Mechanical power and energy expenditure in level treadmill running

The purpose of the present study was to examine the relationship between energy expenditure (EE) and mechanical power (MP) at four speeds of treadmill running; 3.58 (S1), 3.97 (S2), 4.47 (S3), and 5.11 m$\cdot$s$\sp{-1}$ (S4). Twenty-two competitive male runners (age = 28.9 $\pm$ 5.4 yr, mass = 67.8 $\pm$ 6.5 kg) completed a total of four 10-minute steady-rate (SR) runs and one max VO$\sb2$ run on two days. Six subjects repeated the runs to provide estimates of reliability. Physiological measures included VO$\sb2$, HR, and blood lactate (BLa). Seven algorithms based on the work-energy theorem were used to calculate MP. A major difference between the algorithms was the constraints placed on energy transfer (ETr). High speed video (200 Hz) techniques were used to capture sagittal plane kinematics. Five gait cycles were digitized for each subject and processed using a Motion Analysis system. Mean power was expressed relative to the time required for one gait cycle. The mean maxVO$\sb2$ of 69.4 ml$\cdot$kg$\sp{-1}\cdot$min$\sp{-1}$ reflected the high fitness level of these runners. Mean VO$\sb2$ for the SR runs was 58.5, 65.9, 75.2, and 89.4% of max VO$\sb2$ for S1-S4, respectively. Mean HR ranged from 133.4 to 176.9 bpm across speeds. There were no differences in VO$\sb2$ or HR over days and reliability (intraclass R) ranged from R = 0.852 to 0.976. There was a significant increase in mean BLa from 4.4 to 6.8 mmol at S4. Therefore, an oxygen equivalent for BLa of 3 ml O$\sb2\cdot$kg$\sp{-1}$ per mmol BLa (di Prampero, 1981) was incorporated into all subsequent EE values for S4. Mean power increased directly with speed. Extreme values ranged from a high of 3262.4 W for the no transfer model at S4 to a low of 373.3 W for the total transfer model at S1. Reliability for TPos and TNeg ranged from R = 0.967 to 0.994. Differences among slope and intercept coefficients from the individual regressions were apparent for the majority of post hoc contrasts and indicated that the algorithms were unique and differentiated between the methods of ETr. As more power was attributed to energy transfer, the slope coefficients increased. In contrast, an inverse relationship was noted between intercepts and the amount of energy transfer. A major finding of the present investigation was the significant linear relationship between MP and EE independent of speed for six of the seven algorithms. The percentage of variance in EE explained by the limb transfer algorithm was $\sim$50% with SEE ranging from 0.185 to 0.395 l$\cdot$min$\sp{-1}$. This algorithm also remained the most consistent across speeds with ETr changing only 3.2%. Based on the percentage of variance in EE explained by MP, the limb transfer algorithm provides the best estimates of EE from the kinematic methods of energy analysis used in this study