Development of a syndromic surveillance system to enhance early detection of emerging and re-emerging animal diseases

Animal health surveillance plays an important role in protecting animal health, production and welfare, public health and trade from the negative impacts of disease. To address the challenges posed by new, exotic or re-emerging diseases as well as the limitations of traditional surveillance, new approaches, including syndromic surveillance (SyS) and modern communication technologies have been developed to improve early disease detection. SyS is based on the continuous monitoring of unspecific pre-diagnostic health data in order to detect an unusual increase in counts which may indicate a health hazard in a timely manner. An increasing number of studies has been investigating different types of animal health data for a possible use in SyS. Although the potential of cattle mortality data routinely collected in national cattle registers for use in a SyS system was highlighted, the performance of aberration-detection algorithms applied to such data has not yet been investigated. Furthermore, knowledge about the impact of delayed reporting of these data on outbreak detection performance is limited. Clinical observations made by veterinary practitioners reported in real-time using web- and mobile-based communication tools may improve the timeliness of outbreak detection. The willingness of practitioners to report their observations is essential for the successful implementation of such systems. A lack of knowledge about factors that motivate or hinder practitioners to participate in surveillance was found.
The aim of this work was to contribute to the development of a national surveillance system for the early detection of emerging and re-emerging animal diseases in Switzerland, focusing on two Swiss data sources: cattle mortality data routinely reported by farmers to the Swiss system for individual identification and registration of cattle (Tierverkehrsdatenbank TVD); clinical data voluntarily reported by veterinary practitioners to Equinella, an electronic reporting and information system for the early detection of infectious equine diseases in Switzerland.
Time series of on-farm and perinatal cattle deaths, extracted from the TVD, were analysed with regard to data quality and explainable temporal patterns, e.g. day-of-week effect or seasonality. A set of three temporal aberration detection algorithms (Shewhart, CuSum, EWMA) was retrospectively applied to these data to assess their performance in detecting varying simulated disease outbreak scenarios. The effect of reporting delay on outbreak detection was investigated in a Bayesian framework. Participation of veterinary practitioners during the first 12 months of the new internet-based reporting platform of Equinella was assessed. Telephone interviews were conducted to gain insights into factors that motivate or hinder practitioners to participate in a voluntary surveillance system offering non-monetary incentives. Furthermore, the suitability of mobile devices such as smartphones for collecting health data was investigated.
The TVD provided timely cattle mortality data with comprehensive geographical information, making it a valuable data source for Sys. Mortality time series exhibited temporal patterns, associated with non-health related factors, that had to be considered before applying aberration detection algorithms. The three evaluated control chart algorithms adequately performed under specific outbreak conditions, but none of them was superior in detecting outbreak signals across multiple evaluation metrics. Combining algorithms outputs according to different rules did not satisfactorily increase the system’s overall performance, further illustrating the difficulty in finding a balance between a high sensitivity and a manageable number of false alarms. The Bayesian approach performed similarly well in the scenario where delayed reporting was accounted for to the (ideal) scenario where it was absent.
Non-monetary incentives were attractive to sentinel practitioners and overall participation was experienced positive. Insufficient understanding of the reporting system and of its relevance, as well as concerns over the electronic dissemination of health data were identified as potential challenges to sustainable reporting. Mobile devices were sporadically used during the first year and an awareness of the advantages of mobile-based surveillance was yet lacking among practitioners, indicating that they may require some time to become accustomed to novel reporting methods.
This work highlighted the value of routinely collected cattle mortality data for use in SyS, but also the need to carefully optimise aberration detection algorithms for a particular data stream. Alternative methods to the binary alarm system may be chosen for a prospective use of cattle mortality data in a SyS system. The value of evidence framework may be suitable for surveillance systems with multiple syndromes and delayed reporting of data. Before integrating these data into a national surveillance system for the early detection of new, exotic or re-emerging diseases, health authorities need to define response protocols enabling investigation of the data that triggered a statistical alarm and to identify the underlying cause. Possibilities for improving sensitivity and specificity were identified that may be addressed when implementing a future SyS system. In addition, the potential of voluntary reporting surveillance system based on non-monetary incentives was shown. Many of the identified barriers to reporting can be addressed in the future, making the outcome of the pilot project favourable. Continued information feedback loops within voluntary sentinel networks will be important to ensure sustainable participation. Combining reporting of syndromic data and mobile devices in a One Health context has the potential to benefit animal and public health as well as to enhance interdisciplinary collaboration.