Research needs for optimising wastewater-based epidemiology monitoring for public health protection Open Access

The COVID-19 pandemic has had a major impact on public health, well-being, and economic systems globally (Kolahchi et al. 2021). The causative agent, severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has spread rapidly since 2019. Clinical presentation has ranged from asymptomatic cases to hospitalisation or death (Weiss et al. 2020; Zheng et al. 2020; Park et al. 2021). One method for monitoring levels of SARS-CoV-2 that is inclusive, prone to fewer population biases, and less invasive than testing individuals is surveillance of wastewater in sewer systems, where the level of infection in the catchment community is associated with viral copies in sewage (Polo et al. 2020; Sims & Kasprzyk-Hordern 2020). This approach, termed wastewater-based epidemiology (WBE), has been used to complement existing public health data and clinical monitoring procedures (Naughton et al. 2021), with most data coming from high-income countries, for example, in the United Kingdom (Wade et al. 2020), the Netherlands (Medema et al. 2020), and the United States (Kirby et al. 2021). There are WBE studies in low and middle-income countries (LMICs), such as Tunisia (Jmii et al. 2021), Morocco (Amahmid et al. 2021), Brazil (Barbosa et al. 2022), where challenges for conducting this work are greater and complex (Pandey et al. 2021). The response to the pandemic has led to rapid acceleration of WBE research and development, and monitoring is now being performed at national, regional, and local scales (University of California 2022). Additionally, WBE has provided strategically important information through the development of new bioinformatics pipelines for variant detection and advanced algorithms for data interpretation, all of which have helped reduce uncertainty in the data (Karthikeyan et al. 2021; Habtewold et al. 2022). WBE has also provided early warning of local disease outbreaks, thus facilitating public health actions, such as focusing subsequent deployment of face masks, mobile testing, and vaccination facilities (e.g., surge testing; Hillary et al. 2021) and increasing bed capacity at hospitals (Brown et al. 2021; Crits-Christoph et al. 2021; Herold et al. 2021; Swift et al. 2021).

WBE is not a new concept, having been used to monitor illicit drug consumption as early as 2005 (Zuccato et al. 2005; Biello 2007) and the global distribution of poliovirus over the last decade (Hovi et al. 2012). Aside from these applications and monitoring of SARS-CoV-2, WBE has been used on a more ad hoc basis, with the implementation of a wide range of temporally and spatially distinct use cases, such as consumption of the antiviral oseltamivir (Tamiflu) during the 2009 H1N1 pandemic (Singer et al. 2013), the presence of influenza A and H1N1 in wastewater in the Netherlands (Heijnen & Medema 2011), consumption of illicit and licit drugs (including antidepressants) during the 2010–2014 Greek economic crisis (Thomaidis et al. 2016), and the global distribution of antimicrobial resistance (AMR) (Hendriksen et al. 2019).

WBE has demonstrated potential in monitoring different human health and behaviour markers to provide valuable information for public health activities. However, this remains a relatively novel epidemiological tool, and to facilitate the expansion of WBE to other health targets and integration into public health surveillance activities research gaps need to be identified, so further study can be directed in these areas.

The aim of this paper is to review the applications of WBE for public health surveillance, focusing on the following three broad areas: (1) pathogenic viruses; (2) AMR; and (3) chemicals. It describes the additional value WBE contributes to public health surveillance, evaluates the evidence base supporting these applications, and identifies research gaps that need to be addressed to permit large-scale adoption of WBE as a central pillar in public health surveillance and decision-making.

Viruses that cause diseases in humans significantly impact life quality and expectancy, put strain on healthcare settings, and negatively impact economies (Shang et al. 2021). Many novel pathogens to emerge in the past 100 years have been zoonotic in origin, including SARS-CoV-1 and Middle East respiratory syndrome coronavirus (MERS-CoV); human immunodeficiency virus (HIV); novel influenza viruses H1N1, H3, N2, and H2N2; Zika virus; and Ebola virus (Sharp & Hahn 2011; Morse et al. 2012; Recht et al. 2020). Furthermore, the most recent pandemic virus, SARS-CoV-2, is likely to have originated in a zoonotic reservoir (although the zoonotic origin has not been conclusively identified) (Holmes et al. 2021). The conditions for the emergence of novel viruses with pandemic potential remain (Woolhouse et al. 2008; Plowright et al. 2021) and the development of robust plans and resources to monitor the progress of pathogen spread and inform public health interventions is essential to adequately prepare for subsequent pandemics. Endemic pathogens, such as influenza virus, non-influenza RNA respiratory viruses (e.g., respiratory syncytial virus (RSV)), and enteric viruses (e.g., norovirus, rotavirus, etc.) also have a significant health burden, particularly among children, the elderly, and the immunocompromised (Modjarrad et al. 2016; Paget et al. 2019). Enhanced monitoring for pathogenic viruses imposing high burdens of disease by testing wastewater could help to reduce the number of respiratory infections and sequelae (Jartti et al. 2006; Ruotsalainen et al. 2013).

Currently, clinical surveillance for pathogenic viruses is limited to those who seek treatment, which excludes those who are asymptomatic, mildly symptomatic, or who do not present for treatment for other reasons (e.g., lack of access to healthcare or mistrust in healthcare systems; Tsimtsiou et al. 2021). There have been attempts to use data collection methods that would include information on populations not seeking clinical treatment, as well as people who do. This has been done using surveys of symptoms, such as the FluSurvey in Europe (FluSurvey 2017), or Flu Near You in the United States and Canada (FluSurvey 2017; Baltrusaitis et al. 2017), and analysis of digital search engine data on influenza-like illness by FluModel (i-sense-flu) (Lampos et al. 2015; Wagner et al. 2018). However, the results of surveys and search engine data analyses are also affected by biases such as a greater proportion of participants from more affluent and educated backgrounds than the population baseline tend to take part in voluntary surveys (Baltrusaitis et al. 2017; FluSurvey 2017). Models using digital search engine data to track search trends are biased towards populations with regular internet usage, and for whom surveillance data exists to train the model.

For WBE to be useful for viral pathogen surveillance, markers of specific viral infection must be present in faeces, urine, or vomit (Jones et al. 2020). The shedding patterns of viral particles are likely to be affected by several factors, including virus-specific factors (e.g., infection site, incubation period) and host-specific factors (e.g., symptom severity, susceptibility to infection, and age).

The evidence base for faecal shedding of enteric viruses is well established (Supplementary Material, S1), (Allard et al. 1992; Papapetropoulou & Vantarakis 1998; Chitambar et al. 2001; Maunula et al. 2009; Lion et al. 2010; Jeulin et al. 2011; Ganesh & Lin 2013), since the transmission of such viruses is understood to be via the faecal–oral route. However, quantitative estimates of faecal shedding of respiratory viruses are less well explored (Supplementary Material, S1) since their primary transmission route is through respiratory aerosol droplets, rather than water or food contamination.

