An adaptive management framework for optimal response to Foot and Mouth outbreaks

Background/Question/Methods

Optimal decision-making for management interventions during epidemics is often hampered by severe uncertainty. Though uncertainties due to environmental variation and sampling error are difficult to resolve, epistemic uncertainties, due to gaps in knowledge about the underlying dynamics, can be reduced through improvement of the information state. Adaptive management seeks to incorporate monitoring, evaluation, and response into management strategies such that tactics can be modified and updated as information gathered about the outbreak at hand leads to real-time reductions in epistemic uncertainties about outbreak dynamics.

Analysis of the 2001 Foot and Mouth (FMD) outbreak in the UK provided valuable information about both the dynamics of disease spread and the implementation of management actions. However the progression of a novel FMD outbreak will not necessarily follow the same pattern and key aspects of previous FMD spread, and the optimal response cannot be resolved until an outbreak occurs. Predictions of FMD spread and the impact of control efforts are limited by uncertainties about farm-to-farm transmission and the vaccine efficacy. Using a spatially explicit metapopulation model of FMD spread in the UK, we illustrate the application of adaptive management to integrate model uncertainty and update management tactics in response to real-time evaluation of alternative models.

Results/Conclusions

Here we describe a generalized framework for combining dynamic epidemic projections and economic models of the expected cost of interventions to evaluate adaptive strategies for epidemic response. We present a retrospective analysis the 2001 FMD outbreak as a case-study to illustrate the dependence of the optimal intervention strategy on the current state of knowledge and calculate the expected value of active surveillance to resolve model uncertainty and implement changes in outbreak response in real-time. We show that the optimal strategy for farm-based culling is highly dependent on the a priori weights for competing models of the rate and scale of farm-to-farm transmission. Investment in surveillance discriminate among these competing models during the course of an outbreak can allow the implementation of model-dependent optimal strategy and result in significant (>10%) reduction in overall economic costs of the outbreak. The power to discriminate among candidate models under alternative management options can lead to the recommendation of an initially sub-optimal intervention that results in rapid model selection and transition to a model-dependent strategy that minimizes total costs over the duration epidemic.