Dear friends;
We are working on a NMA but our topic is one of the rare ones and it would be better if we can include one arm studies in our analysis. We read about method and "the arm based approach" seems to be working but we have troubles about coding.
If there is anyone who has experience on it, that would be really helpful.
Many thanks :)
I am not sure what you mean unless you give an example. If you have coding difficulties, you may need to abandon the current framework you are planning to use and try the GPM framework in MetaXL (see www.epigear.com)
To be honest, I am not quite sure either. I, myself a kind of lost, right now. We are writing a Network meta-analysis (NMA) and we wanted to include single armed trials into our NMA, there are only a few publications about it. I tried to find a way, and read about methods, I prepaired the data but I couldnt handle with running the codes in WinBUGS. For now I am doing the analysis just with including RCTs but I will keep reading and trying.
Thank you for your usefull advice, I will have a look at webpage detailed, it seems quite helpful.
I am not sure if I understand: is this a WinBUGS coding problem (not my area of expertise) or a conceptual problem with NMA? If the latter, I wonder what would single arm studies contribute to the network? As far as I understand the purpose of NMA is to be able to make indirect comparisons. Could you specify, please?
We read about the conceptual part, it seems possible to include single armed studies in NMA, and in our area of research there are only a few RCT publications, in this case; we thought maybe including single armed studies in our NMA makes our sample size larger and provides us more reliable results. That is why we wanted to try this method so we will have a disconnected network and connect it, but we are having troubles running codes in the programme (which we use WinBUGS), it is not about WinBUGS. It is about us, couldn't figure out the right coding to have the results. It is hard to understand the model and to write the codes because there is no standard/fixed model and codes for this analyse. It is quite new and it seems like we have so much to learn.
It is possible, the statistic is not my area of expertise actually, we are in it as a team. I share the publication and we keep trying. Thank you, Dear Mr. Bjoern Holzhauer, for your helpful guidance.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis