Session (CTI)

Issue: A fundamental precept of medical practice is that treatments should align with patients’ needs, supported by adequate information to guide decision-making. Patient-reported outcome (PRO) data collected during clinical trials provides critical insights into patients’ experiences, allowing them a voice in the evaluation of treatments. Given the effort required to collect and report PROs—particularly as illness progresses—it is important to recognise this burden, which may lead to missing data and potential bias. It is consequently incumbent on analysts and assessors to use PRO data effectively while accounting for missingness. Overview: Taking account of the importance of PRO data, this issue panel will discuss how to appropriately analyse PRO data accounting for missing data and the need for guidelines. Although analytic methods for handling missing data have advanced, guidance on their application in health technology assessments (HTAs)—including the EU Joint Clinical Assessment (JCA)—remains limited. It will address technical and practical challenges, draw on lessons from previous HTA assessments, and consider implications for the JCA. The audience will be invited to share perspectives on their experience of analysing PRO data and the necessity and scope for further guidelines around the appropriate analysis of missing data. Miranda Cooper will outline challenges in handling missing PRO data from a health technology developer’s perspective (10-minutes). Professor Neil Hawkins will review recent NICE oncology appraisals, focusing on approaches used to handle missing PRO data, HTA agency feedback, and outcomes from reviewers and committees (15-minutes). Professor Olivier Chassany will highlight varying evidence requirements for PROs across major European HTA bodies in oncology, underscoring the need for harmonisation for HTA agencies and applicants (15-minutes). Professor Gianluca Baio will present methodological approaches to handle situations where substantial amount of missingness is present, with particular relevance to PRO data accounting for the longitudinal nature of the data (15-minutes).