Prediction of maximum oxygen uptake over time in adults: analysis from the FRIEND registry

Maximum oxygen uptake (V̇O2max) equations from developed countries are inaccurate for developing countries. Accordingly, we aimed to develop equations to predict treadmill V̇O2max over time based on variables other than exercise test in adults from the USA and Brazil undergoing cardiopulmonary exercise testing (CPET). We analyzed data from 2,170 adults who underwent two CPETs (1,307 men; 20-85 years) from the USA (n=1,880) and Brazil (n=290) with a second test after 2.0±1.7 years on average. We fit linear mixed-effects models to develop equations using 90% of the sample, randomly selected. In the remaining 10% of the cohort, we used the coefficient of variation, intraclass correlation coefficient, and the Bland and Altman plots to cross-validate the optimal equation. Our best linear mixed model equation was as follows: V̇O2max (mLO2·kg-1·min-1) = 62.01 - (0.23×Ageyears) - (0.001×Age×Age) - (0.65×Body mass indexkg/m 2) + (5.47×Sexfemales=0; males=1) + (2.78×CountryBrazil=0; USA=1) - (0.68×Arterial hypertensionno=0; yes=1) - (0.45×Hyperlipidemiano=0; yes=1) - (2.02×Smokingno=0; yes=1) - (4.36×Insufficiently activeno=0; yes=1) - (1.67×Beta-blockersno=0; yes=1); R2=0.566. Our main equation was reliable at baseline according to Bland and Altman plot results (mean difference, 0.01 mLO2·kg-1·min-1: 95%CI, -13.94 to 13.98; P=0.966) and over time (0.44 mLO2·kg-1·min-1: 95%CI, -13.5 to 12.4; P=0.439). Demographic and anthropometric attributes, cardiovascular risk, and beta-blockers are valuable for predicting V̇O2max at baseline and over time. The developed equations may apply to countries with socioeconomic and demographic characteristics such as Brazil and the USA.