Quantifying the hidden costs of imperfect detection for early detection surveillance

The global spread of pathogens poses an increasing threat to health, ecosystems, and agriculture worldwide. As early detection of new incursions is key to effective control, new diagnostic tests which can detect pathogen presence shortly after initial infection hold great potential for detection of infection in individual hosts. However, these tests may be too expensive to be implemented at the sampling intensities required for early detection of a new epidemic at the population level. To evaluate the trade-off between earlier and/or more reliable detection and higher deployment costs, we need to consider the impacts of test performance, test cost, and pathogen epidemiology. Regarding test performance, the period before new infections can be first detected and the probability of detecting them are of particular importance. We propose a generic framework which can be easily used to evaluate a variety of different detection methods and identify important characteristics of the pathogen and the detection method to consider when planning early detection surveillance. We demonstrate the application of our method using the plant pathogen Phytophthora ramorum in the UK, and find that visual inspection for this pathogen is a more cost effective strategy for early detection surveillance than an early detection diagnostic test.