The surveillance of antimicrobial resistance in wastewater from a one health perspective: A global scoping and temporal review (2014–2024)

Abstract

Surveillance of antimicrobial resistance (AMR) via a One Health approach must consider the interconnectivity between humans, animals, and the environment. Traditionally, AMR surveillance has relied upon patient-based surveillance in healthcare settings. Wastewater surveillance (WWS) has recently been demonstrated for monitoring AMR to and/or from wastewater treatment plants (WWTPs) which represent a point of intersection between humans, animals and the environment. WWS can be associated with AMR presence and dissemination across entire communities or WWTP catchments, as well as the transfer of AMR to agricultural lands and receiving waters via genes and/or organisms. In this review, the various methodologies used for WWS of AMR and their interpretative significance are identified and discussed, in addition to the potential approaches and outcomes associated with AMR monitoring within WWTPs. A total of 177 reports were identified covering the period 2014 to October 2024, with 136 (76.8 %) appearing after 2019. These recent papers show a distinct emphasis on qPCR and sequencing-based approaches. Surveillance is now global in scope, albeit with a current emphasis on WWTPs in high-income countries. To achieve more effective, global WWS of AMR under a One Health lens, all relevant sectors must understand the principles and capabilities of available methodologies and technologies. Overall, this review seeks to illuminate the diverse interpretations that can be made from WWS of AMR in a One Health context and identify how best to inform future directions regarding AMR monitoring and prevention efforts.

1. Introduction

AMR has become increasingly recognized as a global priority, with drug-resistant pathogens emerging as a significant threat to human and animal health worldwide [1,2]. The evolution of resistant pathogens, including bacteria, viruses, fungi, and parasites, is linked to the aggressive emergence of AMR infections that have become increasingly difficult to treat [[1], [2], [3], [4]]. Projections have AMR-related deaths outnumbering those associated with cancer by 2050, resulting in trillions of dollars in lost productivity for global economies [2,5]. Additionally, AMR poses a significant threat to livestock and the agri-food industry, with dissemination of drug-resistant microorganisms to the environment through soil and water being prevalent. Animals can shed drug-resistant microorganisms and/or genes into the natural environment and may also become exposed to AMR already present in these reservoirs [4,6].

Antibiotic resistance genes (ARG) can be passed on through bacterial reproduction (vertical transmission) or by horizontal gene transfer between bacteria, the latter including uptake of external DNA from environmental reservoirs including water and soil [7]. Non-pathogenic bacteria containing ARG are also important due to transmission of ARG into pathogens [8]. These ARG and bacterial hosts may spread via human and/or animal contact, food, water, sanitation and waste infrastructure, or soil contaminated by human or animal fecal material, including manures [7,9]. Accordingly, contamination of the natural environment by agricultural runoff, pharmaceutical/medical waste, sewage overflows, and wastewater effluent results in a reservoir of antibiotics, ARG, and antibiotic-resistant bacteria (ARB), where AMR can easily evolve and spread [7,8].

1.1. AMR surveillance and wastewater treatment plants

The influent to a WWTP (Fig. 1. node a) is a key sampling point that can provide insight into AMR trends within the associated catchment area (often called the “sewershed”) of the WWTP. This captures a diverse array of inputs including hospital and community sources [[10], [11], [12], [13], [14], [15], [16]], with possible additional sources if a WWTP receives waste material from, for example, industry (including pharmaceutical production), agriculture, livestock operations, and septic tanks [[17], [18], [19], [20]]. Hospital wastewater presents an important contribution owing to concentrated antimicrobial consumption and the prevalence of resistant pathogens in hospitals [21], contributing antibiotic-resistant organisms (ARO), ARG, and mobile genetic elements (MGE) if released without pretreatment [17,22,23].

Fig. 1.

Schematic representation of AMR transmission through the One Health framework, highlighting interactions between A. Human Health, B. Agriculture and Livestock Health, and C. Environment Health, with arrows illustrating the flow of AMR through non-wastewater pathways (blue) and wastewater pathways (red). Critical AMR wastewater pathways are illustrated as nodes (red crosses). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

WWTPs, when present, are often the sole component of the wastewater system with the potential to limit dissemination of AMR to the environment, depending on treatment efficacy [[24], [25], [26]]. WWTP design has been optimized for the removal of solids, organic matter, and nutrients, but not for the removal of antibiotics or ARO, ARG, and MGE [27,28]. Regular sampling from WWTP processing points (Fig. 1. node b) can reveal how AMR may be inadvertently enriched or transformed during treatment [29], with WWTPs previously characterized as “hotspots” for AMR [30,31].

WWTP effluent can release AMR into receiving bodies including lakes and rivers [32,33], and to other environment locations through land applications of biosolids and water reuse for irrigation or drinking water [19,[34], [35], [36]]. Therefore, AMR surveillance of WWTP effluent (Fig. 1. node c) is also recommended [37,38]. It is important to note that where data are available only for influent, AMR in the effluent can still be estimated via characterizing fate during treatment. Similarly, if data are available only for effluent, then influent AMR can also be estimated.