Interest in WBE for the pandemic respiratory virus, SARS-CoV-2, has resulted in numerous studies producing quantitative and temporal information related to faecal shedding of SARS-CoV-2 RNA (Cheung et al. 2020; Jones et al. 2020; Hoffmann & Alsing 2021). Evidence suggests SARS-CoV-2 shedding is greatest early in the course of the infection cycle and then decreases and continues for several days after symptoms and respiratory shedding have ceased (Gupta et al. 2020b; Van Doorn et al. 2020; Xu et al. 2020; Cavany et al. 2022; Hoffmann & Alsing 2021; Zhang et al. 2021b). For most non-enteric viruses, shedding studies are limited to a binary positive/negative result and are conducted when gastrointestinal symptoms are also experienced and may only be detected in a minority of confirmed cases (von Listow 2006; Minodier et al. 2015). Furthermore, the faecal and urinary shedding dynamics of respiratory viruses may not align with shedding at other sites. For example, prolonged shedding of H7N9 Avian Influenza has been observed in faecal and urine samples after respiratory samples became negative and viral detection was more common in the urine of severely ill patients (Hu et al. 2013a; Zhu et al. 2015).

For most viral pathogens, including SARS-CoV-2, our knowledge of shedding rates in faeces and urine have been obtained from hospitalised individuals, thus biasing the results towards cases presenting with more severe symptoms or comorbidities. Additionally, there is scant understanding of how shedding varies with the age of the patient, comorbidities, vaccination status, or re-infection. Data are therefore not representative of the general population, and future research needs to address this bias by ensuring diversity in symptom phenotypes and immunological states is represented in study cohorts. Limited available evidence (e.g., from norovirus, Sabrià et al. 2016) underscores the importance of understanding the differences in shedding of different viruses in faeces and urine from both symptomatic cases and asymptomatic carriers (Jones et al. 2020), who will both contribute waste to sewer systems. Further data are also needed on shedding dynamics, as existing evidence rarely captures the pre-symptomatic phase when shedding rates may be high (Hoffmann & Alsing 2021). The ability to detect outbreaks before cases present in healthcare settings is extremely valuable for public health planning and infection control. Information is also needed on the persistence of faecal shedding following the infectious period to improve the ability to detect active cases (Santos et al. 2020). Understanding how immunity (whether attained through vaccination or previous infection) impacts shedding is also necessary, as this could potentially influence the interpretation of epidemiological estimates through different phases of an outbreak.

In general, urine samples provide lower sensitivity for viral detection, particularly in enteric pathogens such as Enterovirus and Parechovirus (>90% in faeces vs. <60% in urine) (Crom et al. 2013). However, variability exists and for some pathogens shedding in other excretions may be the predominant source: the measles virus (measles morbillivirus) has been found in the urine of ∼88% of infected patients in one study (van Binnendijk et al. 2003), and viral markers may be present in body fluids which also enter sewer systems (Crank et al. 2022), such as sputum (e.g., SARS-CoV-2 (Li et al. 2022)), saliva (e.g., human herpes virus (Fox et al. 1990), Epstein–Barr virus (Niederman et al. 1976)), semen (e.g., HIV-1 (Kariuki et al. 2020), hepatitis C virus (Cavalheiro et al. 2008), and Zika virus (Atkinson et al. 2016)). This is particularly relevant when developing the sampling framework to account for differences where samples may be more heterogenous (e.g., building-level surveillance; Crank et al. 2022).

The stability of pathogen markers in the environment and in wastewater are highly variable (Bofill-Mas et al. 2006). In general, DNA targets are more stable than RNA targets, although physiochemical characteristics of the environment influence the persistence of both (Ward & Ashley 1980; Bofill-Mas et al. 2006; Muirhead et al. 2020). Enveloped viruses (viruses with a lipid envelope that include influenza viruses, coronaviruses, and parainfluenza viruses) are usually less stable in the environment than non-enveloped viruses (without a lipid envelope, including enteroviruses, noroviruses, and adenoviruses). Noroviruses and adenoviruses can survive up to months in the environment (Kotwal & Cannon 2014) and have frequently been detected in wastewater and surface waters (Lago et al. 2003; La Rosa et al. 2010; Rajko-Nenow et al. 2013; Guillois et al. 2016; Farkas et al. 2018; Falman et al. 2019; Mancini et al. 2019; McCall et al. 2021). However, very limited information currently exists on the persistence of the genetic material of enveloped viruses in the environment, especially in wastewater. It is only recently that Ebola (Zaire ebolavirus) was found to persist in water for 8 days (Bibby et al. 2015), whereas it was previously assumed that enveloped viruses were quickly deactivated in the environment (Wigginton et al. 2015; Ye et al. 2016).

Due to the requirement for biosafety level (BSL) 3 laboratories to perform infectivity studies, surrogate viruses such as Phi6 bacteriophage have been used to assess the potential for survivability (Aquino de Carvalho et al. 2017). The persistence of Phi6 bacteriophage under different temperatures, biological activity, and liquid media was found to vary significantly, with temperature being particularly important (Aquino de Carvalho et al. 2017). Even though temperature greatly affects viral survival, most WBE studies do not take temperature or seasonality into account and risk mis-quantifying results at different times of the year, particularly in temperate regions (Hart & Halden 2020). The presence of stormwater in combined sewers can also result in dilution which can limit the accurate quantification of microbial communities (Wade et al. 2022). Even in separated systems, fluctuating groundwater infiltration has the potential to impact observed results, in a way that is not immediately apparent (Choi et al. 2018).

Matrix composition has repeatedly been found to influence virus detection efficiency and survival in wastewater. For example, organic matter has a great impact on molecular detection as they inactivate the enzymes used for DNA amplification (Matheson et al. 2010; Hata et al. 2015). Ammonia can inactivate viruses through genome degradation and viruses could be aggregated or adsorbed into liquid media (Bibby et al. 2015; Aquino de Carvalho et al. 2017). A study found that aggregation of viruses in the solid fraction of the wastewater was higher for enveloped viruses than non-enveloped viruses at 26 and 6% of virus adsorbed, respectively (Ye et al. 2016). Therefore, sampling only the liquid fraction may exclude a large proportion of the viral target.

Biotic factors may also impact detection of viral targets in wastewater. Wastewater is a complex and dynamic mix of microbial communities and competition, predation and selective pressures are not well understood (Choi et al. 2018). In some locations, urban wastewater may also contain animal viruses (e.g., influenza A; Heijnen & Medema 2011), which can confound the signal due to difficulties in separating human and animal contributions without sequencing-based approaches.

It is evident that mechanisms of virus marker inactivation in wastewater environments for different viruses of interest require further investigation. The observed differences between faecal and sewage samples indicate environmental conditions that alter microbial communities and further research is required to be able to interpret viral signals in WBE accurately.

