Detection frequency, classification and distribution of antibiotics
Fig. 1: Detection frequencies of antibiotics identified in 2021 (red, n = 35 biologically independent samples) and 2023 (blue, n = 35 biologically independent samples) at Site A (n = 21 biologically independent samples for both years) and Site B (n = 14 biologically independent samples for both years).
Source data are provided as a Source Data file.
The proportional abundances for each drug class across all wastewater treatment plants (WWTPs) are depicted in Fig. 2A. Quinolones were identified as the dominant category in wastewater samples in both 2021 and 2023, contributing a significant range between 18.36% and 96.51% to the overall antibiotic composition. In accordance with the WHO AWaRe (Access, Watch, and Reserve) metrics, a recommended target is for countries to consume at least 60% of their antibiotics from the Access group. However, our analysis of the studied regions revealed that Access group antibiotics (nine in total) merely comprised 5.84%–22.05% of all detected antibiotics, while Watch group antibiotics (twelve in total) accounted for a staggering 77.95%–98.68% (Fig. 2B), a trend echoed in a review of antibiotic usage among COVID-19 patients. Alarmingly, 2023 saw an even greater percentage of the Watch group relative to the Access group. This shift points toward potential misuses of antibiotics such as quinolones and macrolides after COVID-19 restrictions were lifted. The WHO had previously advised against antibiotic use for mild COVID-19 cases. Disturbingly, several antibiotics on the WHO ‘Watch’ and ‘Reserve’ lists were commonly prescribed in certain countries, particularly in low- and middle-income nations.
Fig. 2: Proportional abundances of antibiotics in wastewater samples.
A Proportional abundances of each drug class. B Distribution of antibiotics according to WHO AWaRe categories (Access, Watch and Reserve). Source data are provided as a Source Data file.
Concentrations, daily mass loads and PNDLs
The concentration, daily mass load, and population-normalized daily load (PNDL) of antibiotics by drug class across all WWTPs are illustrated in Fig. 3 and Table S2. The levels of compounds classified as macrolides, quinolones, sulfonamides, phenicols, and tetracyclines in WWTPs are detailed in supplementary materials (Fig. S3). In 2023, there was an overall increase in concentration, daily mass load, and PNDLs compared to the restricted conditions of 2021. This escalation is primarily linked to the increased exposure to antibiotics during this period. A longitudinal study that tracked various antibiotics in the urine of the general Chinese population also revealed a spike in antibiotic consumption during the pandemic. Additionally, instances of inappropriate antibiotic use and high rates of prescriptions among COVID-19 inpatients were observed, particularly in several low- and middle-income countries. These findings align with various studies that reported a rise in antibiotic usage for COVID-19 patients and widespread empirical antibiotic therapy among hospitalized patients, raising serious concerns about antibiotic resistance.
Fig. 3: Concentration, daily mass load and population-normalised daily load of antibiotics in all WWTPs.
A and B: sampling sites, 1–5: WWTPs, 2021 and 2023: sampling periods (A2 and B5: domestic WWTPs. A1, A3, and B4: mixed WWTPs). Source data are provided as a Source Data file.
Results detailing concentrations classified by the studied compounds are available in the supplementary material (Fig. S4). Wastewater types analyzed included domestic contributions alongside combined domestic and industrial inputs. Antibiotic concentrations were significantly higher in 2023, constituting a serious public health concern. These levels point to wastewater as a route of human exposure to animal antibiotics, a consequence of their extensive use in livestock and aquaculture in China.
Several antibiotics—including azithromycin, ciprofloxacin, clarithromycin, florfenicol, norfloxacin, ofloxacin, tilmicosin, and trimethoprim—were detected at concentrations exceeding the Predicted No Effect Concentration (PNEC) benchmarks as derived from Minimal Inhibitory Concentration (MIC) data. Such findings suggest a potential risk for antibiotic resistance developing within the sewer systems. The continuous use of these critically important antibiotics heightens the risk of emerging multidrug-resistant (MDR) strains, threatening treatment efficacy for serious infections and conditions like neonatal sepsis. The alarming rise in resistance to azithromycin and other broad-spectrum antibiotics could drastically reduce treatment options available, making effective therapies unaffordable, particularly in low- and middle-income countries. Persistent usage and introduction of these antibiotics into the environment present a significant pollution threat and could exacerbate antibiotic resistance issues. Given the daily applications of antibiotics in the medical and agricultural sectors, a thorough, coordinated effort across various sectors is crucial to limit environmental exposure and prevent the clinical misuse of these drugs.
