Application of modelling to determine the absence of foot-and-mouth disease in the face of a suspected incursion

Abstract

AIM: To use disease modelling to inform a response team about the number of animals per herd/flock to be examined, and the start date and duration of clinical surveillance required to be confident that foot-and-mouth disease (FMD) was not present on an island in New Zealand with a population of approximately 1,600 cattle, 10,000 sheep and a small number of pigs, goats and alpacas.

METHODS: Because the probability of detecting clinical disease in (the) primary case(s) in larger herds and flocks was extremely low, deterministic and stochastic mathematical SLIR (susceptible, latent, infectious, recovered) models for the transmission of infection were constructed to estimate the date when clinical lesions in herds and flocks would be detected with 95% confidence. Surveillance targeted the first wave of infections following a suspect index case.

RESULTS: If 70 cattle in herds of about 400 cattle were examined it was estimated it would take approximately 13 (90% stochastic range 9–19) days from first exposure before it would be possible to achieve 95% confidence for detecting clinical signs for a low-virulence virus, and 9 (7–14) days for a high-virulence virus. The duration of sufficiently accurate clinical detection was 17 (15–19) days and 13 (12–14) days for low- and high-virulence viruses, respectively. A sample of 70 sheep from flocks of >1,000 would be required to achieve clinical detection at about the same time but with a shorter period of detection than for cattle. The duration of effective detection could be increased by examining a larger sample in most sheep flocks, however the small size of many cattle herds in the study population limited the confidence of detecting group-level disease in cattle, therefore necessitating repeated herd inspections. The model suggested that group-level detection was not feasible if it was based on elevated body temperature alone because of short durations of fever in infected animals.

CONCLUSION AND CLINICAL RELEVANCE: Simulation modelling is a useful and powerful tool for informing ongoing surveillance activities in the face of an exotic disease incursion. Results of modelling suggested to start clinical inspection activities at 4 days and to continue regular inspection twice a week for about 35 days after the date of first exposure, to satisfy the required 95% confidence threshold of clinical detection of FMD in cattle herds and sheep flocks.

KEY WORDS:

• Foot-and-mouth disease
• incursion response
• surveillance
• stochastic state transition modelling

Acknowledgements

The continued support of the head of the incursion response team, Matthew Stone, and his colleagues, throughout the development of the surveillance strategy is highly appreciated. The study also greatly benefitted from the specialist knowledge of Reinhold Kittelberger and Rick Clough on technicalities and accuracy of antibody and antigen follow-up tests which were vital to achieve 100% specificity for the follow-up examination of animals and herds/flocks suspected to be infected with FMD virus.