Wastewater sampling for viral pathogens may use flow-based interval approaches, but is typically achieved through grab or composite sampling. Grab samples are collected at a single point in time, often at a time of day that is estimated to be ‘peak flow’ to maximise the target load from human sources (Polo et al. 2020). However, this can vary greatly depending on the location of sampling sites as network travel times of wastewater can typically range between 2 and 12 h (Castiglioni et al. 2013) and peak shower usage may dilute the signal (Ahmed et al. 2021). Some markers can have greater daily variability than others (Ahmed et al. 2021). Due to the temporal variability of wastewater concentrations, this is often not representative of the entire population input over 24 h, therefore a composite sample may be used. Composite sampling is a means to try and overcome this, typically via the use of autosamplers, which sample at set intervals (ranging from minutes to hours) and can be used to develop a time series or mixed to make a composite sub-sample. Grab samples appear to be suitable for enteric viruses in wastewater effluent (i.e., treated wastewater) (Sidhu et al. 2017; Farkas et al. 2018), however, variation in wastewater influent means daily composite samples are the best approach for enteric viruses (∼4 samples per day; Farkas et al. 2018) and SARS-CoV-2 detection in influent wastewater (24 samples per day; Ahmed et al. 2021). Further temporal studies to determine enteric pathogen variation over 24 h are recommended (Farkas et al. 2018) and whether these patterns are consistent across various wastewater systems. There is a lack of research comparing sampling techniques for other respiratory viruses. The use of autosamplers requires specific infrastructure and is associated with a higher operational workload and cost, as well as the need for on-site sample refrigeration. These facilities are not always available, particularly in resource-poor areas.

Passive samplers are absorbent devices that can accumulate markers of interest when submerged in wastewater and may offer a good alternative to composite sampling. They do not require specialised equipment to be deployed (Habtwold et al. 2022; Schang et al. 2021), and can be used for sampling in the sewer network where autosampler use is impractical (Corchis-Scott et al. 2021) and are considerably less costly (Petrie et al. 2016). Passive samplers have been successfully used to detect viruses, such as hepatitis A and E, norovirus (Vincent-Hubert et al. 2017), poliovirus (Sattar & Westwood 1977; Cassemiro et al. 2016), SARS-CoV-2 (Schang et al. 2021). However, they may saturate quickly, and daily exposure times may be too long to obtain representative results. Furthermore, as passive samples are submerged and accumulate biological material over time, often to the point of saturation, viral concentration data obtained from passive samplers cannot be easily normalised for factors such as wastewater flow. Thus, the results are considered semi-quantitative. Overall, insight into wastewater dynamics and temporal variation of the marker of interest is needed for different pathogens to determine the most accurate and cost-effective sampling method. This could vary significantly between different sites depending on the pathogen of interest, the availability of infrastructure to permit composite sampling, and whether quantitative data are required for public health use.

There is currently limited standardised guidance on wastewater sample storage conditions and duration prior to molecular marker testing for viruses. It has been suggested that samples be kept refrigerated and analysed as soon as possible after collection, although temperatures and duration of storage vary, depending on the type of viral analyte (e.g., RNA/DNA, enveloped vs. non-enveloped viruses). For example, the US Centers for Disease Control and Prevention (CDC) recommends immediate sample refrigeration at 4 °C, and processing with 24 h (CDC 2022), whereas the US Environment Protection Agency (USEPA) recommends that samples be stored and chilled for up to 48–72 h prior to viable bacteriophage testing (USEPA 2001). For enterovirus viability and norovirus RNA testing in drinking water, the USEPA suggests storage for up to 24 h prior to analysis (Fout et al. 2010). Standardised methods need to be developed and agreed upon to allow comparability between studies, times and locations, and improve quality assurance.

Different methods for optimising RNA extraction, concentration, and quantification of SARS-CoV-2 have been explored. Concentration methods for viral pathogens in wastewater, primarily developed for non-enveloped enteric viruses, have since been applied for SARS-CoV-2 surveillance (Ahmed et al. 2021). These include precipitation with salt or polyethene glycol (PEG) (Farkas et al. 2021; Kevill et al. 2022), electrostatically charged membrane filtration (Ahmed et al. 2021), ultrafiltration (Izquierdo-Lara et al. 2021), adsorption-extraction (AE) with aluminium chloride, silica (Randazzo et al. 2020) or magnesium chloride (MgCl₂) (Ahmed et al. 2021), or using an automatic concentrating pipette (CP) instrument (Ahmed et al. 2021; McMinn et al. 2021). Concentration methods have been found to be one of the greatest sources of variation in the quantification of SARS-CoV-2 in the English wastewater monitoring programme, based on the recovery of Phi6 bacteriophage, which was used as a process control (Wade et al. 2022).

To improve the efficiency of wastewater sampling, testing samples for the presence of multiple viral markers would necessitate optimising viral recovery for multiple pathogens. However, a study by Ahmed et al. (2021) evaluated two concentration techniques (AE supplemented with MgCl₂ and CP) against concentration of SARS-CoV-2 and enteric pathogens. They found the CP method achieved the greatest overall recovery of SARS-CoV-2, however recovery did vary and was not consistent across all viral targets. Another study found viral recovery was reduced when CP was applied to high turbidity samples (Kevill et al. 2022). Further research should investigate concentration methods for different pathogens in wastewater and recovery efficiencies. This is key to optimising multi-pathogen recovery methods, minimising equipment requirements, and streamlining lab throughput. Further method development could additionally focus on optimisation in terms of concentration methods to enable both RNA and DNA extraction, where the latter could also be analysed using culture-independent methods for AMR surveillance, for example.

The gold standard for the targeted detection of viral pathogens in wastewater is quantitative polymerase chain reaction (qPCR). Although ubiquitous in much of WBE programmes, the method is sensitive to biological and chemical inhibitors, which are often present in wastewater extracts (Volkmann et al. 2007). Commercially available reaction mixes provide different levels of resistance to inhibitors, which may result in false-negative results and generate inter-laboratory variation (Farkas et al. 2017). The validation and implementation of strict quality control measures for each qPCR assay are crucial to achieving reliable results (Bustin et al. 2009). For example, one meta-analysis found that only 26% of studies published on SARS-CoV-2 detection in wastewater reported qPCR quality control details and where they were reported did not meet the minimum criteria (Bivins et al. 2021), questioning the reliability of the results.

Emerging targeted technologies such as digital droplet PCR (ddPCR), isothermal amplification (e.g., LAMP) assays and biosensors have also been developed; however, some of these are often less sensitive than qPCR (Haramoto et al. 2018; Farkas et al. 2020b). To date, besides qPCR, only ddPCR has been routinely used for WBE (Jahne et al. 2020; Al-Duroobi et al. 2021; Bivins & Bibby 2021), as this technique may be more sensitive than qPCR (Zhao et al. 2016), particularly for low viral loads or in matrices with more variable contamination (Taylor et al. 2017; Falzone et al. 2020).