Flow rates exhibited notable variability in relation to each WWTP’s capacity, with minor inter-annual fluctuations in WWTPs observed throughout the study duration. In 2021, overall daily mass loads ranged between 3.661 g/day and 553.30 g/day, reflecting variation impacts across different WWTPs. The 2023 sampling campaign revealed significantly higher daily mass loads, indicating an astonishing increase of 2348% compared to 2021. WWTP-B4 (mixed WWTP) documented the peak daily load in winter 2023, primarily attributed to contributions from ofloxacin (H/VA, 2482.74 ± 474.39 g/day), azithromycin (HA, 888.36 ± 445.74 g/day), roxithromycin (H/VA, 728.49 ± 92.99 g/day), and florfenicol (VA, 190.82 ± 177.72 g/day). The detection of tilmicosin exclusively in WWTP-B4 in 2023 highlights the intricate influent compositions stemming from diverse sources including communities, hospitals, and animal husbandry. The elevated tilmicosin load (VA, 734.32 ± 361.95 g/day), along with a 32% increase, signifies the concerning extent of antibiotic application within industrial livestock and poultry sectors in the sampled region during this timeframe.
The size of the population served by selected WWTPs was approximated using local census data. The surge in population-normalized daily loads (PNDLs) was evident between 2021 (471.10 ± 272.51 mg/day/1000inh) and 2023 (9403.31 ± 5349.66 mg/day/1000inh), indicating increases ranging from 63% to 3734% in 2023. WWTP-A3, which supports industrial parks and townships, exhibited the most pronounced rise (mixed WWTP: 3734%), followed closely by WWTP-B5 (domestic WWTP: 3537%) achieving the highest PNDLs at 5094.75 ± 3826.26 mg/day/1000inh. Little demographic variation between 2021 and 2023 appears plausible, reflecting the strict controls thereby enacted during the “zero-COVID” policy. Consequently, the PNDLs shifts cannot be attributed to any changes in population numbers. The analysis of PNDLs for the studied compounds is elaborated further in the corresponding section (Patterns of antibiotics). Confirming the temporal increase in overall antibiotic concentrations and PNDLs in settled populations, we measured NH4-N concentrations, a reliable control compound and population indicator that consistently appears in wastewater. NH4-N levels remained stable throughout both sampling years, with no statistically significant differences detected between groups.
Patterns of antibiotics
The variations in antibiotic levels between 2021 and 2023 are categorized by drug classes. Physicians, lacking robust data and confronted with the absence of effective treatments, may have adopted combined antiviral drugs and antibiotics for COVID-19 management. Many patients with ordinary influenza and other common respiratory infections likely resorted to self-medication with these antibiotics, as many are available over the counter in China. Therefore, a notable increase in the population-normalized daily load of antibiotics was documented in 2023.
Fig. 4: PNDL (population-normalised daily loads) for macrolides, quinolones, sulfonamides, phenicols and tetracyclines in WWTPs.
A and B: sampling sites; 1–5: WWTPs; 2021 and 2023: sampling periods. Source data are provided as a Source Data file.
Macrolides
Quinolones
Distinctive usage patterns of ciprofloxacin and lomefloxacin were identified in the two sampling regions during the COVID-19 pandemic in 2023. While these antibiotics were not detected in the purely domestic WWTP of Site A (WWTP-A2), they appeared in WWTP-B5, indicating divergent prescribing habits between the locations. Lomefloxacin, categorized as a Reserve group antibiotic, is banned in food-producing animals and recommended as a last-line treatment option for human patients. The consistently high levels of lomefloxacin in WWTP-B5 across both sampling years suggest a persistent misuse of the drug at this site, underscoring the imperative need for rigorous monitoring and regulation.
