Estimation of sample size required to detect an outbreak of highly pathogenic avian influenza in a poultry farm during emergency surveillance for secondarily infected farms.

Abstract

The Asian lineage of H5 highly pathogenic avian influenza (HPAI) virus is causing a large number of outbreaks globally. This study was conducted to evaluate the current Japanese emergency surveillance policy of HPAI in poultry farms. Under the current regulation, five chickens, including three dead chickens if available, are sampled from neighbor non-notified poultry farms of an outbreak farm. A susceptible-exposed-infectious-death mathematical model describing HPAI transmission within a broiler farm was developed. Using the model, sample sizes of live and dead chickens, respectively, to detect an HPAI outbreak of the virus with high, moderate, and low transmission coefficients and progression rates from infectious to death, at 9, 14, and 24 days after starting the outbreak, with detection sensitivities from 50% to 99% were calculated using deterministic models. Sampling of five live chickens allowed outbreak detection with a detection sensitivity of 95% only 24 days post-infection for a virus with high transmission and a moderate or low progression rate from infectious to death. Cumulative mortality increased more than 10% at 9 days post-infection for a virus with a high transmission rate. However, sampling of three sick or dead chickens was successful in detecting an outbreak only on or after 14 days post-infection for such a virus. The results suggested that the current sampling framework for HPAI emergency surveillance does not provide a high detection sensitivity in most scenarios, especially in the early stage of an HPAI outbreak. It is important to maintain a high alert in detection and reporting from poultry farms, and the design of emergency surveillance requires careful discussions.