Background: The Health Technology Assessment (HTA) evidential landscape is changing - randomised-controlled trials (RCTs) are involving fewer patients, with shorter follow-up, and often use intermediate or surrogate endpoints. At the same time Real World Evidence (RWE) studies, e.g. observational studies or patient registries, are increasing in both number and scale. These changes are presenting decision makers with a considerable challenge. Whilst RCTs may be considered less biased, and RWE studies potentially more biased (due to treatment-selection effects and confounding), RCTs are not always considered as relevant for real-world decision making because of patient selection.

Objectives: Can we simultaneously address both relevance (to a target population) and rigour (risk of bias) of the available evidence in order to aid health decision making for a target population?

Methods: Using a bivariate power prior approach, to simultaneously down-weight potentially biased studies and up-weight more relevant studies (based on study-level covariates), the evidence from both RCTs and RWE studies for a variety of treatments for patients with multiple sclerosis (MS) is synthesised using a network meta-analysis (NMA) approach.

Results: Adding additional evidence from RWE studies, although increasing the overall evidence base, increased uncertainty surrounding specific treatment-effect estimates in this MS case study as between-study heterogeneity was also increased. However, predicting the treatment-effect estimates for a specific target population ameliorated this increase in uncertainty.

Conclusions: This case study illustrates that a bivariate power prior approach to evidence synthesis can simultaneously address both relevance and rigour, and enables more appropriate tailored treatment-effect estimates to be obtained.