Presentation (CTI)

OBJECTIVES: The National Institute for Health and Care Excellence (NICE) 2022 Manual and TSD26 recommend structured expert elicitation (SEE) when expert judgement is needed, as it can reduce bias and characterize uncertainty. We assessed how frequently SEE was used in NICE technology appraisals (TAs) and highly specialized technologies (HSTs), for what purposes, and how NICE Committees viewed them.
METHODS: On 1/6/26, using TAcrawlR, we searched NICE guidance documents for SEE-‍related terms and extracted elicitation approach, quantities elicited, and Committee discussions.
RESULTS: We identified 24 TAs and five HSTs that used SEE - all except two published within the past 5 years. Eight were in oncology and 13 were in rare or ultra-rare conditions. Methods included modified Delphi (12), SHELF (8), STEER (4), MRC (1), and IDEA (1); three records did not report any information on methods, and it was unclear whether these were formal SEEs. Expert numbers ranged from 3-‍21 (not reported in 10 records); most SEEs (13) enrolled 5-‍10 experts, three enrolled < 5 experts, and three enrolled > 10 experts. Elicited quantities included overall survival (OS)/mortality (12), other clinical inputs (21), and healthcare resource use (10). Committee discussion of SEE appeared in seven TAs and one HST guidance document. SEE was accepted (fully or for selected inputs) in 6/8. In TA891, SEE was not accepted due to uncertainty in SEE methods and results, and their use within the model - informing efficacy inputs in one treatment arm only. In TA1115, the Committee preferred the EAG’s OS curves, despite the company stating SEE support for their OS extrapolation; SEE was not discussed further by the Committee.
CONCLUSIONS: The use of SEE has increased in recent years. SEE can strengthen modeling when evidence is uncertain, but careful design, transparent reporting, and appropriate application within the model are key to maximizing NICE Committee acceptability.