Wastewater-based epidemiology (WBE) represents an effective complementary tool for the surveillance of infectious diseases when the capacity of conventional clinical testing is limited or when there is a significant proportion of infectious asymptomatic cases in the community being monitored. Two of the current challenges of WBE is understanding how the wastewater matrix affects virus recovery and how to translate the wastewater results into a number of infected individuals.

To address these challenges, wastewater surveillance was undertaken at a large regional hospital (Ysbyty Gwynedd) in Wales, UK. The samples were analysed using qPCR for three viruses of public health concern: influenza A virus, norovirus GII (NoVGII) and SARS-CoV-2. Additionally, the concentrations of ammonium-nitrogen and orthophosphate and the faecal indicator, namely crAssphage, were trialled as estimates for human urine or faecal load (i.e. population captured). The mathematical models to estimate the population numbers included empirical equations following the Central Limit Theorem premise (EMCLT) and three separate Monte-Carlo-Bayesian approaches (MCBA). Additionally, pH, electrical conductivity (EC) and turbidity were also measured to characterise the wastewater matrix.

Modelling based on an adaptive neuro-fuzzy interference system was used to assess the impact of wastewater parameters on virus recovery, and literature research was used to interpret the relationship. Ammonium and phosphate had a positive modelled association on virus recovery, likely due to their concentrations being correlated with the number of individuals shedding into the wastewater system. SARS-CoV-2 and crAssphage detection was greatest in the pH range 7.6 – 8.5 and from 6.8 – 8.5, respectively. Turbidity was not associated with SARS-CoV-2 but was with crAssphage, which was attributed to the association of the latter with faecal matter. EC was positively associated with the recovery of both viruses, starting at >0.75 mS/cm due to soluble organic matter and ions inhibiting virus sorption. However, it also reflects the capture of human effluent, which has a high EC value (ca. 20 mS/cm).

CrAssphage-based total population estimates were within the expected boundaries in 35.7% of cases, whereas ammonium and phosphate gave significant overestimates with considerable fluctuations likely due to unknown sources of discharge. One MCBA provided realistic results for NoVGII and influenza A virus, whereas two other MCBA and the EMCLT provided promising results for SARS-CoV-2 and crAssphage. For validation, the SARS-CoV-2 estimates were compared with COVID-19 clinical cases in the region of Gwynedd, whereas the influenza A virus estimates were compared with national cases. The infection numbers were converted into proportion of the total population by dividing the estimates with the crAssphage-based estimates or the clinical numbers by the census data of Gwynedd or Wales. For influenza A, RNA concentrations in wastewater did not coincide with the increase of cases in Wales, probably due to low virus detection rates. Hence, the clinical validation was not successful. When the SARS-CoV-2 estimates were overestimated, it was likely due to underestimated crAssphage estimates of the total population. When converting to percent changes, there was a significant correlation between the estimated and reported trends, pointing out to the potential suitability of the models. However, validation with clinical data may be difficult in a hospital setting due to a dynamic population, a residential area being more suitable.

The study has expanded on the usefulness of MCBA and EMCLT for WBE, and providing their adjustments, particularly normalising virus concentration, they could be implemented for routine surveillance. They are useful in settings with dynamic populations, such as a hospital, where predictions cannot be made about the population infection susceptibility. Despite the assessed impact of physico-chemical parameters on virus recovery, the wastewater matrix is complex and requires a more comprehensive study that explores other parameters (e.g., bio-chemical oxygen demand, total organic carbon) and investigates mathematical relationships.