Standardizing and predicting results from cardiopulmonary exercise testing in patients with heart failure

PURPOSE: Cardiopulmonary exercise testing is a common prognostic tool in heart failure, yet it is not standardized. The purpose of this study was to evaluate a means of standardizing oxygen consumption (VO(2)) measurement and to evaluate the ability to predict peak VO(2) from submaximal exercise. METHODS: Fifty consecutive exercise tests with a respiratory exchange ratio > or =1.10 were evaluated. VO(2) was graphed against respiratory exchange ratio and the peak VO(2) was determined with logarithmic, linear, power, and exponential regression lines. To predict a peak VO(2), each patient's submaximal exercise data (respiratory exchange ratio < or =0.98) were fitted to each regression line. The mean of the last 30 seconds of un-averaged breath-by-breath data was used as the reference value. Peak VO(2) assessments are also provided from the metabolic cart, a rolling time average, and the graphical method. RESULTS: Logarithmic regression best standardized peak VO(2). Mean absolute bias (mL x kg x min) was 0.60 +/- 0.44 for logarithmic, 0.61 +/- 0.47 for linear, 0.85 +/- 0.67 for power, and 1.44 +/- 2.22 for exponential. The mean absolute bias between the peak logarithmic predicted VO(2) and the reference peak VO(2) was 1.62 +/- 1.20 mL x kg x min (9.5% of the peak VO(2)). CONCLUSION: Among the methods studied, logarithmic regression analysis was the best method to standardize and predict peak VO(2) in this cohort of patients with heart failure.