Re: st: Breslow-Day test - meta-analysis

[Roger Harbord <roger.harbord@bristol.ac.uk>]

--On 17 May 2006 07:08 +0100 Maarten buis <maartenbuis@yahoo.co.uk> wrote:

> --- 27026 <27026@bol.com.br> wrote:
> > I have a problem when performing my meta-analysis:
> > Instead of Q-Statistic, I would like to use the option "BRESLOW' to
> > perform the Breslow-Day test for homogeneity of the odds ratio.
> > However, my "METAN" command is not working.
> In my version of -metan- there is no breslow option, so the breslow
> option not working would make sense. The version I got is pretty old so
> it is possible that there is a version roaming around in an other
> archive that -adoupdate- did not pick up. Which version of -metan- are
> you using?
>
> I typed -which metan- and got:
> c:\ado\plus\m\metan.ado
> *! version 1.0 5May98  STB-44 sbe24
The latest version of the -metan- command is only available as part of the 
-metaaggr- package. Type -ssc install metaaggr, replace- to install. -which 
metan- should then give

*! version 1.86 1Apr2004

However the command 27026 was trying is still not valid:

. metan a b c d, or breslow random
Breslow-Day heterogeneity option not available for D+L OR

The Breslow-Day test is in fact only available for fixed-effect 
Mantel-Haenszel meta-analysis of odds ratios, i.e.

. metan a b c d, or breslow fixed

or

. metan a b c d, or breslow

- should work ("fixed" is a default). Note that the Breslow-Day test 
statistic is still labelled "Heterogeneity chi-squared".

Hope that helps,
Roger.

*
*   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/