I am currently planning to conduct a Bayesian network-meta-analysis, thus in a non-treatment related context. I am now coming across several questions and problems and am curious what opinions and approaches others would recommend.

1. I would like to conduct a network meta-regression for the covariates. However, not all covariates are necessarily the same level within a study (the level of the covariate might f. ex. differ across different arms within a multiarm study). So far, I mainly found worked examples where covariates were treated on a study level and just one paper by Donegan (2018) which states that an analysis on trial level would require iteratively weighted least squares estimation with transformed observations. Is there any other experience/opinion on this topic?

2. I usually only find meta-regressions for a random effects model with a comparison of each arm vs. placebo. In a paper by Dias et al. (2011). I even found that a placebo comparison is necessary because the treatment and covariate interaction effects only then act solely in the treatment arm, not in the control. Now I am wondering for the possibilities of RE meta regressions in non-control trial network meta-analysis. In my specific case the analysis shall be applied in a non-clinical context, where there is no placebo group or any group in which I can assume an effect to be null. Do you have other suggestions or came across alternative approaches? (Anyhow, I am wondering: Theoretically speaking, isn’t it pretty farfetched to assume any effects being null in placebo? -or am I getting something totally wrong here?)

3. I am still a little unsure about my outlier detection, I found very view examples and papers on this topic in a Bayesian framework. Only mainly Zhang et al. (2015) and Dias et al. (2018, pp. 75–90) and so far, no R-package or anything similar. Most NMAs I read don’t even report any outlier detection. Any recommendations/thoughts/hints on this maybe?

I am happy about all type of input. Many thanks!

Dias, S., Ades, A. E., Welton, N. J., Jansen, J. P., & Sutton, A. J. (2018). Network Meta-Analysis for Decision Making. In Network Meta-Analysis for Decision Making. Wiley. https://doi.org/10.1002/9781118951651

Dias, S., Sutton, A. J., Ades, a E., & Welton, N. J. (2011). NICE DSU Technical Support Document 2: A Generalized Linear Modeling Framework for Pairwise and Network Meta-analysis of Randomised Controlled Trials. Dsu, 0, 1–98. http://www.nicedsu.org.uk

Donegan, S., Dias, S., Tudur-Smith, C., Marinho, V., & Welton, N. J. (2018). Graphs of study contributions and covariate distributions for network meta-regression. Research Synthesis Methods, 9(2), 243–260. https://doi.org/10.1002/jrsm.1292

Zhang, J., Fu, H., & Carlin, B. P. (2015). Detecting outlying trials in network meta-analysis. Statistics in Medicine, 34(19), 2695–2707. https://doi.org/10.1002/sim.6509