Mapping the EQ-5D index from the cystic fibrosis questionnaire-revised using multiple modelling approaches

Area of Science:

• Health Economics
• Patient-Reported Outcomes
• Biostatistics

Background:

• Estimating health utility values is crucial for economic evaluations in cystic fibrosis (CF).
• The Cystic Fibrosis Questionnaire-Revised (CFQ-R) is widely used, but direct EQ-5D utility values are not always available.
• A mapping algorithm can bridge this gap, enabling utility estimation from CFQ-R data.

Purpose of the Study:

• To develop and validate a mapping algorithm for estimating EQ-5D utility values using CFQ-R data.
• To compare the performance of different regression models (OLS, Tobit, TPM) at item and domain levels.
• To identify the most accurate and reliable mapping functions for use in economic evaluations.

Main Methods:

• A cross-sectional survey of 401 adults with CF in the UK collected CFQ-R, EQ-5D, and demographic data.
• Eight regression models were evaluated using ordinary least squares (OLS), Tobit, and two-part models (TPM).
• Model performance was assessed using intraclass correlations, information criteria (BIC, AIC), and root mean square error (RMSE).

Main Results:

• The best item-level model included all CFQ-R items; the best domain-level model included specific CFQ-R domains and squared terms.
• The item-level TPM with age and gender covariates showed the best within-sample performance.
• OLS and TPM domain models with squared terms demonstrated superior out-of-sample predictive performance.

Conclusions:

• Developed mapping functions, particularly domain-level models, demonstrate acceptable predictive performance, especially in out-of-sample validation.
• These algorithms can reliably estimate EQ-5D utility values from CFQ-R data for economic evaluations of CF interventions.
• Further validation in an independent sample is recommended to confirm model generalizability.