Tuesday, August 10, 2010

Estimation of functional threshold pace based on blood lactate data: the Dmax method

by Andrew R. Coggan, Ph.D.

In response to my previous post:


a reader of this blog wrote to ask if using the Dmax method was also a good way to estimate Dr. Steve McGregor's functional threshold pace (http://home.trainingpeaks.com/articles/running/determining-functional-threshold-pace.aspx). Since I am not a runner and this blog is primarily about the use of bicycle-mounted powermeters, this question had not occurred to me. A quick PubMed search, though, turned up at least two studies of running performance in which the Dmax method has been compared with other approaches (there may be others as well):

1. Nicholson RM, Sleivert GG. Indices of lactate threshold and their relationship with 10-km running velocity. Med Sci Sports Exerc 33:339-342, 2001.

2. Papadopoulos C, Doyle JA, LaBudde BD. Relationship of running velocity of 2 distances and various lactate parameters. Int J Sports Physiol Perform 1:270-283, 2006.

As with the cycling research I cited in my previous post, in both of these studies LT as defined using the Dmax method was found to be more closely associated with actual performance, i.e., 10 km running pace, than other definitions of LT (although again some other definitions of LT were nearly as good). However, unlike with cycling pace at LT as determined using the Dmax method appears to be somewhat higher than functional threshold pace. The latter is defined a bit more fluidly than functional threshold power, but in performance terms it is defined as:

"Actual performance from a recent race or hard training run of 10-15 km.

a. If 10 km time was greater than 45 min, use 10 km

b. If 10 km time was less than 45 min, use 15 km or half marathon"

Since all of the athletes in these two studies were capable of running 10 km in under 45 min, their functional threshold pace would be determined using definition b, i.e., their pace during a 15-21.1 km race or hard training run. However, Nicholson and Sleivert (1) found that the Dmax point coincided almost exactly with 10 km race pace, whereas Papadopoulos, Doyle, and LaBudde (2) found that it was, on average, slightly (i.e., 0.2 km/h) albeit not significantly faster still. More specifically, the results of the latter study were:

Pace at Dmax: 16.8 ± 1.1 km/h

10 km pace: 16.6 ± 1.1 km/h

21.1 km pace: 15.5 ± 1.2 km/h

The reason for this apparent difference between cycling and running is not clear, but those wishing to use the Dmax approach to analyze blood lactate data from runners (or triathletes) should not lose sight of this offset or bias, especially when attempting to use the Dmax point as a surrogate for functional threshold power/pace.

The paper by Papadopoulos, Doyle, and LaBudde (2) provides me the opportunity to discuss another issue, which the precision of predicting performance based on blood lactate data, regardless of how the latter is analyzed. As indicated previously, a number of studies have found that LT defined using the Dmax method is well correlated with endurance performance ability - in fact, it is often the very best predictor (compared to, e.g., OBLA, etc.). Despite this fact, when it comes to prescribing training for a given individual there can be significant differences between what data from blood lactate testing predicts that an athlete can do versus what that individual can actually do. For example, in the study referenced above pace at Dmax was closesest to 10 km pace in only six of the 13 athletes tested - the others raced 10 km either slower (n=4) or faster (n=3) than predicted based on the pace at Dmax. What this means is that if training were prescribed relative to Dmax pace, not all of the runners would be expected to perform at the same effort relative to their actual performance ability. Whether this would prove to be a significant issue in the long run would depend upon the exact circumstances, but recognition of this fact illustrates why for several decades I have ascribed to the belief that:

The best predictor of performance is performance itself.

(Note that the above consideration is not unique to running. For example, although the unpublished data of Drs. Jobson and Edwards that I presented previously indicates that there is no significant difference between power at Dmax and functional threshold power and that the two are highly (i.e., R^2 = 0.94) correlated, the standard error of the estimate when attempting to predict functional threshold power based on the blood lactate data is ±17 W. In turn, this means that only half the time would the predicted value be expected to fall within ±7 W of the true value. My point here is not to claim that blood lactate data cannot be successfully used to prescribe and guide training, as clearly it can - rather, I simply wish to point out the true magnitude of the uncertainty involved, so that people do not overemphasize or overinterpret such data.)

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.