Exploiting strain diversity to expose transmission heterogeneities and predict the impact of targeting supershedding

When a few individuals generate disproportionately many secondary cases, targeted interventions can theoretically lead to highly efficient control of the spread of infection. Practical exploitation of heterogeneous transmission requires the sources of variability to be quantified, yet it is unusual to have empirical data of sufficient resolution to distinguish their effects. Here, we exploit extensive data on pathogen shedding densities and the distribution of cases, collected from the same population within the same spatio-temporal window, to expose the comparative epidemiology of independent Escherichia coli O157 strains. For this zoonotic pathogen, which exhibits high-density shedding (supershedding) and heterogeneous transmission in its cattle reservoir, whether targeting supershedding could be an effective control depends critically on the proposed link between shedding density and transmissibility. We substantiate this link by showing that our supershedder strain has nearly triple the R(0) of our non-supershedder strain. We show that observed transmission heterogeneities are strongly driven by superspreading in addition to supershedding, but that for the supershedder strain, the dominant strain in our study population, there remains sufficient heterogeneity in contribution to R(0) from different shedding densities to allow exploitation for control. However, in the presence of substantial within-host variability, our results indicate that rather than seek out supershedders themselves, the most effective controls would directly target the phenomenon of pathogen supershedding with the aim of interrupting or preventing high shedding densities. In this system, multiple sources of heterogeneity have masked the role of shedding densities-our potential targets for control. This analysis demonstrates the critical importance of disentangling the effects of multiple sources of heterogeneity when designing targeted interventions.