The Dangers of Estimating V˙O2max Using Linear, Nonexercise Prediction Models

    loading  Checking for direct PDF access through Ovid

Abstract

Purpose

This study aimed to compare the accuracy and goodness of fit of two competing models (linear vs allometric) when estimating V˙O2max (mL·kg−1·min−1) using nonexercise prediction models.

Methods

The two competing models were fitted to the V˙O2max (mL·kg−1·min−1) data taken from two previously published studies. Study 1 (the Allied Dunbar National Fitness Survey) recruited 1732 randomly selected healthy participants, 16 yr and older, from 30 English parliamentary constituencies. Estimates of V˙O2max were obtained using a progressive incremental test on a motorized treadmill. In study 2, maximal oxygen uptake was measured directly during a fatigue limited treadmill test in older men (n = 152) and women (n = 146) 55 to 86 yr old.

Results

In both studies, the quality of fit associated with estimating V˙O2max (mL·kg−1·min−1) was superior using allometric rather than linear (additive) models based on all criteria (R2, maximum log-likelihood, and Akaike information criteria). Results suggest that linear models will systematically overestimate V˙O2max for participants in their 20s and underestimate V˙O2max for participants in their 60s and older. The residuals saved from the linear models were neither normally distributed nor independent of the predicted values nor age. This will probably explain the absence of a key quadratic age2 term in the linear models, crucially identified using allometric models. Not only does the curvilinear age decline within an exponential function follow a more realistic age decline (the right-hand side of a bell-shaped curve), but the allometric models identified either a stature-to-body mass ratio (study 1) or a fat-free mass-to-body mass ratio (study 2), both associated with leanness when estimating V˙O2max.

Conclusions

Adopting allometric models will provide more accurate predictions of V˙O2max (mL·kg−1·min−1) using plausible, biologically sound, and interpretable models.

    loading  Loading Related Articles