Sequencing-based approaches have been trialled and implemented in WBE to generate data describing viral communities, including variant-level identification of circulating viruses (Fernandez-Cassi et al. 2018; Adriaenssens et al. 2018, 2021; Guajardo-Leiva et al. 2020) and identification of novel viruses and their variants (Rios et al. 2021). However, existing methods are time-consuming and often biased, tending to over-represent abundant or large DNA genomes and miss rare, shorter genomes (Garner et al. 2021). Amplicon sequencing increases sensitivity and accuracy for known viruses and has been effective in wastewater research to describe certain viral groups, including astroviruses (Hata et al. 2018), adenoviruses (Ogorzaly et al. 2015; Iaconelli et al. 2017), saproviruses (Mancini et al. 2019), enteroviruses (Tao et al. 2020), and noroviruses (Fumian et al. 2019). A similar approach has been used for SARS-CoV-2 by amplifying its whole genome or certain genes using the Illumina ARTIC pipeline (Avgeris et al. 2021; Dharmadhikari et al. 2021; Hillary et al. 2021).

Ultimately, the methodology for detection and quantification of viral pathogens will depend on the requirements of the data for public health. Research on the application of these technologies to wastewater samples for various pathogens is necessary to demonstrate their detection and quantification capability. Additionally, if it is necessary to detect several pathogens in one sample, development of high-throughput quantification tools for multiple targets would be necessary.

Clinical data, such as microbiological diagnostic data, may be used to assess how well markers in sewers correlate temporally and spatially with specific health outcomes in a community of interest. Comparison of WBE data should encompass two broad aspects: (1) that the markers reliably measure the intended target of surveillance (i.e., sensitivity and specificity) and (2) that data are collected at a temporal and spatial resolution that enable public health decision-making.

For SARS-CoV-2 in the UK, WBE data could be compared with public health data, including hospital admissions (Kaplan et al. 2020; Peccia et al. 2020; Galani et al. 2022), case data from the national and regional surveillance programmes (Wade et al. 2022), the Coronavirus Infection Survey (CIS) (Pouwels et al. 2020), and Real-time Assessment of Community Transmission studies (REACT) (Elliott et al. 2021). Comparisons between datasets also reveals potential population behaviours driving transmission, including social response to the pandemic, temporally resolved variations in population size (e.g., commuting, weekend breaks), sampling biases in wastewater, or biases, gaps, or reporting lags in clinical testing (Galani et al. 2022).

While the extensive clinical testing available throughout the pandemic meant comparable data was readily available (noting limitations within these data (Elliott et al. 2021)), similar comparisons might not be as feasible for pathogens like norovirus or RSV, where monitoring is dependent on data from healthcare settings. Therefore, to ensure the data collected through WBE is an accurate reflection of infection rates, future research must ensure these comparisons are designed into studies (e.g., by collecting samples or data from healthcare institutions, the wider community, and open data sources, such as social media).

Asymptomatic carriage of pathogenic viruses by members of the community is currently under-estimated but can play an important role in disease transmission. WBE has the potential to provide additional information on viral transmission dynamics that are currently missed by existing surveillance programmes, which are biased towards symptomatic and severely ill patients. However, knowledge gaps remain that must be addressed before WBE can be useful as a stand-alone surveillance tool to inform public health practice.

• Further quantitative studies on the shedding of viral pathogens (particularly for enveloped viruses such as influenza) are needed. Studies should ideally monitor patients throughout the entire course of the infection, with a range of variants, and recruit diverse cohorts to ensure generalisability of results. Shedding should be evaluated in faeces and urine. Impacts of immunity against viruses (achieved through vaccination or infection) on viral shedding dynamics need to be elucidated.

• Further studies on the persistence of various viruses in wastewater are necessary (particularly enveloped viruses), including evaluation of the impact of factors such as temperature, matrix composition, and biotic processes on viral detection. Researching this within the context of WBE may inform data interpretation for public health.

• Different sampling procedures and analytical techniques should be explored for pathogens of interest to optimise methods. The extent to which viral pathogen detection in wastewater samples can pre-empt or match clinical and community-level data must be evaluated to ensure that WBE data are an accurate reflection of clinical outcomes of interest.

AMR, where microbes develop resistance against antimicrobial drugs, poses a threat to public health that is only likely to worsen. Antimicrobials are becoming less effective over time with continued overuse (Prestinaci et al. 2015) and fewer are being brought to market. Economic impacts are expected due to reduced productivity, resulting from lost days-of-work and longer hospital admissions (WEF 2021).

Compared to infectious viral pathogens described in the previous section, monitoring AMR is complicated given there is no one singular infectious agent responsible, nor single reservoirs for resistant organisms. There are numerous pathogenic antibiotic-resistant bacteria (ARBs) of concern, and antibiotic-resistant genes (ARGs), which can also be released in faecal matter and transferred between even distantly related bacteria by horizontal gene transfer (HGT). Beyond bacteria, resistance to antimicrobial treatment arises in other microorganisms such as azole antifungal resistance in pathogenic fungi, like Candida albicans (Whaley et al. 2017), antiviral resistance, including resistance to antivirals for hepatitis B (Locarnini 2004) and influenza virus (Lampejo 2020; Van Poelvoorde et al. 2020) and resistance among parasites, like Plasmodium spp.

AMR is an issue that is now recognised as extending beyond healthcare settings, with resistance determinants emerging in environmental settings (Larsen et al. 2022; Poirel et al. 2002) and selected for by antimicrobial use in agriculture systems (Woolhouse et al. 2015; Cheng et al. 2019; Pollock et al. 2020), the release of resistance in sewage into waterways and general environmental pollution (Harris et al. 2014; Agga et al. 2015; Murray et al. 2018; Hiller et al. 2019; Mughini-Gras et al. 2019). Therefore, surveillance and management of AMR are often considered through a One Health lens including environmental and animal monitoring (Mendelson & Matsoso 2015; WHO 2021a).

With current monitoring focussing on healthcare settings and a small group of key pathogens (e.g., English Surveillance for Antimicrobial use and Resistance (ESPAUR), (UKHSA 2021); European Antimicrobial resistance surveillance (EARS-Net), and WHO Global AMR Surveillance System (GLASS), (WHO 2021a)), the true scale and potential burden of AMR may be under-estimated (Burnham et al. 2019). ARGs, which are abundant in stool and wastewater samples, are the main way of indirectly monitoring AMR prevalence at large scales. The most comprehensive resistance monitoring study quantified ARGs in wastewater metagenomes in over 70 countries and found major differences in relative AMR abundances and classes in community sewage, reflecting large regional differences in ‘local guts’ (Hendriksen et al. 2019). Furthermore, a recent review by Pruden et al. (2021) described a range of applications of WBE for AMR monitoring including the following: (1) identification of hotspots to guide policy action on interventions; (2) inform medical and veterinary practitioners of the best antibiotics to use to avoid therapeutic failure; (3) inform risk assessments for discharge of antibiotics/ARB/ARGs into the wider environment; and (4) provide data for forecasting models and key drivers. Overall, the use cases for AMR surveillance using wastewater demonstrate the potential to add significantly to public health monitoring and therefore warrant further exploration.

