Presentation (CTI)

OBJECTIVES: Accounting for heterogeneity is critical when combining different sources of information; this is important especially in the accurate estimation of cost-effectiveness and subsequent health technology assessment (HTA) decision-making (Murphy et al. 2021). The UK’s National Institute for Health and Care Excellence (NICE) also now recommends that “when heterogeneity between groups within a population is a concern, any assumptions about homogeneity or heterogeneity⋯ must be clearly presented, tested and fully explored” (NICE 2020).
METHODS: The traditional use of Bayesian Hierarchical Models (BHM) is in the context of basket trials. BHMs can also be used to extrapolate time-to-event outcomes to inform cost-effectiveness analysis for HTA payer dossiers as in McCarthy et al. (McCarty et al. 2024). This work will examine a method to use BHMs in simulated data to extrapolate time-to-event outcomes in parametric distributions including the Weibull, Exponential, Lognormal, and Gamma accelerated failure time (AFT) parametric distributions.​ Survival functions will be estimated via the cumulative distribution function extrapolated for a specific follow-up time; variance-covariance matrices will be estimated, and survival and hazard functions plotted.
RESULTS: BHMs allow the treatment effect in any specific basket to be informed by the effects in other incorporated baskets, thereby: a) maximising the information available, b) increasing the precision of estimates and c) reducing the probability of obtaining unreliable estimates in baskets with only a few patients (Murphy et al. 2021)
CONCLUSIONS: BHMs are a valuable tool especially for sparse data in oncology, informing cost-effectiveness modelling.