Sulfonamides
Sulfamethoxazole, a human antibiotic predominantly used in China, was consistently present in all analyzed samples from both 2021 and 2023, with a remarkable 1647% increase observed in 2023 (PNDLs: 208.60 ± 150.60 mg/day/1000inh) compared to 2021. The data showed a stable ratio of sulfamethoxazole to its metabolite (SMX/aSMX) in all WWTPs, ranging from 0.61 to 1.41. The levels and trends for sulfamethoxazole and its metabolite remained similar over seven days within both sampling periods, indicating that inappropriate disposal of unused sulfamethoxazole did not occur.
Phenicols and tetracyclines
Chloramphenicol and doxycycline, both human-use antibiotics, were only detected in 2021, with PNDLs at 0.71 ± 1.07 mg/day/1000inh and 36.21 ± 52.53 mg/day/1000inh, respectively. Both drugs were absent in WWTP-B5 in 2021, suggesting that their contributions were primarily from pharmaceutical factories at Site B. Numerous studies highlight prolonged tetracycline resistance even after halting drug use. While doxycycline remains one of the world’s most prescriptive antibiotics, the data indicates that tetracyclines—doxycycline included—were not significantly produced or utilized in Eastern China between 2021 and 2023.
Correlations of antibiotics/metabolites: co-prescribing patterns
Correlation analysis of weekly averages in wastewater analyses was conducted to explore potential co-prescribing patterns in 2021 and 2023. Several antibiotics within wastewater samples exhibited significant correlations, including roxithromycin and ciprofloxacin (r = 0.94), roxithromycin and lomefloxacin (r = 0.84), erythromycin and ciprofloxacin (r = 0.84), clarithromycin and ciprofloxacin (r = 0.67), and erythromycin and lomefloxacin (r = 0.71). A strong correlation exhibited between ofloxacin and roxithromycin (r = 0.89, p
Fig. 5: Correlation analysis among all antibiotics in sampled wastewater in 2021 and 2023 (n2021:35 and n = 35 are biologically independent samples) White: no statistical correlation; red: positive correlation; blue: negative correlation; numbers in squares: r value.
Source data are provided as a Source Data file.
The relationships between parent compounds and metabolites were examined, revealing strong positive correlations between sulfamethoxazole and its principal metabolite (r = 0.85, p
Community-wide intake calculations of antibiotics
Back-calculations based on population-normalized daily intake (PNDI) were performed for 12 human and veterinary antibiotics using correction factors (CFs) and PNDLs. These CF values enabled the estimation of the mass of each drug consumed at the community level. The calculations utilized quantified antibiotic concentrations in wastewater and adjusted for the proportion excreted through urine or feces.
Fig. 6: Box plot comparisons of antibiotic PNDIs for 2021 and 2023.
a PNDIs across the surveyed catchment (Site A and Site B, n = 70 biologically independent samples) community, b PNDIs for Site A (n = 42 biologically independent samples), and c PNDIs for Site B (n = 28 biologically independent samples). The blue boxes display data from 2021, while the red boxes show 2023 data. Box plots highlight median (line within box), 25% and 75% percentiles (box limits), and minima and maxima (whiskers). Statistical differences between 2021 and 2023 were evaluated using a two-sided Wilcoxon signed-rank test, marking significant compounds.
A study in Shanghai showed that urinary antibiotic concentrations during the COVID-19 pandemic were significantly elevated compared to pre-pandemic levels. A meta-analysis indicated a prevalence of bacterial infection around 8.6% during the early pandemic phase. Despite this relatively low prevalence, antibiotics were administered to an average of 64% of patients during the pandemic. Notably, the WHO recommended against the empiric use of antibiotics in mild COVID-19 patients. Analysis revealed that despite decreasing antibiotic prescriptions over time, many continued to receive unnecessary treatments.
Total PNDIs noted in 2023 were similar across Sites A and B, although higher rates of increase were observed in Site A. The agreement in PNDIs between sites was notable for roxithromycin, clarithromycin, erythromycin, ciprofloxacin, and sulfadiazine. Higher PNDIs at Site B signified different consumption patterns, indicating more conservative antibiotic use in Site A.
