Spatiotemporal estimation of actively shedding mpox virus, clade IIb, cases using wastewater signals in British Columbia, Canada

Wastewater surveillance programs collect sewage influent to monitor for pathogens or chemical signatures. Using wastewater surveillance to detect viral genetic material can offer less biased prevalence estimates of viral infections than test or case-based surveillance systems, but existing methods often overlook key factors like spatiotemporal autocorrelation and the duration of viral shedding. Here, we incorporate wastewater signals within a hierarchical Bayesian spatiotemporal model to estimate active mpox cases, defined as individuals who are actively shedding the virus. Our model allows for estimation of active mpox cases within community health service areas, the most precise geographical level of health administration in British Columbia, Canada over time. The inclusion of a time-varying wastewater signal improved the model fit to active cases (WAIC = 3569, Δ 51), when compared to a null model fit without the wastewater signal (WAIC = 3620). The mean absolute error of mpox active cases was ∼2 cases (95 %CI, 1-4) per community health service area. Our study demonstrates the use of hierarchical Bayesian spatiotemporal models as essential tools in wastewater-based infectious diseases surveillance, emphasizing the importance of spatial and temporal autocorrelation in understanding the patterns in both wastewater signals and actively shedding mpox cases.