How do you customize the forest plot from metan to include the confidence intervals for the heterogeneity metrics

[Mah Tab]

Hi everyone

I am using metan (Stata 18) to do a meta-analysis and create forest plots. I have data on hazard ratios with their confidence intervals from individual studies. I have incorporated the tau- squared and I-squared as metrics of heterogeneity to the forest plot. However, the confidence intervals for them are not shown in the forest plot, even though they are given on the Stata output in the result window. Can anyone help me how I can modify my code so that the heterogeneity metrics along with their confidence intervals be reported in the forest plot.
This is the code I used.

metan logHR loglowerCI logupperCI, eform effect(Hazard ratio) random(ebayes) forestplot(hetinfo (isq tausq) extraline(yes))

Best regards,
Mah

[Tiago Pereira]

Hi, Mah.

The confidence intervals for tau^2 and I^2 are typically of little interest, given their imprecision, unless you have a very large number of studies/estimates.
What it would be really informative is to plot the 95% prediction interval. To plot it, see example 11 here:

https://www.stata.com/manuals/metametaforestplot.pdf

To answer your question, I believe the most straightforward way to add 95% CIs is to manually edit your graph via graph editor:

use https://www.stata-press.com/data/r18/bcgset, clear
meta forestplot
gr_edit .plotregion1.column1.het_stats[1].text = {}
gr_edit .plotregion1.column1.het_stats[1].text.Arrpush Heterogeneity: {&tau}{sup:2} = 0.31 (95% CI: XX to YY), I{sup:2} = 92.22%, H{sup:2} = 12.86
gr_edit .set_spositions
gr_edit .set_graphwidth

[Tiago Pereira]

Sorry about that; I mistakenly read "-metan-" as "-meta-." My suggestions are applied to -meta-, not -metan-.