Skip to content

Understanding the Bio-Passport: Horner Values During Vuelta 2013

A large amount of attention has been given to Chris Horner’s release of his biological passport records from 2008-2013.  This is, to my knowledge, the most extensive, longest term set of bio-passport records ever released by a single athlete.  They have generate a lot of interest, with a number of blogs (Velo-clinic, Bike-rack-heads, JournalVelo, and VeloNation) and even Outside Magazine pointing out what they feel are things that Mr. Horner should explain.  Several of these analyses (though not all), share two common themes, themes that I believe are in error.  These problematic themes are:

  1. Many focus largely on the notion that Hemoglobin (Hg) should demonstrate a gradual decline over the duration of a grand tour, and that any “rebound” during the latter half of the tour is indicative of suspicious activity, or at best indicator that a “rebound” or “V-shaped” Hg trend is the characteristic of a small minority.
  2. They fail to look at the interaction between Reticulocyte Percentage (%R) and Hg when raising concerns about low-%R an blood-doping.

The first idea, that a rebound is suspicious or rare, is contradicted by the study of over 200 athletes during the Giro Bio (mentioned in a previous blog) which suggests that a rebound is, in fact, the norm rather than the exception.  The second, looking at low %R in absence of the corresponding Hg values is one that gets to the heart of bio-passport profiling and the OFF-score (a bit on that here).

Interplay Between Reticulocytes and Hemoglobin
I am going to attempt to provide some objective context to the 2nd issue, that of evaluating the Hg and %R in isolation of one another by describing the trends in the two data sets (Hg versus %R) and how they relate to one another.  I am not a hematologist, so I will try to refrain from making a conclusion about this data.  However, as a data analyst, I think that it is possible to put forth an objective set of observations about these two series and how they pertain to what is commonly understood as the basic underlying processes of red blood cell production.  I will refrain from discussing the absolute values of these numbers in context of what is “normal”, except to note that Horner’s OFF-scores in this 5 sample set ranged from 90.91 to 106.53, all within the current WADA guidelines for normal variation.  I will first present my description of individual trends, then present my observations about the interaction of those two trends, finally, I will present my thoughts about how that interaction relates to basic concepts of RBC production in the form of a question to hematologists or blood doping experts.

Descriptions of overall data trends

The following observations are pertaining to Horner’s anti-doping tests around the time of the 2013 Vuelta (depicted in Figure 1 above).  The first test occured 2 days prior to the start of the race, and the subsequent 4 tests occurred between the 5th and 20th days of racing.

  1. Both Reticulocyte % (R%) and Hemoglobin (Hg) values are at their highest (0.85 & 15.2) at the first sampling date, 2 days before the start of the 2013 Vuelta.
  2. Reticulocyte percentage (%R) are at their lowest observed level on day 5 of the race (0.39), by the 9th day of the race they have climbed back up to 0.55 and then stabilize 0.55-0.56 for the remainder of the race.
  3. Hg values are at their lowest observed level on the 9th day of the race at 13.5.
  4. Hg values climb  back up by the 13th day of the race and stabilize somewhat, varying between 14.3 and 14.6 during the last 8 days of the race.

Interactions between the 2 data series

  1. Decreases in %R occur before Decreases in Hg – The lowest %R sample was 3 days before Hg reaches its lowest point.
  2. Increases in %R occur before increases in Hg – %R begins to rebound on the same day as Hg values are at their lowest level, with the largest increase in Hg taking place 4 days after the largest increase in %R.

The Chicken and the Egg
The following are based on a single underlying concept about Red Blood Cell (RBC) production:

Given – It is understood that it takes approximately 4-7 days (as described here & here) for the full life cycle from initial production in the bone marrow to RBC, with the transformation from the stage known as a “reticulocyte” to the mature, fully functioning RBC taking about 2 days (as noted in Guyton’s “Physiology of the Human Body”).

Observations/Questions for Hematologists
If this life-cycle time period is true, then I would suggest that following:

  1. A lag-time between the production of new RBC’s and increasing Hg concentration, as shown in the latter part of the race, seems to conform to the basic principles of RBC production.
  2. Since RBC-enriched blood transfusions suppress reticulocyte production, and Horner’s Vuelta drop in %R occurs along with a drop in Hg, wouldn’t this be counter to the effect that would be expected if Horner were transfusing enriched blood?
  3. Has there been some follow-up research that shows approx. 95% of athletes in Grio-Bio having an Hg rebound in the latter part of a grand tour to be irrelevant or flawed?

