Leveraging an established neighbourhood-level, open access wastewater monitoring network to address public health priorities: a population-based study

Abstract

Background

Before the COVID-19 pandemic, the US opioid epidemic triggered a collaborative municipal and academic effort in Tempe, Arizona, which resulted in the world's first open access dashboard featuring neighbourhood-level trends informed by wastewater-based epidemiology (WBE). This study aimed to showcase how wastewater monitoring, once established and accepted by a community, could readily be adapted to respond to newly emerging public health priorities.

Methods

In this population-based study in Greater Tempe, Arizona, an existing opioid monitoring WBE network was modified to track SARS-CoV-2 transmission through the analysis of 11 contiguous wastewater catchments. Flow-weighted and time-weighted 24 h composite samples of untreated wastewater were collected at each sampling location within the wastewater collection system for 3 days each week (Tuesday, Thursday, and Saturday) from April 1, 2020, to March 31, 2021 (Area 7 and Tempe St Luke's Hospital were added in July, 2020). Reverse transcription quantitative PCR targeting the E gene of SARS-CoV-2 isolated from the wastewater samples was used to determine the number of genome copies in each catchment. Newly detected clinical cases of COVID-19 by zip code within the City of Tempe, Arizona were reported daily by the Arizona Department of Health Services from May 23, 2020. Maricopa County-level new positive cases, COVID-19-related hospitalisations, deaths, and long-term care facility deaths per day are publicly available and were collected from the Maricopa County Epidemic Curve Dashboard. Viral loads of SARS-CoV-2 (genome copies per day) measured in wastewater from each catchment were aggregated at the zip code level and city level and compared with the clinically reported data using root mean square error to investigate early warning capability of WBE.

Findings

Between April 1, 2020, and March 31, 2021, 1556 wastewater samples were analysed. Most locations showed two waves in viral levels peaking in June, 2020, and December, 2020–January, 2021. An additional wave of viral load was seen in catchments close to Arizona State University (Areas 6 and 7) at the beginning of the fall (autumn) semester in late August, 2020. Additionally, an early infection hotspot was detected in the Town of Guadalupe, Arizona, starting the week of May 4, 2020, that was successfully mitigated through targeted interventions. A shift in early warning potential of WBE was seen, from a leading (mean of 8·5 days [SD 2·1], June, 2020) to a lagging (−2·0 days [1·4], January, 2021) indicator compared with newly reported clinical cases.

Interpretation

Lessons learned from leveraging an existing neighbourhood-level WBE reporting dashboard include: (1) community buy-in is key, (2) public data sharing is effective, and (3) sub-ZIP-code (postal code) data can help to pinpoint populations at risk, track intervention success in real time, and reveal the effect of local clinical testing capacity on WBE's early warning capability. This successful demonstration of transitioning WBE efforts from opioids to COVID-19 encourages an expansion of WBE to tackle newly emerging and re-emerging threats (eg, mpox and polio).

Funding

National Institutes of Health's RADx-rad initiative, National Science Foundation, Virginia G Piper Charitable Trust, J M Kaplan Fund, and The Flinn Foundation.

Introduction

Triggered by the SARS-CoV-2 pandemic, the use of wastewater-based epidemiology (WBE) as a potentially powerful, rapid, and inexpensive tool to inform public health decision making has seen a remarkable increase globally. For decades, WBE has been used to track chemical and biological threats, with numerous studies underscoring its efficacy and usefulness for understanding and managing community health.1, 2 At the onset of the SARS-CoV-2 pandemic, substantial delays in conventional and individualised clinical testing, due in part to an overwhelmed health-care system and resource limitations,³ positioned WBE as a promising supplemental tool for assessing SARS-CoV-2 spread at the population level.4, 5, 6, 7 Early data from the beginning of the pandemic suggested that SARS-CoV-2 levels in wastewater and sewage sludge offered early indication of clinical confirmed infections, disease, and mortality in a community.8, 9

Research in context.

