Exploring the longitudinal dynamics of herd BVD antibody test results using model-based clustering

Publisher
Nature Publishing Group
Abstract
Determining Bovine Viral Diarrhoea (BVD) infection status of cattle herds is a challenge for any control and eradication programme. A herd's BVD status is typically based on the outcome of a single cross-sectional antibody test. Given the dynamic nature of BVD antibodies, this apparent status is likely to change if the herd is tested sometime later. Therefore, potentially, the BVD status of the herd will change over time. Classifying farms based on the longitudinal behaviour of their BVD antibodies could offer a stable approach to categorisinge farms as it is more likely to capture or represent the true infection or exposure dynamics of the farms. This paper describes the dynamics of BVD antibodies obtained from 15,500 adult cows between 2007 and 2010 from thirty nine cattle farms located in Scotland and Northern England. It explores alternative approaches of classifying herds based on their longitudinal antibody patterns and investigates the epidemiological similarities between farms within the same cluster based on information on exposure factors available on each farm. Average BVD antibody longitudinal trends of the thirty-nine farms over the four years were clustered in two ways: firstly by their linear trends, by fitting a simple linear model on each farm, and secondly, by using two model-based clustering methods based on their magnitude and shape using model-based clustering. Analyses indicate that patterns of antibody response to BVDV and/or vaccine varied between farms. The models clustered farms with similar longitudinal antibody magnitude or pattern. Different farm cluster memberships were obtained depending on the type of clustering approach used. However, strong concordance was found between the two model-based clusters. The approaches isolated groups of farms likely to require quite different BVD control interventions. These approaches can help assess changes due to the impact of control or management measures over time. In addition, these approaches can detect high risk farms (defined by their antibody longitudinal patterns) for targeted control measures thereby saving resources that would have been expended on on-farm investigations if all farms were investigated.
Year
2019
Category
Refereed journal
Output Tags
WP 2.2 Livestock production, health, welfare and disease control (RESAS 2016-21)