None of this is to say anything about whether Mr. Horner was doping during the 2013 Vuelta, or if he has ever dabbled in the Dark Arts.  But it is important, I think, to view data as an integrated whole, as opposed to in isolation, which is the whole point of the Bio-Passport.  Context is key.

Disclaimer: I am not a physician, hematologist, anti-doping expert, or even a veterinarian.  I am just a coach and athlete who has a passion for data analysis, visualization and clean and fair sporting.

Being Normal: Understanding the Bio-Passport

As mentioned previously in “You’re 1 in 1,000 Baby: Understanding the Bio-Passport OFF-Score”, there are a range of OFF-scores that are defined by the WADA BioPassport guidelines that describe the upper and lower values of OFF that are considered “normal”, or more precisely, the range of values that describes 99.9% of the variations in the normal population.  The OFF-score equation and this “normal range” implicitly states that there is a causal relationship between a persons percentage of Reticulocytes (%R) and their amount of Hemoglobin (Hg), in plain language it says “for a given Hg value there is a normal range of %R that would be expected”.

A Normal Probability Space
Figure 1 is a “Probability Space”, it shows a band of %R that are expected to produce a given level of Hg, as defined by the upper and lower bounds of the OFF-score from 85 to 100.  In other words, the area shaded in pink corresponds to the combination of Hg-R% that fall within that 99.9% of the population range, and therefore do NOT produce suspicious OFF scores, or are considered “normal”.  The trend in this “Probability Space” is that it gets wider, and inclines upward as Hg increases.  This trend says that if you have a high Hg, you are likely to have a higher R%, but that other factors can produce higher Hg scores.  Normal range of Hg for adult males is said to be in the range of 14/18 g/dL, however, stbiop_hg-R_rangesudies of athletes and non-athletes places the largest portion of the population in the range of 14.5-16.5 (ref).

One thing that would be important to keep in mind is that t while this may be a good description of the “probability space” of the population of humans as a whole, it might not be necessarily inferred that this also would be the same shape of space that should occur for a given individual.  In other words, an individuals variations might hover about a certain median value for Hg, and their %R values may also have a certain distribution relative to that level of Hg.

Disclaimer: I am not a physician, hematologist, anti-doping expert, or even a veterinarian.  I am just a coach and athlete who has a passion for data analysis, visualization and clean and fair sporting.


Blood Value trends in a Stage Race: Understanding the Bio-Passport


Figure 1: Ivan Basso blood test results showing Hematocrit and Hemoglobin variations over a 21-day stage race from the Giro d’Italia 2010 (source:

Many simplify the notion of stress and blood changes to say that we should expect a downward trend in Hg/Hematocrit during a stage race – you’re body accumulates fatigue/stress and it supresses these values, makes sense right?  People hold up examples of this as indicative of a “clean” performance (see Figure 1 and the linked article discussing Ivan Basso’s 2010 GdI values).

A slight variation on the “downhill ride” shown in the Basso dataset, was that one in Figure 2, which shows Bradley Wiggins blood test values during his 2009 near-podium at the TdF.  Many commentators (not, to my knowledge, the Bio-Passport panel of experts) were skeptical of the “bump” in Wiggins values (see here for one such commentary in NYVelocity).  Some still argued that it was OK, since his overall trend was downward, while others made great issue of the 3rd value (labeled “Sion” in the graph) of 4.

Figure 2: Plot of OFF-score and Hemoglobin values for Bradley Wiggins from the 2009 Tour de France.

