exposed to other potential causes of headaches (e.g. alcohol, smoke, smooth jazz, whatever), then we'd have reason to question whether sushi was the real cause.