37
Sampling Plan: Examples
Alternative
plans for a sepsis trial comparing 28 day mortality rates with
90% power to detect a 7% improvement using N=1700
–Fixed sample
study:
•Gather data on 1700
patients and analyze data
–Group sequential study
(OBF efficacy, P=0.8 futility):
•Perform analysis after
425 patients
•If test statistic very
low or very high, stop
•If test statistic
intermediate, accrue another 425
•Repeat, as necessary,
until maximum of 1700 patients