Final Report: Impact of Residual Pharmaceutical Agents and their Metabolites in Wastewater Effluents on Downstream Drinking Water Treatment Facilities

EPA Grant Number: R829014
Title: Impact of Residual Pharmaceutical Agents and their Metabolites in Wastewater Effluents on Downstream Drinking Water Treatment Facilities
Investigators: Weinberg, Howard S. , Singer, Philip C. , Sobsey, Mark D. , Meyer, M. T.
Institution: University of North Carolina at Chapel Hill , United States Geological Survey
EPA Project Officer: Page, Angela
Project Period: August 27, 2001 through August 26, 2004 (Extended to August 26, 2006)
Project Amount: $524,992
RFA: Drinking Water (2000) RFA Text |  Recipients Lists
Research Category: Drinking Water , Water Quality , Water


This study incorporated a multipronged approach to evaluate the fate and transport of pharmaceutically-derived chemicals in the aquatic environment. Due to their high end-use, antibiotics were singled out for a major investigation which included (i) isolating them from surface waters and sediments impacted by wastewater treatment plant effluent, (ii) determining whether antimicrobial resistance traits in bacteria found in those waters were correlated to environmental levels of those compounds, (iii) evaluating their removal and transformation during drinking water treatment, and (iv) how natural photolysis impacts their survival. Sensitive and reliable solid phase extraction followed by analysis with liquid chromatography and tandem mass spectrometric methods (SPE-LC-MS/MS) were developed for their analysis throughout the drinking water treatment process. Additionally, for representative components of several different groups of drugs, ultra-violet (UV) treatment of waters was investigated as a tool for remediation since such treatment is already in use for clean-up of groundwaters and is being considered in the U.S. for incorporation into drinking water treatment.

Summary/Accomplishments (Outputs/Outcomes):

Methods Analytical methods were developed for analyses of 25 antibiotics, including tetracyclines, sulfonamides, macrolides, quinolones, fluoroquinolones, trimethoprim, and lincomycin in surface waters impacted by upstream wastewater treatment plant (WWTP) discharges, sediments in the water basins, and finished drinking waters using SPE-LC-MS/MS. The method detection limits of the target analytes are generally below 10 ng/L in source water and below 5 ng/L in finished water. Since the co-extracted natural organic matter in sample matrices caused signal suppression for most of the analytes, the method of standard addition was used for quantitation to compensate for the matrix effect on signal variation. Results of a sampling holding time study indicated that source water samples should be analyzed as soon as possible after being collected to avoid appreciable analyte loss, while quenched chloraminated finished water samples should be analyzed within 7 days of sample collection.

Chlorine residuals in drinking water can react with some antibiotics but ascorbic acid was found to be an effective chlorine quenching agent without affecting the analysis of the antibiotics (except for lincomycin) and their stability in water.

Validation and application of methods for determination of clofibric acid, ibuprofen, naproxen, and diclofenac included modifications to the derivatization practice (silylation instead of methylation) instead of diazomethane) for analysis in surface water samples. This avoids the disadvantages of using diazomethane such as its toxicity, lack of stability, and difficulties in preparation. A one-step SPE method for analysis of the X-ray contrast media iohexol was validated at ng/L levels that permitted monitoring remediation using UV treatment.

Fate of Antibiotics Lincomycin was demonstrated to react very rapidly with free chlorine with apparent second order rate constants (kapp) ranging from 4.1 × 104 to 7.7 × 105 M-1s-1 in the pH range of 6 to 9, while erythromycin reacted more slowly with free chlorine with kapp ranging from 1.2 to 9.1 M1s-1 in the pH range of 6 to 10. Second order reaction models were developed based on measured kapp values and speciation of reactants. The reaction kinetics of erythromycin with free chlorine appeared to be faster in settled water (SW) than in laboratory grade water (LGW) probably due to catalytic effects of some matrix components. The rate constants obtained at 1 μM (734 μg/L) initial concentration of erythromycin were validated at the environmentally relevant level of 500 ng/L in LGW. The reaction rate of erythromycin chlorination appeared to increase with the mixing intensity of the reaction system, indicating that in full-scale water treatment the chlorination kinetics of erythromycin would be expected to vary under different mixing conditions.

