Bayesian models for cost-effectiveness analysis in the presence of structural zero costs

Area of Science:

• Health Economics
• Biostatistics
• Statistical Modeling

Background:

• Cost-effectiveness data analysis presents challenges due to complex relationships between clinical benefit and costs.
• Standard statistical assumptions like normality are often violated, especially for skewed cost data.
• The presence of structural zeros in cost data requires specialized modeling approaches.

Purpose of the Study:

• To extend the application of hurdle models to cost-effectiveness data within a Bayesian framework.
• To develop a comprehensive Bayesian specification for analyzing cost-effectiveness data with excess zeros and skewed distributions.
• To provide a practical framework for health economic evaluations using a novel Bayesian approach.

Main Methods:

• Development of a full Bayesian specification incorporating three components: a model for the probability of null costs, a marginal model for costs, and a conditional model for effectiveness.
• Application of hurdle models to account for structural zeros in cost data.
• Utilizing a working example to illustrate the model's features and application.

Main Results:

• The proposed Bayesian hurdle model effectively handles the complexities of cost-effectiveness data, including skewed cost distributions and structural zeros.
• The model provides a unified framework for analyzing both cost and effectiveness measures simultaneously.
• Demonstrated the practical utility and features of the model through a case study.

Conclusions:

• The Bayesian hurdle model offers a robust and flexible approach for cost-effectiveness analysis, particularly when dealing with non-standard data characteristics.
• This methodology enhances the accuracy and reliability of health economic evaluations.
• The study contributes a valuable tool for researchers and practitioners in health economics and biostatistics.