I put some tables in this that were munged by formatting. I am trying to correct that.

As fast as you can for as long as you can? Isn’t the protocol for the AnT test: go as fast as you think you can go for 30 minutes? My AnT field tests tend to have plateaus. The last one I did began with a 14 minute plateau with HR=161, then a 5 minute plateau with HR=162 then a 7 minute plateau with HR=167. (The missing 4 minutes were between the plateaus.) The overall HR average was 162. I conclude that 162 is a good lower bound for my AnT HR at that elevation and that 167 is likely a good upper bound. I used 162 in my training.

The drift test has some fairly serious issues independent of altitude. I personally don’t think it is trustworthy.

Issue 1 is rounding error. This is minor and can be easily fixed.

For whatever reason Training Peaks, Polar, and Suunto all report heart rate averages to the nearest beat per minute. They shouldn’t do that. They seem to think they are suppressing something they call “false precision” which doesn’t actually mean anything. They should report the average heart rate to four or five significant figures. The reason is that if you’re doing a drift test you’re going to take a difference and the difference you get has a two bpm spread because of rounding. That two beat spread in the difference can affect the conclusion. Here’s an example of what I am talking about. I like to think in terms of the actual number of heart beats as I am fairly certain those devices report it.

TABLE 1

Number of heart beats true reported

first 30 minutes HR (bpm). HR(bpm)

4215 140.5 141

or

4214 140.46. 140

second 30 minutes

4425 147.5 148

or

4424 147.46. 147

In table 1 the true HR is the number of heart beats over 30 minutes divided by 30 minutes. The reported HR is the integer HR average after rounding. The calculated difference between the rates that rounded is between 6 and 8 beats per minute.

TABLE 2

beats in 1st interval beats in 2nd interval reported % difference true % difference

4214 4424 (147-140)/140=5.0% 4.98%

4214 4425 (148-140)/140=5.7% 4.98%

4215 4424 (147-141)/141=4.25% 4.95%

4215 4425 (148-141)/141=4.96% 4.96%

The true per cent difference is, to good approximation, 5% in all 4 cases, as it should be. The reported per cent differences run from well below 5% to well above whereas the true percent difference is very close to 5% in all cases. Talk about false precision!

This is easily remedied though. All they have to do is report the average heart rate correctly without rounding. Seems like a big deal to me because if you see a drift of 5.7% you’re likely to think you were too fast and if you see a drift of 4.25% you’re likely to think you were too slow. You might get by saying if the drift is between

4% and 6% then your starting heart rate is an adequate estimate of your heart rate at AeT. But then there’s Issue 2.

ISSUE 2 this one is more serious and why the procedure is not trustworthy.

The drift rate drifts. Say you stay on the treadmill for say 100 minutes. Take the heart rate time series that began at T=20 minutes and ends at T=80 minutes and compute the two 30 minute heart rate averages. Then assume your warmup ended at T=25 minutes and take the two 30 minute averages beginning then. In that 5 minutes my drift dropped from 5% to 3.5%. Which one should I use and why? The choice is arbitrary and is going to have a significant impact on my training.

All the treadmills that I’ve been able to use had time limits of 100 minutes which also causes problems if you want to look at how much your drift drifts. What I do is warm up for 25 minutes then shut the treadmill off and restart it as quickly as I can and warmup for another 5-10 minutes until I’m back at the same work load and my HR has stabilized again. Then I can do 3 30 minute intervals to see how much the drift changes over 90 minutes. The drift between the final two 30 minute intervals tended to drop to near zero with the work loads I’ve used. The first 30 minute drift was usually substantially higher than the second.

Below is some data from one of my drift test attempts.

# r(t) is defined as the 30 minute average heart rate from t to t+30 minutes

# dr is defined as r(t+30)-r(t)

#

t(min) r(bpm) dr(bpm) dr/r(%)

20 138 7 5

25 141 5 3.5

30 141 5 3.5

33 142 6 4.2

35 143

50 145

55 146

60 147

63 148