Statistics
We calculated that a sample size of 80 patients should have
80% power to detect a 90-second improvement in exercise
endurance between NVA237 and the placebo, under the
assumption that the variability in the measurement, in terms
of within-patient standard deviation, was approximately
200 seconds. A blinded, interim sample size re-estimation
was performed after 24 patients completed the study, ie, when
24 patients had completed both treatment periods. Because
the blinded estimate of the standard deviation was less than
200 seconds, no change was made to the planned sample
size of 80 patients. However, to allow for patients dropping
out of the study, recruitment into the study continued