Data and Statistical Analysis
For each stage of exercise, a representative value of each variable (ie, Vo2, Vco2, Vt, respiratory rate, Ve, and HR) was obtained by averaging the breath-by-breath results over the last 30 s of the stage. Since there was only one determination of systolic and diastolic BP during each stage, there was no need to calculate a representative value for these two variables. Then, for each stage, the representative values of a variable from all tests were averaged and the mean value together with its standard deviation were calculated. Reproducibility was assumed to be represented by the coefficient of variation (CVAR) that was obtained by dividing the standard deviation by the mean. In a similar fashion the CVAR was also calculated for AT, Vo2max, and exercise duration. Finally, to test for a systemic time difference in the exercise response of each variable, an approach similar to that suggested by Wallenstein et al was used.
SYSTOLIC BLOOD PRESSURE
That is, for each patient (1) the response of a variable to increments in work was assumed to be linear and a representative slope computed for each exercise test, and (2) the slopes of all tests were correlated with the corresponding times the tests were administered (ie, number of months from the first exercise test). To test for a time trend in the end-exercise variables, linear regression analysis was used to determine whether the correlation coefficients between these variables and the time they were acquired was statistically different from zero. In both cases, a significant correlation coefficient was evidence of a significant time trend.