1.2. AMR and wastewater surveillance in the one health context

WWS is a comprehensive surveillance mechanism for monitoring AMR, as it inherently connects the human, environmental and agricultural sectors, enabling a One Health lens [40]. Historically, AMR surveillance has been patient-based, relying on clinical AMR data consequent to patient/healthcare interactions [39]. Clinical AMR data is derived during direct patient care, with the primary aim of adequately and appropriately treating the clinical infection, which in turn creates the opportunity to collect AMR related data. Data generated in support of clinical management can be merely leveraged for surveillance purposes. WWS is a strategic surveillance mechanism without the assumptions or limitations of clinical data, thereby enabling a greater and more comprehensive collection of data, not just from individuals but from entire communities [[41], [42], [43]]. Accordingly, WWS has emerged as a powerful tool, in association with other available surveillance tools, for outbreak detection and possibly prediction, and can inform protective public health measures [41,44,45], policy, investment by governments, as well as pharmaceutical development efforts [39,40]. The relevance of WWS was highlighted during the COVID-19 pandemic, whereby WW data proved a useful tool for monitoring SARS-CoV-2 virus trends, including for asymptomatic shedders and individuals who did not seek medical attention [42,45].

In wastewater, a multitude of bacteria, including ARB, originate from various sources including residential areas, industries, and institutions including hospitals [46] (see Fig. 1). At each point in the transport of wastewater from these sources to the environment, surveillance can provide unique information regarding AMR in or between various settings within the One Health paradigm. Human health is reflected in the transmission of AMR from hospitals and communities into wastewater, contributing to the agriculture setting in regions with land application of biosolids and, both directly and indirectly, to the environment. Agriculture and livestock health impacts non-wastewater pathways, whereby AMR can spread to environmental settings through agriculture runoff and manure application. Environmental health refers to environment settings being a reservoir for AMR and potentially a site for further AMR development, becoming an AMR source for both human settings and agriculture and livestock settings [7].

Various methodologies can be employed for WWS of AMR across the One Health settings, influencing research outcomes that yield distinct interpretations and conclusions about resistance patterns. This scoping review seeks to examine the methodologies utilized for WWS of AMR over the 2014–2024 period and synthesize research approaches and results to inform AMR risks and decision making from a One Health perspective.

2. Methods

2.1. Search strategy

This scoping review focused on the following research questions:

1. What methodologies for AMR surveillance in wastewater have been used in the 2014–2024 (up to and including October 2024) period, how do they work, and what are the strengths and limitations of each method?

2. How are WWS data used to inform AMR risks from a One Health perspective?

3. How can the WWS and AMR risk information be used in decision-making, and what are the current knowledge gaps?

The “PCC” framework (population, concept, and context) was employed to provide a structured method for combining and analyzing essential concepts and inform the search strategy and inclusion criteria [47]. Global literature focusing on wastewater samples from various sources was used to represent the “population” parameter within this framework. The presence, identification, and characterization of AMR in wastewater samples signified the “concept” of the review, with the “context” focusing on spatiotemporal aspects, including the types of indicators and reservoirs of AMR globally, in addition to AMR monitoring methods and their associated temporal trends over the 2014–2024 period.

Identification of research articles published worldwide in English between 2014 and October 2024 was carried out through a keyword search, including titles and abstracts, based on search strings indicated in Table 1 using PubMed, PubMed Central, Science Direct, and SpringerLink databases between June 1 and October 31, 2024. One search strategy was employed for all three research questions by three of the authors, and searches were duplicated using common acronyms. The search was case-independent and captured plurals where applicable. Based on previous article identification by the authors, known includes (numbers 1,4,5,9,15,17,26,35,57,59,62,70,71,98,106,167 in the study list in Supplementary Materials) were noted to be captured by this search strategy.

Table 1.

Search terms used for literature identification.

Primary term	Search strings
Wastewater	(antimicrobial resistance OR antibiotic resistance OR amr) AND (wastewater treatment OR wastewater OR WWTP) AND (farm OR agriculture OR livestock OR clinical OR hospital OR industry OR production) (antimicrobial resistance OR antibiotic resistance OR amr) AND (wastewater treatment OR wastewater OR WWTP) AND (biological OR aerobic OR anerobic OR activated sludge OR sludge OR digestion OR rotating biological contactor OR reactor OR trickling filter OR membrane bioreactor OR secondary treatment) (antimicrobial resistance OR antibiotic resistance OR amr) AND (wastewater treatment OR wastewater OR WWTP) AND (clarifier OR sedimentation OR membrane OR filtration OR pond OR lagoon OR osmosis OR constructed wetland OR water reuse OR disinfection OR chlorination OR UV OR ultraviolet OR advanced oxidation) (antimicrobial resistance OR antibiotic resistance OR amr) AND (wastewater OR surveillance OR wastewater surveillance OR monitoring)
One Health	((“antimicrobial resistance” OR “antibiotic resistance” OR AMR) AND (wastewater OR “wastewater surveillance” OR “wastewater based epidemiology” OR “wastewater-based epidemiology” OR WBE OR WWTP OR “wastewater treatment plant”) AND “One Health”)
Methods	(antimicrobial resistance OR antibiotic resistance OR amr) AND wastewater AND (broth microdilution OR Kirby-Bauer OR phenotype OR epsilometer test OR e-test) ((“antimicrobial resistance” OR “antibiotic resistance” OR AMR) AND (wastewater OR “wastewater surveillance” OR “wastewater based epidemiology” OR “wastewater-based epidemiology” OR WBE OR WWTP OR “wastewater treatment plant”) AND (metagenomic OR PCR OR qPCR OR genotypic OR sequencing OR WGS))

