Question

Investigators wish to design a randomized trial to compare low-intensity versus full-dose warfarin for the prevention of recurrent venous thromboembolism (VTE). From a previous study, investigators believe that the risk of recurrent VTE in patients on low-intensity warfarin over a follow-up period of 3 years will be about 8%. They also expect full-dose warfarin to have a lower risk of recurrence (although possibly with additional side-effects such as bleeding complications). Using equal numbers of subjects in each treatment group, they wish to test whether the two treatments differ in prevention of recurrent VTE. Based on previous studies, the investigators would like to be able to detect a 4% risk difference between the two treatments.

a. If the investigators assign 250 subjects to each group (group 1= low-intensity, group 2= full-dose), what is the power of the test to detect a 4% difference? Use a two-sided test with α = 0.05.

b. With equal allocation of subjects to each group, how many subjects in each group are needed to have 80% power to detect the relevant difference, based on a 2-sided test with alpha = 0.05.

c. Suppose now the investigators are willing to allocate twice as many patients to the full-dose group of the trial. How many patients will they need to achieve 80% power? Again, use a two-sided test with α = 0.05. Is equal allocation of patients to the two groups preferable? (In general, we prefer the allocation with the smaller total sample size requirement).

Answer 1

n Suns bxohing proceuu.