Strategies for two-stage sampling designs for estimating herd-level prevalence.

Area of Science:

• Veterinary epidemiology
• Biostatistics
• Disease surveillance

Background:

• Accurate disease prevalence estimation in animal herds is crucial for effective control strategies.
• Diagnostic tests often exhibit imperfect sensitivity and specificity, complicating herd-level assessments.
• Intracluster correlation (ICC) within herds can significantly influence test result interpretation.

Purpose of the Study:

• To develop a herd-level sample-size formula accounting for imperfect diagnostic tests.
• To investigate the impact of ICC on herd-level sensitivity and specificity.
• To provide a framework for optimizing sampling strategies in disease surveys.

Main Methods:

• Development of a novel herd-level sample-size formula.
• Utilization of Monte Carlo simulations to assess ICC effects.
• Application to a real-world scenario of ovine progressive pneumonia prevalence estimation.

Main Results:

• Herd-level sensitivity is highly influenced by ICC at low prevalence, potentially leading to false positives.
• The impact of ICC on herd-level sensitivity is reduced at higher prevalence levels.
• The proposed formula enables balancing herd and animal sample sizes by adjusting test characteristics.

Conclusions:

• The developed formula provides a robust method for sample-size calculation in imperfect diagnostic scenarios.
• Understanding ICC is vital for accurate disease surveillance, particularly in low-prevalence populations.
• This approach offers flexibility in designing efficient and cost-effective animal disease surveys.