Another Nail in the Coffin of Correlation and Causation

Surprise: Zetia doesn't reduce heart attacks and strokes, even though it lowers cholesterol. The whole "cholesterol hypothesis" turns out to be exactly that: merely a hypothesis that now appears on its way into the stack labeled "Disproved."

It turns out that statin drugs like Zocor (simvastatin) and Lipitor (atorvastatin), the first drugs that meaningfully reduced cholesterol with a reasonable side effect profile (a patient asked me about cholestyramine last week; I asked if she really wanted to eat four packets of sand a day) can lower cholesterol significantly while also markedly reducing the risk of heart attacks and strokes. When I began reading evidence that it also had an impact on reducing the risk of Alzeimer's dementia, though, I said to myself, "There's probably more going on here than just cholesterol lowering."

Turns out to be exactly right.

Statins lower cholesterol. Statins decrease the risk of cardiovascular disease. Both of these are observations. Making the jump to "Lowering cholesterol lowers the risk of cardiovascular disease" turns out to be an error of confusing correlation with causation. The rooster crows and the sun comes up. Let's make even more noise; oh wait: it's midnight and the sun didn't come up? Something wrong with that formulation.

Now, of course, aren't all those "Cholesterol goals"' in the P4P schemes rendered meaningless? If I get someone's moderately elevated cholesterol down to "goal" with Zetia instead of a statin, have I really done them any good? Apparently not, but I can still get rewarded for it. If this doesn't highlight the perils of P4P, I don't know what does.

(I expected Orac to weigh in first on correlation-causation angle, though Dr. Centor has beaten him (and me) to it. On the other hand, I thought the good Dr. Centor would be the one jumping on the "Quality" issues. Ah well; the unexpected is what makes life interesting.)