Preliminary evaluation of diagnostic tests for avian influenza using the Markov Chain Monte Carlo (MCMC) method in an emergency surveillance

Abstract

In June 2005, an outbreak of avian influenza (AI) caused by a low pathogenic H5N2 virus was identified in Japan. A serological surveillance was conducted because the infected chickens did not show any clinical signs. The Markov Chain Monte Carlo Method was used to evaluate the performances of serological HI and AGP tests because there was not enough time when the surveillance was initiated to conduct a test evaluation. The sensitivity of the AGP test (0.67) was lower than that of the HI test (0.99), while the specificities were high for both tests (0.96 for AGP and 0.90 for HI). Based on the low sensitivity of the AGP test, the HI test was used for primary screening in later stages of the epidemic.