Disease mapping in veterinary epidemiology: a Bayesian geostatistical approach

Area of Science:

• Veterinary Epidemiology
• Geostatistics
• Spatial Statistics

Background:

• Veterinary epidemiology benefits from model-based geostatistics and Bayesian approaches for analyzing point data from well-designed studies.
• Urban epidemiological surveillance requires robust methods to assess disease risk in complex environments.

Purpose of the Study:

• To apply and compare Bayesian geostatistical and hierarchical Bayesian models for spatial risk prediction of dog parasite infection.
• To identify high-risk areas for zoonotic parasitic diseases in an urban setting using a two-stage sampling design.

Main Methods:

• Bayesian Gaussian spatial exponential models and Bayesian kriging were employed to predict continuous risk surfaces.
• A two-stage sampling design involving transects was utilized for data collection in Naples.
• Hierarchical Bayesian models on areal data were used for comparative analysis.

Main Results:

• Bayesian geostatistical and hierarchical Bayesian models yielded consistent results.
• The Bayesian geostatistical approach demonstrated higher accuracy in identifying areas at risk.
• Increased risk areas were predominantly found at city borders with mixed wild and domestic dog populations.

Conclusions:

• Bayesian geostatistics is a valuable tool for accurate spatial risk assessment in veterinary epidemiology.
• Urban fringe areas with dog population mixing pose a higher risk for zoonotic parasitic diseases.
• Effective surveillance strategies should consider spatial patterns and potential interfaces between domestic and wild animal populations.