Sample size estimation to substantiate freedom from disease for clustered binary data with a specific risk profile.

Area of Science:

• Veterinary epidemiology
• Statistical modeling in animal health
• Disease surveillance and control

Background:

• Disease outbreaks often exhibit clustering within animal groups (herds/flocks).
• Existing sample size formulas rely on the intra-cluster correlation coefficient (ICC), an average measure of heterogeneity.
• ICC-based estimates may lack precision for subgroups with distinct heterogeneity levels.

Purpose of the Study:

• To propose the variance partition coefficient (VPC) as a more accurate measure for sample size estimation in clustered disease data.
• To introduce a VPC-based predictive simulation method for substantiating freedom from disease.
• To optimize resource allocation by tailoring sample size estimates to specific subgroup heterogeneities.

Main Methods:

• Utilizing the variance partition coefficient (VPC) to quantify disease clustering within risk-profiled groups.
• Developing a VPC-based predictive simulation approach for sample size determination.
• Applying the VPC method to real-world case studies involving Mycobacterium avium subsp. paratuberculosis in Danish dairy cattle and Salmonella cross-contamination in Greek pork slaughterhouses.

Main Results:

• The VPC offers a more nuanced measure of heterogeneity compared to the ICC for specific subgroups.
• VPC-based sample size estimation allows for optimized resource allocation, particularly in heterogeneous populations.
• The proposed simulation method provides a robust framework for assessing disease freedom.

Conclusions:

• The variance partition coefficient (VPC) enhances the accuracy of sample size calculations for clustered disease data in veterinary epidemiology.
• Tailoring sample size estimates using VPC improves study precision, power, and resource efficiency.
• The VPC-based simulation method is a valuable tool for disease eradication programs and risk assessment.