Interpreting Results from Bayesian Network Meta-Analyses (Nma): A Guide for Non-Statisticians

  • Pacou M
  • Taieb V
  • Belhadi D
  • et al.
Citations of this article
Mendeley users who have this article in their library.


Objectives: Bayesian NMAs enable us to report results based on probabilities, treatment ranking and predictions and are increasingly used to support decisionmaking in HTAs. However, there is a lack of guidance on how to report and interpret results from Bayesian analyses. In addition, the complexity of this type of analyses makes these findings difficult to understand by analysts not trained in Bayesian statistics. We aim to define in simple terms the key concepts behind Bayesian methodology and present a guide to help non-statisticians understand and interpret findings from Bayesian NMAs. Methods: Majorguidelines (incl. NICE, IQWiG, CADTH, HAS and EUnetHTA) were reviewed to identify recommendations made forBayesian NMAs in the context of HTAs. Examples of HTA submissions from manufacturers were used to identify how Bayesian results are reported in practiceIn order to ensure clarity and simplicity, a guide to interpret these results was developed in collaboration with analysts not trained in Bayesian statistics. This guide is illustrated with an example of NMA. Results: Bayesian analyses are often used in the conduct of NMAs meant to inform cost-effectiveness models. Results are generally reported as median or mean of the posterior distribution, standard deviation, 95% credible intervals and forest plots. Additional results include the probability for each treatment of ranking first, the SUCRA (Surface Under the Cumulative Ranking) and the probability for the intervention to perform better than its comparators. Although it could help interpret the findings, graphical representation of the posterior distribution is not commonly reported in HTA. Sensitivity analyses are also often reported, mainly to assess the robustness of the results. Conclusions: Our guide is useful to analysts not trained in Bayesian statistics for decision-making purposes in the context of HTA submissions. More specifically, it is a straightforward reference tool for using NMA results to populate cost-effectiveness models.




Pacou, M., Taieb, V., Belhadi, D., Mesana, L., & Gauthier, A. (2015). Interpreting Results from Bayesian Network Meta-Analyses (Nma): A Guide for Non-Statisticians. Value in Health, 18(7), A718–A719.

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free