Presentation (CTI)

OBJECTIVES: Cost-effectiveness analyses (CEAs) for chronic diseases often rely on clinical trial data with limited follow-up, needs extrapolation of outcomes and costs over longer time horizons. This extrapolation introduces significant uncertainty, which is typically explored through probabilistic sensitivity analysis (PSA) using often simple distributional assumptions. Real-world evidence (RWE) offers an opportunity to reduce this extrapolation uncertainty and enhance the robustness of CEA for chronic conditions. This study aims to find methodologies for integrating real-world long-term outcome data into the PSA framework of CEA for chronic diseases, thereby reducing extrapolation uncertainty and providing more reliable estimates of cost-effectiveness.
METHODS: There are several methods of incorporating RWE into PSA such as informing parameter distributions with long-term RWE, externally validating model extrapolations against RWE to potentially revise model structure or parameters, developing RWE-driven alternative long-term scenarios and quantifying their uncertainty within the PSA, and employing Bayesian updating to refine prior parameter distributions using RWE. These approaches impacting the ICER, cost-effectiveness acceptability curves, and value of information analyses.
RESULTS: The integration of long-term real-world outcomes into PSA is expected to give more realistic and potentially low uncertainty ranges for key parameters, enhance the validity and credibility of long-term cost-effectiveness projections, and generate more informed cost-effectiveness acceptability curves that reflect real-world variability.
CONCLUSIONS: The crucial step towards reducing extrapolation uncertainty and escalating the reliability of cost-effectiveness analyses for chronic diseases is to integrate real-world long-term outcome data into PSA. The methodologies explored in this study offer practical approaches for the richness of RWE to provide more robust and relevant evidence for healthcare decision-making, particularly in the context of long-term value assessment.