Interesting article about statistics & testosterone testing

[Keith A]

I ran across this article "Floyd Landis and Doping: Do Labs and/or the Media Understand Statistics? (http://www.knowledgeproblem.com/archives/001707.html)" that help shed some light on the whole testosterone/epitestosterone issue and thought you guys/gals would be interested as well. The article references another article by the American Statistical Assocation on false positives in testosterone testing (http://www.amstat.org/publications/chance/172.berry.pdf) which is a somewhat lengthy and detail report on the problem (which I haven't read yet).

[keno]

Great stuff, Keith. Thanks.

keno

[11.4]

One of the basic problems with testing for performance enhancing drugs is that the compounds they are looking for are identical to or similar to those found naturally in the body. So there isn't a pure binary yes-it's-there-so-he's-guilty/no-it-isn't-there-so-he's-innocent decision. Instead, sporting authorities have to look at a range of possible data points and decide at what point to constitute a reasonable cutoff. In statistics a 95% confidence level is typically considered quite high (usually sufficient for most scientific research) but it still means that in a 200-man peloton, 10 individuals will be deemed out of bounds. Even at 99% confidence level, which is considered exceedingly high in a research or medical context, two riders are in trouble who simply fell outside the defined standard. So much drug testing relies on such statistical validation, which forces a rider to argue that he/she is simply displaying a minority physiology. Hard to defend and in the current journalistic environment, very unfair.

The ASA article is quite popular and has seen wide distribution. It has a few issues, but the basic case it makes is very sound. Too bad the French and the UCI and WADA don't know how to read.

[keno]

What's even worse is that this is a trivial example of a possibly corrupt statistical conclusion. You don't have to be a Leibniz to see the possibilities here. (I keep thinking that if I were to submit the bare statistics to most people that baseball player A has a .310 batting average and player B has a .375 average, many would conclude, prematurely, that B was a better batter if they hadn't first thought to learn that A had 10x at bats as B, and that A's slugging percentage was considerably higher than B's.)

I wonder what was behind the decision to drop the acceptable ratio from 6:1 to 4:1.

[CalfeeFly]

So much of this ASSUMES that the endocrine system of the body is consistent and easy to test. Well...I spent the last year watching doctors shrug, smile, laugh, joke, frown, get frustrated and you name the emotion as I was tested virtually every week and a wide variety of my scores would go up and down like a yo-yo. They had no idea why and now with "normal" results two tests in a row after over a year there is only a hypotesis.

To give you an idea of another hormone since I'm not as sure of what my testosterone was doing...my TSH which measure for the thyroid...would go from being critically low to 5 and 6 times normal in week. Now the kicker is I had my thyroid removed years ago and I take Synthroid. The dose of Synthroid would stay the same but the levels went up and down. This was after 10 years of dead on normal.

My sodium would stay chronically and at times dangerously low. This again made little sense since there was no reason. This from a guy who salts pepperoni pizza...

I could go on through all the blood work but in a nutshell it is not all that exact of a science, normal varies, crap happens and doctors do not know why, and other factors can cause the reactions. In my case, it now appears it could be starting in the pituitary and in turn causing a host of problems. Last kicker now is after four days of testing in the hospital all that is still a mystery.

With this background of mine, I like Floyd hope that he is not “convicted” without full testing and the most complete information possible.