Nonexercise regression models to estimate peak oxygen consumption

The purpose of this study was to develop a VO2peak prediction model derived from nonexercise (N-EX) based predictors. VO2peak was measured using a walking treadmill protocol with 229 females and 210 males between 20 and 79 yr of age (mean +/- SD: 38.62 +/- 10.36 ml.kg-1.min-1). Subjects were randomly divided into validation (V) (85% of total; N = 374) and cross-validation (CV) (15% of total; N = 65) groups. The V group was used to validate generalized and gender-specific models using stepwise multiple regression procedures with gender, age and age2, percent body fat, and a physical activity code (AC). The generalized ml.kg-1.min-1 (R2 = 0.77, SEE = 4.90 ml.kg-1.min-1, SEE% = 12.7%) and gender-specific (females: R2 = 0.72, SEE = 4.64 ml.kg-1.min-1; males: R2 = 0.72, SEE = 5.02 ml.kg-1.min-1) models were highly accurate relative to N-EX and exercise based models in the literature. Cross-validation procedures were used to evaluate model stability. The generalized model was stable across the total CV group and various CV subsamples (by gender, decade-wide age groups, and AC groups), but not across groups similar in VO2peak. These results suggest that N-EX models can be valid predictors of VO2peak for heterogenous samples.