Validation of a Time-to-Exhaustion-Derived V ˙ O 2 max Prediction Equation and Secondary Criteria Agreement: The...

Measuring maximal oxygen consumption $\left(\dot{\text{V}}{\text{O}}_{2\text{max}}\right)$ during a graded exercise test (GXT) poses challenges in population-based studies. We examined the relationship between GXT duration (GXTd) and $\dot{\text{V}}{\text{O}}_{2\text{max}}$ to create a prediction equation for estimating $\dot{\text{V}}{\text{O}}_{2\text{max}}$ from GXTd. We also assessed the agreement among secondary $\dot{\text{V}}{\text{O}}_{2\text{max}}$ criteria: respiratory exchange ratio (RER; ≥1.10), maximal heart rate (mHR; ≥85% age-predicted), and ratings of perceived exertion (RPE; >17).

Data include 580 Coronary Artery Risk Development in Young Adults (CARDIA) participants (55.9% female, 38.6% Black) who attended the year 35 exam (age 54-66 yr) and completed a GXT with gas analysis. Pearson correlation coefficients were computed between GXTd (seconds) and $\dot{\text{V}}{\text{O}}_{2\text{max}}$ (mL·kg^-1·min^-1). A regression equation was developed to predict $\dot{\text{V}}{\text{O}}_{2\text{max}}$ from GXTd. Secondary criteria agreement was assessed using percent agreement, kappa, sensitivity/specificity, and receiver operating characteristics (ROC) analysis, excluding 38 participants on beta-blockers.

Mean (standard deviation) GXTd and $\dot{\text{V}}{\text{O}}_{2\text{max}}$ were 341.6 (137.0) s and 26.6 (7.1) mL·kg^-1·min^-1, respectively, exhibiting a strong correlation $(r=0.83)$ similar for male and female participants ( $r=0.81$ and 0.80, respectively). The final prediction model included GXTd, body mass index, and physical activity status as significant $\dot{\text{V}}{\text{O}}_{2\text{max}}$ predictors ( $R^2=0.72$ ; bias = 0.14 mL·kg^-1·min^-1; 95% confidence interval (CI), -1.24 to 1.52). RER, mHR, and RPE thresholds were met by 90.8%, 85.2%, and 42.8%, respectively. Agreement between mHR and RER was 82.9% (kappa = 0.178 (95% CI, 0.07-0.29); 34.0% sensitivity, 87.9% specificity). The agreement between RPE and RER was 45.3% (kappa = 0.018 (95% CI, -0.02 to 0.06); 62.5% sensitivity, 43.6% specificity). ROC analyses showed a larger area under the curve for mHR and RER (0.762 (95% CI, 0.705-0.820)) compared to RPE and RER (0.598 (95% CI, 0.522-0.674)).

GXTd serves as a reliable measure of cardiorespiratory fitness in the absence of gas analysis, and a corresponding $\dot{\text{V}}{\text{O}}_{2\text{max}}$ prediction equation can be effectively utilized. Discrepancies in secondary criteria highlight challenges in determining maximal effort, especially based on participant reports.