Spatiotemporal Modelling and Risk Evaluation of Avian Influenza Spreading in a Multi-host System in Denmark

PhD Student: Yangfan Liu
Email: yali@sund.ku

THE PROJECT
Avian influenza (AI) is a global health threat triggered by avian influenza viruses (AIVs), which can be differentiated into highly pathogenic avian influenza viruses (HPAIVs) and low pathogenic avian influenza viruses (LPAIVs). Wild birds and poultry are the mainly affected species by AIVs regarding population declines and economic losses. Denmark is located in flyways and fly routes of abundant migratory birds, and numerous Anseriformes birds, for example, geese, swans, and dabbling ducks are active species engaging in latitudinal migration moves and can be found on Danish ground.

Since late 2020, a series of outbreaks in Danish poultry caused by HPAIV belonging to a novel clade 2.3.4.4b occurred, along with a drastic surge of HPAIV-positive wild birds, indicating the start of HPAI epidemic seasons. Previous studies analysed Danish AI surveillance data before 2020 and provided epidemiological insights into landscape effects and molecular characteristics, but the novel epidemic situation requires evaluation of the current risks and gaining a better understanding of HPAI transmission among and between wild birds and poultry. Moreover, significant gaps remain in monitoring and control strategies facing the new AI situation. Lastly, epidemic preparedness lacks a system for predicting HPAI dynamics in Denmark, including multiple host species, spatial heterogeneity, and seasonal variability.

THE PURPOSE
The purpose was to develop and implement a system to predict AIV occurrence in wild birds and poultry farms in Denmark. Such a system includes determining risk factors for disease situations in wild birds and poultry, developing a mathematical transmission model for wild birds, and quantifying the spill-over risk and association of wild birds with poultry outbreaks.

THE RESULT
The significant negative effect of the distance to wetlands on the probability of disease occurrence in wild birds (odds ratio: 0.95, p = 0.012) were identified, suggesting a decrease in AIV occurrence probability by 5% for every one km distance from wetlands. Another significant risk factor was the order of wild birds, and pairwise comparisons reported that the order Anseriformes (e.g., dabbling ducks and geese) was of higher concern regarding AIV detection.

The discrete-time stochastic model was established to elucidate the transmission mechanism of HPAI in Danish wild birds. By calibrating the model using disease surveillance data of the 2020/21 season, the model predicted a time-dependent HPAIV risk map. It revealed consistent spatial heterogeneity with a high risk in coastal and water areas and a low risk in urban and inland. Barnacle goose (Branta leucopsis) was the most affected species according to the model. The model was further developed for spill-over risk mapping for poultry farms. The result showed Southern Zealand was a hot spot regarding potential HPAI spill-over by wild birds.