Scale break in forest plot after using metan

[Michael Solomon]

Hi all,

I hope that someone can help me? I'm using Stata 12 on a mac and I have performed a meta-analysis using the metan command on 18 studies. One of the studies has outlier data with a SMD of -3.9 (owing to very small sample size and big changes in the treatment arm). I want to include this data in the forest plot but when I graph it, the other 17 studies are very squashed and the graph looks very odd. Is there a way of creating a scale break to include the outlier without squashing the distribution of the other studies?

Many thanks and best regards,
Michael

[David Fisher]

Hi Michael,
Stata in general is not set up to allow scale breaks; see http://www.stata.com/support/faqs/graphics/scale-breaks.
metan allows confidence limits to be outliers by displaying an arrow at the axis limit (using e.g. the force option); but if the effect size itself is an outlier, that's another matter.
When I next have a few minutes to spare I might have a quick play around and see what might be possible, e.g. applying any of the tricks in the link above to forest plots.
Regards,
David.