AMR is represented in stool by a diverse array of ARGs and mobile genetic elements (MGEs), primarily associated with bacterial hosts. Resistance carriage in humans varies substantially internationally due to different environmental exposures, diet, and antimicrobial use (Hu et al. 2013b; Pal et al. 2016; Feng et al. 2018). The abundance and diversity of ARGs also varies between body sites/tissues, with one study using metagenomic sequencing on oral and stool samples in different countries and finding a higher abundance, but lower diversity, of ARGs present in oral samples compared to stool samples (Carr et al. 2020). Studying stool samples may provide information not captured from data such as antibiotic prescription rates, which do not always correlate with ARGs present in faeces (Collignon et al. 2018; Hendriksen et al. 2019; Carr et al. 2020).

Quantification of ARGs present in stool using metagenomics and qPCR has been well studied (Hu et al. 2013b; Rose et al. 2017; Feng et al. 2018) and in urine mostly studied in the limited context of urinary tract infections (Li et al. 2020; Kaye et al. 2021). However, the clinical relevance of ARG detection in metagenomic data is poorly understood since the presence of ARGs in faecal samples does not necessarily result in AMR infections, even in high-risk patients (Tacconelli et al. 2009; Wryes et al. 2021). Furthermore, some genes (e.g., tetracycline genes) perform multiple functions unrelated to AMR, including signal tracking and hydrophobic protein transport (Martínez 2008; Dantas & Sommer 2012; Feng et al. 2018). There is also the question as to whether ARGs present in the gut and stool are an accurate indication of resistance at other parts of the body (Piddock 2016; Carr et al. 2020), the duration of colonisation, and risk factors extending shedding in faeces (Leonard et al. 2018). Further research in this area is also required for ARGs in urine samples, since there is evidence to suggest enhanced resistance following kidney transplantation is reflected in urine (Rani et al. 2020). Overall, the link between ARG detection in stool and urine to treatment-resistant infections among various at-risk groups needs further exploration.

Like viral markers, the stability and abundance of AMR indicators in wastewater are complicated by many abiotic and biotic factors that influence microbial populations and resistomes (collection of AMR genes within the microbiome). As with viral pathogens, animal hosts of AMR can contribute ARGs to wastewater, making it difficult to precisely identify which AMR genes in a sample originated from humans (McCann et al. 2019). Biofilms present in sewers may also serve as a source of ARGs in wastewater independent of human waste inputs (Medina et al. 2020), and dispersion of ARG-bearing microbes may be triggered by the conditions in sewer environments (Kaplan 2010). In addition, bacteria and ARGs are capable of multiplying in the environment, and the presence of antibiotic residues, heavy metals, and biocides in wastewater can influence the selection of certain bacteria and their genes, and therefore their abundance, outside the human body (Baquero et al. 2008; Knapp et al. 2008; Martinez 2009; Gullberg et al. 2011; Murray et al. 2018).

There are many additional factors in wastewater that have an impact on the microbial community and resistome dynamics, including local residence times, geographic area, antibiotic prescribing, temperature, salinity, dissolved oxygen, micro-ecological factors such as predation, and the presence of organic matter (Caucci et al. 2016; Sun et al. 2016; Jiao et al. 2018; Lamba et al. 2018; Jong et al. 2020; Pallares-Vega et al. 2021). Additionally, further research on different wastewater treatment plant types and maintenance and impacts of combined sewer overflows (CSOs) on AMR detection is critical in places with aging or poorly maintained sewer systems.

Due to the higher antibiotic usage in hospital settings, previous studies have investigated the impact of hospital wastewater mixing with community wastewater on resistomes, with mixed results (Islam et al. 2017; Lamba et al. 2017; Buelow et al. 2018, 2020; Quintela-Baluja et al. 2019, 2021; Li et al. 2021). Future work should focus on evaluating the factors that affect the influence of sewer conditions and inputs to the sewer systems that impact measures of AMR. Such research will aid interpretations of AMR data derived from wastewater samples.

As with viral markers, methods for the collection and storage of samples must be evaluated to ensure robust data are extracted from samples. A systematic review by Chau et al. (2021) compared composite and grab samples to evaluate performance according to association with clinical AMR data and found greater agreement with composite samples. The authors observed that several studies did not report their sampling methodologies, limiting comparability of studies and highlighting the urgent need for standardisation of approaches (Hassoun-Kheir et al. 2020). There are also few studies on the use of passive samplers for AMR surveillance, which may provide a more cost-effective and practical approach to sampling than composite samplers. Further research should seek to determine the most accurate and cost-effective sampling methodologies to inform standard practice, to allow data to be compared regionally and internationally. This has commenced through the EMBRACE-WATERS (rEporting antiMicroBial ResistAnCE in WATERS) initiative, which is aiming to develop reporting standards for AMR research in wastewater and other aqueous environments (Hassoun-Kheir et al. 2021).

Poulsen et al. (2021) compared storage conditions on the microbiome and resistome of sewage samples and pig faeces. They observed that sample storage conditions had an influence on microbial community and resistance genes, where freezing wastewater changed the microbiome in comparison to processing the samples immediately. AMR abundance in frozen sewage samples was also overall significantly higher than samples stored at 22 or 5 °C (Poulsen et al. 2021). Therefore, it was recommended that sewage samples are processed the same day, frozen immediately, or stored in the fridge and processed during the day or on the following day. However, the opposite finding was observed for pig faeces, indicating that stool samples, or perhaps even near-source samples have different storage requirements (Poulsen et al. 2021). Further research on the storage requirements of near-source samples would inform standard practice and standard protocols.

Access to storage facilities and appropriate alternatives need to be considered, to ensure sampling is consistent and samples are comparable.

The use of multi-array high-throughput qPCR arrays has shown promise because they can be designed to capture many different indicator genes at once (Wang et al. 2014). These have already been used successfully for wastewater applications (Quintela-Baluja et al. 2019).

In recent years, metagenomic sequencing has proved to be a powerful approach for the detection of AMR. Metagenomic sequencing has the potential to reveal the diversity of ARGs, from natural ocean and soils to samples from the guts of animals, humans, and sewage (Hendriksen et al. 2019; Zeng et al. 2019). Factors included in metagenomic pipelines such as the MetaCompare's relative resistome risk (Martínez et al. 2015), can assign a value to the mobility of ARGs or the potential to occur in pathogens which could be useful for determining clinical risk using metagenomic data. Sequencing-based approaches can also be used with and without a culture phase: targeted metagenomics permits the quantification of thousands of ARGs in culturable and potentially pathogenic bacteria (Leonard et al. 2018). As with viruses, detection of novel and rare genes will be limited.