Lincomycin was oxidized to isomeric sulfoxides which subsequently hydrolyzed to form isomeric hydrolysis products. Isomeric sulfanilic acid esters were also identified as intermediates from lincomycin oxidation. Erythromycin underwent chlorine attack on the tertiary amine group in its structure to form an N-chloro intermediate which hydrolyzed to form N-demethylation and di-N-demethylation products. The transformation products react with free chlorine at relatively slower rates than the parent antibiotics, suggesting that these compounds are likely to persist in finished drinking water. Doxycycline undergoes competitive initial reaction pathways to form chlorine-substituted, di-N-demethylated, dehydrogenated, and hydroxylated intermediates which further react with free chlorine. Appreciable amounts of organochlorine products were produced from chlorination of doxycycline.

The occurrence studies revealed the presence of a variety of residual antibiotics at concentrations up to 95 ng/L in source waters of selected drinking water treatment plants mostly located downstream of wastewater effluent discharge. Most of the antibiotics detected in source waters were either below the practical quantitation level or were found at greatly reduced levels in finished water, indicating their partial removal in full-scale water treatment. Sulfamethoxazole was the most often detected antibiotic in source waters in this study probably because of its high usage and stability in aqueous systems. Tetracyclines were the least detected antibiotics in this study probably because they are more likely to partition out of the water column and into sediments and, therefore, not be transported to drinking water supplies.

Analysis of samples at different stages of treatment in drinking water treatment plants show that coagulation/flocculation and subsequent sedimentation can remove erythromycin, trimethoprim, and ciprofloxacin to some degree, but were not effective in removing sulfamethoxazole. Carbon adsorption, however, is very effective in removing sulfamethoxazole and trimethoprim, and can also remove erythromycin, lincomycin, and ciprofloxacin to some degree. Chlorination can partially transform these compounds depending on their individual reactivity and contact time with free chlorine.

Transformation products/metabolites of lincomycin and erythromycin were tentatively detected in samples collected at different stages of two treatment plants and appeared to persist through drinking water treatment up to the consumer’s tap.

In photolysis reactors, sulfamethoxazole was found to be susceptible to degradation. However, the presence of natural organic matter and particulates in the water column increased the persistence of sulfamethoxazole in the aqueous phase. Sorption to sediments was negligible in reactors except in the presence of high organic soil when 15% partitioned over 20hrs of recirculation. Since most streams and rivers have low organic content in the sediments, the concentration of sulfamethoxazole is expected to be low in comparison to other xenobiotics. On the other hand, photolysis byproducts were identified in the aqueous phase, including an isoxazole isomer which was also detected in the effluent discharge of a WWTP and further downstream when low pressure mercury lamps were used to disinfect the effluent.

Sulfamethoxazole, trimethoprim, tetracycline, ciprofloxacin, and levofloxacin were extracted from sediments located at the point of WWTP discharge and 725m downstream. Concentrations in the sediment were higher than those found in the surface water and effluent, particularly for ciprofloxacin (at least 400 times more concentrated in the sediment). While there have been many studies regarding the occurrence of antibiotics in WWTP effluent and surface water, less is known regarding the fate.

Antibiotic Resistance The presence of antibiotics and antibiotic resistant bacteria in human and animal fecal wastes and in ambient waters has been documented. Of particular concern are those bacteria that not only infect humans and animals, where they can get exposed to antibiotics in these hosts, but are also able to proliferate in the environment, where they can not only acquire antibiotic resistance bit also amplify and possibly spread the resistance traits they possess. The aeromonads, including Aeromonas hydrophila, which is on the U.S. EPA Candidate Contaminant List, are bacteria that may have these abilities and properties. Therefore, they were studied for antibiotic resistance properties and other properties of human health concern (e.g., the presence of virulence genes) in relation to their occurrence in municipal sewage and in ambient waters impacted by sewage effluents. Fifty isolates from each sampling point at various locations around wastewater plant discharges from three plants into receiving waters were tested for resistance to several commonly employed antibiotics.

