Abstract

Both obesity and a low level of cardiorespiratory fitness (VO_2max) are strong predictors of morbidity and mortality. Therefore, in any lifestyle intervention study, it is important to ascertain if any changes in VO_2max are clinically important, while considering any parallel changes in body mass and/or composition. Most researchers adopt a simple ratiometric approach to normalising VO_2max (ml·kg⁻¹·min⁻¹), even though this is associated with several important assumptions, which are rarely confirmed by researchers. Using data from the Activity Counselling Trial (ACT), we developed and applied a novel allometric approach for comparing normalised changes in VO_2max between different treatment arms. Overall, 874 participants were randomly allocated into advice only (n=292), assistance (n=293) and counselling group (n=289), respectively. A repeated-measures allometric model was adopted to adjust the changes in VO_2max for the concurrent changes in body mass and fat-free mass from baseline to 24-month follow-up. Magnitude thresholds for standardized differences in VO_2max of 0.20, 0.60, 1.20, 2.0 and 4.0 were considered as a small, moderate, large, very large and extremely large effect, respectively. The within-subjects allometric exponent (90% CL) for body mass was 0.46 (0.40 to 0.53) and 0.37 (0.31 to 0.43) in men and women, respectively. The respective fat-free mass exponents were 0.79 (0.71 to 0.87) and 0.64 (0.56 to 0.73). Ratio-scaled VO_2max increased 4.5% (2.1% to 7.0%) more in the assistance vs advice group. This increase was slightly smaller when allometrically scaled to body mass [3.9% (1.7% to 6.1%)] and fat-free mass [3.7% (1.5% to 6.0%)], although standardised mean differences were similar between approaches (≈ 0.2). Similar findings were obtained for the counselling vs advice groups. Changes in VO_2max were small in all study arms for men. Contrary to the original study report and subsequent citations, scaled intervention effects on cardiorespiratory fitness in the ACT are unlikely to be clinically important. Given the non-linear relationship between changes in body size and VO_2max, a within-subjects allometric modelling approach is recommended for accurately quantifying any intervention-induced effects in future studies.