“V” is for Victory
Despite the allure of the simple notion, “if they have a downward trend, they are clean, if not, they’re doping”, it seems that the reality may be a bit more complex.  A recent study was released, which tracked values of over 250 cyclists through the “Baby Giro” a 10-day stage race.  This study found a trend that contradicted this simplistic “downhill” trend, stating that

Compared to baseline values, erythrocyte, hemoglobin, hematocrit, MCHC, platelet and reticulocyte counts were all consistently lower at mid-race, but returned to normal by race-end, while leukocytes were increased in the final phase

In other words, it was a “V” shaped curve (see figure 3).  What does this mean?  Well, this is of course open to speculation.  Some might say that since these are cyclists, who are therefore by definition dopers, that it is evidence of how it looks when you are doping.  Others might say, no, this is how the body responds to stress.  Still another point of view might be that this is only a 10-day stage race, so we cannot infer too much from this to a race of 21 stages.  Whatever the perspective, it’s complicated, and it if we accept that the majority of Baby Giro participants were clean, then it paints the “downhill = clean” notion as far too simplistic.  From my point of view, I don’t necessarily know what it means, but I would love to see this data from the TdF, Giro or Vuelta published year in and year out.  This would give yet another lens into comparison amongst individuals and trends within the peloton as a whole.


Figure 3: Plots of median value of Hematocrit (Ht) and Hemoglobin (Hg) for 253 riders from the 2010 and 2012 “Giro Bio”.

Disclaimer: I am not a physician, hematologist, anti-doping expert, or even a veterinarian.  I am just a coach and athlete who has a passion for data analysis, visualization and clean and fair sporting.

You’re 1 in 1,000 Baby: Understanding the Bio-Passport OFF-Score

Trends in Reticulocyte Percent (R%), R-count, and OFF-score for 3 blood samples taken over 10 days of racing for 253 riders participating in the 2010 and 2012 GiroBio,

Figure 1: Trends in Reticulocyte Percent (R%), R-count, and OFF-score for 3 blood samples taken over 10 days of racing for 253 riders participating in the 2010 and 2012 GiroBio,

Disclaimer: I am not a physician, hematologist, anti-doping expert, or even a veterinarian.  I am just a coach and athlete who has a passion for data analysis, visualization and clean and fair sporting.

The “Athlete’s Biological Passport” (ABP) is used in cycling to detect changes in an athlete’s balance of haemotological parameters that might be indicative of doping. The passport if a set of blood values and associated metrics that attempts to describe an athlete’s “normal” or “equilibrium” point, a way of separating natural variations due to genetics, environment and training stimuli from the variations that result from prohibited pharmacological stimuli. It is a complex topic, with the experts in interpreting the passport often expressing their beliefs in terms of probabilities, likelihoods, and ultimately, taking very conservative approaches to the responsibility that they have: their judgement can ultimately be a substitute for a “positive” or “negative” doping charge. One feature of the ABP is what’s known as the “OFF-Score” – a score which has an upper and lower range of “normal” values, in other words, a too high or too low OFF-score is considered to be indicative of doping. Though the experts reiterate that the OFF-score is merely one part of their ABP assessment, perhaps due to it’s wonderful name, and easily interpreted range (85-110 is considered “non-suspicious”), it seems to be a pretty cut and dried matter for arm-chair doping analysts.

OFFs, ONs, and ESAs
At first glance, the name “OFF-score” could give the impression that it was a suspicion score, that is, a number that says “something is off here”, but it’s not. In fact, the OFF-score could better be described is a an indication that a persons red-blood generative process has been switched “OFF”, which is noted as an effect of blood-transfusion, or “coming OFF” a round of EPO or other “erythropoiesis-stimulating agent” (ESAs), or activity (like coming down from altitude). In contrast to the OFF-score, there is also an “ON-score”, which refers to the opposite case, that is a score that indicates the likelihood that a person’s RBC production has been switched “ON”, perhaps as a result of a round of EPO or some other stimulus for the generation of red blood cells. Below is a quote from an article entitled “Blood doping and its detection (Jelkman & Karsten 2011)“:

Some blood parameters, such as the concentration of Epo and reticulocytes (Ret), increase on administration of ESAs (ON-score), whereas they decrease after RBC transfusion or after the cessation of ESA administration (OFF-score)

A Jury of Your Peers
Now, the OFF-score is not a score of cheating, or even of suspicion – at it’s core, it is simply a number that is derived from the ratio of hemoglobin (oxygen carrying red blood cells – Hg) and reticulocytes (%R – baby red blood cells). In general, the expectation is that the more baby-RBC’s you have, the higher your Hb since baby-RBCs grow up to hold a lot of Hg. The formula for OFF-score is as follows;

Eq 1: ([Hb] (g/L) − 60 × √ (reticulocyte percentage)

There are several things to note when considering the meaning of this score:

