Quantification of the Relationship Between Distiller's Grains Co-Products and Escherichia Coli O157:H7 Load in Real-World, Commercial Feedlots

Non-Technical Summary: Researchers consistently report that use of distiller's grains approximately doubles E. coli O157:H7 (EcO157) shedding. An important yet oft overlooked consideration is that their studies housed cattle in small pens of 20 animals or fewer. Our data demonstrate that in commercial feedlots, substantial pen-level clustering of EcO157 exists resulting in extra-binomial (or even bimodal) variation in pen-level pathogen burden. Consequently, the pen (or environment) is the primary determinant of burden within agent-host-environment triad. One should not assume, therefore, that EcO157 burden remains constant when the environment (e.g., pen size) is modified. Pen sizes are typically 10-fold larger in commercial feedlots than in research settings (i.e., ~200 versus 20 animals/pen). A critical need, therefore, is to determine if the distiller's grains-EcO157 association is evident in commercial feedlots. In our research, we will evaluate an association of use of distiller's grains and EcO157 in real-world, commercial feedlots. These data will provide informative decision making tools for industry on the value and risk of distiller's grains use. <P> Approach: We will conduct a prospective cohort study in which we will enroll a convenience sample of 5 feedlots that use distillers' grains in their ration3 (exposed cohort) and 5 feedlots that do not (unexposed cohort). Inclusions criteria for each cohort are: * Exposed cohort Continuous incorporation of distillers grains co-products into the finishing ration for at least 6 months. * Unexposed cohort No use of distillers' grains co-products for at least 6 months. We will visit each feedlot on 4 occasions at approximately 6-week intervals from June through October. During each visit, we will collect freshly voided, pen-floor fecal samples from pens of cattle that are within 2 weeks of their projected harvest date. At each visit, we will identify all the pens of cattle that have a projected harvest date within the subsequent 2 weeks. We will enroll 4 pens that meet this criterion; if more than 4 pens are available as is typically the case, we will randomly select 4 using a random number generator. Within each of these 4 pens, we will identify 25 freshly-voided, well-defined fecal pats and collect approximately 100 grams of feces from each into pre-labeled specimen containers. Label information will include date, feedlot code, lot number associated with pen of cattle, and sample sequence number. In this study, our primary outcome measure is the within-pen burden and as such, we will not attempt to establish a link between fecal pat to its specific source animal. Rather the fecal pat will be associated only with the pen of cattle. We will use standard microbial methods to estimate prevalence (yes/no) and concentration of E. coli O157 in fecal samples. A unique aspect of our research will be the data analysis. At the core of the analysis is the need to evaluate the association between use of distillers' grains and burden of E. coli O157:H7 in feces. The data are, however, complex in structure in that they represent multilevel data with pens clustered within feedlot and feedlots clustered over time. Another way to view the design is that we will sample feedlot, within feedlots we will sample pens, and within pens we will sample fecal pats - at each level of sampling, we will need to model dependency of outcome. Another complexity of the data is that they will likely violate assumptions of various standard regression techniques (e.g., logistic regression) in that there is considerable extra-binomial variation to the point the data are bimodal. A variety of advanced modeling techniques will be used to provide informative analyses such as generalized linear mixed models, zero-inflated models, and finite mixture analysis.