The 16s rDNA molecular analysis gave very low resolution species due to the lack of specificity of the primers and the amplification of many different species. The method was unable to determine species’ differences. This was discovered while analyzing several well characterized control species and strains. All but one of the controls was positive on visualization using gel electrophoresis. The results of this study suggest that Aeromonas strains may be remaining in the treated sewage effluent after municipal wastewater treatment and reclamation and are then discharged into environmental waters. Another explanation for their presence in receiving waters is that their resistance traits were acquired in the water due to the selective pressure of the antibiotics presence in the discharged waste water and the creek, or they were acquired by genetic transfer from other bacteria. Because these antibiotic resistant Aeromonas bacteria were isolated from receiving waters and because aeromonads have previously been shown to proliferate in water and sediments, the potential for dissemination and proliferation of these potentially multi-drug resistant pathogens via the aquatic environment is high. This is because antimicrobial resistance in Aeromonas spp. is primarily plasmid-mediated, leading to a high potential for Aeromonas spp. to transfer these resistance properties to microorganisms of other genera.

Sewage effluent-impacted sediment, which is known to concentrate both bacteria and certain antibiotics, was examined as a possible arena for the acquisition by bacteria of increased resistance to antimicrobial pharmaceuticals. Environmental bacteria, Aeromonas spp., were isolated from these sediments, characterized for resistance to 4 antimicrobials and these data were used to measure the differences in up- and downstream drug susceptibility. This study concluded that the impacts of a wastewater treatment plant discharge contribute markedly to the antimicrobial resistance of Aeromonas spp. in the environment. This was shown by an approximately 30% increase in the number of downstream isolates demonstrating either absolute or intermediate resistance to a number of antibiotics.

UV Treatment The use of low pressure (LP) and medium pressure (MP) mercury lamps was evaluated for direct and indirect (by addition of hydrogen peroxide) UV treatment of ketoprofen, naproxen, carbamazepine, ciprofloxacin, clofibric acid, and iohexol. Overall, MP lamps proved to be more effective at maximizing the degradation of the selected group of compounds by both direct and indirect photolysis. Fundamental direct and indirect photolysis parameters obtained in LGW were reported and used to model the LP and MP direct and indirect photolysis of the pharmaceuticals in surface water. Direct (MP and LP) and indirect (MP) photolysis modeling predicted the experimental results obtained very well while the LP indirect photolysis model results underestimated the removal of naproxen and carbamazepine. Using LP lamps the highest removal rate was obtained by iohexol (kf,LP,SW=0.0058cm2/mJ) followed by clofibric acid (kf,LP,SW=0.0013cm2/mJ), naproxen (kf,LP,SW=0.0005cm2/mJ), and carbamazepine (kf,LP,SW=0.00003cm2/mJ) while with MP lamps clofibric acid showed the highest removal rate (kf,MP,SW=0.0067cm2/mJ) followed by naproxen and iohexol (kf,MP,SW=0.0025cm2/mJ), and carbamazepine (kf,MP,SW=0.0002cm2/mJ). Removal of carbamazepine from surface water using direct photolysis is not likely to be feasible due to its low quantum yield (0.0006 and 0.0023mol.einstein-1using LP and MP lamps, respectively). Indirect photolysis however, will increase the degradation of carbamazepine and naproxen considerably in surface water. As an example, at the UV fluence of 100mJ/cm2the MP removal of carbamazepine will increase from negligible to 13% while the removal of naproxen will increase from 36% to 52%. The indirect photolysis fluence-based rate constants obtained using LP and MP lamps varied between 0.0017≤kf,LP/H2O2,SW≤0.0064cm2/mJ and 0.0027≤kf,MP/H2O2,SW≤0.0076cm2/mJ, respectively.

The direct and indirect photolysis models were also used to predict the degradation of the compounds under different experimental conditions. In general, the rate constants increase with decreased light source path length and increased hydrogen peroxide concentration although at high concentrations the increase of the overall degradation rate constants is slightly attenuated due to the scavenging of light.

