OBJECTIVES: Respiratory syncytial virus (RSV) is associated with high clinical and economic burden, especially in infants under one. Palivizumab and nirsevimab are licensed in this population for RSV prevention. Estimating the cost-effectiveness of these options is critical to inform reimbursement decisions. However, obtaining accurate numbers of RSV-driven hospitalizations and associated mortality – key drivers of cost effectiveness – is challenging due to coding practices and the impact of the COVID-19 pandemic.

METHODS: Structured expert elicitation (SEE) is recommended by the UK National Institute for Health and Care Excellence to address uncertainty where empirical evidence is lacking. Several SEE frameworks exist, using different approaches to the elicitation and aggregation (e.g., behavioral and mathematical) of expert judgements. We used the Sheffield Elicitation Framework (SHELF), given its focus on expert discussion and consensus building, to elicit judgements from clinical experts from different backgrounds on the number of annual UK RSV-related hospitalizations across several subgroups of interest.

RESULTS: Individual-level probability distributions were successfully elicited. The subsequent consensus workshop was critical for experts to share experiences and rationales for why the number of RSV-related hospitalizations may differ from estimates in published literature. Key points of discussion included that: not all RSV cases are laboratory-confirmed, as diagnosis does not alter treatment; false-negatives are more common than false-positives; literature was outdated; and, since the COVID pandemic, more sensitive tests (PCR) have become available and routine diagnostic testing has increased. After the workshop, group-level probability distributions were generated considering the collaboratively established multiple facets of uncertainty.

CONCLUSIONS: While mathematical aggregation is an efficient method to address uncertainty of a large group of experts, behavioral aggregation using SHELF can be valuable to explore more complex quantities that require between-expert interaction with a smaller group of experts when it is expected that the knowledge of the whole is greater than its sum.