Improving wastewater-based epidemiology through strategic placement of samplers

Abstract

Wastewater-based epidemiology (WBE) is a fast emerging method for passively monitoring diseases in a population. By measuring the concentrations of pathogenic materials in wastewater, WBE negates demographic biases in clinical testing and healthcare demand, and may act as a leading indicator of disease incidence. For a WBE system to be effective, it should detect the presence of a new pathogen of concern early enough and with enough precision that it can still be localised and contained. In this study, then, we show how multiple wastewater sensors can be strategically placed across a wastewater system, to detect the presence of disease faster than if sampling was done at the wastewater treatment plant only. Our approach generalises to any tree-like network and takes into account the structure of the network and how the population is distributed over it. We show how placing sensors further upstream from the treatment plant improves detection sensitivity and can inform how an outbreak is evolving in different geographical regions. However, this improvement diminishes once individual-level shedding is modelled as highly dispersed. With overdispersed shedding, we show using real COVID-19 cases in Scotland that broad trends in disease incidence (i.e., whether the epidemic is in growth or decline) can still be reasonably estimated from the wastewater signal once incidence exceeds about 5 infections per day.