Determining associations and predictability of pen level management factors and health outcomes related to respiratory disease in the first 45 days of the feeding phase

Abstract

Bovine respiratory disease (BRD) is an important health issue in the beef cattle industry. This syndrome is a multifactorial disease and continues to be a leading cause of morbidity and economic loss in feedlot cattle operations. Research has identified risk factors associated with BRD; however, knowledge gaps remain. Three studies were conducted to evaluate potential associations and predictability of pen level management factors and their impacts on BRD-related health outcomes during the first 45 days on feed (DOF) in the feeding phase. The first study evaluated potential associations between pen level management factors, such as area per head and bunk space per head, combined with cattle attributes and their impact on BRD morbidity incidence in the first 45 DOF. Our results showed pen housing characteristics related to pen area per animal and bunk space per head were significantly associated (P < 0.05) with BRD incidence in the first 45 DOF; however, effects of these associations were modified by cattle attributes. The second study had two objectives. The first objective was to evaluate the diagnostic performance of five predictive models using area under the curve (AUC) to classify incoming groups of cattle into high- (≥15% BRD incidence 45 DOF) and low-risk (<15% BRD incidence 45 DOF) groups based on the BRD morbidity within the first 45 DOF. The types of models used for this analysis were logistic regression, decision tree, random forest, discriminant linear, and naïve Bayes models. The second objective was to evaluate the models using an economic analysis to determine if the models were economically advantageous compared to a control scenario that represents feedlot personnel classifying expected risk. AUC of the models ranged from .682 to .789 for with a random forest model having the highest AUC. The model with the best economic performance compared to the person classifying expected risk was determinant on the proportion of high-risk cohorts in the population. The decision tree model displayed the highest potential economic advantage when the proportion of high-risk cohorts reaches approximately 40% or less compared to the person. The third study evaluated potential associations between pen level management factors related to the number of water sources, shared water sources, and shared fence lines combined with cattle attributes and their impact on BRD morbidity incidence in the first 45 DOF. Shared fence lines, number of water sources, and their interactions with cattle demographics included in the models, were significantly associated with BRD morbidity in the first 45 DOF (P < 0.05). The interaction between shared water sources and total cohort size at arrival was found to have statistical significance (P < 0.05), but did not display biological significance. Several pen level management factors were tested for associations and predictability related to BRD morbidity in the first 45 DOF. Understanding how these factors and their impact on BRD can potentially lead to increased animal welfare, increased sustainability, and lessen the burden of economic losses in the feedlot cattle industry.