Tracing antibiotics in sewers: Concentrations, measurement techniques, and mathematical approaches Open Access

Surface water pollution has traditionally been a field of research due to the potential effects of pollutants on the environment, ranging from direct impacts on ecosystems (Felis et al. 2020) to public health implications (Jones et al. 2005; de Jesus Gaffney et al. 2015). The identified issues have led environmental authorities to establish minimum water quality standards aimed at detecting and controlling both conventional pollutants, such as nitrogen, and unconventional pollutants, such as heavy metals. Despite efforts made at the regulatory level, the protocol for monitoring the so-called emerging pollutants remains unclear (Fatta-Kassinos et al. 2011). These pollutants include personal care and hygiene products, hormones, illegal drugs, and pharmaceutical residues such as antibiotics. The latter compounds have recently received particular attention due to their potential impacts on the alteration of aquatic organisms (Danner et al. 2019), including acute toxicity and changes in bacterial community dynamics (Grenni et al. 2018; Felis et al. 2020), as well as concerns about human exposure through drinking water (Wellington et al. 2013; Deo 2014).

Antibiotics are divided into several classes based on their chemical structure, mechanism of action, and range of activity against bacteria. Common classes include macrolides (e.g. azithromycin, clarithromycin, and erythromycin), which inhibit bacterial protein synthesis; tetracyclines (e.g. doxycycline and oxytetracycline), which interfere with protein synthesis by binding to the bacterial ribosome (Chopra & Roberts 2001); and fluoroquinolones (e.g. ciprofloxacin, levofloxacin, and norfloxacin), which inhibit bacterial DNA gyrase and topoisomerase IV, essential for DNA replication (Hooper 2001). Antibiotics have been critical in treating bacterial infections in humans, animals, and plants (Felis et al. 2020; Han et al. 2021), partly due to the advantages of using them in modern clinical procedures (Flasche & Atkins 2018; Hutchings et al. 2019) and livestock to guarantee the demand for animal products (Cycoń et al. 2019). However, because of their extensive and widespread use, these compounds have accumulated in the environment, contaminating surface and groundwater (Kümmerer 2009; Jelic et al. 2015; Liu et al. 2017; Mirzaei et al. 2019; Jari et al. 2022). The presence of these pharmaceutical compounds and their metabolites in the environment has adverse effects such as ecological contamination (Bilal et al. 2020), persistence in groundwater (Cycoń et al. 2019; Mirzaei et al. 2019; Zainab et al. 2020), and the potential development of antibiotic-resistant bacteria (ARBs) (Kümmerer 2009; Le-minh et al. 2010; Patel et al. 2019; Felis et al. 2020; Boogaerts et al. 2021; Gałązka & Jankiewicz 2022). As an effort to curb the rise of antibiotic resistance, the World Health Organization (WHO) has classified antibiotics into three main categories: Access, Watch, and Reserve (AWaRe) (WHO 2023). ‘Access’ antibiotics are first-line treatments for most infections and should be widely available. ‘Watch’ antibiotics show a higher resistance potential and should be used with caution. ‘Reserve’ antibiotics are used as a last resort for severe infections when other treatments are ineffective (WHO 2023).

Foul sewers collect wastewater containing a wide variety of pollutants, including antibiotic residues. In urban areas, these residues originate mainly from hospital and industry discharges, as well as human excretion. In humans, only a fraction of the antibiotic intake is absorbed by the body (between 30 and 90%), while the remainder is excreted through the urine and feces, eventually reaching the sewer system (Kümmerer 2009; Li et al. 2019; Kairigo et al. 2020). Despite being present in low concentrations, these residues show minimal degradation or transformation in sewers and can desorb from sediments (Gao et al. 2023), resulting in elevated concentrations and consequently increasing the load entering wastewater treatment plants (WWTPs). The presence of these pollutants in sewers can promote the spread of ARBs in surface waters due to the incomplete remotion in WWTPs (Zhang et al. 2023) and the potential retransformation of metabolites in their parent compounds (Delli Compagni et al. 2020; Han et al. 2021; Cheng et al. 2022).

