Estimating utility data for patient symptom severity in chronic spontaneous urticaria
OBJECTIVES: To obtain utility estimates suitable for use in economic models for chronic spontaneous (idiopathic) urticaria (CSU).
METHODS: Patient-level data from three randomised clinical trials: ASTERIA I, ASTERIA II, and GLACIAL were analysed. Health states were derived from Urticaria Activity Score (UAS7), a patient-completed diary of signs and symptoms which calculates an average daily score over 7 days. Higher score means more severe symptoms. UAS7 scores for the health states were: Urticaria-free: 0; Well-controlled urticaria: 1-6; Mild urticaria: 7-15; Moderate urticaria: 16-27; Severe urticaria: 28-42. Mean EQ-5D utilities were calculated for each health state. Individual trial analyses showed inconsistent utilities across the UAS7 health states due to small subsample sizes. A mixed model was used to predict EQ-5D according to UAS7 health states in a pooled dataset containing all treatment arms and time-points from the three trials. The predictor variable was UAS7 health state and the dependent variable was EQ-5D utility. Fixed/random effects for trial and patient were included and the following covariates: UAS7 health state at baseline (Moderate or Severe), presence of angioedema at baseline and during follow-up, duration of CSU, number of previous CSU medications, and gender of the patient. A parsimonious model was selected using the approach of backwards elimination; UAS7 health state was forced into the model. The validity of pooling trials was considered through visual comparisons and interaction terms.
RESULTS: There was a consistent improvement in EQ-5D utilities as severity of urticaria improved. Mean utilities at Week 12 ranged from 0.712 in patients with severe urticaria to 0.897 in patients who were urticaria-free. Sensitivity analysis confirmed the robustness of results.
CONCLUSIONS: The results suggest that EQ-5D utility score increased with decreasing severity of urticaria. EQ-5D utility scores allow the comparison of HRQoL across diseases by calculating QALYs in economic models.