Risk assessment for livestock disease outbreaks between farms

Applied statistical approaches that make use of detection data arising in disease outbreaks

We developed and applied statistical approaches that make use of detection data arising in disease outbreaks, enabling better control by inferring when farms become infectious.

Image
sheep gathered around feed in a snowy field

The need to control livestock disease outbreaks

Livestock epidemics can give rise to significant economic, welfare, and social costs. However, these costs may be limited by efficiently targeting control measures, e.g. vaccination, culling, or isolation of infected animals and high-risk farms.

The need to identify at-risk farms

During an outbreak that spreads between farms, it is critical to identify infected farms and those at greatest risk of subsequent infection to better target limited resource. A key difficulty in practice is failure to identify infected farms resulting "silent" spread of the disease.

 

BioSS contribution: Statistical risk assessment from historic and ongoing epidemics

We developed and applied advanced statistical modelling techniques, that use only data on farm locations and detected cases, to estimate unobserved quantities, such as the period between infection and detection, and when farms become infectious. This enables inference of silent spread and the better targeting of control options.

Comprehensive testing using simulated scenarios designed to represent field data show that these techniques can be reliably applied to both ongoing and historic outbreaks, and the estimates obtained can be used to enable quantitative risk assessments to enable the targeting of controls.

Impact of work

We applied this methodology to a Classical Swine Fever outbreak (East Anglia UK, year 2000) in which just 16 farms were infected1. And this work informed a comprehensive assessment of the vulnerability of the British swine industry to a classical swine fever outbreak2.

We have recently extended and applied our methodology to understand a 2015 outbreak of Highly Pathogenic Avian Influenza (H5N2) in US poultry farms. This enabled inference of the time taken by surveillance activities to detect infection on farms and the number of cases resulting from non-local disease incursions relative to between farm spread.

Prospects for the future

The methods developed are widely applicable to data from outbreaks of different diseases at a range of scales, e.g. to other livestock diseases and the spread of disease in crops or forestry at the field or plot scale, to the landscape scale and beyond.

Acknowledgements

This work was developed at BioSS by Kokouvi Gamado and Glenn Marion in collaboration with Thibaud Porphyre at the University of Edinburgh. It was funded under the Scottish Government Centre for animal disease outbreaks (EPIC) and the Scottish Government's Strategic Research Programme in environment, agriculture and food.

Further information

Gamado, K.M., Marion, G. and Porphyre, T. (2017). Data-driven risk assessment from small scale epidemics: estimation and model choice for spatio-temporal data with application to a classical swine fever outbreak. Frontiers in Veterinary Science 4, 16. Frontiers.

Porphyre, T., Correia-Gomes, C., Chase-Topping, M.E., Gamado, K.M., Auty, H.K., Hutchinson, I., Reeves, A., Gunn, G.J. and Woolhouse, M.E.J. (2017). Vulnerability of the British swine industry to classical swine fever. Scientific Reports 7(42992). Nature Publishing Group.

Image
cow with mountains behind
Glenn smiling wearing a green shirt

Further details from: