I emailed one of the authors of the study linked in the OP, raising the criticisms I posted upthread. He got back to me very quickly.
I’m not going to post his reply, but below I am posting my reply to him, which will make clear what his points to me were, as well as my response. He also sent me some of his statistical analysis, and a new paper, "
The Epo Fable in Professional Cycling: Facts, Fallacies and Fabrications", a continuation of his argument that the effect of EPO is exaggerated.
There are several arguments made in this new paper: 1) a meta-analysis of 17 studies of EPO shows relatively little effect on performance. That’s based on
another study of his, and I comment on this in my reply, posted below. ; 2) historical studies of speeds in stage races and time trials, including the data in the Texas Sharpshooter paper as well as studies by others referred to in that paper; and 3) LA's time trial speeds, data originally presented in
a paper about two years ago. That is actually the first of three papers in a series comparing LA's performance to historical ones, the others are at:
http://pubs.sciepub.com/ajssm/2/5/4/ajssm-2-5-4.pdf
http://pubs.sciepub.com/ajssm/1/4/3/
Note that the third paper is a study of mountain time trials. I also comment on this in my reply to him, posted below:
Thanks so much for your quick response. I appreciate it.
You say, “First of all, I am not saying 'there was no doping' (after all we do not know that, do we?). I am only saying there is no discontinuity in performance in the epo years. So my conclusion that the effects are probably overrated.”
But what counts as a discontinuity? Your analysis shows that there is a 1.5% increase in speeds in the 1990s
above and beyond the 0.160 kph per year increase observed over the previous sixty years.
You continue, “How do you know this 1.5% increase is due to the effect of doping? I dare you to prove it. I can name a host of alternative factors that may explain this increase which are as just as plausible. Moreover, look at the steady increase in performance over the years. Why should the progress in the epo years be due to the effect of epo? What about the progress in the 40s , 50s, 60s, 70s, 80s, 90s? Are they due to other doping substances?”
I emphasized that I can’t prove this increase resulted from doping. But it’s certainly consistent with it. In fact, your meta-analysis of EPO studies, which I will get to in a moment, shows that this is just the kind of increase we would associate with doping. But the point here is that the burden of proof is on you to account for this increase, since you’re the one making the statement “there is no discontinuity in performance in the EPO years”. There isn't a huge discontinuity, but there does appear to be some.
Also, I think at least some of the increase during the earlier years probably was due to doping. Just because riders didn’t have EPO available doesn’t mean that the substances that were available couldn’t help them. Any drug that increased power, recovery or alertness, as substances of that era generally did, would help a rider ride faster.
With regard to the Perneger years, I can only say again that I think you have flipped the difference (0.89 kph) around. You reduce the difference in speeds between the EPO era and the previous era by subtracting out the steady increase in kph expected over those years. The 1.5% difference is what remains after this correction (and a smaller one for distance). To be consistent, you should do the same in a comparison between the Perneger era and the EPO era. Since the raw data indicate the speeds in those two periods are the same—this is also evident in Fig. 2, and you yourself speak of a leveling off beginning in the EPO era--the corrected data must indicate that speeds were lower in the Perneger era. In other words, if you assume that this steady increase continues throughout the early part of this century, the speeds of the riders benefit from this, which means if they didn’t have this benefit, they would be slower than the speeds of the 1990s, in just the same way that the speeds of the 1990s are concluded to be not much greater than the speeds of the previous era. If there is some other way you can come up with a greater speed in the Perneger era, you have not been clear about it at all in your paper.
Finally, with regard to the effects of EPO. As I’m sure you’re aware, the worst abuse of this drug was prior to the 50% HT rule, when riders could take as much as they wanted, far more than would be used in any academic study. Even after the rule was imposed, in 1997 I believe, there was a period of several more years before a test for EPO was developed, which meant riders with a naturally low HT could still take very large amounts of it. After the test was instituted, riders tended to switch to blood transfusions, with EPO frequently being used to mask the suppression of reticulocytes normally accompanying such a transfusion.
So I think there are reasonable questions about whether most of the available studies gave volunteers dosages of the drug comparable to what riders took at the height of that era. However, even with that limitation, your meta-analysis found an increase of 6-7% in V02max and 7-8% in power. You seem to think this is trivial, as it will result only in a maximum gain in speed of around 1 kph in a TT, roughly 2 – 2.5%. But again, because of the power/speed relationship in TTng, one would not expect a large speed effect. This increase is in fact fairly consistent with the 1.5% increase—to repeat, above and beyond the steady 0.16 kph per year increase—your own analysis found for the 1990s.
Moreover, an increase of 7-8% in power would correspond to the same 7-8% increase in speed in a climb. This could reduce the time of a 40 minute mountain top finish by more than three minutes, which is an enormous benefit. It’s the difference between winning the stage and not even finishing in the top 10, and added up over a Grand Tour, the difference between winning or podiuming, and again, not top 10. And this does not even take into account at least one study that reported that time to exhaustion is increased far more by EPO—up to 50%--than V02max, nor the effect in conserving energy over time in a stage race.
I understand you did analyze a few mountain time trials, but the number is relatively small and includes widely different gradients, even within some ITTs, as well as between different ones. You did develop a climbing index, but as far as I can see, this does not take into account that variation in slope has a major effect on wind resistance, which in turn means that power/speed relationships can be very complicated. About half of the 19 TT you analyzed had a climbing index much smaller than the mean slope, which if I understand correctly means there was a large portion of flat or gentle slope riding (e.g., the first half of so of Armstrong’s mountain TT in 2001 was relatively flat or at most gently sloping). The problem becomes even worse when weather conditions are not taken into account.
So while your analysis is certainly a welcome addition to the debate over the performance effects of doping, and no doubt the best possible using time trials, I think the best measures of climbing remain stage finishes which have a fairly steep and constant gradient. You are surely aware of the enormous differences between Alpe D’Huez times in the 1990s, and those before or after. While some of the differences, particularly with respect to earlier years, can be ascribed to different racing strategies, no rider after the opening years of this century has come close to the best times of the EPO period. To the best of my knowledge, the top 15 or so fastest times all occurred between 1994 and 2004, with no rider since within two minutes of the fastest time.
The final TT in your table in that analysis is Armstrong’s ride up ADH, one of the fastest, and his speed there does exceed his predicted speed, which is based on a steady increase of about 0.2 kph year over the period of analysis. You argue there that the difference is not that large, but if that’s the case, why have all subsequent climbs up ADH been so much slower? There is the fact that they occur at the end of a long stage, whereas Armstrong in 2004 climbed ADH as a TT, but if riders today can’t come close to the ADH times of the EPO era, which were also at the end of a long stage, that surely strongly suggests that they would not come close to Armstrong’s 2004 time trial, either (when he raced ADH three years earlier, at the end of a stage, he was only about 30” slower). Which means, at the very least, that the steady increase in climbing speeds over time has not simply leveled off, but reversed. And it underscores my earlier point that much of the effect of EPO and other forms of blood manipulation are likely to be in greatly enhanced ability to conserve energy, during a stage as well as between stages.