Baseline characteristics were compared using chi square for categorical variables and one-way ANOVA for continuous variables. The primary endpoint was assessed using sequential pairwise analysis of covariance adjusting for clinical site, baseline PWT, and cilostazol use (adjustments done to increase precision of the statistical comparison). The second baseline treadmill test was used for the comparison. Separate pairwise models were fit using the given two groups being compared. First, supervised exercise and stenting were each compared with optimal medical therapy with a one-sided 0.025 level of significance. Given significance of both comparisons, supervised exercise and stenting were then compared with a 2-sided 0.05 level of significance.
The secondary endpoints of change in free-living daily step activity measured by pedometer use, biomarkers, and quality of life indicators were assessed by pairwise analysis of covariance, adjusting for baseline cilostazol use and study center but with a two-sided significance level of 0.05 for each comparison without adjustment for multiple comparisons. Pedometer activity was normalized to steps per hour to account for differences in hours of pedometer use during the assessment period. All analyses were conducted according to intention-to-treat. Results are based on available data. Multiple imputation of missing primary endpoint data was also performed.
We estimated the PWT would improve by 60% in OMC, 125% in SE, and 164% for ST, based on published data (7 (link);9 (link);30 (link)). Given baseline mean (SD) PWT estimate of 5.0 (3.8) minutes, with 63 evaluable participants in both the ST and SE groups, or 158 participants total between ST, SE, and OMC, the study had 80% power to detect the difference between SE and ST, >99% power for ST vs. OMC comparison, and 98% power for SE vs. OMC. Allowing for 30% premature withdrawal and inclusion of an exploratory arm of ST plus SE, a sample size of 252 was planned. The sample size was adjusted to 217 after removal of the ST plus SE arm due to slow enrollment. Although the study did not meet conservative pre-specified stopping rules, recruitment was stopped early on recommendations of the DSMB due to slow enrollment after review of interim results.
The secondary endpoints of change in free-living daily step activity measured by pedometer use, biomarkers, and quality of life indicators were assessed by pairwise analysis of covariance, adjusting for baseline cilostazol use and study center but with a two-sided significance level of 0.05 for each comparison without adjustment for multiple comparisons. Pedometer activity was normalized to steps per hour to account for differences in hours of pedometer use during the assessment period. All analyses were conducted according to intention-to-treat. Results are based on available data. Multiple imputation of missing primary endpoint data was also performed.
We estimated the PWT would improve by 60% in OMC, 125% in SE, and 164% for ST, based on published data (7 (link);9 (link);30 (link)). Given baseline mean (SD) PWT estimate of 5.0 (3.8) minutes, with 63 evaluable participants in both the ST and SE groups, or 158 participants total between ST, SE, and OMC, the study had 80% power to detect the difference between SE and ST, >99% power for ST vs. OMC comparison, and 98% power for SE vs. OMC. Allowing for 30% premature withdrawal and inclusion of an exploratory arm of ST plus SE, a sample size of 252 was planned. The sample size was adjusted to 217 after removal of the ST plus SE arm due to slow enrollment. Although the study did not meet conservative pre-specified stopping rules, recruitment was stopped early on recommendations of the DSMB due to slow enrollment after review of interim results.
