Every single one of those recommendations as of 2007 was the result of correlational epidemiological studies. It cannot be emphasized enough that correlation is not capable of proving causation. Period. All that correlation (or the lack thereof) is capable of doing is disconfirming hypotheses, not confirming them. When it comes to confirming a hypothesis it's randomized, controlled trials or GTFO.