The effectiveness of contact tracing in heterogeneous networks

Abstract

Case isolation and contact tracing is a widely-used intervention method for controlling epidemic outbreaks. Here, we argue that the effectiveness of contact tracing and isolation is likely underestimated by existing studies because they do not take into account the different forms of heterogeneity and sampling biases from the network structure. Specifically, we show that contact tracing can be even more effective than acquaintance sampling at locating hubs. Our results call for the need for contact tracing to go both backward and forward, in multiple steps, to leverage all forms of positive biases. Using simulations on networks with a power-law degree distribution, we show that this deep contact tracing can potentially prevent almost all further transmissions even at a small probability of detecting infected individuals. We also show that, when the number of traced contacts is small, the number of prevented transmission per traced node is even higher---although most traced individuals are healthy---than that from case isolation without contact tracing. Our results also have important consequences for new implementations of digital contact tracing and we argue backward and deep tracing can be incorporated without the important sacrificing privacy-preserving requirements of these new platforms.