THE IMPORTANCE OF A CORRECT STUDY DESIGN TO DIFFERENTIATE BETWEEN TWO OPPOSING MODELS

Abstract

Dear Editor-in-Chief: The study of McLaughlin et al. (2) used the outcomes of four different running tests (V˙O2max test, LT test, running economy (RE) test, and peak treadmill velocity (PTV) test) to predict performance in a 16-km time trial run. Linear regression analyses were used to determine which parameter best predicts 16-km running performance. Velocity at V˙O2max (r = −0.972) was found to be the strongest predictor of 16-km running performance followed by V˙O2max (r = −0.907) and PTV (r = −0.902). As velocity at V˙O2max is predicted from the V˙O2max and RE measurement, the authors concluded that the cardiovascular model (CVM) (1) explains endurance performance better than the central governor model (CGM) (4), which they argue requires that the PTV should be the premier performance predictor. Our original suggestion that PVT is linked to muscle power rather that to maximal oxygen transport (5) resulted in the development of the CGM (3,4). This model argues that exercise performance is regulated to maintain homeostasis in all bodily systems (3,4) and is not simply the result of superior muscle power as suggested by McLaughlin et al. (2), nor does this model predict that exercise performance is limited by oxygen delivery to the muscle, as suggested by the CVM. Instead the CGM argues that by recruiting the largest possible number of skeletal muscle fibers with the greatest contractile capacity, superior athletes will have the shortest foot contact times and the longest flight times (6), resulting in the highest maintainable running speeds and superior endurance performance. On the basis of their finding that "velocity at V˙O2max" had the strongest relationship with 16-km running performance, McLaughlin et al. (2) conclude that the capacity for oxygen delivery to the muscle together with RE determines endurance running performance. However, as their cross-sectional research design cannot prove causation, they cannot conclude that the CVM explains running performance better than the CGM. Furthermore, the overlap in 95% confidence intervals for velocity at V˙O2max (r = 0.92-0.99) and PTV (r = 0.74-0.96) suggests that it cannot even be reasonably concluded that velocity at V˙O2max is a better predictor of the 16-km running time than is the PTV. The main point of this letter is to point out that the study of McLaughlin et al. (2) was not designed to determine causality. To compare the predictions of the CGM and the CVM, measurements such as muscle recruitment, muscle contractility, muscle force production, foot contact, and flight times need to be measured in addition to traditional CVM variables such as V˙O2max and its derivates. In the absence of these measurements, McLaughlin et al. seem to have overinterpreted the significance of their findings to find support for the CVM. This example emphasizes the importance of designing the correct experiment to prove causality and the danger of the overinterpretation of findings from an inappropriate research design. Robert Patrick Lamberts Timothy David Noakes UCT/MRC Research Unit for Exercise Science and Sports Medicine Department of Human Biology University of Cape Town Cape Town, South Africa