2.2. Eligibility and selection of sources

Fig. 2 outlines details of the literature screening process, including eligibility and exclusion criteria [48]. Records identified from the original database searches (n = 8724) were aggregated into a spreadsheet and then underwent de-duplication (n = 149 removed), followed by title and abstract screening, with records retained based on content that contained clear links between AMR and wastewater (n = 707). Full-text articles were excluded if papers were not a primary research article (e.g., reviews, n = 304), if samples were not directly from wastewater processes (n = 52), if the paper analyzed data from a previously reported publication (n = 11), if samples were not focused on community wastewater (e.g., food waste, n = 49), if sample size was less than 10 or not stated (n = 56), if the paper was solely for method development and/or validation (n = 26), if the paper focused solely on chemical quantification of antibiotics (n = 45), or if the focus was synthetic wastewater and/or benchtop studies (n = 25). Additional supplementary records were then added from previous identification by authors or citation “snowballing” from initial identified records, including several articles (n = 13) where sample number was <10 but significant phenotypic characterization had been done, (n = 51). After screening, the total number of records included in the review was n = 177 (Fig. 2). The complete list of analyzed articles is provided in Supplementary Material. Articles that discussed AMR in wastewater but were omitted from further analysis due to low sample number or lack of details are listed on a separate spreadsheet in Supplementary Material. This analysis was repeated independently by the three first authors and 26.1 % of results cross-checked to ensure consistency and avoid accidental exclusion of relevant sources. No grey literature, for example, government reports, was included in this review.

Fig. 2.

Preferred reporting items for scoping reviews and meta-analyses (PRISMA) flowchart of literature search and study selection process.

3. Results and discussion

3.1. Data extraction and synthesis

A total of 177 studies were identified as describing WWS for AMR over the 2014-Oct. 2024 period, with 136 (76.8 %) of these reported after 2019, highlighting the recent nature of this field. The extraction of these articles was cross-checked by the first three authors for consistency and correction of errors. As shown (Fig. 3), European countries (9.2 % global population) accounted for 42.3 % of studies (n = 75), Asian countries (59.0 %) accounted for 29.9 % of studies (n = 53), African countries (18.3 %) comprised 13.2 % of studies (n = 23), North America (4.7 %) 7.5 % (n = 8), South America (8.2 %) 5.4 % (n = 15), and Oceania (0.6 %) 1.3 % of all studies surveyed (n = 3) [49]. With AMR representing a global problem, the notable gaps in the literature, highlighted by the continental distribution of studies, and the country specific breakdown in Fig. 3, draw attention to the need for standardized and appropriately-implemented monitoring of AMR in wastewater.

Fig. 3.

World map showing the spatial distribution of all identified wastewater-based studies which utilize phenotypic and genotypic AMR surveillance methods.

The most prevalent sample points in identified studies were a combination of influent and effluent (27 %), various locations within WWTPs (27 %), post-release (11.4 %), influent only (10.8 %), effluent only (9.6 %), lab-based (supplementary records with additional characterization of various sample types, used for procedure-based studies, 6.9 %), sludge only (5.4 %), and community samples (such as building-level passive samples, 1.5 %). The mean sample number was 62.6 per study, with the standard deviation being 95.4, indicating a wide range of sampling regimes. Samples were taken over a mean time-period of 10 months, with a standard deviation of 17 months, indicating that sampling period duration was also highly variable. The mean time-period of less than a year is a point of concern, as this limits potential insight into seasonal trends.

3.2. Methods of AMR detection in wastewater

AMR detection methods for WWS are traditionally divided into i) phenotypic methods, using growth and conventional bioassays, and ii) genotypic methods, using nucleic acid detection-based methods (see Table 2). These methods produce different and complementary information regarding AMR in each sample. Among the 177 studies identified on WWS of AMR, 41.8 % (n = 74) employed at least one phenotypic method for the detection and characterization of ARB, 84.2 % (n = 149) utilized at least one genotypic (molecular-based) method, and 28.8 % (n = 51) used both a phenotypic and genotypic method (Table 2).

Table 2.

Current phenotypic and genotypic methodologies used in the literature to identify AMR in wastewater environments.

