« Hillary care will cost $110 billion and will mandate coverage | Main | Pharm quick links »

September 17, 2007

The flip flop rhythm of epidemiology

I think you can subtitle this post "The weakness of observational studies versus the randomized clinical trial (RCT)."

Do We Really Know What Makes Us Healthy? is the cover story by Gary Taubes published in yesterday's NY Times' magazine about the utility of epidemiology in providing the public with guidelines for maintaining health.

He starts off by talking about the hormone replacement therapy (HRT) controversy where HRT was supposed to prevent heart disease supported by observational studies. Then in the 90s, HRT was generally recommended for all women, but it was only was randomized clinical trials, such as WHI and HERS, looking at secondary and primary prevention respectively, that showed that HRT is not heart protective.

The point is women who take HRT are a special population, with unquantifiable characteristics, that tend to bias an observational study. You never know what you can miss, Taubes says. This confounding variables are better managed with a RCT. It also what's needed in attempting to prove causality.

I try to follow this up with further discussion.

UPDATE: The WSJ has an article published today Report Sheds Light on Hormone Therapy which suggests a reason why HRT might not have the cardioprotective effect for some women.

Researchers at University of Texas Southwestern Medical Center in Dallas have discovered that a compound related to cholesterol can block the hormone estrogen from performing functions in blood vessels that keep them healthy and free of disease that can lead to heart attacks.

This team found a molecule called 27-hydroxycholesterol, or 27HC, which is a byproduct created as the body processes cholesterol. The found in mice that 27HC and estrogen target the same receptors in blood vessels.

Following this hypothesis, premenopausal women are protected by the high levels of estrogen and low levels of 27HC, but during menopause, therapeutic estrogen is not effective because it can't compete with 27HC.

The researchers claim that this could account for the results found in the above-mentioned WHI and HERS studies.


TrackBack URL for this entry:

Listed below are links to weblogs that reference The flip flop rhythm of epidemiology:


blog comments powered by Disqus