Exploring the surveillance potential of mortality data: nine years of bovine fallen stock data collected in Catalonia (Spain).

• Main Authors: Anna Alba, Fernanda C Dórea, Lucas Arinero, Javier Sanchez, Ruben Cordón, Pere Puig, Crawford W Revie
• Format: Article
• Language: English
• Published: Public Library of Science (PLoS) 2015-01-01
• Series: PLoS ONE
• Online Access: http://europepmc.org/articles/PMC4398401?pdf=render

The potential of fallen stock data to monitor the health status of animal populations has been noted in previous studies. However, further research is required to implement these systems for surveillance. This work presents a novel approach to determining the baselines associated with bovine fallen stock, comparing patterns between subpopulations and identifying subpopulations in which an abnormal event may occur. This study was based on data from 193,873 disposal visits carried out between 2004 and 2012 across a total of 2,991 bovine farms. Proxy measurements such as the number of collections carried out and the weight of carcasses collected were used. Both outcomes were aggregated weekly at different geographical scales for three production types (beef cattle, dairy cattle and heifer fattening). The analysis of these data combined autoregressive integrated moving average modelling and hierarchical time series methods.The three production types exhibited historical baselines that differed notably from one another. Based on the 757 beef cattle farms monitored, the mean number of collections registered per week at the regional level was 37 (range: 10-83). This series was relatively constant over time and showed a marked yearly seasonality. In contrast, for the 426 dairy cattle farms the mean number of disposal visits registered weekly was 121 (range: 71-180), showing half-yearly and yearly seasonality and a marked increase over the period monitored. From the 1,808 heifer fattening farms the mean number of disposal visits was 248 (range: 166-357) and the pattern presented a marked alternating trend over time. These patterns were assessed and compared at regional, provincial, county and municipal levels. The use of hierarchical time series approaches appeared to be a useful tool for comparing the patterns within different subpopulations over time as well as for assessing the spatial extent to which various abnormal events could be detected.