Choice of Instrument Impacts Healthcare Decisions: Effect of Source of Utility Derived from the Same Patient Population on Cost-Effectiveness

OBJECTIVES: Most HTA guidance recommend deriving health state utility values (HSUVs) for cost-utility analyses from generic instruments such as EQ-5D. However, HSUVs may also be derived using mapping algorithms and disease-specific utility measures. The potential impact of the instrument used to derive HSUVs on healthcare decision-making has been sparsely studied. In this post-hoc analysis, we derive HSUVs for an economic model using different measures collected from the same patients to evaluate a hypothetical treatment for severe asthma and the impact of those HSUVs on cost-effectiveness. METHODS: MUSCA (GSK/ClinicalTrials.gov identifier 200862/NCT02281318) was a 24-week randomised, double-blind, placebo-controlled trial of mepolizumab in severe eosinophilic asthma. EQ-SD-5L, Saint George's Respiratory Questionnaire (SGRQ), Asthma Symptom Utility Index (ASUI), and Asthma Control Questionnaire (ACQ-5) were collected at baseline, week 12, and week 24 visits. HSUVs were derived using the EQ-5D-5L index, EQ-5D-5L cross-walked to EQ-5D-3L, SGRQ mapped to EQ-5D-3L, and ASUI. The cost-effectiveness model was based on the model cited in the NICE evaluation of omalizumab in severe asthma and was replicated and adapted to fit the availability of utility data. The model included a stopping rule whereby patients required a 0.5 ACQ-5 score improvement to continue treatment. The hypothetical therapy was assumed to yield a 20-40% relative reduction in exacerbations, depending on type, and a 40% greater response rate. RESULTS: 452 patients completed all three questionnaires at the week 12 and 24 visits. The estimated costs per QALY were £101,000 (EQ-5D-5L index), £91,000 (EQ-5D-5L crosswalk), £68,000 (mapped SGRQ), and £51,000 (ASUI). CONCLUSIONS:The instrument chosen to derive HSUVs had a considerable impact on cost-effectiveness and may ultimately influence which medicines are made available to patients. Using generic instruments such as EQ-5D in line with HTA guidance for deriving HSUVs could lead to considerably unfavourable estimates of cost-effectiveness compared to other sources.