Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Stratified Sample size calculation in Stata IC 15.0

    Hello fellow members of this forum,

    I am planning to perform a randomized trial for an intervention versus placebo, in a 1:1 ratio. The outcome is the binary presence of a side effect the intervention is supposed to prevent, calculated by either Chi squared or Fisher exact tests.
    The problem is: I know for a fact that there is a potential confounder variable (presence of diabetes), that could increase the risk of this event and maybe where intervention could have a different effect. For this reason, i plan to stratify the randomization by taking into account the presence or absence of diabetes at baseline.

    I created the sample calculations on a general randomization model, where the prevalence of de outcome in the control group is expected to be 72.3, in the intervention, 42.3, with a power of 80% and two-sided p value of 0.05:

    . power twoprop 0.723 0.423, power (0.80) alpha (0.05)

    Performing iteration ...

    Estimated sample sizes for a two-sample proportions test
    Pearson's chi-squared test
    Ho: p2 = p1 versus Ha: p2 != p1

    Study parameters:

    alpha = 0.0500
    power = 0.8000
    delta = -0.3000 (difference)
    p1 = 0.7230
    p2 = 0.4230

    Estimated sample sizes:

    N = 84
    N per group = 42

    My question is: how can I account for stratification in the sample size calculations in Stata? I looked into the option and there was just cluster sampling, which is not the case...

    Thank you,

    Ligia Macedo
    Last edited by Ligia Macedo; 04 Mar 2018, 13:59.

  • #2
    Originally posted by Ligia Macedo View Post
    how can I account for stratification in the sample size calculations in Stata?
    You already did. What you've powered the study is: when randomization is stratified on the basis of the presence of a diagnosis of diabetes mellitus, the difference between control-treatment group and experimental-treatment group that you would be loathe to miss is -30%, assuming that the control-treatment group rate is 72.3%.

    Comment

    Working...
    X