So About Those Test Results…
It’s been a good several weeks since Selena Roberts’ and David Epstein’s article The Case Against Lance Armstrong was published by Sports Illustrated, in both online and print forms. For those who aren’t committed fans of cycling, their article lays out in great detail a saga of doping, or alleged doping, by the 800-pound gorilla of American cycling. Well, not that Lance actually weighs 800 pounds. But he does wield quite a bit of power both in North American cycling circles and at the international level. And I’m not talking about the kind of power that one could measure with a PowerTap or similar type of training device, either.
Lance Armstrong certainly doesn’t need any help defending himself against these most recent allegations. And I’m not about to take a position on his guilt or innocence. But it’s worth looking at the Sports Illustrated article, because at least part of the article illustrates a problem in how certain types of articles are reported.
Roberts and Epstein may be close to whatever the truth of Armstrong’s athletic feats happens to be, at least when it comes to the big picture. So many people have told similar stories about the world-famous cancer survivor that there must be some grain of truth to what these people are saying. That, or there’s a massive anti-Lance conspiracy designed to take the 7-time Tour champion down a few notches.
I have a hard time believing that so many people would conspire in such a manner, but stranger things have happened. Especially in the world of professional cycling. Even though the article may come close to the heart Armstrong’s seemingly legendary athletic feats, there is at least one part of the article where the reporters wade into the weeds. That’s the part where they present what seems like compelling scientific evidence that yes, Virginia, our esteemed hero of the Grande Boucle committed the heinous act of doping.
What I’m referring to is this:
From 1990 to 2000, Armstrong was tested more than two dozen times by Catlin’s UCLA lab, according to Catlin’s estimate. In May 1999, USA Cycling sent a formal request to Catlin for past test results—specifically, testosterone-epitestosterone ratios—for a cyclist identified only by his drug-testing code numbers. A source with knowledge of the request says that the cyclist was Lance Armstrong. In a letter dated June 4, 1999, Catlin responded that the lab couldn’t recover a total of five of the cyclist’s test results from 1990, 1992 and 1993, adding, “The likelihood that we will be able to recover these old files is low.” The letter went on to detail the cyclist’s testosterone-epitestosterone results from 1991 to 1998, with one missing season: 1997, the only year during that span in which Armstrong didn’t compete. Three results stand out: a 9.0-to-1 ratio from a sample collected on June 23, 1993; a 7.6-to-1 from July 7, 1994; and a 6.5-to-1 from June 4, 1996. [emphasis added] Most people have a ratio of 1-to-1. [ed. Not exactly true. Most white males of western European background have this ratio. Men -- and women -- from other parts of the world can exhibit higher natural ratios than what is suggested here by the writers.] Prior to 2005, any ratio above 6.0-to-1 was considered abnormally high and evidence of doping; in 2005 that ratio was lowered to 4.0-to-1.
What the casual fan needs to know about the testosterone test is this: When an athlete’s A sample is tested, the lab actually subjects at least three portions of that sample to the T/E ratio test. A sample only comes up positive if three results occur in the same general ballpark, give or take a test’s measurement uncertainty (which we’ll come back to shortly).
[Catlin] said that because he tested by code and not by name, he has “no clue which sample belonged to Lance,” but he admits the data are disturbing. He explains that one failed confirmation would be a “once-in-a-blue-moon” occurrence.
Interesting assertion, and Catlin certainly is one who knows from drug testing. But I don’t know that the assertion holds up. Revisiting a the Landis case, his original A sample T/E ratios were all over the place. Sure, you probably heard about the 11-to-1 ratio that came up in the A sample, but there was also a reading above 14-to-1 and several in the range of about 5-to-1 or 6-to-1. (The results on the B sample were more consistent, however, but that’s a different story.)
The point is, fluctuations occur. How often they occur is something we don’t get a glimpse of, because when results are reported, we almost always see a single ratio, rather than the set of ratios reported.
As for the three high T/E ratio results detailed in the letter, [Catlin] says, “that’s very strange.” When [the] letter was read to [Andreas] Breidbach recently, he too expressed concern, saying, “Wow, that should not happen. If you find a nine and can’t confirm, then something is very wrong with your screening test.”
There’s an interesting point. How good is the screening test that’s performed? Unfortunately, the article doesn’t really tell us. It could well be that Breidbach is inadvertently letting us all in on a big secret: the screening test is not very good. But an actual admission that the testing isn’t good would make the testers look, well, not very competent. And that would cast the whole anti-doping system in a rather bad light.
