Sunday, December 28, 2008

The Winnowing: Duckstrap

Duckstrap is a consultant to pharmaceutical companies on setting up statistically useful and regulatory compliant development evaluation. He's looked at the available data along with us and heavily participated in the discussion. He too makes reference to the Berry argument.

[Back to the Introduction]

How does the Landis case look from the vantage of two years’ perspective? Muddy. I submit that nobody knows for sure whether Floyd took testosterone on that fateful day in July except Floyd. Not me. Not the folks from the ex-LNDD, and certainly not USADA or CAS. Or perhaps more pertinently, that nobody (again excepting Floyd) who strenuously purports to know, actually knows what they, in fact, do not know. I believe that CAS verdict was influenced by the politics of anti-doping much more than a strong scientific case; that by the time the case got to that point, the science was almost an afterthought. That is wrong in my view.


Here are the things I am convinced of:

1) The statistics of the anti-doping testing are flawed. This is Don Berry’s argument, and, for me, it is convincing. The essence of it is this: if you believe that a relatively small fraction of riders dope with a particular agent, then the test for that agent needs to be very very specific, with a very low false positive rate. From screening to confirmatory T:E ratio to IRMS, LNDD’s procedures never came close to that standard. The new tests for CERA, where there is a specific molecular signature to the molecule allowing its distinction from endogenous EPO may approach that standard.

2) Overall, LNDD’s assay procedures and methods were inexcusably slipshod. The lack of adherence to procedures for sample custody were symptomatic of a broader scientific sloppiness that saw really bad chromatography in the screening procedure, unidentifiable peaks in the “confirmatory IRMS chromatography”, and finally, actually different chromatography conditions in the IRMS chromatography, so there was no definitive identification of the molecule(s) whose IRMS signature was measured. These are systemic, methodological issues with the assays as they were run in that lab. It does not mean that this tecnology could not produce a definitive answer, but in the hands of these hacks it did not. It doesn’t mean they didn’t get the right answer, but it also means they couldn’t know that they had—hence neither can we.

As for Floyd, whether he did or didn’t, I wish him all the best. If he did it, then he has taken us all for quite the ride, but has also paid a huge price for it. If he didn’t do it, then I believe he has done about all a person could do to exonerate himself, and appears, wisely, to have chosen to move on. It is couragous to fight the fight, but also couragous to put it behind you and get on with the business of living life. I’m glad to have played a small part in the drama, and am even more grateful to TBV for providing a forum for civil and knowledgable discourse of the subject.

Kevin Dykstra, PhD
aka “Duckstrap”


Cub said...

With Lance Armstrong being tested more than once every other week, you have to wonder if he's nearing the point of bucking the odds of a false positive.

TBV's retirement might be shorter than expected.

DBrower said...

Don't bet on a return.