Diamond plots for meta-analysis of proportions

[Alexandra Roddy Mitchell]

Hi,

I am doing a meta-analysis of proportions using "metaprop". It is a large meta-anlysis and too large for the forest plot to be legible. I want to use a diamond plot to visualise pooled estimates and graph subgroup analyses. Does anyone know if Stata is able to diamond plots? And if so, what the command is?

Thank-you.

[Leonardo Guizzetti]

I don't recognize diamond plot as any commonly used terminology in the meta-analysis literature. Can you describe how this differs from a typical forest plot (either with or without subgroups) ?

[Alexandra Roddy Mitchell]

Hi Leonardo,

thanks for your assistance. What I have come across is the literature are plots that use diamonds to represent the pooled estimate, 95% CI and weight, much like the diamond at the bottom of a forest plot. For example:

If anyone has any idea if this is possible in Stata, it would be greatly appreciated.

Thanks

[ericmelse]

Dear Leonardo,

Have you tried the user community contributed packages lpdforest or metaan?

There publications available (a search for these authors will guide you to more on the subject):

Kontopantelis, E., & Reeves, D. (2013). A Short Guide and a Forest Plot Command (Ipdforest) for One-Stage Meta-Analysis. The Stata Journal, 13(3), 574–587. https://doi.org/10.1177/1536867X1301300308

Kontopantelis, E., & Reeves, D. (2010). Metaan: Random-effects Meta-analysis. The Stata Journal, 10(3), 395–407. https://doi.org/10.1177/1536867X1001000307

Kontopantelis, E., & Reeves, D. (2020). Pairwise meta-analysis of aggregate data using metaan in Stata. The Stata Journal, 20(3), 680–705. https://doi.org/10.1177/1536867X20953575

[Leonardo Guizzetti]

Hi Leonardo,

thanks for your assistance. What I have come across is the literature are plots that use diamonds to represent the pooled estimate, 95% CI and weight, much like the diamond at the bottom of a forest plot. For example:[ATTACH=CONFIG]n1655373[/ATTACH]

If anyone has any idea if this is possible in Stata, it would be greatly appreciated.

Thanks

Thanks for the explanation. It would then seem that these diamond plots as you call them are nothing more than graphical summaries of the pooled estimates. I'm not aware of any such software that will produce this plot exactly with the diamond markers, but if you aren't fixed on this precisely, coefplot (SSC) can be used to plot any such point estimate and their CIs. It was intended for regression coefficients, but it is flexible enough for arbitrary values. I'm not sure if it would allow for annotations like the point estimate and CI, number of studies, etc. I don't use coefplot so I can't provide better guidance there.

ericmelse has also given some good pointers they might help guide you to other routes using existing meta-analysis tools. In addition, there is also David Fisher's ipdmetan which has its own forest plot capability.