Methods	Mechanism	Contribution to AMR knowledge network	References
Phenotypic methods
Kirby-Bauer	Evaluates bacterial susceptibility by measuring inhibition zones around antibiotic disks.	Can provide qualitative data for resistance patterns and supports large-scale surveillance and stewardship.	n = 31 [26,54,[58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86]]
Epsilometer test (E-test)	Uses gradient strip to determine the MIC of antibiotic for a bacterial isolate.	Can provide MIC values to define resistance levels, which is useful in both clinical and research settings	n = 4 [82,[87], [88], [89]]
Broth Microdilution	Determines the MIC and MBC by serially diluting antibiotics in liquid medium to assess growth inhibition and bactericidal activity, respectively.	Can provide quantitative MIC and MBC data to establish resistance profiles and validate phenotypic methods globally.	n = 50 [12,34,50,56,57,62,63,69,[79], [80], [81], [82],84,85,88,[90], [91], [92], [93], [94], [95], [96], [97], [98], [99], [100], [101], [102], [103], [104], [105], [106], [107], [108], [109], [110], [111], [112], [113], [114], [115], [116], [117], [118], [119], [120], [121], [122], [123], [124]]
Genotypic methods
PCR	Amplifies DNA fragments within a sample to allow for the detection of a specific target gene.	Can be used for the detection of target genes within a wastewater sample, specifically to identify and/or monitor the presence and/or spread of a gene of interest within an environment.	n = 21 [66,67,70,71,73,75,77,80,82,83,85,[87], [88], [89],102,104,107,[125], [126], [127], [128]]
Quantitative PCR (qPCR)	Using previously determined sequences against a standard curve, detects and quantifies ARG of interest, useful for detection of low abundance genes.	Can be used for population-level analyses and high-throughput methods can detect hundreds of ARG of interest. Provides an accessible and comparable AMR surveillance method in a multitude of environments.	n = 80 [10,11,13,[16], [17], [18],20,[26], [27], [28],32,34,36,55,58,61,62,64,68,69,86,[93], [94], [95], [96],[98], [99], [100], [101],105,120,124,[129], [130], [131], [132], [133], [134], [135], [136], [137], [138], [139], [140], [141], [142], [143], [144], [145], [146], [147], [148], [149], [150], [151], [152], [153], [154], [155], [156], [157], [158], [159], [160], [161], [162], [163], [164], [165], [166], [167], [168], [169], [170], [171], [172], [173], [174], [175], [176]]
Digital PCR (dPCR)	Provides an absolute quantification of nucleic acid by partitioning samples to detect and quantify ARG of interest. High sensitivity for low abundance or rare gene detection.	Can provide population-level analyses, higher sensitivity and absolute quantification allows for the detection of rare ARG in any environment without relying on a previously validated standard curve.	n = 6 [15,27,58,[177], [178], [179]]
16S rRNA gene sequencing	Allows for the identification of bacterial taxonomy from a given sample by using a universal bacterial gene to sequence and analyze.	Often used in AMR studies to analyze bacterial diversity in a sample or after isolation of a pathogen of interest; can validate bacterial taxonomy to the genus or species level from a clinical or environmental sample. Can be used to compare diversity over time or space.	n = 61 [20,[25], [26], [27], [28],34,54,58,61,64,75,80,90,91,95,[98], [99], [100], [101], [102],105,120,[123], [124], [125], [126], [127],[132], [133], [134],[137], [138], [139], [140],[142], [143], [144], [145], [146],149,151,154,156,157,160,161,163,164,[166], [167], [168], [169],171,172,[174], [175], [176], [177],[180], [181], [182]]
Digital multiplex ligation assay (dMLA)	Probes allow PCR amplification of DNA from a bacterial isolate, attaching a barcode to be pooled, sequenced, and analyzed.	Can be used for surveillance programs for AMR to detect point mutations in ARG in isolates of interest; provides a cost-efficient method that can be scaled up to accommodate for large screening assays.	n = 3 [69,124,183]
Multilocus sequence typing (MLST)	Characterizes bacterial isolates into groups based on the sequence of housekeeping genes.	Can be used in AMR studies to separate bacterial isolates in “sequence type” groups, detect ARG and compare genetic diversity among isolates from wastewater samples.	n = 6 [27,85,102,106,156,184]
Whole-genome sequencing (WGS)	Uses DNA barcoding and sequencing from bacterial isolates to characterize entire genomes, including non-coding regions.	Provides high-resolution insight into bacterial genomes from detected isolates, useful for linking pathogenicity and ARG from samples of interest.	n = 33 [12,27,30,34,50,[55], [56], [57],60,79,80,83,[90], [91], [92],97,99,104,106,121,130,133,134,146,156,162,164,171,177,[184], [185], [186], [187], [188]]
Metagenomics	Allows for the simultaneous characterization and quantification of genes from all organisms present in a sample using short-read sequencing data.	Can be used for high-throughput, broad AMR surveillance to characterize and quantify thousands of genes from a single sample, compare relatedness, and provide insight into pathogenicity and community composition.	n = 35 [14,15,25,33,93,123,126,131,136,137,148,153,168,179,180,182,187,[189], [190], [191], [192], [193], [194], [195], [196], [197], [198], [199], [200], [201], [202], [203], [204], [205], [206]]

Of those studies that used at least one phenotypic method, 67.6 % (n = 50) used broth microdilution, 41.9 % (n = 31) utilized the Kirby-Bauer disk method, and 5.4 % (n = 4) used the E-test method. Within the first group, 2.0 % (n = 1) of studies investigated only the minimum bactericidal concentration (MBC), 32.0 % (n = 16) determined minimum inhibitory concentration (MIC), 32.0 % (n = 16) assessed both MIC and MBC, and 34.0 % (n = 17) did not specify their testing approach. Overall, the broth microdilution method was the most frequently applied phenotypic method and demonstrated significant value in characterizing resistance profiles in environmental isolates. For example, a 2023 study used broth microdilution to evaluate resistance in Klebsiella pneumoniae from sewage surveillance in Norway, revealing substantial resistance including colistin-resistant K. pneumoniae and ESBL-producing clones, further establishing the method's ability to detect AMR in wastewater and related environments [50].

