(15 pts) A Randomized Clinical Trial was designed to compare the effectiveness of

two different antiretroviral (ARV) drugs in patients co-infected with TB and HIV in

Korea. A sample size of 660 per group was chosen to have 90% power to detect a

difference in mean CD4 at 1 year of 25 based on an estimated standard deviation of

140.

a. Which of the following is an accurate interpretation of the 90% power of this

study? (circle 1)

1. We will have a 90% chance of finding a significant difference if the

observed mean of the two treatment groups differs by 25 and the standard

deviation is 140.

2. There is a 90% chance that the observed difference will be 25 or more if

the standard deviation is 140.

3. At the end of the study, if the standard deviation is 140 then on average

90% of the data will be contained in a 90% confidence interval.

4. We will have a 90% chance of finding a significant difference if the true

mean of the two groups differs by 25 and the standard deviation is 140.

b. If the true standard deviation in this population is 210, will the study have more

or less power than expected?

c. All participants will be observed for exactly one year, at which point their CD4

count is measured. Once the study is over, what type of test can be used to

compare the observed differences in CD4 count between the two groups?

d. The investigators are considering adding a third and fourth group who would

receive two other ARVs. If they do so, what test would you use to simultaneously

compare the four ARVs with respect to CD4?

e. What is the problem with comparing all pairs of ARVs using a t-test at

alpha=0.05 for each?