Modeling SF-6D Health Utilities: Is Bayesian Approach Appropriate?

Area of Science:

• Health Economics
• Preference-Based Health Measures
• Cross-Country Data Analysis

Background:

• Preference-based health measures, such as the SF-6D (Short Form 6-Dimension), are crucial for health economics evaluations.
• Conducting valuation studies in low- and middle-income countries (LMICs) is often cost-prohibitive due to small populations.
• Existing valuation data from other countries offers a potential solution for generating reliable estimates in data-scarce regions.

Purpose of the Study:

• To investigate the feasibility of using existing health valuation data from a high-income country (UK) to inform estimates in a low- and middle-income country (Lebanon).
• To develop a more precise Lebanon SF-6D value set by incorporating UK data as informative priors.
• To assess the utility of this method for improving health valuation in resource-limited settings.

Main Methods:

• Utilized standard gamble techniques to extract health state values from samples in Lebanon (49 states) and the UK (249 states).
• Employed a nonparametric Bayesian model to estimate a Lebanon value set, using the UK SF-6D data as informative priors.
• Compared the resulting value set against one derived solely from Lebanese data using various prediction criteria.

Main Results:

• Incorporating UK data as prior information significantly improved the precision of the Lebanon health valuation estimates compared to using Lebanon data alone.
• The Bayesian model demonstrated that prior information from a larger dataset (UK) effectively enhanced the estimation process.
• All prediction criteria favored the model that utilized the UK data, highlighting the benefit of cross-country data sharing.

Conclusions:

• Existing health valuation datasets can be effectively merged with smaller, country-specific datasets to generate more robust and precise value sets.
• This approach offers a cost-effective strategy for LMICs to develop their own national health valuation tools.
• The findings support the broader application of cross-country data pooling for health economics research, particularly in resource-constrained environments.