Understanding the current state of antibiotics in sewers is essential for assessing their environmental impact and potential risks to public health. This paper presents a critical review of research on the presence of these compounds in urban sewers and hospital effluents. The latter was included as one of the main sources of contamination by pharmaceutical compounds and promotion of ARBs are hospital effluents (Zillien et al. 2019; Alexander et al. 2022; Rolbiecki et al. 2022). The main purposes of this review are to (i) quantify the number of studies focused on detecting antibiotics in sewer networks, including concentrations and chemicals compounds reported, and the variability of their presence worldwide; (ii) outline the techniques currently used to quantify these pharmaceutical compounds; and (iii) present the water quality models and mathematical approaches employed for studying and analyzing the compounds in sewer pipes.

This paper is expected to provide researchers, government departments and agencies, environmental professionals, and water utilities with comprehensive knowledge to develop more effective strategies for controlling, managing, and mitigating antibiotic contamination in sewer networks. In addition, the dataset presented here can be used to create a consolidated database of antibiotic concentrations in sewer systems. The rest of the paper is organized as follows: First, the review methodology used to select and classify the papers is presented. Next, the main findings are highlighted. Furthermore, the knowledge gaps and future research directions are suggested. Finally, the main conclusions are provided.

This review was carried out by searching keywords, such as ‘antibiotics’, ‘sewer’, ‘wastewater-based epidemiology’, ‘risk quotient’, and ‘wastewater surveillance’, on indexed databases like Scopus and Web of Science. The search was limited to articles published over the past 10 years (2014–2024) to incorporate the most recent advances and knowledge in the field. By focusing on recent literature, the review captures the latest trends, conducted research, measurement techniques, and mathematical approaches, providing a current and comprehensive overview of the topic. While we cannot guarantee that all available studies in the literature have been reviewed, the information presented here outlines the current trends in this field of research.

In this study, the following factors are considered to classify and analyze the 91 papers found in the literature: (i) the range of concentrations, class, and AWaRe classification of the antibiotics reported in hospital effluents and sewer networks (discussed in Section 3.1); (ii) the existing measurement techniques employed for detecting the pharmaceutical compounds (Section 3.2); and (iii) the water quality models/studies/approximations available for studying reaction of compounds, defining environmental and health risk and estimating patterns of drug use, among others (Section 3.3).

Figure 1(b) presents the number of antibiotic compounds identified by the class and categorized under the AWaRe classification. Sulfonamides, fluoroquinolones, and penicillin showed the highest number of reported compounds, while several classes, including aminocyclitols, amphenicols, beta-lactamase-inhibitors, carbapenem, fourth-generation cephalosporins, glycopeptides, glycylcyclines, lipopeptides, nitrofurans, oxazolidinones, thioamides, triclosan, and trimethoprim derivatives were represented by only one compound each. Sulfonamides are predominantly ‘Unclassified’, indicating that many of these antibiotics, such as sulfabenzamide, sulfachinoxaline, and sulfachloropyridazine, among others, either lack the data necessary for classification or are not commonly used in clinical practice. In contrast, fluoroquinolones showed several compounds classified under the ‘Watch’ category, highlighting their importance and the need for conservative use due to the potential risk of resistance. At the same time, a considerable proportion of penicillin falls into the ‘Access’ category, showing that these compounds are frequently used for treating common infections.

