Integrating More Individualised Estimates of Costs and Effects in Economic Evaluations in Healthcare.

The proposed research aims to i) Evaluate subgroup heterogeneity at both early and late translation stages to inform personalised economic evidence of new therapeutics; ii) Examine treatment effect heterogeneity across multiple health outcomes using different decision criteria.

Evaluating Subgroup Heterogeneity to Inform Personalised Economic Evidence Identifying baseline covariates in which treatment response may differ from the population average treatment effect is of central interest to policy makers, industry and reimbursement agencies as well as informing clinical decision making in an age of precision medicine.

Treatment-covariate interactions are frequently used to test for the homogeneity of treatment effects but the lack of statistical power to detect sub-group differences increases the probability of Type-II errors in many studies.

There are a range of methods that can be applied to test for heterogeneous treatment effects in the statistical and econometric literature, yet their utility to inform health economic evaluations remains ambiguous.

This PhD will evaluate the different methods to determine their suitability for deriving economic evidence for key healthcare stakeholders at specific points in the translational process.

The project will examine how Bayesian hierarchical linear models with a sceptical prior can be used to identify individuals who benefit most from treatment.

The use of sceptical priors within a novel approach called Bayesian Credible Subgroups (BCS) ensures evidence of large heterogeneous effects are unlikely but also does not rule out the possibility that they may exist through the posterior distribution.

Using existing data from both large multicentre randomised controlled trials (RCTs) and early-phase studies, Bayesian inference will be adopted to classify individuals into more credible subgroups based on a decision rule of three possible class memberships - responders, non-responders as well as an uncertain responder group in which further evidence is needed to inform clinical decision making.

The PhD will explore whether key baseline covariates used to determine credible subgroups simply shift the central tendency of the distribution of costs and effects using flexible regression methods.

Generalized Additive Models for Location, Scale and Shape will be compared with standard generalized linear models to assess whether key covariates influence the mean, variance and/or shape of the distribution of costs and effects.

Data driven approaches on treatment effect heterogeneity will be compared to assess clinical and economic utility for both early and late-phase evaluation of healthcare innovations. This will include, but not limited to, Bayesian Networks, Random Forests and Extreme Gradient Boosting.

Results from the alternative applied methods will be disseminated through mock Health Technology Assessment panels, comprising clinical and non-clinical academic project staff alongside representatives from OHE to critique the robustness and evidential value of the research methods.

Re-examining Treatment Effect Heterogeneity Across Multiple Health Outcomes Defining health benefits for a single endpoint can be straightforward but treatment effects may not necessarily be confined to a single outcome.

RCTs generate information on multiple outcomes to inform decision making on the clinical effectiveness, safety and cost-effectiveness of new therapeutics.

Identifying those who benefit from treatment across multiple outcomes is challenging due to the potential heterogeneity in treatment response across differing endpoints whilst simultaneously accounting for multiplicity adjustment.

Deriving more personalised estimates of treatment response across multiple outcomes would provide decision-makers with transparent decision rules on the clinical translation of therapeutics and prioritisation on data collection within enrichment designs. The PhD will re-examine treatment effect h