Good Practices on the Evaluation of Surrogate Endpoints for Health Technology Assessment Decision Making from the ISPOR Taskforce

Surrogate endpoints now underpin around 60% of new FDA drug and biologic approvals of the past two decades, yet their validation is often weak, leaving health technology assessment agencies and payers to decide under considerable uncertainty. Oriana Ciani and Dawn Lee, co-authors of the ISPOR Good Practices Task Force report, present its recommendations: the three levels of evidence for surrogacy (biological plausibility, individual-level association and trial-level association) and the statistical methods for assessing them, including multivariate meta-analysis accounting for within-study correlation, Bayesian borrowing across treatment classes and indications, and the role of real-world data. The second part addresses building surrogate relationships into health economic models, judging plausibility, and characterising parameter and structural uncertainty through probabilistic, scenario, threshold and value-of-information analyses, followed by discussion.