Assessing spatial disparities: a Bayesian linear regression approach

Area of Science:

• Epidemiology
• Biostatistics
• Spatial Analysis

Background:

• Detecting spatial health disparities in regionally aggregated data is crucial for public health.
• Spatial dependence in health outcomes complicates the identification of significant disparities.
• Defining and inferring spatial disparities statistically presents significant challenges.

Purpose of the Study:

• To develop a robust statistical framework for detecting spatial health disparities.
• To enhance the Bayesian linear regression model with spatial autoregression for improved analysis.
• To enable model-based detection and delineation of boundaries between regions with disparate health outcomes.

Main Methods:

• Enrichment of the Bayesian linear regression framework with spatial autoregression.
• Development of analytical tractability for accelerated computation.
• Application to US county-level lung cancer mortality rates from the Institute of Health Metrics and Evaluation (IHME).

Main Results:

• The proposed method allows for model-based detection of spatial disparities.
• Significant computational acceleration was achieved through derived analytical tractability.
• Simulation experiments on a US county map demonstrated the method's effectiveness.

Conclusions:

• The enhanced Bayesian regression model provides a statistically robust approach to identifying spatial health disparities.
• The method facilitates the delineation of boundaries between regions with differing health outcomes.
• This approach offers effective and computationally efficient analysis of spatial health data.