I read with interest the article entitled "Altman DG, Bland JM:
Interaction revisited: The difference between two estimates. BMJ 326:
219, 2003". This describes test for interaction for subgroup analysis
in a meta analysis. For ex: In a meta-analysis of randomized
controlled trials of Aspirin vs. control for end point of mortality,
to test the effect of say gender.
I have the following questions:
1. The article describes evaluating tests of interaction when there
are two groups involved. For example to test the effect of age on
mortality in a randomized trial of active treatment vs. placebo when
age is considered as 2 groups...>65 or not. How can I perform the
same test for interaction for multiple groups (across the groups).
For ex: if age is divided into 4 groups.
2. Is there a way to perform the test of interaction using STATA,
while performing the calculation for pooled risk using say the metan
program (peto OR) without manually calculating it as outlined in the
table of the article.
thanks,
Sripal.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
