Friday, November 21, 2008

PMean - STaTS reborn?

I have referred to the STaTS pages in the past for statistical references. Sadly, Children Mercy hospital has temporarily taken down the pages, but Steve Simon has resurrected some of his content (and hopes to be able to get all of his content back) and more at pmean.

Thursday, November 13, 2008

Autism and rainfall -- a sadly serious case of lying with statistics

So when I heard of the autism and rainfall study a few days ago, I thought it was a joke out of The Onion or similar publication. I realized today that it was an apparently serious study. Orac dissected the statistical details of the study, so I won't repeat that effort. But I wonder what the next correlate will be. Even more of a stretch is to try to pin this on mercury poisoning by claiming that rain is washing all the mercury out of the sky, with no mention of the seriously circuitous route the mercury would have to take to get from the sky to some small child's brain.

While we don't understand the mechanism of autism nor the relationship between heredity and environment in autism etiology, and while it's certainly a good idea to reduce environmental mercury in every reasonable way we can, suffice it to say that nothing good comes from such a missapplication of scientific and statistical principles to these important issues.

Wednesday, November 12, 2008

Fearmongering in the natural medicine blogosphere

If you peruse my personal blog, you'll probably find that I have a fairly open minded view of alternative medicine. A couple of times, I've taken MD bloggers over at scienceblogs to task over their applications of logical fallacies to arguments against alternative medicines. Unfortunately, logical fallacies, fearmongering, and conspiracy theory are too common in the natural medicines blogosphere. One site has commented about the recent JUPITER study (where rosuvastatin/Crestor(r) was shown to cut cardiovascular events even in people with normal LDL/HDL cholesterol) that (somebody?) was going to force everybody to take statins whether they are healthy are not. This commentary serves nobody, including the consumers of natural or alternative medicine modalities, and is so ridiculous that a moment's thought ought to dismiss the whole commentary.

What questions are raised by the study concern whether the cholesterol levels are enough, or whether inflammation (as measured by C-reactive protein) ought to be considered as well when assessing the risk of heart disease. Another question is whether this effect is common to all statins, or shows up in particular statins. (Could generic simvastatin create the same effect.) A third question is whether the cost is worth the benefit, as the NNT for 1.9 years of treatment (to prevent a cardiac event) is on the order of 120 people. In short, more questions are raised by the JUPITER study than are answered, and public policy toward the consumption of statins is unlikely to change based on this one study.

Update: a newly released study about Lipitor suggests kidney function may be involved, so far at the "association," but not "causation," level.

Edits: some grammar mistakes and clarifications.

Sunday, November 9, 2008

Study: Cholesterol drugs could help those with healthy levels - CNN.com

Study: Cholesterol drugs could help those with healthy levels - CNN.com

Sounds like the primary mechanism of action in statins isn't lowering cholesterol, but rather reducing C-reactive protein or other inflammation intermediates.

Cholesterol -- total, LDL, and HDL -- are imperfect biomarkers. I think the relationship between heart disease and cholesterol deserves a closer look.