Health Scores for Farmed Animals: Screening Pig Health with Register Data from Public and Private Databases

There are growing demands to ensure animal health and, from a broader perspective, animal welfare, especially for farmed animals. In addition to the newly developed welfare assessment protocols, which provide a harmonised method to measure animal health during farm visits, the question has been raised whether data from existing data collections can be used for an assessment without a prior farm visit. Here, we explore the possibilities of developing animal health scores for fattening pig herds using a) official meat inspection results, b) data on antibiotic usage and c) data from the QS (QS Qualität und Sicherheit GmbH) Salmonella monitoring programme in Germany. The objective is to aggregate and combine these register-like data into animal health scores that allow the comparison and benchmark of participating pig farms according to their health status. As the data combined in the scores have different units of measure and are collected in different abattoirs with possibly varying recording practices, we chose a relative scoring approach using z-transformations of different entrance variables. The final results are aggregated scores in which indicators are combined and weighted based on expert opinion according to their biological significance for animal health. Six scores have been developed to describe different focus areas, such as "Respiratory Health", "External Injuries/ Alterations", "Animal Management", "Antibiotic Usage", "Salmonella Status" and "Mortality". These "focus" area scores are finally combined into an "Overall Score". To test the scoring method, existing routine data from 1,747 pig farm units in Germany are used; these farm units are members of the QS Qualität und Sicherheit GmbH (QS) quality system. In addition, the scores are directly validated for 38 farm units. For these farm units, the farmers and their veterinarians provided their perceptions concerning the actual health status and existing health problems. This process allowed a comparison of the scoring results with actual health information using kappa coefficients as a measure of similarity. The score testing of the focus area scores using real information resulted in normalised data. The results of the validation showed satisfactory agreement between the calculated scores for the project farm units and the actual health information provided by the related farmers and veterinarians. In conclusion, the developed scoring method could become a viable benchmark and risk assessment instrument for animal health on a larger scale under the conditions of the German system.