power calculation for two groups (parallel)

[rony443]

Hi
I have got 400 and 600 samples in control and test groups respectively. And in each of these two groups, i have an effect variable that occurs at the rate of 30%. this is a retrospective data analysis.

I want to investigate whether the effect variable impairs the patient outcome (response rate as well as median survival) in control and treatment groups and if so, whether such effect is any different in treatment and control groups.

Any ideas which variables i need for this calculation ? What formula can i use to calculate the power for this study. thanks

 

[hlsmith]

So you have two groups (test and control) with 600 and 400 people respectively in them. You have an exposure variable which is equal in proportion for both groups and you want to see if it explains median survival (as a binary variable) and response rate (please better define this).


So it sound like you will be running a survival analysis, correct?

 

[rony443]

So you have two groups (test and control) with 600 and 400 people respectively in them. You have an exposure variable which is equal in proportion for both groups and you want to see if it explains median survival (as a binary variable) and response rate (please better define this).


So it sound like you will be running a survival analysis, correct?

Thanks, hlsmith.
Median survival for each group can be computed by number of years of disease free survival for each case. So it is a continuous variable but can be dichotomized in case it helps analysis.

Response rate is percentage of patient responded in each group, which is based on dichotomous variable for each patient- responded or not.

I want to see whether the effect variable (which is discrete) in control and/or test groups can influence survival.

I want to know if the number of cases are adequate based on power calculation.