Evidence before this study

We searched PubMed using the terms “wastewater-based epidemiology” AND “SARS-CoV-2” AND “COVID-19”, for articles that reported SARS-CoV-2 detection in wastewater published from January, 2020, to April, 2020. No language restrictions were applied to the search. The field of wastewater-based epidemiology (WBE) has been active for decades, using analytical tools to monitor a wide range of chemical and biological agents indicative of various aspects of human health, behaviour, and exposure in support of public health strategies and interventions. In response to a demand for enhanced data resolution early in the COVID-19 pandemic, researchers realised the benefits of spatially discrete sampling campaigns within urban sewersheds that could target study populations at the neighbourhood, campus, and even building (dormitory) level. This new approach to population health monitoring by WBE has raised important questions regarding what constitutes ethical practices, and how to shield the emerging science from concerns of privacy, inappropriate use of data, and the potential stigmatisation of populations.

Added value of this study

This study highlights pioneering work in the use of wastewater-based epidemiology conducted at the neighbourhood level for monitoring both substance use (opioids) and the spread of COVID-19, showcasing the early adoption of community-forward WBE to manage health risks. A guiding principle in study design was maintenance of ethical standards in data collection and data use by engaging community members early on, obtaining buy-in before commencing monitoring efforts, and sharing data equitably with all stakeholders. This academic and municipal partnership showcases an early-established (2018) neighbourhood-level wastewater monitoring effort for opioid use accompanied by an online data dashboard that served to rapidly launch a similar model for SARS-CoV-2 and COVID-19 to effectively aid in the public health response using a non-invasive and inclusive approach to community health protection.

Implications of all the available evidence

The time lag observed between recognition of a new public health threat (eg, SARS-CoV-2) and the development and deployment of public health response (eg, development and scale-up of clinical testing and vaccination capacity) will not be limited to the COVID-19 pandemic, as already evidenced by mpox (formerly known as monkeypox). The lessons learned in this study serve to inform municipalities and other institutions on how to develop a wastewater-based surveillance system capable of adapting to future public health threats, including chemical exposures, infectious disease agents, and antibiotic resistance, while maintaining ethically sound data collection, dissemination, and management strategies.

The City of Tempe, Arizona, with a residential population of around 200 000, had been an early adopter of WBE for the purpose of tracking opioid use, which began in May, 2018, and led to the launch of a fully interactive, public-facing, open access WBE dashboard in February, 2019.¹⁰ In this municipal and academic partnership, Tempe and Arizona State University (ASU) first engaged with community stakeholders in a series of public town halls and workshops to achieve a high level of ethical practice and community acceptance of the use of WBE.¹¹ Once established, monthly wastewater samples were shared, with subsequent analysis and joint reporting of opioid use-trends within the community (oxycodone, codeine, heroin, fentanyl, and metabolites; μg per day per 1000 people) in five urban sewersheds.¹⁰ From this effort, a routine was developed that served to enhance the public health response by integrating Tempe Fire Medical Rescue, crisis intervention services, and other stakeholders into a workgroup that relied on WBE data to guide resource deployment by community need (figure 1 ). With this existing framework in place, Tempe and ASU were in a unique position at the start of the COVID-19 pandemic to leverage this previously established collaboration to rapidly monitor SARS-CoV-2. Quantitative assessments of virus levels in wastewater soon followed, with objectives to identify hotspots of infection early, and implement targeted interventions where appropriate. The pre-existing, neighbourhood-level wastewater monitoring network offered an opportunity to test the potential of WBE to serve as an early warning system that might reveal virus presence and spread before clinical case data from testing of individuals.12, 13 Thus, the overarching goals of our study were (1) to compare WBE data to newly reported clinical cases, related hospitalisations, and associated deaths at a high temporal and spatial resolution (ie, county, city, zip code [postal code], and neighbourhood levels), and (2) to determine whether the concurrent pandemic monitoring by WBE produced data and information not available or obvious from clinical testing.

Figure 1.

Schematic of the Arizona State University and City of Tempe academic and municipal partnership

A WBE monitoring network established in 2018 for monitoring opioid use was leveraged to enable a rapid transition to monitoring SARS-CoV-2 during the COVID-19 global pandemic (starting in 2020) with work products including the world's first WBE-informed public-facing interactive online dashboards to combat the opioid and COVID-19 epidemics through a data-driven targeted public health response. ASU=Arizona State University. LC-MS/MS=liquid chromatography-tandem mass spectrometry. RT-qPCR=reverse transcription quantitative PCR. WBE=wastewater-based epidemiology.

Methods

Study locations

This study was conducted within the City of Tempe, Arizona, and the Town of Guadalupe, Arizona, (ie, Greater Tempe), with an estimated residential population of approximately 200 000, and home to Arizona State University. For opioid monitoring before the pandemic, the Greater Tempe area was divided into five sewer catchments (Areas 1–5; appendix p 8) as determined by recurring compliance monitoring. Two additional catchments from an adjacent municipality that contributed to Tempe wastewaters were collected to isolate the Tempe-associated sewage signal. To improve spatial resolution, additional sampling locations were identified based on ease of collection (Area 6). Three other permanent locations needed infrastructure modifications or approvals before onboarding, including Area 7, the Town of Guadalupe (appendix p 8), and Tempe St Luke's Hospital, which had an active COVID-19 ward at the time this study took place (appendix p 9). In total, 11 wastewater catchments were analysed in this study.

The Institutional Review Board of Arizona State University determined this study was not research involving human participants and was exempt from formal review (STUDY00006069).

Sample collection, processing, and analysis

The predefined sampling strategy for opioid monitoring consisted of 7 consecutive days of sample collection each month across variable weeks from permanent, subsurface sampling stations. Although this sampling strategy was sufficient for long-term trend assessment of opioid use, SARS-CoV-2 monitoring required increased temporal resolution based on infection rate dynamics. Accordingly, flow-weighted and time-weighted 24 h composite samples of untreated wastewater were collected at each sampling location within the wastewater collection system for 3 days each week (Tuesday, Thursday, and Saturday) beginning in April, 2020, (Area 7 and Tempe St Luke's Hospital were added in July, 2020). Samples were collected either with an automated refrigerated or portable sampler (Teledyne ISCO, Lincoln, NE, USA) using a mixture of wet and dry ice for cooling (appendix p 10). Wastewater flow was monitored by an ISCO LaserFlow meter (Teledyne ISCO, Lincoln, NE, USA), located within a nearby manhole, or estimated based on historic data (appendix p 11). Composite samples were transferred to high-density polyethylene bottles stored in coolers with ice for transport and processed on the same day to minimise degradation.

Samples were processed and analysed according to published studies.5, 14, 15 Briefly, raw wastewater samples were analysed for SARS-CoV-2 following sequential steps of filtration (0·45 μm polyether sulfone membrane), viral concentration by ultracentrifugation (10 kDa molecular weight cutoff centrifugal filters), nucleic acid extraction (50 μL resultant RNA extract), and TaqMan-based (Invitrogen, Waltham, MA, USA) reverse transcription quantitative PCR (RT-qPCR) targeting the SARS-CoV-2 E (envelope) gene. Spike-and-recovery experiments were performed using murine hepatitis virus (as a surrogate) based on previously published methods.5, 16, 17 Full method details for sample processing and analysis, including primer and probe sequences and RT-qPCR methods, are in the appendix (pp 2–3, 6–7).

Population estimates

Resident populations for each sewershed were estimated using 2010 census block group data. Employment estimates were obtained from the Maricopa Association of Governments (MAG) 2019 employment data and included the following classifications: employees living outside of Tempe (non-resident, employed) and Tempe residents (resident, employed). To correct for changes in employment numbers during lockdown events (commercial closures) and telecommuting activities, we used available MAG average weekday traffic volume (compared with non-lockdown events) in Maricopa County. This percentage was used to correct the non-resident (Tempe employed) employment numbers. Student population estimates were obtained from publicly available campus resident data¹⁸ and changes in wastewater flow volume.

Clinical data

Newly detected clinical cases of COVID-19 by zip code within the City of Tempe, Arizona, were reported daily by the Arizona Department of Health Services. The City of Tempe began extracting and archiving these data on May 23, 2020. Daily case data are not available before this date (data are in aggregate as total cases from the start of the pandemic). Data on Maricopa County-level new positive cases, COVID-19-related hospitalisations, deaths, and long-term care facility deaths per day are publicly available and were collected from the Maricopa County Epidemic Curve Dashboard.¹⁹ Data on clinical testing capacity at both the city level (Tempe) and state level (Arizona) were acquired through publicly available datasets.19, 20

Statistical analysis

Measured concentrations of SARS-CoV-2 (genome copies per L) in each sewer catchment (x) were transformed to viral load per day (genome copies per day) using the following equation:

$$\text{Viral}\text{load}(\text{genome}\text{copies}\text{per}\text{day})={\text{C}}_{\text{x}}\times {\text{Q}}_{\text{x}}$$

where C_x (genome copies per L) is the measured concentration in a given wastewater catchment and Q_x is the total daily volumetric flow rate (L per day) for that wastewater catchment (appendix p 11). In cases where one sewer catchment flowed into another (appendix p 12), a mass balance was performed whereby the viral load (genome copies per day) from the contributing catchment was subtracted from the receiving catchment to isolate individual signals.

Statistical assessments were conducted in MATLAB R2021a (MathWorks, Natick, MA, USA). Root mean square error, a statistical test that measures the standard deviation of the residuals (ie, the predicted data), was used to calculate the offset between different compared data categories using the following equation:

$$\text{Root}\text{mean}\text{square}\text{error}=\sqrt{{\sum }_i^n{({\text{x}}_{\text{i}}-{\text{y}}_{\text{i}})}^2}$$

where n is the number of observations in each specific dataset, x _i is the viral load (genome copies per day) of SARS-CoV-2 in wastewater, and y _i refers to external datasets: either the newly detected clinical cases, COVID-19-related hospitalisations, or COVID-19-related deaths. Data were assessed between April 1, 2020, and March 31, 2021, using individual waves of infection corresponding to up to three events peaking in June–July, 2020, August 2020, and December, 2020–January, 2021. Data were shifted from 0 to 20 days in both directions (forwards and reverse) for each of the comparisons. The data resolution between clinical cases and wastewater testing were different (daily vs three times per week); thus, clinical results that did not have a corresponding wastewater data point were omitted from the assessment post-shift.

Role of the funding source

The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results

Between April 1, 2020, and March 31, 2021, a total of 1556 wastewater samples were analysed across the Greater Tempe area. The number of samples collected per catchment during the study varied from 155 samples in Area 1 to 103 samples in Area 7 and 101 samples in Tempe St Luke's Hospital, with observed differences in the number of samples resulting from occasional sampler malfunctioning and staggered onboarding of sampling locations. The mean total number of SARS-CoV-2 detections per catchment throughout the study was 66 (SD 36), with a minimum of four (Area 3) to a maximum of 116 (Area 6). When detected, the mean SARS-CoV-2 concentration was 617 000 (SD 2·075 million) gene copies per L (median 251 450 gene copies per L [IQR 3500–260300]), indicative of great fluctuations in virus levels over time. Detailed concentration information is provided in the appendix (p 9).5, 14

SARS-CoV-2 viral loads were calculated for each sample using wastewater flow data provided by Tempe (appendix p 11). Flow rates in catchment Areas 1–7 had data recorded at 2-min intervals in real time using permanent laser flow meters, and the Town of Guadalupe and Tempe St Luke's Hospital had only historical flow data available. Flow varied from a maximum in Area 1 of 54·5 million L per day (SD 6·6) to a minimum of 0·106 million L per day (historical estimate) for the hospital location; median 3·0 million L per day (IQR 0·4 million–16·8 million). Mass balances that corrected wastewater sample co-mingling in Areas 1–3 (appendix p 12) resulted in viral loads ranging from 6 × 10¹⁰ to 1 × 10¹³ genome copies per day (appendix p 13). The distributions in viral load varied between each location, with Areas 1, 2, and 6 having smoother distributions in viral load over time, and other locations such as Areas 3, 5, and 7 showing more isolated single-day spikes in activity. Most locations showed two waves in viral levels peaking in June, 2020, and December, 2020–January, 2021. However, in catchments close to ASU, an additional unique wave of viral load was seen (Areas 6 and 7) at the beginning of the fall (autumn) semester in late August, 2020 (appendix p 13).

Estimated Tempe (Areas 1–7) sewer catchment subpopulations ranged from a minimum of 6698 in Area 5 and a maximum of 142 520 in Area 1, the largest geographical catchment area (appendix p 5). Variability in Tempe data was a function of the total numbers of residents, employed individuals, and the number of students in the catchment areas. The population of the Town of Guadalupe (6500) was determined using US census data,²¹ and the hospital location was omitted from this population analysis as the number of hospital staff and patients was unknown.

The result of these efforts ultimately culminated in eight collection locations viewable online by the public (Areas 1–7 of Tempe and the Town of Guadalupe) on an interactive dashboard that went live the week of May 4, 2020 (appendix p 8), with the Town of Guadalupe displayed on a separate tab (appendix p 8). The dashboard displays each catchment area overlaid on a street-level city map so that users can geospatially identify contributing locations to each catchment. Data are shown as the logarithm of genome copies per L of wastewater and are presented as a weekly average consisting of the Tuesday, Thursday, and Saturday collected samples. Text and infographics accompany these data, such as: overview of WBE basics, how to properly interpret the data, and how data are used by the city. Additionally, the SARS-CoV-2 wastewater dashboard is nested in a Community COVID-19 Health Site that contains pertinent information provided by the Centers for Disease Control and Prevention, city demographic information, and positive clinical cases reported by zip code. During the initial lockdown from May to early June, 2020, the wastewater-informed data dashboard revealed remarkably elevated virus concentrations (genome copies per L) in the Town of Guadalupe beginning the week of May 11, 2020, to the week of June 8, 2020, that later decreased after prompt interventions such as face-mask mandates and community education in town halls beginning in June, 2020 (appendix p 8).

Measured viral loads per day of SARS-CoV-2 within each sewer catchment area in Tempe were aggregated and partitioned to their respective zip codes according to their geographical distribution (appendix p 14). The five-digit, Tempe, Arizona zip codes used in this study are 85281, 85282, 85283, and 85284, and will be referred to here as ZC-1, ZC-2, ZC-3, and ZC-4, respectively. From this analysis, wastewater-derived SARS-CoV-2 peaks in activity correlated with newly detected clinical cases per day in three distinct waves of activity: June 2020, late August, 2020 (fall [autumn] semester 2020), and December, 2020–January, 2021. Comparisons between spikes of SARS-CoV-2 viral loads in wastewater and clinical case data showed that peaks in wastewater preceded positive clinical cases by 7 days (ZC-1), 6 days (ZC-2), 11 days (ZC-3), and 10 days (ZC-4) mean of 8·5 days (SD 2·1), during the first wave of the pandemic (June, 2020), and again during the late-August, 2020, wave (6 days; figure 2 ). Tempe aggregated viral loads were compared to Maricopa County Public Health data²⁰ (figure 3 ), which again showed peaks of SARS-CoV-2 viral loads in wastewater preceding new clinically reported cases by 2 days, COVID-19-related hospitalisations by 16 days, and deaths by 18 days during the first wave of the pandemic in June, 2020. During the December, 2020–January, 2021 wave, wastewater was no longer a leading indicator in any region in Tempe. Trends either directly aligned with newly reported clinical cases (ZC-1) or lagged behind clinical case data by 2 days (ZC-2 and ZC-3), 4 days (ZC-4; figure 2), and 2 days (aggregate); mean of −2·0 days (SD 1·4; figure 2). At the county level, wastewater lagged behind clinical results, peaking at 3 days behind Maricopa reported cases, 3 days behind SARS-CoV-2-related hospitalisations, and 1 day behind SARS-CoV-2-related deaths (figure 3).

Figure 2.

