We are pleased that our recently published study has stimulated scientific interest for further discussion. Unfortunately, Goulet et al. (4) have misread our paper and have drawn incorrect conclusions. Specifically, the speeds of the ad libitum trial (AD) and the prescribed drinking trial (PD) presented in the Results section and in Figure 1 (1) have been misunderstood by the authors of the letter to the editor. They mistakenly concluded that the AD resulted in better performance and faster finishing time than the PD, but even visual inspection of Figure 1 suggests the opposite. As stated in the Results section, under the subheading cycling performance, “Cycling speeds during the first two 5-km hill cycling were not different between the AD trial (30.3 ± 2.3 and 29.2 ± 2.7 km·h−1) and the PD trial (29.8 ± 2.1 and 29.2 ± 2.4 km·h−1, P > 0.05). However, during the third bout of the protocol, the cyclists completed the 5-km climb by 31 ± 30 s (4.7% ± 4.4%) faster in PD compared with AD by cycling at a faster speed (PD = 30.2 ± 2.4 km·h−1, AD = 28.8 ± 2.6 km·h−1, P < 0.05; Fig. 1).” Goulet et al. incorrectly used the following data for AD: 30.3, 29.8, and 28.8 km·h−1 (average 29.6 km·h−1) and for the PD: 29.2, 29.2, and 30.2 km·h−1 (average 29.5 km·h−1) to calculate mean cycling speed for the three 5-km bouts. On the basis of this error, they reported that the AD trial would have finished faster, which is incorrect. Actually, during the third all-out 5-km bout, the cyclist on the PD finished 31 s faster and 248 m in front of the AD, whereas during the first and the second 5-km test, cyclists on the PD finished only 8.5 and 0.8 s slower than the AD (P > 0.05) or 70 and 6 m behind the AD trial, respectively (P > 0.05).
The authors of the letter also compared our findings with a previously published study that used a nonvalid statistical analysis because there was no reliable estimate of experimental error for comparing six conditions with six subjects (3). Interestingly, when the experiments were grouped into low versus high fluid intake, leading to hypohydration of −3.75% and −1.5%, respectively, greater hypohydration was associated with lower power output, slower exercise performance, but similar core temperature.
In our previous experiment (2), we examined the effect of mild hypohydration while cyclists started and maintained a mild hypohydration of −1% of body weight during the same 30-km protocol. In the present study, we hypothesized that AD would lead to progressive dehydration and impaired performance at the later stage of the protocol because subjects started euhydrated. Indeed, after the third 5-km bout, subjects reached −1.4 ± 0.5 kg, whereas no significant water deficit (−0.2 ± 0.2 and −0.5 ± 0.4 kg) was observed after 5-km bouts 1 and 2. Hence, with significant water deficit only apparent during the last exercise bout, averaging of all three conditions was deemed unnecessary.
We thank the authors of the letter for helping us clarify some issues in this very important field of hydration and performance.