Of those studies using at least one genotypic method for WWS detection, quantification, and/or characterization of AMR, 53.7 % (n = 80) used qPCR, 40.9 % (n = 61) used 16S rRNA gene sequencing, 23.5 % (n = 35) used metagenomics, 22.1 % (n = 33) used whole genome sequencing (WGS), 14.1 % (n = 21) used PCR, 4.0 % (n = 6) used dPCR, 4.0 % (n = 6) used MLST, 2.0 % (n = 3) used dMLA, and 0.7 % (n = 1) used metatranscriptomics. In contrast with phenotypic methods, these can detect DNA from damaged or viable-but-not-culturable (VBNC) bacteria [51,52], in addition to environmental DNA in the sample matrix. Also, unlike phenotypic methods using cultivation, genotypic methods normally cannot associate detected resistance genes with specific host bacteria, nor can they confirm if those genes are expressed within the microbial population of the tested sample [39,53].

When considering WWS of AMR, the requirements of surveillance programs, including emphasis on simple, rapid, and low-cost methods, it may be advantageous to include a pre-culture step prior to nucleic acid detection to select for specific bacteria and/or improve the detection limit for certain resistance genes in complex WW samples [12,[54], [55], [56], [57]], although this will increase the time, complexity and cost of the surveillance. For example, Al-Mustapha et al. (2024) isolated methicillin-resistant Staphylococcus aureus (MRSA) from municipal wastewater through culture-based approaches and detected drug resistance to clindamycin, sulfamethoxazole/trimethoprim, tetracycline, fusidic acid, erythromycin, and vancomycin using broth microdilution. The study subsequently employed WGS to identify specific strain types, plasmid profiles, and resistance genes, detecting the mecA gene in all isolates [57]. Therefore, phenotypic and genotypic methods can act as complementary tools, providing a more comprehensive approach in WWS efforts of AMR.

Fig. 4 presents a Sankey diagram based on all studies showing the relationship between the focus of studies, the surveillance methods used (Table 2), and the classification of these studies as phenotypic or genotypic. Studies were grouped based on primary focus and conclusions, with sub-sets of influent-focused trends being community and municipal (investigating typical municipal sanitary sewage and community trends, 28.8 % of studies per surveillance type), hospital and clinical (impacts on wastewater microbiome trends, 20.1 %), agricultural and livestock (such as waste streams from animal husbandry, 6.9 %), and industry (primarily pharmaceutical production wastewater, 2.1 %). Subsections of WWTP specific foci were influent-effluent (investigating trends in inputs and outputs of WWTPs including sludge and effluent, 10.8 %), treatment efficacy (effectiveness at removing AMR markers as a proportion of individual treatment types, 5.4 %), constructed wetlands (tertiary wetland treatment effect, 3.6 %), post-treatment (effluent only, 2.1 %), sludge treatment (studies limited to investigation of dewatered sludge only, 2.1 %), and disinfection (such as chlorination, 1.2 %). Finally, studies investigating post-WWTP trends concerned impact on receiving bodies (AMR levels in lakes, rivers, etc., that are impacted by WWTPs, 9.3 %) and studies focused on general trends in wastewater investigated seasonality (trends at a determined point in the wastewater system over time, 3.3 %) and the resistance profile of samples (such as determining ARO and MGE, 3.9 %).

Fig. 4.

Sankey diagram based on the 177 identified studies, which shows the breakdown of method type (left), followed by surveillance method, key study topics, type of sample, study duration, and number of samples on the right. Note that where studies reported more than one surveillance method or sample type, each method and sample type was included individually so column numbers total more than the number of studies.

An analysis of the compiled research articles revealed consistent advances over time in the methodologies employed, with an increasing emphasis on the integration of phenotypic and genotypic methods. Fig. 5 depicts the temporal trends of methodologies employed for WWS of AMR between 2014 and 2024. During this time-period, the frequency of published studies increased dramatically from n = 2 in 2014 to n = 41 in 2024 (available by Oct. 2024), with a strong correlation between time and identified study number (R² = 0.855). This increase coincided with a 47.9 % rise in genotypic methods for surveillance (n = 1/3; 33.3 % in 2014, n = 69/85; 81.2 % in 2024), specifically using quantitative PCR (qPCR), 16S rRNA gene sequencing, WGS, and metagenomics. It was noted that the use of molecular methods increased over time (from 2014 to October 2024); the use of qPCR increased from zero to 21.2 % (n = 0/3 in 2014 and n = 18/85 in 2024), 16S rRNA sequencing increased from zero to 20.0 % (n = 0/3 in 2014 and n = 17/85 in 2024), WGS increased from zero to 15.3 % (n = 0/3 in 2014 and n = 13/85 in 2024), and metagenomics increased from zero to 11.8 % (n = 0/3 in 2014 and n = 10/85 in 2024). These temporal trends indicate a surge of studies using WWS of AMR in the years leading up to 2024, especially using genotypic methods.

