Sensitivity of wastewater-based epidemiology for detection of SARS-CoV-2 RNA in a low prevalence setting

Abstract

To assist public health responses to COVID-19, wastewater-based epidemiology (WBE) is being utilised internationally to monitor SARS-CoV-2 infections at the community level. However, questions remain regarding the sensitivity of WBE and its use in low prevalence settings. In this study, we estimated the total number of COVID-19 cases required for detection of SARS-CoV-2 RNA in wastewater. To do this, we leveraged a unique situation where, over a 4-month period, all symptomatic and asymptomatic cases, in a population of approximately 120,000, were precisely known and mainly located in a single managed isolation and quarantine facility (MIQF) building. From 9 July to 6 November 2020, 24-hr composite wastewater samples (n = 113) were collected daily from the sewer outside the MIQF, and from the municipal wastewater treatment plant (WWTP) located 5 km downstream. New daily COVID-19 cases at the MIQF ranged from 0 to 17, and for most of the study period there were no cases outside the MIQF identified. SARS-CoV-2 RNA was detected in 54.0% (61/113) at the WWTP, compared to 95.6% (108/113) at the MIQF. We used logistic regression to estimate the shedding of SARS-CoV-2 RNA into wastewater based on four infectious shedding models. With a total of 5 and 10 COVID-19 infectious cases per 100,000 population (0.005% and 0.01% prevalence) the predicated probability of SARS-CoV-2 RNA detection at the WWTP was estimated to be 28 and 41%, respectively. When a proportional shedding model was used, this increased to 58% and 87% for 5 and 10 cases, respectively. In other words, when 10 individuals were actively shedding SARS-CoV-2 RNA in a catchment of 100,000 individuals, there was a high likelihood of detecting viral RNA in wastewater. SARS-CoV-2 RNA detections at the WWTP were associated with increasing COVID-19 cases. Our results show that WBE provides a reliable and sensitive platform for detecting infections at the community scale, even when case prevalence is low, and can be of use as an early warning system for community outbreaks.

Keywords: COVID-19; New Zealand; PEG precipitation; RT-qPCR; Viral shedding.

Graphical abstract

Fig. 1

Model to estimate viral shedding relative to estimated symptom onset date (model 3). Peak shedding at symptom onset date (or estimated onset date), with proportionally less for 3 days prior, and 9 days after, symptom onset date. Model 3 was adapted from He et al. (2020).

Fig. 2

History of each symptomatic (Panel A) and asymptomatic (Panel B) COVID-19 case at the managed isolation and quarantine facility (MIQF). Closed red circles indicate the date reported as symptom onset, open red circles indicate the imputed symptom onset date for asymptomatic cases. Closed black circles indicate the date the case entered the MIQF. Grey lines indicate dates each case was present in MIQF. Black crosses indicate the date the case exited the MIQF.

Fig. 3

Number of RT-qPCR replicates positive for SARS-CoV-2 RNA in the MIQF (Panel A) and WWTP wastewater (Panel B) samples. Mean RT-qPCR Cq values are shown with an inverted y-axis (as lower Cq values represent higher viral concentrations).

Fig. 4

Detection of SARS-CoV-2 RNA in wastewater collected at the MIQF (Panel A) and at WWTP (Panel B). Red triangles indicate MIQF positive samples, yellow triangles indicate MIQF negative samples, red circles indicate WWTP positive samples, and yellow circles indicate WWTP negative samples. On the y-axis, log₁₀ genome copies/L of 1 or 2 have been imputed to represent the 1 or 2 positive RT-qPCR replicates. Blue shading indicates COVID-19 cases at the MIQF on given day (model 1), purple shading estimated infectious cases (model 2), beige shading estimated relative infectious cases (model 3), and green shading new daily cases (model 4). The period when community cases were identified (28 July to 11 September 2020) is indicated by orange bar on x-axis.

Fig. 5

Logistic regression of the probability of detecting SARS-CoV-2 RNA at the WWTP, estimated using A) MIQF infectious cases (model 2) for the whole study period; B) MIQF relative infectious cases (model 3) for the whole study period; C) New daily cases (model 4) for the whole study period; D) MIQF infectious cases (model 2) excluding the community cluster data; E) MIQF relative infectious cases (model 3) excluding the community cluster data; F) New daily cases (model 4) excluding the community cluster data.

Fig. 6

The proportion of symptomatic cases to asymptomatic cases (model 1) at the MIQF by the number of RT-qPCR replicates positive for SARS-CoV-2 RNA at the WWTP.