In addition, 77.8% of the sulfonamide reports are classified as ‘Access’ (e.g. sulfadiazine and sulfamethoxazole), while the remaining 22.2% correspond to the ‘Unclassified’ category (e.g. sulfadoxine and sulfamonomethoxine). These compounds are widely used to treat several bacterial and protozoal infections in human and veterinary medicine (Chopra & Roberts 2001; Munir et al. 2011). In the case of fluoroquinolones, 14 compounds were reported 205 times in 64 papers. Of these, 57% are classified as ‘Watch’ (e.g. ciprofloxacin and levofloxacin), while the remaining 43% are ‘Unclassified’ (e.g. danofloxacin and marbofloxacin). Ciprofloxacin (39.5%) was the most frequently monitored fluoroquinolone, consistent with its prevalence in WWTPs (Van Doorslaer et al. 2014; Ajibola et al. 2021). These compounds have shown limited biodegradability (Mayoudom et al. 2018), often persisting through conventional wastewater treatment processes (Serna-Galvis et al. 2019), affecting downstream environments. The few reports of antibiotics classified such as amphenicols (e.g. chloramphenicol), aminoglycosides (e.g. gentamicin and kanamycin), carbapenems (e.g. meropenem), aminocyclitols (e.g. spectinomycin), and glycopeptides (e.g. vancomycin) may be due to their limited clinical usage or specific application contexts. For example, amphenicols have restricted clinical use due to potential side effects and carcinogenicity (Reis et al. 2020), while carbapenems are last-resort antibiotics reserved to treat severe bacterial infections (Sharma et al. 2023). When detected in sewers, these compounds often indicate specific contamination sources, such as hospital effluents or pharmaceutical manufacturing sites (Felis et al. 2020), which can serve as hotspots for antibiotic pollution and ARBs (Rizzo et al. 2013).

The AWaRe classification showed that the ‘Watch’ and ‘Access’ categories show the highest number of antibiotic reports. These categories include antibiotics that can be prescribed more frequently than those in the ‘Reserve’ group, according to the WHO recommendations. However, it also comprises antimicrobials in the ‘Watch’ group, which are used as a first option to treat severe infections or diseases whose causal agent is highly likely to be resistant to ‘Access’ antibiotics (WHO 2022). Although antibiotics in this group are closely monitored to avoid inappropriate use that might lead to resistance emergence, our results suggest that these compounds are commonly present in sewer systems. These findings are concerning due to the potential for increased resistant bacterial populations. The ‘Unclassified’ category has significantly fewer reports, with 115 in total, while the ‘Reserve’ has 13 reports. The latter group of antibiotics is typically reserved for critical cases, so their absence in reports may indicate proper management as they are used as a last resort in treating bacterial infections.

Most of the studies included in this review implemented powerful separation techniques of compound mixtures in water coupled with methodologies with high specificity and sensitivity for antibiotic quantification. Prior to compound mixture separation, commonly, antimicrobials are concentrated by solid phase extraction (SPE). Approximately 77.7% of studies preconcentrate water samples through SPE; however, salting-out assisted liquid–liquid extraction (SALLE) (Gezahegn et al. 2019) and QuEChERS (Quick, Easy, Cheap, Effective, Rugged, and Safe) extraction followed by d-SPE Clean-up methods (Ajibola et al. 2021), SPE and ultrasonic-assisted extraction (UAE) (Muriuki et al. 2020), SPE and soxhlet extraction (SE) (Ren et al. 2021), have also been reported. Subsequently, compound separation is conducted through different types of liquid chromatography (e.g. liquid chromatography (LC), high-performance liquid chromatography (HPLC), ultra-high-performance liquid chromatography (UHPLC), ultra-performance liquid chromatography (UPLC), and reversed-phase HPLC (RP-HPLC)). These are coupled with a wide range of detection methods, including diode array detectors (DAD), photodiode array (PDA), mass spectrometry (MS), chromatographic quantification through UV and fluorescence detectors, and tandem mass spectrometry (MS/MS). In addition, most of the equipment includes electrospray ionization (ESI), heated electrospray ionization (HESI), and quadrupole time-of-flight (QTOF). The frequency of each of these concentration, separation, and measurement methods is presented in Table 2. This table also presents the average recovery rates for concentration and separation, and the detection and quantification limits for measurement. Full information is provided in Supplementary Table S1. The results show that detecting antimicrobial compounds in sewer systems requires specialized knowledge and involves high costs and time-consuming processes, which do not facilitate the reporting of these compounds, mostly in developing countries.

