Author(s): SCHIFFERDECKER, I., KELLER, S., JI, S., WAHL, P., Institution: GERMAN SPORT UNIVERSITY COLOGNE, Country: GERMANY, Abstract-ID: 2693

INTRODUCTION:
Ratio scaling (dividing a variable by body mass [BM]) is commonly used in sport science to remove BM effects from BM-dependent parameters. However, the statistical assumptions required for adequate BM normalization by this method are rarely met [1,2], which can distort the statistical relationships between scaled variables and other parameters. Previous research shows that ratio scaling does not adequately control for BM in physiological variables such as maximal oxygen uptake (VO2max), particularly in youths [3], whereas group-specific allometric scaling provides a suitable alternative [1,2]. Thus, this study aimed (1) to model running velocity at the second lactate threshold (LT2) in young athletes, including sex-specific subgroups, and to examine how allometric scaling affects the contributions of key endurance performance determinants (VO2max, oxygen cost of running [OCR], and fractional utilization of VO2max at LT2 [%VO2maxLT2]) to the explained variance in LT2 (R2), and (2) to determine whether stepwise regression retains BM as an independent predictor after adjustment of physiological determinants for BM.
METHODS:
199 youth squad athletes completed an incremental test to determine VO2max, OCR, LT2 and %VO2maxLT2. After data quality checks, 111 data sets remained. To examine BM-dependency, physiological variables were linearly regressed on BM within each group (mixed sex, male, female). When significant relationships were observed (p<.05), group-specific allometric exponents were calculated and variables adjusted accordingly. LT2 was modelled separately for each group using multiple linear regression with VO2max, OCR and %VO2maxLT2 as predictors, scaled either ratiometrically or allometrically. Afterwards, bidirectional stepwise regression was performed with classical endurance determinants and BM as candidates. Commonality analysis quantified the unique and shared contributions of predictors to R2.
RESULTS:
VO2max and OCR were ratio- and allometrically scaled. In females, classical predictors explained ~92% (ratio-scaled) and ~86% (allometric) of R2; in males, ~95% regardless of scaling method. In females, stepwise regression retained BM as a predictor (p<.001) in the allometric but not in the ratio-scaled model. In males, BM was not retained. Across sexes, VO2max showed the largest unique contribution to R2 when ratio scaling was applied. In females, OCR provided the largest unique contribution when predictors were allometrically scaled.
CONCLUSION:
Controlling BM effects using group-specific allometric scaling may prevent misleading conclusions about associations between physiological variables and performance markers. Ratio scaling appears to reduce the discriminative power of OCR, especially in females. Higher BM seems to negatively affect running endurance performance in female athletes.

[1] Notley et al. (2020). Med Sci Sports Exerc [2] Welsman & Armstrong (2019). Pediatr Exerc Sci [3] Armstrong & Welsman (2020). Med Sci Sports Exerc