Using network analysis to analyse review findings

jjrm845

I'm a Masters student and am completing a systematic literature review. I want to analyse the findings across studies specifically looking at subtypes of a condition and their associations with health outcomes across four domains. There is a lot of variation in health outcomes within domains and I have effect sizes, beta values, and ORs. How on earth do I go about this? I don't think there's enough consistency for a meta-analysis/forest plot. I'm thinking a separate analysis for each domain. My supervisor suggested network analysis but I'm really unsure how to set up the data set. Any help would be appreciated!

EJ

I've asked our expert for his insights...

František

Hi,

There are many possible approaches to this - one of them is indeed a network analysis (which is not supported in JASP yet) but you could also do subgroup/meta-regression analyses for each domain/condition and include random effects for additional sources of heterogeneity, like drug/treatment effects etc. This is quite a complex question, though and I can't really provide a more elaborate answer without much more details. I would recommend checking the documentation of the metafor package - pretty much all meta-regression / multivariate models are implemented in JASP and the helpfiles might help you identify what functionality you are searching for.

Cheers,

Frantisek