Supplementary Table S2 presents the models found in the review, classified into four types based on their specific purposes. Wastewater-based epidemiology (WBE) is an approach that uses wastewater to monitor public health (Galani et al. 2021; Duan et al. 2022). This technique estimates the use of illicit drugs (Zillien et al. 2022), alcohol consumption, and exposure to various pharmaceuticals and compounds at a population level (Fallati et al. 2020) by measuring chemicals or biomarkers in wastewater (Yuan et al. 2016). The calculation of risk quotient (RQ) is used for evaluating the potential impacts of several contaminants in ecosystems (Escher et al. 2011; Deo 2014; Aydin et al. 2019; Shokoohi et al. 2020; Ren et al. 2021). Empirical models are focused on forecasting the mass loads and concentrations of pollutants produced and entering the sewer system (Bollmann et al. 2019; Zillien et al. 2019; Pouzol et al. 2020). Finally, kinetic models focus on determining the degradation/transformation (Ren et al. 2021) of pollutants in the environment.

Kinetic models are mathematical representations describing the rates of chemical or biological reactions in water. Despite several investigations having been focused on defining sorption coefficients, photodegradation, hydrolysis, and biodegradation rates of different antibiotics, these have carried out experiments on different environmental media, including WWTP effluents and agricultural runoff (Harrower et al. 2021) or by using batch setups (Nguyen et al. 2017; Sharma et al. 2023). Only one report considering in-sewer degradation of antibiotics was found in this review. Ren et al. (2021) used a 1,200-m pilot experimental sewer to study the biotransformation of several pharmaceutical compounds, including sulfamethoxazole. This compound showed high degradation in water because of the hydrophilicity and the low log K_ow, which leads to biodegradation. In contrast, the adsorption in sediments was low, which may indicate that the reduced concentration is due to biodegradation.

Studies have also focused on predicting antibiotic mass loads in urban wastewater. Bollmann et al. (2019) implemented a model for estimating yearly water discharge and triclosan and erythromycin mass loads from urban sources into the Baltic Sea, obtaining accurate concentration results. In addition, Zillien et al. (2019) proposed a model for predicting the pharmaceutical emissions to hospital wastewater, considering the mass of the active compound and the excretion factor. Furthermore, Pouzol et al. (2020) developed a comprehensive model to calculate daily and hourly sulfamethoxazole loads and concentrations in wastewater at the inlet of WWTP. This model integrated a hydraulic module to simulate wastewater discharges and water routing in the sewer network, along with a module to estimate antibiotic discharges at the inhabitant level by considering factors such as the specific posology, the metabolisms of the inhabitant, and the times at which the inhabitant is excreting into the sewer system. The model showed good accuracy in capturing antibiotic mass load patterns. Full details are presented in Supplementary Table S2.

Based on the search conducted in this study, there are several gaps and potential research directions for improving knowledge in this field. These are presented and discussed as follows:

• (A) Several antibiotic compounds have been detected in sewer systems, but most of the reported data are point measurements using composite (Yuan et al. 2016; Auguet et al. 2017; Azanu et al. 2018; Ngigi et al. 2020; Holton et al. 2023), grab (Al-Mashaqbeh et al. 2019; Ikizoglu et al. 2023), or passive (Duarte et al. 2023) samplings. The above limits the ability to identify temporal variations, which can be particularly significant CSO events (Acosta et al. 2024). These variations often result from meteorological and climatological factors such as rainfall patterns, temperature fluctuations, and seasonal agricultural practices, with diffuse pollution effects playing a crucial role (Xie et al. 2025). For instance, heavy rainfall can dilute certain antibiotic compounds (Zhang et al. 2024) while simultaneously increasing the mobilization of metabolites through CSOs. Temperature variations can also influence biotransformation and degradation rates within sewer networks (Kaeseberg et al. 2018). Few studies have reported fluctuations during annual seasons (Huang et al. 2019; Kosma et al. 2020; Cho et al. 2023; Ikizoglu et al. 2023; Oharisi et al. 2023) or during normal and holiday weeks (Duan et al. 2022). Determining this variation is essential for improving the operation of WWTPs, as non-homogeneous influent pollution affects pollutant removal rates and, consequently, concentrations at the discharge of the systems. Integrating these temporal variations into wastewater management strategies can lead to more effective mitigation of antibiotic contamination in aquatic systems. Future research should prioritize continuous real-time monitoring systems to capture dynamic fluctuations and eventually enable more accurate modeling and operational modifications.