Comparison of WBE to prescription data
Table 1 Comparison of PDNIs to the monthly catchment prescription data
While prescription data provide valuable insights into trends, they do not necessarily reflect true antibiotic consumption within the community. The increase in prescription rates for Azithromycin and Clarithromycin exceeded levels found in wastewater, suggesting non-adherence to prescribed treatments or stockpiling for future use. Conversely, some antibiotics quantified in wastewater were absent from the electronic prescription records, potentially due to acquisitions outside the catchment, over-the-counter purchases, or delays in prescriptions. These aspects highlight the complexity of estimating community consumption and underscore the supplementary role of prescription data in understanding antibiotic use.
Limitations
This study encounters several limitations. First, the average residence time of wastewater within sewers can vary based on multiple factors; however, uniform loss rates from toilets to WWTPs were assumed. Future research should focus on refining these estimates and sampling regions, possibly integrating models to address residence times and environmental conditions. Second, there are no WBE data available on antibiotic levels preceding COVID-19. Third, due to technical constraints, cephalosporin antibiotics’ presence couldn’t be evaluated. Future research will enhance the analysis of cephalosporin antibiotics to ensure sensitive detection. Furthermore, extensive analyses are warranted for deeper insights, especially concerning antibiotic resistance data collected during the same timeframe. Monitoring antibiotic consumption throughout the pandemic, specifically for particular drugs, is essential. This study introduces WBE as a robust tool for estimating community-wide antibiotic exposure, including assessments of both parent compounds and their metabolites. Significantly, it represents the inaugural use of WBE in China to track and compare antibiotic usage patterns before (2021) and after (2023) changes to national policy. Antibiotic consumption saw a marked rise after the relaxation of the “zero-COVID” policy in 2023. Continued attention on antibiotic resistance is critical both in China and worldwide, with a concerning trend visible in 2023 showcasing increased proportions of antibiotic use from the Watch group along with unwarranted prescribing habits, largely propelled by macrolides and quinolones. WBE shows a compelling resemblance and amplification of prescription data, emphasizing its efficacy as a superior monitoring tool during exceptional circumstances.
How do fluctuations in population size and demographics impact the accuracy of population-normalized daily loads (PNDLs) and daily intake (PNDI) estimates for antibiotics?
First, the analysis is based on wastewater samples, which may not fully capture the individual consumption behaviors and patterns of antibiotic use across diverse populations. Variability in wastewater composition can occur due to factors such as population density, demographics, and local healthcare practices, potentially skewing the results. Second, the reliance on population-normalized daily loads (PNDLs) and population-normalized daily intake (PNDI) requires precise population statistics. Fluctuations in population size or transient populations (such as tourists) may affect the accuracy of the estimates.
Additionally, while the study provides an overview of antibiotic usage, it cannot account for regional variations or the impact of local health policies and regulations on prescribing habits. Moreover, the presence of antibiotics in wastewater does not directly indicate misuse, as some levels can reflect therapeutic use. The data may also be influenced by other factors, such as non-compliance with treatment regimens, stockpiling of medications, or access to over-the-counter antibiotics.
There are also potential issues with the detection methods utilized, which may have limitations in sensitivity or specificity, impacting the accuracy of quantified antibiotic levels. while correlation analyses highlight potential co-prescribing patterns, these do not establish causation, and further investigation is needed to understand the underlying factors driving observed associations.
Conclusions
This study reveals significant increases in the population-normalized daily loads of various antibiotics in wastewater treatment plants from 2021 to 2023, particularly during the COVID-19 pandemic. The findings illustrate concerning trends in antibiotic misuse and self-medication, particularly in the context of emerging infectious diseases and potential resistance development. Comprehensive monitoring and regulation, alongside public health education, are critical to curbing inappropriate antibiotic use and ensuring safe and effective treatment for patients.
Future studies should focus on establishing more nuanced methodologies for assessing community antibiotic consumption, including longitudinal studies that track changes over time, the impact of public health interventions, and the integration of prescription data with wastewater analysis for a more holistic understanding of antibiotic usage patterns.