Summary of Findings
A summary of practical implications of this research for water treatment plants, include:

  1. Some pharmaceuticals such as ketoprofen and ciprofloxacin were significantly removed from LGW by direct photolysis using a UV fluence that can typically be used during drinking water treatment (100mJ/cm2). These results need to be validated in surface water due to the matrix competition for UV light that decreased the removal of some of the other pharmaceuticals examined in this study.
  2. Carbamazepine doesn't appear to be amenable to photodegradation in surface water at the UV fluences typically used for drinking water treatment (in the range 30 to 140mJ/cm2). A significant reduction could be obtained if a high UV fluence (such as 1700mJ/cm2) was used in combination with 10mg/L of hydrogen peroxide. If higher hydrogen peroxide concentrations were used with lower UV fluences the compound's removal might also be improved due to higher OH radical formation.
  3. MP lamps performed better for the removal of some pharmaceuticals such as naproxen, clofibric acid, and carbamazepine than LP lamps and this outcome could be predicted by determination of the compounds' absorbance spectra.
  4. The generation of OH radicals in an AOP in surface water doesn't seem to enhance removal of all pharmaceuticals, as was observed for clofibric acid and iohexol.
  5. All the pharmaceuticals tested in surface water (carbabazepine, naproxen, clofibric acid, and iohexol) were reduced to levels below detection using LP or MP lamps with a UV fluence of 1700mJ/cm2and 10mg/L hydrogen peroxide.
  6. For some of the pharmaceuticals studied the formation of chlorination and photolysis products was observed.
  7. Residues of antibiotics are being found at low ng/L levels in some drinking waters generated from surface water impacted by upstream WWTP discharges.
  8. The incomplete oxidation of antibiotics during drinking water treatment may not completely eliminate their biological effects. The health effects of long-term exposure of antibiotics and their transformation products at trace levels via drinking water consumption are unknown. The risk assessment associated with presence of antibiotics in drinking water should take into consideration the effects of the parent antibiotics as well as their transformation products.
  9. It is highly likely and most plausible that the wastewater effluent facilitates the phenomenon of elevated antimicrobial resistance among targeted microbes downstream of the point of wastewater discharge into surface waters.

Quality Assurance (QA)
Field QA Samples collected during the occurrence survey were accompanied with chain of custody documentation that included precise details on how to collect, store, preserve, and ship samples so that they would not lose integrity prior to analysis. The designated preservatives as identified in this research were placed in labeled 500 mL or 1L amber bottles with open top caps containing Teflon-lined seals prior to shipment. At the time of collection, aqueous samples were filtered through 0.45 μm glass-fiber filters into the sample collection bottles until the sample was head-space free. The bottles were stored at 4°C until close to the time of pick-up for overnight shipment packed in protective materials and also kept cool with frozen “blue” ice packs. External QA/QC comprised field and equipment blanks, field spikes, and blind samples. One field blank, equipment blank, field spike, and blind sample was submitted with each sampling event or every 10 samples. Sediment samples were shipped in high density polypropylene 1L bottles without preservative.

Laboratory QA/QC Samples received at the UNC Laboratory were logged in and refrigerated at 4°C until they were extracted usually within 24 hours. Aqueous samples were extracted in sets of 16 that included 11 samples, one duplicate, two standards and two blanks. Duplicate seven point standard curves were generated by extracting 500 mL laboratory grade water (LGW) spiked at 0.02, 0.05, 0.1, 0.2, 0.5, 1.0, and 2.0 μg/L with a standard mix containing 2.5 ng/μl of each analyte. The standards of each sample set were checked against the existing curve. Blanks were used to check for potential sample carryover. Stock solutions of each analyte were made from analytical grade standards at a concentration of 1 mg/mL in methanol. The standard mix was prepared in 100 mL batches from individual stock solutions in LGW with 20 mM ammonia acetate adjusted to pH 6. The standard mix was divided into 20 mL aliquots that are kept frozen until use. Instrument performance evaluation standards were prepared to monitor the condition of the mass spectrometers used in this research. Tolerance within +/-20% response on a targeted ion at a specific on-column concentration (which varied for each analyte) was an acceptable limit for operations to proceed.

Microbial Positive and Negative Controls Microbiological methods for bacteria and their antibiotic resistance properties were carried out according to sandard procedures as documented in such sources as Standard Standard Methods for the Examination of Water and Wastewater, EPA's manual entitled Microbiological Methods for Monitoring the Environment; the Manual of Clinical Microbiology, 6th edition and other handbooks and manuals of standard procedures. For antibiotic susceptibility testing procedures were essentially those described in "Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria That Grow Aerobically, Fifth Edition (NCCLS, 2000. This newly revised standard provides updated reference methods for the determination of minimal inhibitory concentrations (MICs) for aerobic bacteria by broth macrodilution, broth microdilution, and agar dilution. This document contains updated MIC interpretive criteria and quality control parameters tables.