Multiple pipelines for detection and enumeration of AMR exist (Gupta et al. 2020a), but a standardised method is needed for WBE if opportunities to compare data internationally are to be realised. With the high cost of sequencing-based approaches, this approach may not be accessible to some regions, therefore centralised sample processing and data analysis resources and funding to LMICs is needed (Aarestrup & Woolhouse 2020). Development of combined sequencing with multiplex qPCR and culturing methods would make monitoring more accessible across different socioeconomic settings.

There is some evidence that WBE AMR data are correlated with data on regional clinical infections. Hutinel et al. (2019) demonstrated strong and significant correlations between E. coli isolated from wastewater samples and patient samples from Gothenburg, Sweden. Significant correlations were also found between samples taken from 10 wastewater treatment plants in Europe and clinical samples from EARS-Net, where resistance of E. coli to four antibiotics was analysed (Huijbers et al. 2020). Genotypic data gathered via qPCR arrays have also been found to broadly reflect clinical resistance in European sites (Pärnänen et al. 2019). Chau et al. (2021) identified a strong association with phenotypic and genotypic data from WBE to data gathered from clinical settings, indicating WBE provided insight into the resistance profiles of a community.

Comparisons between clinical and wastewater data have also shown WBE can be used to inform colonisation of specific pathogens, such as carbapenem-producing Enterobacteriaceae (CPE). CPEs are a major threat to healthcare settings and screening is currently limited to those who are most at risk (i.e., who have travelled, came into contact with a positive case, or visited a hospital abroad (Brown & Fry 2020)). Studies have found wastewater CPE cultures are reflective of what is observed in clinical samples (Meir-Gruber et al. 2016; Ludden et al. 2017; Flach et al. 2021). Isolates have been identified in WBE prior to identification in clinical samples, demonstrating its potential use as an early warning tool to prevent hospital outbreaks (Flach et al. 2021). WBE has also identified strains which were not detected in clinical samples, interpreted to be due to a carrier that was missed from clinical screening, or a result of persistence in wastewater after previous exposure (White et al. 2016).

While there are correlations between WBE and clinical resistance datasets, there is currently too much uncertainty to enable public health response. An improved understanding of the connection between clinical colonisation and wastewater could be achieved through more comprehensive studies including participants who are deemed low risk, and further investigation into the impact of wastewater on the persistence of resistance.

Surveillance of AMR using wastewater is complicated by the ubiquity of AMR across the One Health triad. Detection of AMR in wastewater may therefore not be from human carriers experiencing or at risk of treatment-resistant infection, but may be there as a result of zoonotic reservoirs, or persistence of AMR in the environment. Further research is needed before this tool can be used to support public health decision-making.

• The link between ARG detection in stool and urine to treatment-resistant infections among various at-risk groups warrants further exploration.

• Further studies should consider the effect of abiotic and biotic factors, selection processes, the influence of seasonality, and different types of wastewater treatment plants as this will aid interpretation of WBE data. The influence of wastewater from healthcare settings should be reviewed on a case-by-case.

• Further research should determine the best sampling and storage methodologies with consideration to resource constraints to inform standard practice and facilitate data comparisons.

• A standardised methodology for wastewater sample laboratory analysis is needed to allow international collaboration and comparisons. This approach needs to be accessible for LMICs.

• Improved understanding of the link between WBE and clinical resistance could be achieved through a comprehensive study incorporating participant and wastewater sampling and further investigation into the impact of wastewater on the persistence of resistance.

Chemical analysis of wastewater provides a broad range of information that can be used to understand multiple parameters influencing public health. Potential insights include but are not limited to (1) illicit drug use (Thomas et al. 2012; Ort et al. 2014; González-Mariño et al. 2020), (2) alcohol and tobacco consumption (Baz-Lomba et al. 2016), (3) consumption of over-the-counter medicines (Harman et al. 2011; Fattore et al. 2016; Melchor-Martínez et al. 2021; Rice et al. 2020), (4) consumption of antimicrobials (Castrignanò et al. 2020; Elder et al. 2021), (5) exposure to chemicals such as pesticides (Rousis et al. 2020), toxicants from everyday products including parabens, UV filters, bisphenols (BPAs), and phthalates (Lopardo et al. 2019a; Senta et al. 2020), and (6) biomarkers for certain non-communicable diseases, for example, cocaine, F2-isoprostanes, Prostaglandin E2 (Thomas & Reid 2011; O'Brien et al. 2019).

WBE has long been recognised as an approach for monitoring illicit drug use. Illicit drug use data are especially challenging in terms of surveillance, due to the behaviour being illegal, and highly stigmatised (Huizer et al. 2021). For this purpose, it fills significant gaps in other monitoring techniques such as surveys, drug-related hospital admissions, and arrest data, which do not capture the extent of consumption and are not temporally and spatially sensitive (González-Mariño et al. 2020; Huizer et al. 2021). This approach has been used widely, including an international study in 2011, where five illicit drugs (cocaine, cannabis, amphetamine, methamphetamine, and methylenedioxymethamphetamine (MDMA)) were analysed in the wastewater of initially 19 European cities, increasing to a total of 73 cities globally by 2017, including locations in LMICs (Thomas et al. 2012; Gonzalez et al. 2020). Ongoing monitoring in 80 European cities is supported by the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) which hosts an interactive website (EMCDDA 2021). Antibiotic consumption has also been investigated within the context of ARG prevalence in wastewater, showing strong correlations between antibiotic levels, ARGs, and population size (Castrignanò et al. 2020; Elder et al. 2021).

Chemical analysis of wastewater could therefore provide spatially and temporally comprehensive epidemiological information to aid public health and healthcare practitioners. Longitudinal analysis would allow regional and temporal comparisons to be made and help to understand principal mechanisms driving public health trends. WBE could also provide information in data-poor regions, where antimicrobials, prescription drugs, and exposure to pollutants may be less regulated.

Chemical residues are excreted through urine and stool and can be detected in wastewater, usually reported as a percentage of the target residue (Baker et al. 2014). Understanding of human pharmacokinetics including excretion rates are therefore essential to back-calculate the approximate total exposure or consumption of chemicals (Zhang et al. 2019).

A comparison between the parent compounds and their metabolites for WBE found a higher certainty using metabolic markers (Baker et al. 2014), while other studies went further to suggest that only metabolic markers should be used in calculations where possible (Kasprzyk-Hordern et al. 2021).