Fig. 5.

Stacked bar plot displaying the frequency of phenotypic and genotypic methodologies for WWS of AMR between 2014 and 2024. R² = 0.855 for total study frequency over time.

3.3. AMR surveillance in wastewater influent

WWTP influent surveillance (Fig. 1. node a; Fig. 6) captures important aspects of the associated catchment or “sewershed”. WWS studies of AMR in influent, which represent 55.9 % (n = 99) of the identified articles, have detected trends that correlate with inputs from the community (31.6 %, n = 56), hospitals (22.0 %, n = 39), or industries such as wastewater from pharmaceutical production (0.6 %, n = 4). For example, a cross-European study utilizing qPCR demonstrated that the influent of urban WWTPs mirrored regional clinical AMR profiles, strongly influenced by local antibiotic prescription patterns [10]. Findings showed that most antibiotic resistance classes were at higher levels in WWTP effluents in countries with higher antibiotic consumption [10]. A study in Norway using city-scale WWTP surveillance found that influent samples contained ESBLs in Escherichia coli isolates, with the profile of ESBLs resembling the pattern in regional clinical cases [12]. Notably, in isolates from 10 influent samples (n = 300), resistance most frequently observed was against ampicillin (16.6 %), sulfamethoxazole (9.7 %), and trimethoprim (9.0 %) [12]. Similarly, detailed monitoring of hospital WW in Finland also using qPCR identified carbapenem resistance genes, specifically detecting bla_KPC levels positively associated with the presence of K. pneumoniae in effluent from one hospital but not a second hospital, indicating that WWS can determine AMR levels emerging from healthcare facilities [11]. Another study reported bla_KPC and vanA detection in hospital WW but not in community WW from a sewershed location not receiving hospital WW. The use of a treatment system reduced or eliminated antibiotics and ARG, including bla_KPC and vanA, further emphasizing their direct link to healthcare facilities. Antibiotic concentrations in hospital WW, including ciprofloxacin and sulfamethoxazole, were found to be over 200 % higher than in community WW [17]. Moreover, prescribing practices in hospitals have been correlated with AMR in hospital effluent and typically contribute to higher levels of AMR than community sources, although this varies by the antibiotic and AMR studied [17,207,208].

Fig. 6.

Stages of a typical wastewater treatment, where influent includes inputs such as community wastewater, hospital wastewater, agricultural and food waste, and septic waste, effluent is discharged to the environment, and sludge is transported for potential reuse as fertilizer. A. Primary treatment includes physical treatment of clarifiers, sedimentation, and grit and fines removal. B. Secondary treatment includes biological treatment such as activated sludge (aerobic, anaerobic, mesophilic), rotating biological contactors, moving bed biofilm reactors, membrane bioreactors, and trickling filters. C. Tertiary treatment refers to filtration (forward and/or reverse osmosis, ultrafiltration, nanofiltration), and disinfection (advanced oxidation, ultraviolet, and chlorination). Additional D. Pond treatment considered aerated lagoons, wastewater stabilization ponds, septic systems, and constructed wetlands. These treatment technologies are viewed as modular and seen in many configurations in application, with any combination of primary, secondary, tertiary, and/or pond treatment.

Given that WWTPs may also process waste from septic systems, agriculture, and livestock operations, further sampling would be required to fully understand the scope of inputs, as noted in 3.4 % (n = 6) of the articles. In a 2015 article, WWTP influent including agricultural runoff had an unusually high percentage of bacterial isolates resistant to meropenem (6.8 %) along with ampicillin resistance in 72 % of isolates [34]. Other unique waste inputs from industrial and institutional contributors to a sewershed can noticeably affect the influent microbiome of a WWTP, as illustrated by a 2023 study where influent samples (n = 15) were determined to have a high relative abundance of tetC, followed by tetG and tetX, which the authors associated with common human and veterinary use of tetracycline antibiotics [27]. Another study [206] examined WW from residential buildings, a companion animal centre, workers' dormitories, hospitals, and an urban WWTP influent. Hospital sewage exhibited the highest abundance of ARG, with β-lactam resistance genes (OXA-1, OXA-10, TEM-117, GES-15) and sulfonamide resistance genes (sul1) contributing to 30 % of total ARG detected. Conversely, animal centre WW exhibited a lower ARG abundance and primarily contained tetracycline resistance genes such as tetA, tetG, and tetW, likely reflecting veterinary antibiotic use. WW from residential buildings and workers' dormitories shared 90 % similarity in their resistome profiles, with ARG such as fosA, macA, and arnA, attributing their distinct resistance profiles to fosfomycin, macrolide, and polymyxin antibiotics. WWTP influent, expected to combine all sources, had the highest diversity of ARG [206]. Collectively, these studies among others [125,126,132], underscore the value of WW from various sources as an important sampling node for AMR surveillance.

