st: Bootstrapping METAN by study identifier: epidemiological study, bootstrap meta-analysis

["Cangelosi, Michael" <MCangelosi@tuftsmedicalcenter.org>]

Dear Statalist:

I have an epidemiological meta-analysis that is examining a few
categories of outcomes with multiple endpoints assigned to these
categories.  I don't care too much about the endpoints themselves.
METAN has been very helpful to get pooled relative-risks of some
outcomes that have only one measurement per study.  

However, I need to modify the usual routine to account for multiple
measurements within a same category (e.g. study X has multiple endpoints
assigned to category (1); so would receive more weighting than a study Y
which has only one endpoint assigned to category (1)).  I am
contemplating bootstrapping METAN, grabbing only one outcome measurement
per study.  While a command something like:

bootstrap ES=r(ES) seLogRR=r(selogrr), metan A B C D by(OutcomeType)

seems close, METAN is still not grabbing only one outcome per study.
Has anyone already performed a similar analysis?  Does METAN have any
capability to address multiple measurements per study? (the .hlp file
seems to suggest no).

Suggestions?

Michael J. Cangelosi MPH, MA

P.S. I Recently posted this message in HTML and suspect message is
trapped, if anyone received both versions, I offer my sincere apoligies.

*
*   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/