st: non-parametric test compare change over time

["Lam, Chun Nok" <ChunNok.Lam@med.usc.edu>]

We have a longitudinal dataset comparing intervention and control. And we measured subjets' HbA1c at baseline and 6 month. We have a small sample size - 90 total and 45 each group at baseline. I want to see I the change in HbA1c differs significantly between the 2 groups. I start off by otaining a delta (6mo - baseline) for each subject, and use -ranksum- to calulate the group diference. However, due to lost to follow-up, the sample size got smaller using this delta approach. What I hope to do is to use a non-parametric test that can ocmpare the median of all the cases available at baselind, and compare to the median of all the cases available at follow-up. So that the median will be observation-wise at each time-point. What would be the test do it? I am aware that repeated measure ANOVA in Stata can do this trick - to compare observation-wise instead of subject-wise; but the test is ANOVA and it's for means, probably not the right test for this non-parametric sample. 

THank you for your help.

Jonathan Lam


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/