HAS THE INTRODUCTION OF NICE TSD26 MEANINGFULLY REDUCED DECISION UNCERTAINTY AROUND LONG-TERM SURVIVAL ESTIMATES?
Author(s)
Alice Dean, PhD1, Ben Thackray, MRes2, Alice Reading, MSc3.
1Costello Medical, Bristol, United Kingdom, 2Costello Medical, London, United Kingdom, 3Costello Medical, Cambridge, United Kingdom.
1Costello Medical, Bristol, United Kingdom, 2Costello Medical, London, United Kingdom, 3Costello Medical, Cambridge, United Kingdom.
OBJECTIVES: The NICE Decision Support Unit Technical Support Document (TSD) 26 details robust methods for structured expert elicitation (SEE) of long-term survival estimates. We investigated the extent of adoption of TSD26 methods in NICE technology appraisals (TAs), and whether adoption has meaningfully reduced decision uncertainty relating to Company-selected survival extrapolations.
METHODS: NICE TAs reporting on cost-utility analyses published since TSD26 (March 2025) and before June 2026 in life-limiting indications were identified. Methods used to elicit or validate survival estimates were extracted and alignment to TSD26 evaluated. Uncertainties raised by External Assessment Groups (EAGs) and Committees regarding survival extrapolations were subsequently extracted.
RESULTS: Of 15 published TAs identified, 10 reported clinical expert input into survival extrapolations. Of these, 3/10 reported SEE methods, with only 1/10 explicitly adopting TSD26 methodology: TA1153. In TA1153, the EAG described SEE methods as robust and comprehensive, ultimately accepting 3/4 Company-selected base case survival extrapolations. Survival extrapolations in two TAs that utilised SEE methods without explicitly adopting TSD26 methodology received more criticism. However, this predominantly related to the clinical evidence base, including immature survival data, rather than SEE methodology. Among seven TAs which conducted pragmatic validation of survival curves, validation methods were only noted to significantly contribute to uncertainty in 1/7 TAs, whereas immature trial data (3/7) and modelling approaches (3/7) were more commonly critiqued.
CONCLUSIONS: Initial analyses suggest limited uptake of TSD26 methods, likely reflecting the substantial time and monetary investment necessary for implementation. Alignment with TSD26 methodology was received favourably. However, as the extrapolations in TA1153 were based on mature survival data, the extent to which alignment to TSD26 reduces decision uncertainty around extrapolations remains to be conclusively demonstrated. Evidence from the remaining TAs utilising SEE methods indicate that SEE can reduce but not eradicate extrapolation uncertainty, which remains dependent on the underlying evidence base.
METHODS: NICE TAs reporting on cost-utility analyses published since TSD26 (March 2025) and before June 2026 in life-limiting indications were identified. Methods used to elicit or validate survival estimates were extracted and alignment to TSD26 evaluated. Uncertainties raised by External Assessment Groups (EAGs) and Committees regarding survival extrapolations were subsequently extracted.
RESULTS: Of 15 published TAs identified, 10 reported clinical expert input into survival extrapolations. Of these, 3/10 reported SEE methods, with only 1/10 explicitly adopting TSD26 methodology: TA1153. In TA1153, the EAG described SEE methods as robust and comprehensive, ultimately accepting 3/4 Company-selected base case survival extrapolations. Survival extrapolations in two TAs that utilised SEE methods without explicitly adopting TSD26 methodology received more criticism. However, this predominantly related to the clinical evidence base, including immature survival data, rather than SEE methodology. Among seven TAs which conducted pragmatic validation of survival curves, validation methods were only noted to significantly contribute to uncertainty in 1/7 TAs, whereas immature trial data (3/7) and modelling approaches (3/7) were more commonly critiqued.
CONCLUSIONS: Initial analyses suggest limited uptake of TSD26 methods, likely reflecting the substantial time and monetary investment necessary for implementation. Alignment with TSD26 methodology was received favourably. However, as the extrapolations in TA1153 were based on mature survival data, the extent to which alignment to TSD26 reduces decision uncertainty around extrapolations remains to be conclusively demonstrated. Evidence from the remaining TAs utilising SEE methods indicate that SEE can reduce but not eradicate extrapolation uncertainty, which remains dependent on the underlying evidence base.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
EE66
Topic
Clinical Outcomes, Economic Evaluation, Health Technology Assessment
Topic Subcategory
Trial-Based Economic Evaluation
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, Oncology