I am looking to do a metaanalysis (a sub-group analysis thereof ) involving two studies on their diagnostic performances. Now I noticed both STATA and openmeta are unable to do analysis on 2 studies. Are there any software(free ones would be preferred) who can do the same?
Animesh Ray, I was skeptical about your statement that Stata cannot perform meta-analysis with two studies, so I tried the following:
// NOTE: This code requires commands that were added in Stata 16.
// https://www.stata.com/new-in-stata/meta-analysis/
clear *
webuse pupiliq
describe studylbl stdmdiff se week1
// Prepare data for meta-analysis
meta set stdmdiff se, studylabel(studylbl) eslabel(Std. Mean Diff.)
// Do the analysis with default settings
meta summarize
// Do the analysis using only the first 2 studies
meta summarize if study < 3
The final command did result in a meta-analysis with 2 studies.
You'll be more likely to get good help if you show exactly what command(s) you tried to execute and exactly what the error or warning message was. HTH.
Also, one could reasonably just do the calculations by hand or in Excel.
Henry Ratul Halder
My meta analysis has multiple subgroup analysis as a part of sensitivity analysis. It is commonly done in a lot of meta analyses - so I am not sure it can be called worthless. Regards
Bruce Weaver Thank you for your interest.
It is a meta analysis involving diagnostic studies using STATA 13.
1. Using metandi module to get HSROC
metandi tp fp fn tn if handheld1_thermography2==1 & site_1forehead==1 & corecriteriafever
Thanks for posting your code and error messages, Animesh Ray. Have you tried contacting the folks who contributed those Stata commands? You can find email addresses for them at the bottom of the respective help files:
- help metandi
- help midas
You could also try posting to Statalist. Some members there may have run into the same issue. HTH.
- https://www.statalist.org/forums/forum/general-stata-discussion/general
Hi Animesh, i am sure Henry did not mean that your specific analysis was worthless, moreso that any meta-analysis on 2 sets of data has so few datapoints that the results are not really generalizable, as you would want a meta-analysis to be. I would recommend you do a narrative interpretation of the studies you wish to analyse, as that might give the reader more value out of the specific research question you want answered from that analysis.
Hi Animesh, I agree with Birk. I am also doing two study sub group meta analysis right now and facing the same question if it is worth discussing about, but my objective involves group and subgroup analysis so i am planning to involve this in my systematic review.
Also, I was able to do this in STATA using admetan. I hope this helps.
@Alka Bishnoi
Thank you. I have not used admetan. Can you share the syntax?
@Thom S Baguley
Do you mean, I do it by the formula
Effect size of study1 * n in study1/n in both studies + effect size of study2 *n in study2/n in both studies.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software