Anaerobic Capacity in Running: The Effect of Computational Method

IntroductionTo date, no study has compared anaerobic capacity (AnC) estimates computed with the maximal accumulated oxygen deficit (MAOD) method and the gross energy cost (GEC) method applied to treadmill running exercise.PurposeFour different models for estimating anaerobic energy supply during treadmill running exercise were compared.MethodsFifteen endurance-trained recreational athletes performed, after a 10-min warm-up, five 4-min stages at ∼55–80% of peak oxygen uptake, and a 4-min time trial (TT). Two linear speed-metabolic rate (MR) regression models were used to estimate the instantaneous required MR during the TT (MRTT_req), either including (5+YLIN) or excluding (5-YLIN) a measured Y-intercept. Also, the average GEC (GECAVG) based on all five submaximal stages, or the GEC based on the last submaximal stage (GECLAST), were used as models to estimate the instantaneous MRTT_req. The AnC was computed as the difference between the MRTT_req and the aerobic MR integrated over time.ResultsThe GEC remained constant at ∼4.39 ± 0.29 J⋅kg–1⋅m–1 across the five submaximal stages and the TT was performed at a speed of 4.7 ± 0.4 m⋅s–1. Compared with the 5-YLIN, GECAVG, and GECLAST models, the 5+YLIN model generated a MRTT_req that was ∼3.9% lower, with corresponding anaerobic capacities from the four models of 0.72 ± 0.20, 0.74 ± 0.16, 0.74 ± 0.15, and 0.54 ± 0.14 kJ⋅kg–1, respectively (F1.07,42 = 13.9, P = 0.002). The GEC values associated with the TT were 4.22 ± 0.27 and 4.37 ± 0.30 J⋅kg–1⋅m–1 for 5+YLIN and 5-YLIN, respectively (calculated from the regression equation), and 4.39 ± 0.28 and 4.38 ± 0.27 J⋅kg–1⋅m–1 for GECAVG and GECLAST, respectively (F1.08,42 = 14.6, P < 0.001). The absolute typical errors in AnC ranged between 0.03 and 0.16 kJ⋅kg–1 for the six pair-wise comparisons and the overall standard error of measurement (SEM) was 0.16 kJ⋅kg–1.ConclusionThese findings demonstrate a generally high disagreement in estimated anaerobic capacities between models and show that the inclusion of a measured Y-intercept in the linear regression (i.e., 5+YLIN) is likely to underestimate the MRTT_req and the GEC associated with the TT, and hence the AnC during maximal 4-min treadmill running.