• (B) The results show a difference between the number of antibiotics detected in developed countries such as Spain, Germany, Italy, and the USA and the amount of these compounds detected in developing countries (e.g. Colombia and some African countries). In the latter, a higher variety of antibiotic classes was identified, likely because of inefficient treatment system implementation and lack of modern technology (Qadir et al. 2010). These types of treatment systems, common in developing countries, result in low removal rates and potentially high discharges of pollutants to surface water bodies, contributing to the presence of a broad spectrum of antibiotics in the environment. The insufficient treatment capacity and surveillance in developing regions increases the risk of antibiotic resistance, which can spread rapidly, constituting a global threat. A comprehensive understanding of the worldwide extent of antibiotic contamination and its impact on public health and ecosystems requires specific studies in developing and developed regions to understand the factors influencing antibiotic pollution in diverse socioeconomic and environmental contexts. International collaboration and funding mechanisms could support the establishment of standardized protocols for antibiotic monitoring.

• (C) Many of the antibiotic compounds reported are classified as ‘Watch’ according to the AWaRe categorization. This is especially important because of the potential of these compounds to generate ARBs in water. Despite their critical implications, there is a lack of comprehensive studies on detecting and quantifying metabolites produced during transformation in sewers (Ren et al. 2021) or by back-transformation processes (Ren et al. 2021; Cheng et al. 2022; Zillien et al. 2022), which increases antibiotic concentrations and hence influent pollution to WWTPs or via overflows to the environment, during heavy rain events. Few studies have reported metabolites for fluoroquinolones, macrolides, lincosamides, sulfonamides, and trimethoprim derivatives (Castrignanò et al. 2020; Kosma et al. 2020; Han et al. 2021; Duan et al. 2022; Zhang et al. 2023). Further studies for detecting wider metabolite concentrations and compounds in sewers should be conducted to identify the persistence in the environment. Additionally, in-sewer biotransformation pathway studies of antibiotic compounds must be carried out to determine their transformation products and environmental fate. The above is crucial as metabolites are often equally toxic as their parent compounds (Bavumiragira et al. 2022; Zhang et al. 2023) and are active compounds that can potentially generate ARBs.

• (D) Currently, measurement techniques for detecting antibiotic concentrations in water are expensive, time-consuming, and require specialized knowledge to obtain accurate results. The above has limited large-scale sampling and continuous monitoring, which hinders the development of effective strategies to mitigate the environmental and health risks associated with antibiotic contamination. As a result, there is a broad research line to develop low-cost sensors to control and measure antibiotics and their metabolites in real time. This type of sensor reduces the monitoring cost (Hamel et al. 2024), which can increase the amount of data collected by water utility companies. In addition, the development and applicability of biosensors, which currently are being developed for ampicillin (Nguyen et al. 2024), roxithromycin (Shuai et al. 2024), and ciprofloxacin (Asmare et al. 2024), among others antibiotic compounds, is essential to quickly detect these pollutants at low concentrations. Further approaches, such as optimal placement of sensors in sewer networks, can potentially help in the early detection of antibiotics and their metabolites, reducing operational and data-acquisition costs and minimizing the uncertainty associated with unmonitored network locations (Banik et al. 2015). Future research should aim to integrate these sensors with automated data processing systems, improving real-time analysis and decision-making capabilities.

