Examination of data normalization procedures for expressing peak VO2 data

Peak O2 uptake (VO2) has traditionally been compared among individuals differing in body composition by dividing measured values (1/min) by fat-free mass (FFM) (i.e., ratio method). However, the ability of the ratio method to mathematically remove the confounding influence of FFM from peak VO2 has recently been questioned. Therefore, we compared the effectiveness of the ratio method vs. regression modeling to normalize peak VO2 in a large cohort of males and females for differences in FFM. Regression modeling adjusts peak VO2 according to the relationship derived from the regression of peak VO2 on FFM. Results showed that peak VO2 was 60% higher in males (3.53 +/- 1.01/min) than in females (2.22 +/- 0.6 l/min; P < 0.01). With the ratio method (i.e., peak VO2/FFM), peak VO2 was 15% higher in males (54.6 +/- 12 ml.kg FFM-1.min-1) than in females (47.4 +/- 11 ml.kg FFM-1.min-1; P < 0.01). In contrast, when a regression-based approach was employed to normalize values, no significant difference in adjusted peak VO2 was observed between males and females (3.04 +/- 0.9 vs. 3.01 +/- 1.0 l/min). In conclusion, dividing peak VO2 by FFM can produce spurious results, because this approach does not take into account the nonzero intercept. Therefore, a regression-based approach should be used to normalize peak VO2.