3.4. Surveillance of WWTP effluent release to the environment

WWTP effluent, whereby AMR can enter natural aquatic environments including lakes and rivers (Fig. 1. node c; Fig. 6), accounted for 9.6 % (n = 17) of the identified articles [32,33]. As WWTPs do not eliminate and may, in fact, amplify ARB and ARG, WWTP effluents have the potential to spread AMR to the broader environment and increase subsequent risks of exposure to humans and animals. This spread may occur through agricultural application of sludge and water reuse for irrigation or drinking in addition to discharge into natural water bodies [19,[34], [35], [36]]. In a 2023 study [180], 165 ARG subtypes from a pool of 677 possible subtypes, found across 15 different resistance classes, were identified in treated WW. These identified subtypes showed significant correlation with factors such as temperature, pH, or flow rate, suggesting that the characteristics and variations within the effluents can be strongly influenced by WWTP conditions. Another study sampled 23 WWTPs in Germany to determine the abundance of ARG and pathogenic bacteria in the effluent [129]. Twelve clinically relevant ARG were identified and categorized by their occurrence rates in WW, while five pathogenic bacteria (E. coli, Pseudomonas aeruginosa, K. pneumoniae, Acinetobacter baumannii, and Enterococci) were monitored by PCR. Results showed that the total ARG and pathogenic bacteria varied widely, with concentrations in some effluents being up to 100 times higher than others, regardless of WWTP size. Particularly, WWTPs including hospital WW showed increased correlations between pathogenic bacteria and ARG like bla_NDM-1 and vanA [129]. Surveillance of river samples downstream of WWTPs in Sweden [176] and the Netherlands [163] highlighted significant levels of ARG and intl1 genes. Another study reported a 543 % increase in ARG concentrations and a 164 % increase in ARG richness in sediments downstream versus upstream of a WWTP [33].

Studies of biosolids from WWTPs have found elevated contamination with ARG, with the relative abundance of AMR highest in dewatered sludges commonly used for land application; however, this abundance is influenced by factors such as digester retention time and the type of biological sludge treatment [20,58,100,173]. Research by Qin et al. (2022) determined that ARG such as tetA, sul1, mefA, and IS6100 were significantly enriched in all soils following long-term treatment with biosolids [160]. Similarly, Yang et al. (2018) reported that ARG including catB8, php, vanB, and str constituted over 50 % of the total ARG load in soils cultivating crops like lettuce [174]. Additionally, Ross and Topp (2015) revealed that while manure and biosolids increase ARG in soil bacteria, bacteriophages in biosolids also serve as a significant reservoir for ARG [127]. Considering this in concurrence with the growing concern about AMR, determining microbiome risks in biosolids is essential prior to land application due to food, soil, and runoff contamination and dissemination risks.

3.5. Fate of AMR in wastewater treatment plants: WWTPs as AMR “hot spots”

A typical WWTP, including influent streams, treatment stages, and discharge to receiving bodies is shown in Fig. 6. Within WWTPs, varied levels and types of treatment impact the removal or potential proliferation of AMR; primary (physical), secondary (biological), tertiary (disinfection and advanced treatment), and pond treatment (lagoons and wetlands).

A 2023 study identified high concentrations of ARB in both WWTP air (i.e., aerosols) and biosolids, with notable levels of azithromycin-resistant bacteria posing potential risks to WWTP workers [119]. Moreover, a study on disinfection processes at WWTPs showed that low doses of chlorine significantly increased ARG transfer via conjugative mechanisms, while high doses effectively suppressed this transfer [128]. Finally, metagenomics sequencing across different WWTP compartments confirmed the widespread distribution of ARG, predominantly on plasmids and integrative conjugative elements, underscoring the role of MGE in the persistence and transfer of ARG within WWTPs [181]. Wastewater treatment technologies and their impacts on ARO, ARG, and MGE were reviewed from studies focusing on the efficacy of treatment and presented in Fig. 7.

Fig. 7.

ARG and MGE fate during WWTP treatment stages, where the x-axis represents types of treatment in terms of primary, secondary, and tertiary treatment levels (tertiary treatment includes advanced treatment (filtration), disinfection, and constructed wetland treatment). The y-axis indicates ARG studied and their fate over individual treatment, as such, only studies focusing on individual treatment stages are included. Dark red indicates a 200 % average increase in gene count and dark blue indicates complete removal over a stage. Dashed cross-hatching indicates a standard deviation of 0.5 to 1, and solid cross-hatching indicates a standard deviation of greater than one. Stars indicate data is limited to one study where the WWTP did not have multiple sequential treatment processes for comparison [134,144,151,152,154,157,164]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Standard deviations in Fig. 7 provide insight into the variability among studies investigating the fate of ARG and MGE within individual treatment stages; mean standard deviations per treatment type were not reported for grit removal, filtration, UV disinfection, and membranes (due to only one study investigating each type), 130 % for clarifiers, 20 % for aeration basins, 30 % for aerobic biological treatment, 100 % for anoxic and/or oxic biological treatment, 70 % for chlorination, and 50 % for constructed wetland treatment.

