How to make subgroup headings more prominent on forest plot made using metan

[Anneka Mitchell]

Hello,

I am very new to using Stata and also to this forum so I'm hoping you can help. I am using Stata 14.0 on Windows 10 and I created a forest plot using metan which I am pretty happy with (see attached). However, I would like to make the subgroup headings (combined, dabigatran, rivaroxaban) bold so that they stand out.

I read the help documetation for metan and have managed to adjust some formatting from there and I have looked at the help documentation for graph_options and text, however I just cannot figure out what option I need to add to make the subgroups stand out.



The code I used to make the graph was: metan loghr loglci loguci, eform lcols(author year age doac_dose) second(random) favours(Favours DOAC # Favours Warfarin) sgweight by(doac) boxsca(60)

This is an example of my data:

* Example generated by -dataex-. To install: ssc install dataex
clear
input str12 author float year byte(doac doac_dose age sex) float(loghr loglci loguci)
"Forslund"     2017 1 1 1 1  .2546422 -.09431065  .6097656
"Forslund"     2017 1 2 1 1  .6830968   .1655144 1.1999648
"Forslund"     2017 1 3 1 1 .08617773 -.34249035  .5007753
"Bengtson"     2017 2 1 1 1  .2311117  .01980261  .4382549
"Avgil-Tsadok" 2016 2 1 1 1 .26236424  .13102825  .4054651
end
label values doac doac_name_lbl
label def doac_name_lbl 1 "combined", modify
label def doac_name_lbl 2 "dabigatran", modify
label values doac_dose doac_dose_lbl
label def doac_dose_lbl 1 "all", modify
label def doac_dose_lbl 2 "standard", modify
label def doac_dose_lbl 3 "low", modify
label values age age_lbl
label def age_lbl 1 "75 and over", modify
label values sex sex_lbl
label def sex_lbl 1 "both", modify

Thank you in advance and sorry if I have added to much information! Anneka

[Eustina Kwon]

Wondering the same thing; I'd like to make the categories (combined, dabigatran, rivaroxaban) and the top row bold. Thank you!

[David Fisher]

Dear Anneka Mitchell

Sorry, I missed your post back in October. I have since replied to Eustina Kwon 's similar query; see https://www.statalist.org/forums/for...n-forest-plots. Hopefully it will be of use to you too.

Kind regards,

David.

[Anneka Mitchell]

Hi David Fisher
Thank you for pointing me towards your answer to Eustina Kwon. Unfortunately, I cannot seem to get admetan to produce a forest plot.

admetan loges loglci loguci, lcols(study doac doac_dose age_sex) by(doac) forest(hetstat(p) leftj) saving(isch_meta) nograph

This keeps giving me the following error:
Error in option lcols() or rcols(): Label name doac attached to variable doac
is reserved for use by ipdmetan, admetan and forestplot.
In order to save the results set, please rename the label attached to this variable.
Error in admetan.BuildResultsSet
(Note: meta-analysis model was fitted successfully)
r(101);

Can you advise as to what I might be doing wrong? I've checked the help file for admetan but cannot work it out.

Best Wishes,
Anneka

[David Fisher]

Hi Anneka,

Unfortunately you have found a bug, which I have now fixed in an update.

However, you may still find that the results aren't quite what you expect. This is because you are supplying the same variable name (doac) to both by() and to lcols(). The subgrouping variable -- that is, the variable in by() -- plays a particular role in creating the layout of the plot, and is designed to only appear as a subgroup heading, not as a column. Therefore, you may find that you obtain a better result if you create a copy of doac to pass to by(), as follows:

clonevar doac2 = doac
admetan loges loglci loguci, lcols(study doac doac_dose age_sex) by(doac2) forest(hetstat(p) leftj) saving(isch_meta) nograph

Hope that helps.

David.

[Anneka Mitchell]

David Fisher thank you for all your help, I've managed to make some nice looking plots using admetan now. Best Wishes, Anneka