Definiëring van waterrijke gebieden in Nederland in relatie tot vogelgriepbesmettingsrisico bij commerciële pluimveebedrijven bij commerciële pluimveebedrijven in Nederland

Summary

Recent WBVR research has revealed various risk factors that influence the risk of infection with highly pathogenic avian influenza (HPAI) in commercial poultry, such as (combinations of) farm types and landscape variables. In various policy processes, such as the intensification plan for bird flu prevention and the associated structuring choice for animal diseases and zoonoses within the National Program for Rural Areas (NPLG), attention is paid to these risk factors in relation to HPAI. An important landscape variable is the proximity and amount of water around poultry farms. Water in the form of ditches, canals, lakes, etc. is a stable factor, i.e. a factor that changes little or nothing in the landscape over time. In order to take this risk factor into account, it is necessary to be able to determine when a location with a poultry farm is rich in water. For that, two models have been developed, a basic model based only on water surface area in a radius of 500 m around a poultry farm and a model that combines water surface and forest-and-tree surface areas in a radius of 500 m around a poultry farm. these models can be used in a practical and unambiguous way to classify the highly pathogenic avian influenza (HPAI) infection risk of poultry farms. It should be noted that the much more comprehensive models with more variables (but not all stable over time) as previously reported by Gonzales et al. (2023), perform better (combination of sensitivity and specificity) than the “simple” models developed here. Different decision trees are presented that provide an overview of the performance in terms of sensitivity and specificity to classify the HPAI-infection risk by a specific choice of a cut-off value for the water surface area and/or the forest and trees area in a radius of 500 m around a poultry farm. The basic model (water alone) has an optimal combination (largest sum of sensitivity and specificity) of 58% sensitivity and 70% specificity at a water surface cut-off value of ≥ 3.2 ha. A higher desired specificity is accompanied by a lower sensitivity by choosing a higher cut-off value for the water surface. The model in which water and forest-and-trees are combined has an optimal combination of 44% sensitivity and 84% specificity with a water surface cut-off value of ≥ 3.2 ha and a forest and tree surface cut-off value of < 0.1 ha. At a similar specificity of 84% to the combined model, the basic water-only model has a lower sensitivity (39%) compared to the combined model (44%). Here too, the following applies: a higher desired specificity is accompanied by a lower sensitivity by choosing a higher cut-off value for the water surface. In summary, compared to the basic model (water alone), the combined model (water and forest/trees) has an improved specificity with a lower sensitivity compromise. Improved specificity of the models contributes to increased certainty in correctly classifying a given area as “high” risk for HPAI-infection.