Fig. 7 shows that clarifier treatment results in a 30 % mean reduction of ARG; however, tetracycline-resistant genes such as tetA, tetG, tetS, and tetX, and mobile genetic element intI1 were found to increase during clarifier treatment. Clarifier results also had a standard deviation of 130 %, indicating a high variation in treatment effects and the possibility that the presence of certain ARG is not impacted by the removal of solids; further, starred clarifier results show that most instances of decreased ARG were derived from a single study. Additionally, one study found that short clarifier times are likely not sufficient for the removal of ARG [100], highlighting the need to optimize these systems if AMR reduction is to be considered.

Secondary treatment yielded removal efficacies with high standard deviations, with retention time reported as a major influence [100]. Of the biological treatments investigated, anoxic conditions were found to increase ARG and MGE by 49 % on average, while aerobic and anaerobic treatments decreased ARG by an average of 59 % and 19 %, respectively. ARG increased by 1.6 % with chlorine disinfection (standard deviation of 70 %) and 20 % with UV disinfection (standard deviation 0 %, indicating a research gap). This latter finding is concerning as disinfection represents the final treatment stage in conventional WWTPs, suggesting additional treatment may be required to decrease dissemination of ARG and MGE to the environment. One study indicated that chlorination specifically increased the abundance of extracellular ARG, suggesting that ARG removal rate depends on the chlorine-susceptibility of host bacteria [154].

Finally, constructed wetland treatment was found to decrease ARG by an average of 56 % (standard deviation of 35 %), indicating these systems have the potential to reduce ARG as a tertiary treatment step, particularly in communities with land availability for these systems. Likewise, filtration was found to decrease ARG by 61 %, emphasizing the efficacy of particular tertiary treatments.

This variability, as well as instances of ARG increasing across some conventional wastewater treatment methods, points to the need for increased and site-specific monitoring of WWTPs to optimize treatment and decrease environmental dissemination from effluents and waste outputs such as sludge. Increases in ARG and MGE, as discussed above, suggest WWTPs can provide conditions for ARB proliferation that depend on bacterial abundance, background antibiotic concentrations, resistance mechanisms, and plant-specific operational parameters such as temperature and retention time [138,143,144,151].

4. Conclusions, limitations, and future considerations

Resistance to antimicrobials poses a complex, multifaceted challenge that requires a proactive, multifactorial approach. This review demonstrates an innovative and scalable method for monitoring AMR across communities and environments using WWS, providing valuable insights into the prevalence and spread of AMR in real time. Unlike traditional AMR surveillance protocols that rely on testing of symptomatic patients in healthcare settings, WWS captures community-level data, thereby reducing biases and offering a more comprehensive view of AMR trends. By integrating patient-independent data, WWS can complement clinical data to enhance the overall effectiveness of AMR surveillance, providing a more representative picture of resistance trends across different populations and regions [62,63]. WWS also extends beyond clinical and community monitoring, revealing resistance mechanisms and patterns within wastewater treatment facilities and their downstream natural environments. WWS therefore likely represents a key tool for a One Health-based understanding of patterns of AMR worldwide, empowering more informed and targeted approaches to address this global health challenge.

While this review has illustrated the importance of WWS for AMR, this field is not without its limitations. As highlighted in this review, for the 2014–2024 time-period, the field of WWS has been established, particularly for AMR surveillance, but is still in the early stages of demonstrating its full potential. As this area of study continues to collect data using a wide range of locations and methods, a more systematic review can be considered. Additionally, WWS primarily focuses on samples from centralized WWTPs, potentially excluding rural or vulnerable communities without access to municipal sewage systems [209]. WWS data may also overestimate ARG abundances in associated communities due to the evolution of resistance mechanisms within wastewater environments [39,40]. To address these challenges, future AMR surveillance efforts should emphasize the use of multiple methods and sampling points, including the integration of WWS with clinical data, to achieve a more comprehensive and precise understanding of AMR. For example, the use of phenotypic techniques, such as culture-based methods, combined with genotypic methods, including qPCR and sequence-based approaches, can provide a suite of data that covers microbial viability, community and/or gene composition, and AMR abundances within wastewater. Moreover, both researchers and health officials should strive for standardization both for method development and data interpretation. Effective implementation of WWS will better allow public health officials and policymakers to prepare for, and respond to, AMR-related illness outbreaks, monitor trends, and make decisions involving funding, education, and healthcare.

CRediT authorship contribution statement

Rhiannon Punch: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Conceptualization. Rayane Azani: Writing – review & editing, Writing – original draft, Visualization, Methodology, Formal analysis, Conceptualization. Claire Ellison: Writing – review & editing, Writing – original draft, Visualization, Methodology, Formal analysis, Conceptualization. Anna Majury: Writing – review & editing, Conceptualization. Paul D. Hynds: Writing – review & editing, Conceptualization. Sarah Jane Payne: Writing – review & editing, Writing – original draft, Visualization, Methodology, Conceptualization. R. Stephen Brown: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Methodology, Formal analysis, Conceptualization.

Declaration of competing interest

All authors indicate that they have no competing interests to declare concerning the submitted review manuscript.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.