  • Units Conversion – In the passport files that you will see, Hb is usually in the range of 13-16, whereas in the OFF-score calculation, that value is effectively multiplied by 10, having a range of 130-160.
  • This value is an expression of the underlying notion that in the average or median athletes body more reticulocytes generally means more hemoglobin.  In other words, this is a way of measuring an individuals ratio of Ret/Hg against that of their peers.  The fact that there is a range of “normal” values means that the actual Ret/Hg ratios in human beings is not fixed … some folks RBC’s may “live longer”, resulting in a lower level of Ret to produce a given Hg level, or vice-versa.
  • The most important thing to note, is that a value that is considered “of concern” or “suspicious” is one that lies outside of the 99.9th percentile (as prescribed by the WADA guidelines).

You’re 1 in 1,000 Baby: Cheaters & Outliers
I have often heard suspicious values in a passport as representing “a 1 in a million chance that this athlete is NOT doping”, or claiming that being outside of the prescribed OFF-score range indicates a 99.9% chance that the athlete IS doping – it should be stated unequivocally that this is not correct. An OFF-score that meets the 99.9% threshold (high or low) is literally a 1 in 1,000 score. That is, the statistics suggest that in the whole population of non-doped athletes that have been tested, 1 out of every 1,000 samples is outside of this prescribed range, in other words 0.1%.

Factors Influencing OFF-score
Plugging numbers into Eq1, it can be shown that if you have a high Hb coupled with a low-%R, that yields a high OFF. The idea being that if you have a high number of mature RBC’s without a correspondingly high %R, that you might have just transfused some RBCs, or recently came off of EPO and your body is now shutting down reticulocyte production to allow you to return to your natural equilibrium level. However, coming back from altitude ( has been show to cause this effect, and the stress of stage racing has been shown to raise %R (see figure 1), which might manage to either raise OR lower OFF-score.

Visualizing Training Duration, Intensity & Performance


Figure 1: Training volume (total hours), average threshold interval performance, number of 30:30 repeats per week, and threshold interval best time by week, July 1, 2013 through October 3, 2013.

Here is an experiment in visualizing what I think are the relevant training stimuli that I have been applying over the last few months as I explore the sport of cyclocross.  My training schema is overall very simple.  I do some long rides on the road bike (reflected in “Total Hrs” column), and I do 2 activities on grass in a local park: 1) “30:30’s”, which are short hill climb repeats, about :30 in duration up a ~15% grade with about :30 rest on the way down in sets of 5-10, and 2) “Loops”, which are Tempo/threshold intervals on a grassy loop of about 2.5 miles (which for me is 10-15 minutes of riding).  The graph in Figure 1 plots aspects of those activities over the period from July 1, 2013 to October 3, 2013.  Below is the table that this plot draws from.  The 4 aspects that I have plotted are as follows:

  • Total Time training, in hours (does not include warmup & warmdown time) – in blue.
  • # of 30:30 intervals completed for the week – in orange.
  • Average Loop interval times for the week – in red.
  • Best Loop interval time for the week – in green.

Looking at threshold Loop interval performance, my times remained fairly static from July through the month of August – hovering around the high 12 minutes to low 13 minute range.  During the first week in September they began to drop fairly substantially, moving to the low 12 minute/high 11 minute range.  A couple of things worth noting, preceding that drop in the first week of September I managed to put in over 8 hours in the saddle, though that was followed by a week where I barely made 2 hours of total time.  The week prior to that I also began doing 30:30 repeats in earnest.  Also, though it is not reflected in the graph, my 30:30 repeat time improved from about :33 seconds per repeat in July to :26-:27 seconds per repeat by the end of September even as I increased the number of 30:30 repeats that I could do in a single session.  The target level of effort that I put into each 30:30 repeat is approximately that which I think I could sustain for a 2:00 all out effort.  The increase in number of repeats is a direct reflection of how fatigued I became after repeated efforts – my ability to recover from successive repeats improved dramatically over time.  Another notable aspect that coincided with this last month of training was my ability to “step on the gas” in a given interval.  Prior to September my fastest repeat and my slowest repeat were fairly similar, maybe :30 faster at the times that I really put the hammer down.  As the month of September wore on into October, my top-end capacity on these loops increased dramatically, with my fastest time for a single repeat dropping to 10:30, about a full minute faster than my sustainable tempo pace which settled in at roughly 11:30 per loop.  I think that much of this performance improvement was due to my ability to recover quickly after pushing it up the short uphill portions of the loop course – I was able to push it harder up the climbs, and recover at a faster pace.

