metan with sample size

[Pallavi Ch]

Hi all,

I am working on meta-anaylsis. I have 4 studies with sample size, coefficients, p values and standard errors (also confidence intervals but only for two studies). Now I want to run meta-analysis. I used the code

'metan Effect_value SE, random'

However, it is not displaying study authors instead it shows 1, 2, 3, 4

I also want to include study sample size using the following code

'metan Effect_value SE, random n(sample_size)'

Study_ID Effect [95% Conf. Interval] % Weight
------------------------+----------------------------------------------
1 0.280 0.190 0.370 34.63
2 0.240 0.150 0.330 34.63
3 0.330 0.052 0.608 3.63
4 0.170 0.068 0.272 27.10
------------------------+----------------------------------------------
Overall,DL 0.238 0.185 0.291 100.00
---------------------------------------------------------------------
I tried with forestplot option, then I found 10% heterogeneity, which was 0 using metan

It would be great if someone could guide me where it went wrong and how to solve it

Thank you

[George Ford]

version number? n() does not appear to be allowed in more recent versions. I think wgt() is the new option.

lcols(Study_ID) will give you labels for the Study_ID. Or, you have Study_Name, you can lcols(Study_Name).

providing sample sizes might be helpful.

[Pallavi Ch]

Hi,
thank you for responding

I am using STATA SE 18.5

Sample sizes are as follows

study 1- 89
study 2- 204
study 3- 202
study 4- 2280

[Pallavi Ch]

Hi George Ford,

Here is the corrected code based on your suggestion, and it worked

metan Effect_value SE, random lcols(Study_ID sample_size)

graph looks like this


Now I wanted to reduce the space between sample size and forest plot also increase the space between forest plot and effect 95%CI. I used graph editor option- Aspect ratio, graph size, scale but was not able to fix it. Could you or someone please help me with this.

Also I tried the following code for the same sample

meta set Effect_value SE, studylabel(Study_ID)

meta update, studylabel(Study_ID) studysize(sample_size)

meta forestplot

however, graph did not show the sample size
Is it because of precomputed effect sizes?


Graph editing in meta forest plot is more flexible though
Is there any specific condition for using meta or metan? I am confused

Any suggestions would be helpful

Thanks