An extensive list of excretion rates of biomarkers for lifestyle (e.g., illicit drugs, alcohol), exposure (e.g., pesticides), health (e.g., pharmaceuticals), and population (e.g., endogenous compounds) is available from Gracia-Lor et al. (2017). These estimates are from studies published after 2000, and primarily after 2010. However, some metabolites only have urine or faecal excretion percentage reported, and rarely both. This means some studies estimating exposure have only used excretion rates from urine in back-calculations (van Nuijs et al. 2015), likely underestimating exposure since often only phase I metabolites of chemicals are excreted in urine. A more accurate approach would be to include both phase I and II metabolites, which can be found in urine and faeces (Daughton 2018; Escolà Casas et al. 2021).

Further research is needed to understand excretion data for metabolites, which are often less complete than for parent compounds and varies depending on individual and medical conditions (Escolà Casas et al. 2021). Unfortunately, further pharmacokinetic studies on excretion rates are not always feasible, due to high costs, time requirements, and ethical considerations (Gracia-Lor et al. 2017). Alternative approaches include in vitro studies using liver enzymes (Mardal et al. 2016) and computer modelling (Reid et al. 2014). However, such data provide information on metabolic products and general insight into metabolism, but not quantitative information on excretion rates that could be used for WBE. Clinical proteomics have also led to the identification of new biomarkers. However, due to the time and expense required for clinical trials, very few are being confirmed clinically (Nkuipou-Kenfack et al. 2017; Rice & Kasprzyk-Hordern 2019). Nevertheless, the identification of human proteins in wastewater offers an opportunity for community-wide monitoring of diseases (Carrascal et al. 2020), and further studies on key biomarkers including metabolites as well as other endogenously formed entities such as proteins are needed. Emphasis should be put on improving understanding of pharmacokinetics to help quantify excretion factors for stool and urine, particularly where population variation is likely.

Chemical stability in wastewater varies depending on the original compound and the conditions in wastewater. The stability of illicit drug biomarkers has been reviewed elsewhere (McCall et al. 2016a). A study investigating 43 pharmaceuticals showed that the majority remained stable in a pressurised sewer of 7.6 km, with a concentration error of ±10% for most pharmaceuticals, with some at ∼60% (Jelic et al. 2015). Temperature impacts biomarker stability, with one study showing that half-lives of biomarkers in colder temperatures were significantly longer than in warmer temperatures (Senta et al. 2014). McCall et al. (2016b) also found that temperature, pH, hydraulics, residence times, sorption processes, and biodegradation rates affect drug stability in sewers. Biofilms may also influence the stability of illicit drug biomarkers (McCall et al. 2016a; Li et al. 2018) and particulate matter in sewage has been shown to affect results, where a large proportion of the analyte may partition onto particulate matter (Baker et al. 2012). A study from Tang et al. (2020) found phthalate metabolites in wastewater were in much higher concentrations than predicted from urinary excretion, therefore whether metabolites are formed from parent compounds within wastewater environments should be investigated.

Further studies are required to estimate correction factors for metabolites that are to be used in WBE, particularly through understanding the conditions in wastewater that may influence interpretations, including the amount and type of sewage biofilm and different environmental conditions (e.g., temperature, pH and biodegradation) (McCall et al. 2016a).

Analysis of chemicals and drugs in wastewater is most accurately conducted through 24-h flow-proportional composite sampling. While both time- and flow-proportional sampling is commonly applied, flow-proportional instead of time-proportional samples (e.g., every 20 min), have been found to provide an accurate indication of peak input of chemicals and biomarkers excreted into wastewater (Ort et al. 2010; Petrie et al. 2017). Grab or spot sampling does not allow for full analyte capture due to common diurnal variabilities in wastewater composition linked with variable human excretion patterns as well as external factors, like rainfall. To preserve chemicals for accurate analysis, storage time and temperature are crucial. Samples need to be transported to laboratories refrigerated (if over a short period of time, e.g., a few hours) or frozen (for overnight transport). If chilled, they must be immediately frozen on arrival to prevent analyte degradation. Acidification can improve the stability of chemicals, by preventing some drugs from partitioning to the water fraction (Baker & Kasprzyk-Hordern 2011).

Passive sampling has been investigated (Moschet et al. 2015; Townsend et al. 2018; Rimayi et al. 2019). A comparison of passive and active sampling undertaken by Petrie et al. (2016), found the concentration of chemicals captured from passive sampling was within a factor of 2 of the known concentration. However, passive samplers might not achieve the same resolution as composite samples (Jones et al. 2015), but may nevertheless be used when analyte detection alone is a suitable measure for public health purposes. Further research is needed before it can be used as a standard alternative sampling method where there is insufficient infrastructure for composite sampling.

To preserve chemicals, understanding storage time and temperature for chemical stability is crucial. Chemicals are likely to degrade under repeated freeze–thawing cycles and over long-term storage. In a study by Fedorova et al. (2014), influent wastewater samples were spiked with 124 pharmaceuticals and personal care products to determine the impact of different storage conditions on analyte detection. Storage conditions at 4 °C resulted in greater amounts of analyte being detected compared to samples kept at −18 °C, with 15 analytes undetected after a freeze/thaw cycle. An understanding of the stability of target compounds and biomarkers, and appropriately designed standards for storage is necessary, to ensure consistency and comparability among studies.

Chemical markers are present in wastewater at low concentrations, spanning from sub-ppt to ppm levels. Ultra-performance liquid chromatography is the method of choice for the separation of structurally wide-ranging groups of analytes with variable polarities, but gas chromatography can be also utilised for more volatile targets, also with common derivatisation steps utilised for more polar analytes.

In order to enable analysis at trace ppt levels, analyte enrichment is required. Solid phase extraction with wide-ranging materials is widely utilised to both clean-up and enrich liquid samples. QuEChERS method as well as pressurised liquid extraction or accelerated solvent extraction are also suitable for the extraction of analytes from solid samples such as suspended particulate matter in wastewater (Nieto et al. 2010; Kachhawaha et al. 2017; Pang et al. 2017). High-throughput methods are increasingly being used and within the context of WBE can provide more insight into multiple targets. Triple quadrupole mass spectrometry is usually applied to undertake multi-residue targeted analysis of >100 markers with one analytical run and is highly useful for large catchment studies (Proctor et al. 2019). High-resolution mass spectrometry can also allow untargeted screening for novel chemicals, with subsequent data mining allowing the potential to expand and develop analytes of interest (Lopardo et al. 2019b).

Further research on high-resolution mass spectrometry approaches is required for data mining and identifying further steps in sensing approaches. High-resolution analysis approaches would enable detection of an expanded biomarker suite, and allow multilevel estimation of public exposure to stressors and resulting effects (e.g., disease status).