Comparison of wastewater-derived viral loads of SARS-CoV-2 (genome copies per day) and clinical cases of COVID-19 in Tempe, Arizona

(A) SARS-CoV-2 genome copies per day in the four zip codes (ZC-1 to ZC-4) and in aggregate, overlaid with newly reported clinical cases. Numbers are the number of days the wastewater signal leads (positive number) or lags (negative number) clinical cases, determined by root mean square error analysis. (B) Reported number of saliva-based tests processed per day for faculty, staff, and the general public (grey), students only (black), and a 7-day trailing average (green dotted line). Vertical grey dotted lines indicate noteworthy enhancements of testing efficacy: laboratory capacity increased to 12 000 tests per day in November, 2020, and January, 2021 was reported to have the highest volume of saliva-based tests received. These improvements are a probable cause of wastewater monitoring transitioning from a leading to lagging or real-time indicator of community viral presence.

Figure 3.

Wastewater-derived SARS-CoV-2 viral load (genome copies per day) compared with county-level and state-level clinical datasets

(A) Peaks in SARS-CoV-2 viral load (genome copies per day) in Tempe, Arizona wastewater compared with Maricopa County, Arizona new clinically detected cases of COVID-19, and COVID-19-related hospitalisations and deaths. Numbers are the number of days the wastewater signal leads (positive number) or lags (negative number) clinical cases. (B) Total number of diagnostic tests for COVID-19 conducted per week (light pink) in the state of Arizona, demonstrating increased testing efficacy starting from 30 available sites state-wide in May, 2020, to more than 500 available testing sites state-wide in October, 2020. (C) Reported timeline of events during time of study serves to support that as clinical testing improved throughout the pandemic, wastewater monitoring transitioned from a leading indicator to a lagging or real-time indicator. *The Arizona State University Biodesign Clinical Testing Laboratory developed a saliva-based test in partnership with the Arizona Department of Health Services to offer free, accessible testing state-wide. †Three universities: Arizona State University, Northern Arizona University, and University of Arizona.

Discussion

We employed WBE to monitor SARS-CoV-2 in Greater Tempe, Arizona, USA, by implementing a unique, high-frequency (three samples per week), and high spatial resolution (ie, neighbourhood-level) sampling approach in conjunction with immediate, open access data sharing with the public via an online dashboard. The present work is unique in that it provides historical context to the use of WBE, and how it can serve to reshape the philosophy of wastewater-informed public data sharing, and displaying public health information pertaining to both the opioid epidemic in the USA and the global COVID-19 pandemic. This study further illustrates that sub-sewershed wastewater monitoring produces actionable data, and can be conducted ethically with support from the community and relevant stakeholders.

The measured values of SARS-CoV-2 in wastewater were in line with those reported from other wastewater monitoring studies,9, 22 with the maximum concentration of 37·6 million gene copies of SARS-CoV-2 per L. This measurement occurred at the hospital location, which had an active COVID-19 ward at the time of collection on Jan 11, 2021, during peak pandemic conditions. Higher relative standard deviations (RSDs) for SARS-CoV-2 concentrations recorded for a given week (three observations per week) occurred in locations with a higher proportion of commercial businesses, including Area 4 (RSD 83%) and Area 5 (RSD 93%), compared with areas with largely residential catchments (Area 1 [58%] and Area 2 [65%]). This finding might explain the relatively smoother trends over time in Areas 1 and 2 compared with areas with higher transient populations that showed isolated single-day spikes in viral presence. These results suggest that a high-frequency sample collection approach should be considered in catchments with a higher proportion of transient populations, which might be susceptible to greater variability in wastewater-derived viral measurements from day to day.

Estimating population size by study area was challenging due to collecting wastewater from within the sewer infrastructure rather than determining population served by a wastewater treatment plant as historically conducted.²³ Because of net importation of people to the city for work, it was important not only to quantify the number of residents but also the number of non-residents employed and transient student populations, a task accomplished using MAG and on-campus student resident data. To date, no other WBE study has reported use of employment data to refine population estimates within collection systems. MAG data needed to be corrected for lockdown activities, for which we used Arizona Department of Transportation arterial traffic flow data (eg, 40% decrease in arterial traffic equated to a 40% decrease in employment populations). Campus resident data was appropriate to assess temporal changes in student populations due to online courses; however, numbers did not account for changes during holiday travel or off-campus housing; thus, percentage changes in wastewater flow were used to estimate population changes. For instance, wastewater flow from Area 6 (university-adjacent catchment) increased by 20% throughout the academic year; therefore, the population in that area was assumed to increase proportionally.