Table 1: Training volume (total hours), average threshold interval performance, number of 30:30 repeats per week, and threshold interval best time by week, July 1, 2013 through October 3, 2013.

Week End Total hrs Loop Avg (min) # 30:30 Loop Best (min)
7/7/2013 3.42
7/14/2013 3.08
7/21/2013 2.38
7/28/2013 1.25 13.50 13.00
8/4/2013 4.13
8/11/2013 3.77 12.92 8 12.92
8/19/2013 4.49 12.80 12.50
8/26/2013 3.73
9/1/2013 3.42 13.28 20 13.28
9/8/2013 7.20 12.02 12.02
9/15/2013 1.30 11.65 6 11.65
9/22/2013 4.03 12.00 25 11.67
9/29/2013 2.50 12.00 27 11.33
10/6/2013 1.06 11.17 10.50

Open Water Stage Races

On Saturday May 18th, we will be hosting a unique open water swimming event – the Ben Hair Memorial Open Water – a “stage” race. This race will feature all the regular mass-start draft legal action of the typical open water event, but will have the added twist of a Team Time Trial and an overall ranking of athletes based on their total time for the individual open water swim and their teams time in the TTT. It is this “total time ranking” that makes a stage race a stage race – that is, athletes compete in multiple events, often different events that play to different athletes strengths, and the overall winner is that person who holds themself together, limiting their losses on stages that do not play to their strengths, and “taking time out of” their competitors on those that do. The Ben Hair Memorial Open Water race has the twist of the Team Time Trial in there, but there are many other ways to run a stage race. Stage racing is very rare in the swimming world, but with the ever-growing numbers of races and racers, we should expect to see more of them in the near future – suspense, teamwork and fun – what’s not to like?

Stage racing is a concept that is most well known in cycling, where competitors race over multiple courses, or stages, and their cumulative time for all stages results in their final ranking for the race. Since these races often give awards for individual stage finishes and for team rankings, this overall time ranking is often referred to as the “General Classification”. These types of races abound, with famous traditional races like the Tour de France, and the Giro d’Italia in Europe, and now domestic races such as the Tour of California, the Tour of the Gila and the USA Pro Cycling challenge.

Swimming Stage Races
Swimming stage races are, however, quite rare. In addition to the 2013 Ben Hair Memorial Open Water race, there are only two other races in the U.S. that fit the bill: the Highland Lakes stage race in Texas, and the Hudson River 8 Bridges Swim in New York.

The Highland Lakes stage race, run in the Austin Texas area, bills itself as “The Worlds First Open Water Stage Race” – This race features 5 stages, in 5 different lakes with the 2013 edition happening October 23rd through 27th. This race was first run in 2007 and has been featured in Swimming World’s online magazine. This race is of somewhat moderate proportions, with its “Monster Challenge” stage competition featuring over 15 miles of racing. The overall winner in 2012, Keith Bell, spent just under 7 hours in the water.

The Hudson River 8 Bridges Swim began in 2011, with a general classification awarded for total time in 2012 – – the stage distances are mind-boggling, with the “shortest” stage at 13 miles, and the longest being 19.8 miles. However, these are all swum down stream, so the individual stages end up under 5-6 hours – still of epic proportions. The 2012 overall winner, Grace Van Der Byl, completed the race in 31 hours and 47 minutes.

This years Ben Hair Memorial stage race is a sprint-fest by comparison, only two stages with the general classification based on the total time for the individual 5k and the 1k Team Time Trial. These stages are swum in a single day, however, and will place a real emphasis on the strength of an athletes team, since a strong athlete with a weak team could easily lose a minute or more to a rival with a deeper team.

Rethinking the Qualification System for U.S. Open Water Nationals

A number of years ago, USA Swimming instituted changes to the Junior and Senior national system of meets to address concerns within the coaching community and within the leadership of our sports governing body.  These concerns were largely based on a desire to have a system of meets that would serve as a series of integrated steps in an achievement pathway that led logically to excellence on the international stage.  Those systemic changes have continued to evolve over the years to the current model of a single long course National championships and a single Junior National meet, both held in summer.

