Combination of variance components estimation, regression trees and logistic regression as a concept for exploring associations between animal husbandry conditions and lesions found at slaughter

Integrated quality control measures have been discussed in farm animal husbandry and veterinary medicine for some time in search of concepts to link information on management factors with disease data gained by veterinary meat inspection at slaughter. Better understanding of relationships between these variables hopefully will help to improve management practices as well as meat quality and safety. In order to analyse data gained on-farm and at slaugherhouses and to investigate relationships between animal husbandry conditions and lesion occurrence at slaughter a biometrical concept has been developed. This concept combines a characterization of variation structures, the generation of regression trees to explore association structures in available animal management and disease data, and logistic regression used to quantify systematic effects of farm management factors. The combination of these three steps is illustrated with an example.