The northern-most zip code of Tempe that encompasses ASU (ZC-1) was the only zip code that showed viral increases in late August, 2020. Contributions to viral load by university students within a given community was not just isolated to Tempe; contributions also occurred in other communities that housed large universities.24, 25 These results align with preliminary assessments of wastewater and clinical case data that suggested monitoring wastewater provided an early warning capacity ranging between 2 and 21 days, demonstrating that wastewater can serve as an early indicator of future clinical case load, morbidity, and mortality.9, 22

Although wastewater monitoring has demonstrated changes as a leading indicator at the wastewater treatment plant level,²⁶ to our knowledge, our study is still the first report of capturing changes in lead-time dynamics of viral presence at the sub-catchment, zip-code level. This consecutive decrease in lead time between wastewater measurements and clinical testing might best be explained by notable capacity improvement in availability and frequency of clinical testing largely driven by saliva-based tests (from 30 sites in May, 2020, to 500 sites in October, 2020) that took place in parallel over the course of this study (Figure 2, Figure 3).18, 27 This unique clinical testing environment differs from other reported studies where the drivers of lead and lag might be affected by other parameters,28, 29 suggesting that the greatest benefits of WBE might be early detection of disease outbreaks in situations where a substantial health-care response has not yet been mounted for many reasons (ie, clinical testing site scarcity, testing fatigue, vaccination campaigns, and widespread use of at-home rapid tests).

The importance of neighbourhood-level sampling was demonstrated in our study and allowed for the identification of isolated hotspots, such as the Town of Guadalupe, Arizona, where the municipalities' wastewater co-mingles with Tempe's Area 3. However, in Area 3 SARS-CoV-2 was only detected four times during the entire year-long sampling campaign (appendix p 13), implying that the elevated SARS-CoV-2 signal originating in the Town of Guadalupe was attenuated beyond detectability at the Area 3 sampling location, and was visible only through high spatial resolution monitoring. Clinical testing in theory might have rendered the Town's infection hotspot visible; however, testing was scarce in the area, thereby potentially obscuring what was happening within the community.

The limitations of our study include the use of a single-gene target (E gene) analysis to determine SARS-CoV-2 presence, which was among one of the first gene targets reported by WHO in 2020. Additional viral targets might have improved our detection frequencies; however, the efficiency of the assay did not change over the course of this study, suggesting it to be reliable for this purpose. Sewage temperature, travel time, and storage are known to influence the stability of labile wastewater-borne biomarkers such as viral RNA;³⁰ however, similar to most WBE studies, data were not corrected for these variables.

This study illustrates that a major challenge to neighbourhood-level monitoring by WBE is not only assay development, but also creating partnerships with city personnel, gaining trust from community members, establishing the sampling network and methodology, understanding which establishments or buildings are contributing to a given collected sample, and how these populations change (eg, on weekdays, at weekends, and during closures or events). These factors lead to detailing and understanding the primary outcomes of this type of investigation—how data should be protected, shared, and used to inform public health decision making. Thus, this work can serve to inform municipalities interested in adopting and implementing WBE programmes to monitor already known and newly emerging public health threats, and further, afford the ability to rapidly transition between threats, chemical or biological, if and when necessary. Our study can therefore be transferable across disciplines for a wide variety of applications in support of further understanding and investigating public health datasets.

Data sharing

All data needed to evaluate the conclusions in the Article are present in the main Article text and in the appendix. The SARS-CoV-2 Wastewater Monitoring Dashboard and COVID-19 Community Health Site in the City of Tempe, Arizona are available at https://covid19.tempe.gov/.

Declaration of interests

EMD is a managing member of AquaVitas, a company working in the field of wastewater-based epidemiology. RUH is a managing member of AquaVitas and founder of the Arizona State University non-profit project OneWaterOneHealth operating in the same intellectual space. All other authors declare no competing interests.