Extra intestinal pathogenic E. coli causes a public health burden by causing infections which can lead to deaths by sepsis. These are commonly found to colonize food animals and are present in retail meat as a contaminant. There is an urgent need to study these zoonotic transmissions to determine precise proportions of human extraintestinal E. coli infections. We aim to use host-specific accessory elements in conjunction with core-genome phylogenetics to understand transmission routes and host-switch events.Here, we present the results from our host element discovery workflow. We performed a pan-genome analysis on a set of inhouse E. coli genomes isolated from human urinary tract infections (N=1,188) and retail meat [chicken (N=1,156), turkey (N=473) and pork samples (N=310)]. We searched for host-specific genes by characterizing their co-occurrence using the Sørensen-Dice method resulting in multiple gene clusters associated with poultry, pork and human clinical UTI infections. The gene clusters were mapped to closed and assembled genomes from our collection to characterize the genetic context of the elements that resulted in a total of 366 genes. Based on the physical location and statistical associations, the individual genes were grouped into 52 multigene elements consisting between 2 and 42 genes (6.75 average). These multigene elements were used along with Bayesian latent class models to estimate the putative host jumps by E. coli of food animal origin.