Investigating methods for incorporating real world data in health technology assessment decision-making: application in rheumatoid arthritis

This thesis addresses the issue of incorporating real world data (RWD) on treatment effectiveness into the health technology assessment (HTA) process. Evidence synthesis methods are evaluated, developed, and illustrated using a case-study assessing the effectiveness of biologic drugs as treatments for rheumatoid arthritis (RA). An extensive simulation study is undertaken comparing existing methods, under a Bayesian framework, for performing a pairwise meta-analysis combining randomised controlled trials (RCTs) and single-arm trials (SATs) using aggregate triallevel data (AD). The simulation study results suggest that prior-based methods provide a  exible and robust approach to incorporate SAT data into a meta-analysis, by de?ning an informative prior distribution for the pooled treatment effect which can be down-weighted according to pre-speci?cation or commensurability with the RCT data. Novel methods are developed for performing a network meta-analysis combining RCTs and SATs using a mixture of individual participant data and AD, with covariate adjustment on the baseline response. The application of the developed methods to data from a network of RCTs, assessing a number of biologic drugs versus placebo as treatments for RA, suggests that the predicted baseline response for SATs can signi?cantly in uence the pooled treatment effect estimates. The impact of methods for incorporating RWD (including disease-registry data) is demonstrated by repeating the cost-effectiveness analysis from a previous HTA report, using the updated clinical effectiveness estimates. The incremental cost-effectiveness ratio estimates, comparing biologic versus non-biologic treatment strategies for RA, were higher when RWD were included and treatment effect estimates based on registry data were lower compared to those based on RCTs alone.

RWD can provide information regarding the potential variability in cost-effectiveness results, and should be incorporated into the evidence synthesis to improve HTA decision-making.