As we begin to think about the growth of the sport of open water swimming, it is natural we are turning a critical eye towards the current system of advancement for our open water athletes.  The governing body and coaches have set up an open water national team, national championships, and support our elite open water swimmers in many positive ways, and our athletes in turn have made their way onto many international podiums … but as the world open water scene grows, we must continue to evolve and improve our system of advancement.

The Road to U.S. Nationals is Through the Pool – Are There Better Ways?
In the U.S., athletes wishing to simply compete in an international distance race such as 5k or 10k face significant obstacles.  Simply put, the demands of the pool season and constraints of weather limit opportunities for racing to late spring (primarily May) and late summer (primarily September).  Perhaps because of this scarcity of races, the only way of qualifying for US Open Water nats is to meet a pool qualifying standards (800 or 1500) or by placing in the top 15 of a world cup event (of which there are exactly Zero held in the US in the 2013 calendar):

Gender Distance 800 LCM 1500 LCM 1000 SCY 1650 SCY
Women 10K 09:03.5 17:18.3 09:59.2 16:43.5
Men 10K 08:23.4 16:14.3 09:13.4 15:39.9
Women 5K 09:09.9 17:38.7 10:11.2 16:55.0
Men 5K 08:32.5 16:36.7 09:25.4 15:56.3

That is not to say that these standards are unreasonable, they are not.  Corresponding roughly to the qualifying times for the summer Junior nationals, they are certain to weed out the pretenders.  Nor is it to say that we should not have pool standards, since there will still be cases where geographic and climatic constraints eliminate opportunity for some athletes, even if we were to expand the open water meet calendar significantly.

However, it is my opinion that there are still a couple of ideas that we should consider as we look to the future:

  1. Since they are based on relatively short events (800-1500), there are possible flaws in the way the pool standards are constructed that merit consideration.   By having relatively short pool races act as a surrogate for open water ability, the current cuts place an emphasis on speed over endurance, and pool swimming over open water.  Moreover, the 10k pool standards are faster than the 5k pool standards, whereas we would expect that athletes who excel at 10k might actually be slower in the 1500/1650 than athletes who excel at the 5k.
  2. A system of open water qualifier meets would help to not only identify those athletes who may have a special propensity for the unique demands of open water swimming, they would also provide a critical proving ground for gaining race experience prior to showing up at nationals or a world cup race.

Some Alternative Models
All of this is not to say that the current approach to qualifying does not insure a high level of competition – it surely succeeds in that regard. But the path to achievement is tipped in favor of older, often post-collegiate swimmers with a record of success in pool events.  At the very least, the current pool system could be enhanced to better reflect the endurance demands of the actual international race distances, and to acknowledge that there is far more to success in open water than simply having a good aerobic engine.  The following are a few ideas that have been put forth for consideration:

  • Open Water Qualifier Meets:
    • Hold regional Open Water Qualifiers for Nationals, or,
    • Provide a certain number of qualifying slots at pre-specified invitational open water meets.
  • Junior Nationals/Sectionals:
    • Add open water events at junior national meets, with top 10 finishers automatically qualifying for the senior level OW nationals
    • Add open water events at sectionals, with top 3 finishers qualifying for Senior Nats, and top 6 finishers qualifying for Junior Nats
  • Pool meets specifically for open-water qualification:
    • Devise a meet framework for a 3-5k pool swim qualifier, and/or:
    • Set qualifying times for the annual postal 3k meet, with mandatory judging/timing criteria equivalent to a dual meet or hosted time trial meet

As we move forward it is important that we consider the pathway that we lay out for our athletes, providing a wide range of opportunity to athletes regardless of their geographic and climatic challenges.  In the end, this may also help to insure that we continue to see those folks with the greatest open water potential find their way to the top step of the podium.

Robert Burgholzer is currently helping with the Ben Hair Open Water Meet which features a 5k and a team time trial, on May 18, 2013.

Digital Geography

We blog about GIS, geodata, webtechnology and have jobs too!

The 512 Project

Learning to eat healthy on a food stamp budget.


ramblings from a former distance swimmer who enjoys hours on the trail

Rees Vineyard

Wine Growing From Ground Zero


Exploring the Body & Mind Through Sport

Athletic Algorithms

Exploring the Body & Mind Through Sport