Abstract
Accumulated evidences are in support of the critical role of migratory wild birds in highly pathogenic avian influenza virus (HAPIV) spread and evolution. An effective surveillance strategy to study HPAIV dispersal in wild birds and identify critical interfaces between wild birds and poultry on the landscape for potential interspecies transmission and virus evolution will be essential. This study integrates conditional likelihood with epidemiological research designs to investigate the risk of poultry farm outbreaks due to the introduction of HPAIV by migratory birds from the Taiwan citizen scientist dataset. The prediction accuracy of poultry farm outbreaks was evaluated by employing the global risk score (GRS) of multiple species of wild birds under different significant level. Although 11 wild bird species demonstrated significant associations (P < 0.05) with HPAIV poultry farm outbreaks, the bird species with P < 0.1 gave the optimal area under the receiver operating characteristic curve. Without prior adjustment of environmental factors, the posterior adjusted likelihood results suggested that the agricultural wetland yields notably strong positive odds ratio estimates for all species. The GRS under adjustment of environmental factors, a linear combination of the 27 migratory species and 27 resident species, was 0.320, yielding a sensitivity of 0.81 and specificity of 0.77. The framework of GRS developed in this study can be applied to different countries for enhanced surveillance of HPAIV.