The Roberts/Epstein article gives us single results for the tests in question, and those results paint a certain picture. But the ratios, themselves, are meaningless without context. The ratios give an idea of the magnitude of the difference between the concentration of testosterone and epitestosterone in a sample. What they don’t tell us is the raw numbers. Were the testosterone numbers elevated? Or were the epitestosterone numbers depressed? That has a bearing on whether you can even begin to suggest that an athlete has been doping, or whether something else is going on in the person’s body.
The numbers present, though, lead to the conclusion that on those occasions, Armstrong was up to his eyeballs in testosterone. Maybe he was, maybe he wasn’t. Since other tests of the same sample didn’t provide enough certainty about the results, the labs involved didn’t declare any adverse findings.
Of the three results the article discusses, one very likely falls into the gray area of measurement uncertainty. Measurement uncertainty is, when you take a measurement of something, how close is the measured value to the real value. It’s expressed in terms of plus or minus X, and depending on how the measurement uncertainty was determined, there is a certain probability that the true value lies within the range of values specified. Another term for this is confidence interval, from the world of statistics.
In the case of the lab testing Floyd Landis’ infamous sample in 2006, the measurement uncertainty was ± 0.8. While the Sports Illustrated article doesn’t tell us the measurement uncertainty for any of the three results attributed to Lance Armstrong’s samples, we can make a somewhat educated guess that the 6.5-to-1 result was probably within the bounds for the top end of the allowable T/E ratio back when it was conducted.
From that point of view, the explanation of why a 6.5-to-1 ratio wouldn’t count as a positive is pretty straightforward. The other two, we don’t have enough information about to draw a conclusion.
By presenting the T/E ratios, the writers give the impression that Armstrong tested positive for testosterone on three occasions. But, according to the rules in place and the way the testing is performed, he didn’t test positive. The reality is that these particular results don’t prove much of anything, and are a weak argument at best that Armstrong was not riding the straight and narrow path to cycling glory.
The more compelling part of the story is that a number of people seem to paint a consistent picture — and not a pretty picture — of Big Tex. While that all falls into the realm of he said/she said, it’s the consistent appearance of smoke from a number of quarters that leads one to suspect that there may be a fire going on somewhere. How big a fire is yet to be seen.
And speaking of bombshell articles…
A week ago, Paul Kimmage published an interview of Floyd Landis in The Sunday Times (London, that is). And then he put a transcript of the interview out at NYVelocity.com. (Hat tip to AustinCyclist for the links.) The transcript clocks in at about 31,000 words and can take a good several hours to read. It gives further context to Landis’ comments in The Sunday Times, and adds additional information, too. Kimmage does a very good job interviewing his subject (he spent seven hours with Landis in November conducting the interview), but like all interviews, some responses beg for follow-up questions.
Sometimes, which questions those happen to be depend on what we bring to reading an interview, and what knowledge we have of the subject already. Even reporters can go back to their own work and find that they missed a follow-up question or two. Still, despite its imperfections, the lengthy transcript of the Kimmage/Landis conversation is one of the best interviews I’ve read in a long while, and certainly one of the best interviews of a professional cyclist that I’ve seen.
What raised my eyebrows were the sections that discuss the — for lack of a better phrase — endemic corruption in the halls of the UCI’s headquarters. It seems as though the people heading the organization view their power as a license to shake down anyone and everyone they can. Just as some commenters here have suggested over the years. And, the leadership of the UCI basically appears to have the attitude, “it’s our way or the highway.” AustinCyclist also provides this link to Cyclocosm.com, where documents backing up Landis’ story about his difficulties collecting from bank guarantees administered by the UCI can be found. The guarantees are required of all UCI-licensed teams, to ensure that riders on teams that collapse will still get paid.
Meanwhile, more stories are surfacing suggesting that former UCI chief Hein Verbruggen was/is a corrupt old sod, and that he ran the UCI more as his personal fiefdom than as a professional organization. Predictably, the UCI’s current head has said, basically, “No corruption here. Move along.” Or, to get the quote right:
“It’s impossible to be corrupt in the way we’re being accused, in terms of bribery and assisting riders cover up doping positives,” McQuaid told Cyclingnews.
“We’d welcome any investigation into the UCI. There has never been corruption in the UCI.”
Be careful what you wish for, Mr. McQuaid. If an investigation did happen, you might be proved a liar. Investigators and reporters have a way of digging up awfully embarrassing information. Just ask former Senator Gary Hart.