• (E) Most of the models have focused on calculating the RQ associated with the presence of certain antibiotic compounds in sewers. The above is expected since those are the most feasible models that can be used due to extensive studies reporting the PNEC value for several algae, bacteria, invertebrates, and fish. Current studies have largely reported RQ values above 1.0, which is an indicator of potential risk for the environment and human health. High RQ values show that currently reported concentrations in sewers are above the threshold for which no adverse effects on aquatic and ecological organisms are expected, highlighting an environment favorable for the proliferation of ARB. Sublethal antibiotic concentrations exert selective pressure, favoring the survival and replication of resistant strains and promoting the horizontal gene transfer between microbial populations (Berendonk et al. 2015). Hence, sewer systems can function as reservoirs for ARB, which can re-enter human communities through water recycling and agricultural runoff, creating pathways for the dissemination of human pathogen resistance genes (Rizzo et al. 2013). In addition, elevated antibiotic concentrations may disrupt microbial community structures, affecting key ecological processes such as nutrient cycling and food dynamics (Zheng et al. 2020; Li et al. 2024). In this context, conducting studies of this nature is essential for defining legislation and regulations to establish maximum antibiotic concentrations in sewers to avoid ecological and health risks.

• (F) There are studies applying machine learning (ML)/artificial intelligence (AI) models but in the context of photocatalytic degradation of tetracycline (Salahshoori et al. 2024) or antibiotic removal in constructed wetlands (Bao et al. 2023), among other studies. We found a lack of ML and AI models for studying antibiotics and their metabolites in sewer networks. This is in part due to the lack of massive data required to develop these models. Along the same lines, the development of a database is essential to collect information on antibiotic concentrations in sewer systems worldwide. This database is necessary to train and develop AI and ML models, useful to identify patterns in data, relationships between different pollutants, and the prediction of concentrations based on socioeconomic, climatological, and cultural characteristics, among others. In this context, the dataset collected in this paper is useful for these purposes, as we are collecting information on sewer pollutant concentrations from different countries and for different classes of antibiotics. Naturally, there is a lot of information that is outside the scope of this study, such as information from government agencies, technical reports from utility companies, and, in general, documents that are not indexed in a database. The data presented here can also be fed with information before 2014, which is currently beyond the scope of this study. In addition, this effort should be complemented with information provided for other pharmaceutical compounds, pesticides, polycyclic aromatic hydrocarbons (PAHs), corrosion inhibitors, and plasticizers, among other organic chemicals (Zhang et al. 2024).

• (G) Existing degradation/biotransformation reported rates have been developed under well-controlled laboratory conditions (especially using batch reactors), but in practice, physics and chemistry might be more complex. Laboratory conditions typically fail to account for diurnal variations in concentrations, which can lead to over or underestimation of these rates. In real-world sewers, advection, dispersion, sedimentation, and resuspension tend to dominate the water quality dynamics (Jia et al. 2021). Studying degradation/biotransformation processes using in-field data and considering dynamic concentrations is essential to outperform the calculation of these rates. With those values, simple first-order kinetic models (similar to those reported for illicit drugs (Li et al. 2019)) can be implemented and calibrated to check changes in concentrations of single pollutants in pipes. Then, these calibrated rates can be useful for complementing existing models such as the IWUS model (Vezzaro et al. 2014) by including other processes like consumption-excretion, deconjugation, and ionization-based partitioning to solids (Delli Compagni et al. 2020). Models including sediment transport processes (Montes et al. 2022, 2021; Vanegas et al. 2022) can also be useful for estimating pollutant loads entering WWTPs.

This paper shows a comprehensive review of antibiotic concentrations, frequent measurement techniques, and mathematical approaches used to predict and model these compounds in sewer systems. The report of 109 compounds, including 27 classes, shows the complexity of detecting and studying these pollutants in urban drainage systems. Key issues include the need for studies that capture temporal variations in antibiotic concentrations, particularly in the context of different meteorological conditions and seasonal fluctuations, which are essential for optimizing WWTP operations. Comprehensive studies on antibiotic metabolites are needed to understand their environmental impact and potential to generate ARBs. In addition, developing low-cost sensors and biosensors for real-time detection is essential for increasing the number of antibiotic compounds monitored and collecting large amounts of data. These findings should be leveraged to train advanced ML and AI models, enhancing the understanding of these compounds in sewers. The database presented in this work can be a first step to guide future research in this field.

This research was supported by the Universidad de La Sabana, under the call ‘Avanza un paso más hacia la última milla’.

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

The authors declare there is no conflict.