The Valid Measurement of Running Economy in Runners

Discussion

The primary aim of the present study was to investigate the validity of O_C as a quantification of RE by comparing this variable with the underlying E_C of running. A disparity was evident between the changes in O_C and E_C as running speed increased, with significant increases in E_C but no change in O_C, which appeared to be explained by a progressive shift toward carbohydrate metabolism. A further finding was that linear ratio scaling, rather than a power function, was the most appropriate way of accounting for BM when using both measures of RE.

In accordance with previous findings, as running speed progressed toward the vLTP, elevations in E_C were apparent for each increment in running speed, increasing by 3.7% across the four speeds assessed, whereas O_C remained consistent, supporting the premise that O_C is independent of running speed. It is therefore clear that a discrepancy exists between these two variables. Fundamentally, the validity of a given variable can be viewed as the extent to which the data represents what it is intended to measure. Given that E_C provides the actual indirect assessment of the underlying energy turnover during running, the differing response of E_C and O_C with running speed questions the validity of O_C as a measure of RE.

The inability of O_C to account for variations in substrate metabolism would appear to explain the discrepancy with E_C. It is well established that RER increases with increments in exercise intensity. Indeed, RER in the present investigation increased from 0.88 to 0.94 across the four monitored speeds below LTP, reflecting a shift toward greater use of carbohydrates to fuel energy turnover. As carbohydrates have a greater energy equivalent per mole of O₂ compared with lipids, the increase in carbohydrate metabolism appears to offset the greater energy required at higher speeds, resulting in a stable O_C. As the RER is inherent within the calculation of E_C, it is not influenced by changes in substrate metabolism, enabling the identification of the true relationship of RE and running speed. However, the mechanisms underpinning the association between E_C and running speed remain unclear. Studies using rodent models and humans suggest a greater efficiency in type I fibers when compared to type II fibers. The elevated E_C might therefore reflect the increased recruitment of comparatively inefficient type II muscle fibers at higher work rates. This could be a consequence of the increasing work rate per se, which requires the recruitment of additional motor units. Furthermore, the changing mechanics of running at higher speeds with greater joint angular velocities and reductions in ground contact times, which both require greater rates of force development, might also promote recruitment of type II motor units. Therefore, as alterations in E_C appear to be metabolically driven, it is proposed that future assessments should be conducted at a uniform running intensity relative to vLTP.

The confounding influence of substrate use on O_C might also compromise the interpretation of interindividual differences and intraindividual changes in RE. Notable interindividual variability in substrate use during exercise has been observed in trained athletes. In addition, a greater use of lipids at a given exercise intensity is apparent after endurance training, which might inflate O_C despite no differences in E_C and potentially be erroneously interpreted as detrimental to performance. However, for prolonged events, an increased use of lipids might benefit performance through enhanced glycogen sparing. It is therefore evident that the inability of O_C to account for variations in substrate use compromise the validity and subsequent interpretation of this commonly used measure of RE.

The accurate comparison of RE between individuals requires an expression that is independent of the confounding influence of BM. In this study, linear ratio scaling had marginally higher R values but lower RMSE than power function scaling, for both E_C and O_C, and therefore appeared to be the most appropriate method to remove the influence of BM on RE in endurance runners. Our findings support recent observations demonstrating ratio scaled RE measurements to be independent of BM. In contrast, previous investigations of heterogeneous cohorts of trained athletes have proposed measures of submaximal oxygen uptake to be proportional to BM and BM. However, unlike the homogenous cohort of endurance runners within the current study, these investigations included athletes from various sporting disciplines as opposed to our cohort from a single discipline (running). Hence, our findings demonstrate the necessity to evaluate the appropriate scaling factor for a given population, which might be sensitive to the body composition of the cohort. Given the better fit of the linear model, the comparative simplicity of its calculation, and the widespread use of this exponent within the literature, the expression of E_C relative to BM would appear preferential. It is therefore proposed that measures of RE in highly trained endurance runners should be expressed relative to BM to accurately account for variations in BM.

In conclusion, our data indicate that O_C does not provide a valid index of the underlying E_C of running. Although E_C increased with increments in running speed, O_C is insensitive to such changes, likely because it does not account for variations in substrate use. Moreover, the expression of E_C relative to BM was found to be the most appropriate method to remove influence of BM on this variable. It is therefore proposed that E_C should be used as the primary measure of RE, scaled appropriately to BM.