Further studies are also necessary on chemical population markers. Population normalisation is necessary for all aspects of WBE, where inputs into the sewer may vary over the course of the day or be diluted by stormwater. Estimation of population may also provide insights into population mobility and travel. Metabolic markers for population normalisation are being explored, such as metabolites for caffeine or nicotine (Senta et al. 2015), or prescription pharmaceuticals such as atenolol (Lai et al. 2011). Further application of these biomarkers in WBE studies, and more information on how these products are metabolised, are necessary to use these to quantitatively estimate population.

Several studies have compared prescription and WBE data (Carballa et al. 2008; Lai et al. 2011; Verlicchi et al. 2014; Escolà Casas et al. 2021). Data on illicit drug market trends and confiscated drugs have been used to sense-check WBE illicit drug data (Kankaanpää et al. 2016; Krizman-Matasic et al. 2019; Gonzalez et al. 2020). Considerations for evaluating WBE for pharmaceutical consumption include ensuring prescription data are at an appropriate temporal and spatial scale, where national averaged data are insufficient for comparison (Carballa et al. 2008; Lai et al. 2011; Verlicchi et al. 2014). An example of a study with a higher temporal resolution is Escolà Casas et al. (2021), which compared pharmaceutical detection via WBE against a monthly aggregated pharmaceutical dataset. Such datasets need to be explored and compared against WBE data to ensure data are correct, identify and explain discrepancies, and account for any variation.

Chemicals can also act as indicators of human health. However, these are known to undergo changes in the body and/or the sewer environment, affecting the choice of chemical markers for surveillance, collection and storage of samples, and data interpretation.

• Further research is needed to understand excretion data for metabolites and endogenously formed entities, such as proteins. Improved understanding of pharmacokinetics is necessary to help find excretion factors in stool and urine, and how these may vary due to treatment received or age group.

• There is a lack of information on biomarker stability, particularly for metabolites. Further studies should investigate conditions in the sewer which may impact interpretations of WBE data, including sewage biofilm and different environmental conditions.

• Improved understanding of passive sampling could support WBE in areas where composite sampling is not feasible. Further understanding of the stability of various biomarkers is necessary to design appropriate standards, to ensure consistency and comparability among studies.

• Expansion of the biomarker suite coupled with high-resolution mass spectrometry approaches could allow multilevel estimation of public exposure and resulting effects. Exploration and application of population level biomarkers in WBE studies are necessary to have an improved estimated of population.

• WBE data need to be compared to appropriate datasets at an appropriate temporal and spatial scale to identify and explain discrepancies and account for any variation.

To be useful for public health monitoring, WBE needs to be integrated and add value to other public health initiatives, including clinical case reports, self-reporting systems, or social media tracking systems like FluDetector (Lampos 2016) or ZoeApp (Menni et al. 2021). Current epidemiological systems typically do not integrate clinical and environmental monitoring together. However, such an approach is necessary to ensure both groups are openly communicating, sharing insights, and harmonising approaches. Communication and collaboration with clinical and public health teams are also necessary for ensuring the data collected are useful in informing practice (e.g., identification of clinically relevant resistance within a community would need to be communicated to veterinary and clinical practitioners). Likewise, people working in healthcare might share with WBE teams requirements for novel targets. Collaboration with organisations collecting animal and environmental data could also be beneficial for developing a One Health surveillance framework.

Further research is needed between people working in WBE and those in healthcare settings to develop improved ways for public health users to receive these data and the best communication tools. Further discussions across clinical and environmental sectors is crucial for the design of WBE to ensure this is integrated into existing monitoring and provides a practical application. In addition, communication across the general public warrants further consideration. Publicly available dashboards during the COVID-19 pandemic showing daily case rates should be considered for additional WBE applications, allowing the public to have transparent access to information, raise awareness of health trends and provide information to make informed decisions. An interactive website or centralised dashboard providing clear, accessible summarised/non-identifiable data, integrating environmental, wastewater, and clinical data would be a potential way to communicate this information.

Having an overall understanding and ability to communicate uncertainties is also important. Within the UK, sewage networks are regionally diverse and can be publicly and privately owned, leading to differences in the way sampling is delivered (Wade et al. 2022). Some of the challenges may be especially evident in smaller rural communities, which may have hundreds of small wastewater treatment sites (Farkas et al. 2020a). Being able to communicate this uncertainty to relevant clinical stakeholders, the media and the general public is necessary to ensure data transparency and guide informed interpretations and responses.

Despite wastewater being considered an ethical way of collecting data, further consideration and engagement of social scientists and the public is required to assess the ethics of WBE in applications beyond the pandemic. It is widely assumed that WBE uses aggregate measurements and cannot be used to identify individual persons. However, concerns have been raised over the privacy of data obtained from small populations or marginalised communities (McClary-Gutierrez et al. 2021), as well as over how the data will be used. Individual consent for the collection of wastewater samples for SARS-CoV-2 analysis has not been deemed necessary (Kwiatkowska et al. 2021). While data should be made publicly available and transparent, it is necessary to navigate how best to ethically and legally handle data, as well as communicate this information. Further research is required to gain community feedback and support open access information and prevent disagreements and stigmatisation (Polo et al. 2020). Regions or marginalised communities may be associated with negative health, and this could have a negative impact on well-being, local economies, and potentially insurance policies (Wang et al. 2020; Harris et al. 2021). These discussions are particularly urgent as the technology and applications for WBE advance.

Research on WBE has proven it is a potentially useful tool in monitoring the presence of health markers, providing a theoretically constant stream of information on the health status of whole communities. Proposed uses include tracking viral pathogens, AMR, and chemicals (i.e., drug and pharmaceutical metabolites, markers of human disease or pollution exposure). However, each application of wastewater has its own challenges and research needs, including varying disease dynamics, non-human reservoirs, instability of markers in sewers and storage and interpretation for public health. Research into these areas would allow WBE to be developed as a useful tool for public health to supplement existing surveillance. However, WBE methods and protocols have been developed in high-income countries with improved sanitation. With greater globalisation, emerging threats to health can occur anywhere, and spread around the world rapidly. International cooperation and coordination is necessary to monitor for future threats, and if WBE is to play a role in this, comparable datasets across communities and countries are needed. In particular, we should consider and support the development and validation of these techniques for lower income settings to ensure improved global health surveillance.

The United Kingdom Government (Department of Health and Social Care) fund the sampling, testing, and data analysis of wastewater in England. The Welsh Government (Technical Advisory Cell) funds the sampling, testing, and data analysis of wastewater in Wales. The following funding sources are acknowledged by the contributing authors: ‘National COVID-19 Wastewater Epidemiology Surveillance programme (NE/V010441/1) UKRI-NERC COVID Response (D.W.G., D.L.J., B.K.-H.); ‘Building an Early Warning System for community-wide infectious disease spread: SARS-CoV-2 tracking in Africa via environment fingerprinting’ (EP/V028499/1) GCRF/Newton Fund (B.K.-H.).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.