Tuesday, June 26, 2007

The Confusing Thing About Association Studies.


Today is a moderately slow day for genetics in medicine (I can't believe I just said that). But if you want, you can sell your genome to some stranger for 5 grand USD (I can only imagine the identity theft issues) you do it at your own risk. Listen, if you are hard up for cash please do not sell your DNA sample!! Who knows what they will use it for..........Perhaps to frame you for a crime.


But what I want to really talk about is how confusing association studies can be. So lets examine some of these.

  1. Long Term Aspirin use prevents cancer incidence in colon 32%, prostate 19%, and breast cancer 17%* (statistically non-significant). There is some molecular evidence of this in colon cancer. But not the others....... The catch is that you have to use aspirin adult dose for >5 years. Why? Like most association studies.....No one knows. What good is that?

  2. Hormone replacement therapy increases Ovarian cancer incidence This study called the Million Women Study is a large cohort of British women. 948,576 postmenopausal women were assessed for ovarian cancer incidence. Users were 20% more likely to develop Ovarian Cancer. 1 in 5, that seems small, but in a million women (well......just 52k shy) that's alot of cancer!!! Especially such a nasty killer. But here's the kicker....

  3. Oral Contraceptive hormones Reduce Colorectal Cancer risk! Wait a second.....Aren't these female hormones too? This study shows an almost 40% reduced incidence of colon cancer in these women from the Women's Health Study. Perhaps this has to do with dosage? But Who Knows....It's an association study!!!

  4. Smoking Cuts Risk of Parkinson's Disease So that is what the media says about this study. Ok so now you have got me flipping out. No mechanism, No pathogenesis, No explanation.... Smoking kills, but at least it reduces your likelihood of ALSO having Parkinson's. Almost a 40% reduction in the likelihood of having Parkinson's. How? Who Cares....It's an association study! This kills me. The people could have predisposition genes for nicotine addiction/taste/etc which also have some salutatory effects. I do not think that smoking is what saves these patients brain cells!!!! But that's not what the press will tell you.

The Sherpa Says: What is sold as a good piece of science is quite often a piece of something else! Just because it was toiled over and hard work to develop it was done does not make it true, correct or even appropriate. I am here to say.....If it sounds fishy it probably smells fishy too. Throw out association studies until you have a reason for the association!!!


2 comments:

Chris said...

It's not really fair to say that we should throw out association studies. They have a role, and that is to inform future research.

What these studies produce is interesting questions: If asprin reduces colon cancer incidence, how might that work? That thread can then be picked up by some research scientists, who may hypothesize mechanisms and do testing.

The problem, as you point out, is that the press gets wind of these studies and grossly misreports the underlying science. Put the blame where it belongs, on the journalists who fail to do their homework.

Steve Murphy MD said...

Ok,
Fair enough. I blame the press. But I also blame an article if it doesn't have a good discussion that delineates the shortcomings and inability to draw direct conclusions.
-Steve