Risk-based surveillance programmes and early warning systems for efficient detection of threats in the animal food production chain

The aim of the project is to improve the cost-effectiveness of surveillance programs in the animal food production chain by using a risk based approach, i.e. by examining samples from animals, herds and products that have a higher risk of being infected by a disease agent or contaminated by undesired chemical compounds. A risk based approach will be applied in the design to make the surveillance programs more effective. The production chain for cattle and cattle products is used as an example and threats to food safety and animal health for this species are studied. In a workshop with the participation from the industry and the government, improving the surveillance of paratuberculosis was considered important. Using governmental and industry registers, we collected information on factors that may indicate a higher risk for paratuberculosis and information on events that may occur with higher frequency when paratuberculosis is present in dairy herds, so-called risk indicators. Mixed stalling with goats and emaciated slaughter animals are examples of such indicators. Data from a ten year period was analysed with multivariate statistical process control (MSPC). The analysis identified herds that had an increased frequency of risk indicators for paratuberculosis. By prioritizing examination of these herds rather than randomly selected herds, the surveillance targets herds where there is a higher probability of finding the disease. The method is now used in the routine surveillance of cattle for paratuberculosis. There are many severe infections in cattle that may cause reproduction failures as decreased non-return percentage, abortions and stillbirths. Details on insemination, pregnancy tests, calvings, abortions and stillbirths are recorded for most Norwegian dairy herds in the Norwegian Dairy Herd Recording System (Tine Norwegian dairies). In this project we study whether this information can be used to design targeted surveillance programs for cattle diseases where reproduction failures are an important part of the symptoms. Bovine viral diarrhoea (BVD) is a cattle disease that gives reproduction failures. In 1992, Norway started an eradication programme and from 2006 and onwards no herd with BVD virus has been detected. It is important to prevent re-establishment of the disease in Norway. By using stochastic analysis of scenario trees, we have calculated that in 2011 it was 99.6 % confidence that Norway was free from disease based on surveillance data since 2007. We are continuing the work by exploring the possibilities for designing a risk based surveillance program for BVD virus. By modelling Norwegian data from the '90s, we have identified risk indicators under Norwegian conditions. Based on these risk indicators, various alternative surveillance programmes for BVD virus have been designed. Thereafter, the different programmes where evaluated by the probability of detecting at least one infected herd (sensitivity) and by the proportion of the herds that where selected for sampling (coverage). By considering both these factors, the surveillance programme can be optimised with regard to both freedom from disease where high sensitivity is important, and early detection where high coverage is important. In risk based surveillance the sample size will usually be reduced, while the cost of data and collection of samples will usually increase. We are establishing a framework for calculating the cost-efficiency of risk based surveillance strategies compared to randomised strategies. Although the initial cost of implementing risk based surveillance is relatively high, preliminary results show that risk based surveillance will be cost efficient for BVD. In addition to BVD, infections with infectious pustular vaginitis virus, bluetongue virus, Schmallenberg virus, Brucella abortus and certain serotypes of Salmonella and other infections may cause reproduction failures in cattle. The project explores how reproduction parameters may be used to target the surveillance and early warning for several diseases simultaneously. By using principal component analysis on risk influence measures obtained in the MSPC analysis, several clusters of herds with reproduction failures could be identified. These clusters differentiate between different groups of syndromes (early abortion, late abortion, and decreased non-return percentage). This is promising for using the method to differentiate between various disease complexes when performing risk based or syndromic surveillance. This work has strengthened the knowledge on methods for use in design and evaluation of surveillance programmes for infectious diseases in production animals. An evaluation of the sensitivity of the surveillance programmes for enzootic bovine leukosis in cattle, Brucella melitensis in sheep and Aujeszky's disease in swine has been performed. The results from the evaluation are used to optimise the sample size of these surveillance programmes.

Norway is currently recognised as having a favourable situation with regard to infectious diseases in animals and a low exposure of health hazards through Norwegian animal products. Increased international trade increases the risk of introduction of infec tious diseases and other threats into Norway with a potentially negative impact on human and animal health and on the economy. Good surveillance that enables early detection and rapid response to emerging diseases and other threats is crucial to contain, control or eliminate these threats. This project aims at improving the cost-efficiency of surveillance programs within the food production chain by using a risk-based approach, i.e. by collecting samples from animals, herds or products with a higher ris k of being infected or contaminated. Thereby, a more efficient surveillance strategy may be designed. This will be combined with syndrome-based surveillance and developed further into early warning systems. The food production chain of cattle and cattle products will be used as an example and threats to both food safety and animal health will be studied. By using information in Governmental and Industry registers research will be performed to identify appropriate syndromes and risk factors for Norwegia n conditions. Methods will be applied to identify events deviating from normal that can be used as warning signals to trigger surveillance activities or early warning messages. The new strategies will be evaluated by their sensitivity, early detection and cost-efficiency. The outcome will be suggestions for new efficient surveillance programmes and early warning systems for cattle and cattle products. Furthermore, the methodological framework that have been developed, will be transferable to other areas of the food production. This will support the Government and the Industry in their efforts to obtain food safety, detect threats at an early stage and in efficient control of animal diseases.