ARCHITECTURE OF RARE-DISEASE MODELS: STRUCTURAL CHOICES AND CLINICAL TRIAL ENDPOINT USE IN NICE HST EVALUATIONS
Author(s)
Sandra Milev, MSc1, Benjamin White, MSc1, Monique Martin, MBA, MSc, PharmD2;
1Red Nucleus, HEOR, Yardley, PA, USA, 2Red Nucleus, HEOR, London, United Kingdom
1Red Nucleus, HEOR, Yardley, PA, USA, 2Red Nucleus, HEOR, London, United Kingdom
OBJECTIVES: To characterize economic model structures in NICE HST evaluations and assess how primary clinical trial endpoints were incorporated into these models.
METHODS: All NICE HST appraisals (HST1-HST33) were reviewed. Extracted information included model type and approach and NICE commentary on appropriateness of model structure and incorporation of clinical endpoints. Quantitative summaries were generated across model categories.
RESULTS: Among 31 appraisals, 30 used cohort-based models and 1 used individual patient simulation. Overall, 69% used Markov cohort, 13% used semi-Markov, 9% used partitioned-survival structure, and 9% used combined decision-tree + Markov approaches; among them 1 DALY-based and 3 milestone-based models were identified. Health-state complexity ranged from 2 to 12 alive states, with a mean of 6, and ~70% of models defining 4-8 alive states. Neuromuscular and neurodevelopmental indications, as well as multisystem metabolic diseases with validated progression scales, used the largest number of states, whereas conditions with binary clinical pathways or sparse natural-history data used the fewest. NICE generally accepted structural choices. Explicit structural concerns were raised in only 4 of 31 appraisals, typically when the model failed to reflect clinically meaningful functional pathways. Use of primary clinical trial endpoints varied substantially: 14 models (45%) used the primary endpoint directly to define transitions or health states, while 17 models (55%) did not, instead relying on clinically meaningful disease stages or functional outcomes more suitable for long-term modelling. Surrogate endpoints were often accepted but consistently characterized as high uncertainty due to limited validation.
CONCLUSIONS: NICE demonstrates substantial flexibility in model-structure selection but consistently prioritizes clinical relevance. Primary trial endpoints are frequently unsuitable as structural drivers, reinforcing the need for health-state structures aligned with real-world functional status and long-term disease trajectories, and underscoring the importance of incorporating HTA-relevant, meaningful outcomes at trial design to maximize European launch success.
METHODS: All NICE HST appraisals (HST1-HST33) were reviewed. Extracted information included model type and approach and NICE commentary on appropriateness of model structure and incorporation of clinical endpoints. Quantitative summaries were generated across model categories.
RESULTS: Among 31 appraisals, 30 used cohort-based models and 1 used individual patient simulation. Overall, 69% used Markov cohort, 13% used semi-Markov, 9% used partitioned-survival structure, and 9% used combined decision-tree + Markov approaches; among them 1 DALY-based and 3 milestone-based models were identified. Health-state complexity ranged from 2 to 12 alive states, with a mean of 6, and ~70% of models defining 4-8 alive states. Neuromuscular and neurodevelopmental indications, as well as multisystem metabolic diseases with validated progression scales, used the largest number of states, whereas conditions with binary clinical pathways or sparse natural-history data used the fewest. NICE generally accepted structural choices. Explicit structural concerns were raised in only 4 of 31 appraisals, typically when the model failed to reflect clinically meaningful functional pathways. Use of primary clinical trial endpoints varied substantially: 14 models (45%) used the primary endpoint directly to define transitions or health states, while 17 models (55%) did not, instead relying on clinically meaningful disease stages or functional outcomes more suitable for long-term modelling. Surrogate endpoints were often accepted but consistently characterized as high uncertainty due to limited validation.
CONCLUSIONS: NICE demonstrates substantial flexibility in model-structure selection but consistently prioritizes clinical relevance. Primary trial endpoints are frequently unsuitable as structural drivers, reinforcing the need for health-state structures aligned with real-world functional status and long-term disease trajectories, and underscoring the importance of incorporating HTA-relevant, meaningful outcomes at trial design to maximize European launch success.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
EE383
Topic
Economic Evaluation
Disease
SDC: Rare & Orphan Diseases