Presentation (CTI)

OBJECTIVES: Cost-utility analyses depend on validated quality-of-life measures, but EQ-5D can show ceiling effects in certain symptom states, biasing utilities and responder analyses. In the National Institute for Health and Care Excellence (NICE) appraisal of dupilumab for severe chronic rhinosinusitis with nasal polyps (CRSwNP), the committee acknowledged that EQ-5D inadequately captured disease impact but could not methodologically disregard it. This analysis used patient-level trial data to identify a symptom threshold resolving this concern.
METHODS: Pooled patient-level data from the SINUS-24/-52 phase 3 trials (NCT02912468/NCT02898454) were analysed to relate baseline 22-item sinonasal outcomes test (SNOT-22) score to baseline EQ-5D utility (EQ-5D-5L mapped to EQ-5D-3L, Hernández-Alava) and to responder status (≥8.9-point SNOT-22 plus ≥1-point nasal polyp score improvement). A SNOT-22 ≥50 entry threshold was applied to the NICE base-case population (adults with ≥1 prior sinus surgery) within a lifetime Markov cost-utility model (NHS and Personal Social Services [PSS] perspective; 3.5% discount rate; dupilumab plus established clinical management [ECM] versus ECM alone). Deterministic and probabilistic sensitivity analyses were performed.
RESULTS: Admitting patients with low baseline SNOT-22 produced two artefacts. First, lower SNOT-22 correlated with higher baseline EQ-5D, inflating baseline utility and limiting measurable improvement. Second, patients with lower baseline SNOT-22 could also not achieve the ≥8.9-point responder reduction and were non-responders by construction, inflating non-responder rates. Restricting entry to SNOT-22 ≥50 removed both, leaving a population in which EQ-5D retained discriminative validity and more plausible utilities. In this threshold-defined population, accepted by NICE in TA1134, the committee’s preferred probabilistic ICER was £24,846 per QALY gained, within the range NICE considers cost-effective.
CONCLUSIONS: Patient-level analysis of disease-specific symptom scores can define populations in which generic preference-based instruments retain validity, giving HTA bodies a defensible basis for submissions questioning generic-instrument validity. The approach may extend to other symptom-dominated indications affected by EQ-5D ceiling effects.