Metan (no box)

[Adam Mitchell]

Hi, I am using stata 17 and using the following code for a meta analysis to create a forest plot. I do not want any weighted boxes just the point estimate so I use the nobox command. However using this removes any marker of the estimate so it is jus the confidence interval line. Does anyone know why this is and how I can fix it to get a point estimate marker.

set more off
metan beta se if Outcome=="FI", lcols(Bio) rcols(pvalue) classic scheme(s1mono) ///
xlab(-4.0, -2,0, 0, 2.0, 4.0) force effect(Estimate) textsize(100) nobox nowarning nowt nohet nooverall ///
plotregion(style(none)) diamopt(lcolor(black)) pointopt(msymbol(S) msize(small)) olineopt(style(black) lpattern(shortdash)) xsize(15) ysize(10) ///
t2title(Estimate (95% CI), size(small) placement(center) position(6))

[David Fisher]

Hi Adam,
The problem is the use of the two options classic and nobox together. The option classic is intended to replicate the look of an older version of metan, and gives solid black (weighted) boxes without point estimate markers. Therefore, when nobox is added, you are left with no markers at all!
I would suggest removing the option classic. I don't think this option affects anything apart from the boxes, so you should find that the rest of your plot is unaffected.
Thanks,
David.

[Adam Mitchell]

Hi David, thank you for such a quick response. This worked perfectly and I now have the graphs I need. Really appreciate the help.
Best,
Adam