Modelling a preference-based index for EQ-5D-3L and EQ-5D-3L + Sleep using a Bayesian framework.

Area of Science:

• Health economics
• Biostatistics
• Psychometrics

Background:

• Frequentist methods are traditionally used for health state valuation.
• Emerging research explores Bayesian approaches for modeling health valuation data.
• This study investigates Bayesian modeling for EQ-5D-3L and EQ-5D-3L+Sleep.

Purpose of the Study:

• To present and demonstrate the superiority of a Bayesian framework for modeling EQ-5D-3L and EQ-5D-3L+Sleep health state valuation data.
• To compare Bayesian models against conventional frequentist methods.

Main Methods:

• A valuation study involving 160 participants using the time trade-off technique for 18 EQ-5D-3L and 18 EQ-5D-3L+Sleep health states.
• Development of three Bayesian models (linear regression, random effect, random effect with covariates) using Markov chain Monte Carlo simulation.
• Model performance evaluation based on predictive accuracy (mean predictions, RMSE) and model fit (DIC).

Main Results:

• The random effects with covariates model demonstrated the best performance for both EQ-5D-3L and EQ-5D-3L+Sleep.
• Achieved superior predictive performance (lower RMSE and DIC) compared to frequentist models.
• Bayesian models provided more accurate predictions of observed health state values.

Conclusions:

• Bayesian methods offer an improved modeling approach for EQ-5D-3L valuation data, including the EQ-5D-3L+Sleep extension.
• This framework provides greater flexibility in characterizing the uncertainty associated with health state valuations.
• The findings support the adoption of Bayesian methods for more robust health economic evaluations.