Statistical methods for the geographical analysis of rare diseases

Area of Science:

• Epidemiology and Biostatistics
• Spatial Analysis
• Disease Surveillance

Background:

• Accurate detection of disease clusters is crucial for public health.
• Traditional methods may not fully capture spatial variations or account for rare disease data.
• Advanced statistical techniques are needed for precise disease cluster identification.

Purpose of the Study:

• To provide a comprehensive summary of methods for disease cluster detection.
• To introduce spatial scan statistics and their extensions for covariate inclusion.
• To illustrate the application of these methods using real-world disease data.

Main Methods:

• Summary of relative risk estimation methods with spatial smoothing.
• Discussion of spatial autocorrelation and general clustering tests.
• Introduction to spatial scan statistics and Generalized Linear Models (GLMs) with covariates.
• Application of zero-inflated models for rare disease analysis.

Main Results:

• Demonstrated smoothed relative risk estimates accounting for spatial variation.
• Presented scan methods for locating disease clusters effectively.
• Showcased the utility of GLMs and zero-inflated models in cluster detection.
• Applied methods to Systemic Lupus Erythematosus and brain cancer data.

Conclusions:

• The summarized methods offer robust approaches for disease cluster detection.
• Spatial scan statistics and GLMs enhance the ability to identify disease hotspots.
• Zero-inflated models are valuable for analyzing rare diseases.
• These techniques are applicable to various public health surveillance scenarios.