In general, the precision and accuracy of microbiological data for samples was based upon replicate analyses. For precision, the standard deviation (SD) of replicate measurements fell within the 95% confidence interval (CI) of the mean (X-bar) plus or minus 1.96 SD, or for few replicates, X-bar plus or minus t, where t is the critical value of Students t distribution. Accuracy of microbial assays was based upon whole units, which are individual cells, colonies, or other countable units for enumerative methods and positive and negative units (culture tubes, etc.) for quantal methods. For enumerative methods, accuracy was generally estimated to be plus or minus <5% because a minimum of 20 units (colonies) were counted. For quantal assays, accuracy was estimated from the Poisson distribution, based on the number of analytical units per dilution, the number of dilutions and the dilution interval used. Typically, quantal infectivity assays were carried out using 5 units (e.g., cultures) per dilution and three decimal (10-fold) dilutions, allowing the use of Most Probable Number (MPN) tables giving both the MPN values and its upper and lower 95% limits. For completeness, replicates were run for enumerative methods and multiple units per dilution were run for quantal methods. All laboratory tests on each experimental variable or condition were carried out in duplicate or more.

Sample points and times were standardized. For representativeness in field samples, sample volumes were sufficiently large to allow separate analysis of duplicate samples. To achieve comparability standardized sampling points, sample volumes, sampling procedures and analytical methods were used. Data are reported in standard units to permit direct comparisons. Samples are were analyzed by trained laboratory staff. Samples were shipped and/or stored at refrigerated temperature on ice packs in insulated containers or in laboratory refrigerators and analyzed or used within 24 hours of collection. Samples were shipped by approved carriers and are accompanied by chain-of-custody forms which document sample handling conditions over time.

Sterility indicator tape was used for each load of material autoclaved, and biweekly checks of autoclave performance were made using heat-resistant spores of B. stearothermophilus. Sterility checks were made for each batch of microbiological media and diluent. Laboratory strains of test microbes were verified for purity by cultivation, biochemical reactions and staining reaction and by immunofluorescence and other analytical methods. For antibiotic susceptibility testing, positive control bacteria were included as reference strains and antibiotics susceptibility reagents purchased from reliable commercial sources meeting NCCLS and ISO requirements. Incubator and water bath temperatures were checked and recorded daily. Reagent grade water was tested for quality weekly. All instruments were calibrated weekly or more often and results recorded. Each batch of microbiological media and reagents was checked according to standard quality control procedures. Positive and negative control samples were included in each analysis. The positive controls for GYR A, GYR B, PAR C and PAR E genes for quinolone resistance in Aeromonads were GYR A MC3 DS1 1, GYR B MC4 DS1 6, PAR C MC3 DS1 4, and PAR E MC3 DS1 11 and were sequenced at the UNC Genome Analysis Facility. They were compared against published sequences from GenBank for confirmation prior to use as a control. The negative controls used DNA/RNA free water. The positive control for the 16s rDNA PCR method was A. caviae ATCC 15468. For the isolation method, the positive control was A. hydrophila ATCC 7966, A. caviae ATCC 15468, and A. sobria ATCC 43979 while the negative controls were P. Aeruginosa ATCC 27853 and E. coli B.

Personnel Changes
This project supported the graduation of two doctoral candidates and three masters students over a span of 5 years. Some of the students graduated prior to the completion of the project period but after they had completed their aspect of the study. Two additional Masters level students began their studies in the fifth year of this project and will complete their degree requirements after completion of the project. Investigator responsibilities and emphases shifted during the course of this study. Philip Singer was involved in the selection of the utilities for the occurrence survey while Jan Vinje joined the project in the latter stages bringing his molecular biology expertise to the project extension. Karl Linden, though not supported from this grant, lent his expertise to the work carried out to investigate UV transformation and was a participant in one of the student’s thesis committees.

Journal Articles on this Report : 3 Displayed | Download in RIS Format

Other project views: All 12 publications 3 publications in selected types All 3 journal articles
Type Citation Project Document Sources
Journal Article Pereira VJ, Linden KG, Weinberg HS. Evaluation of UV irradiation for photolytic and oxidative degradation of pharmaceutical compounds in water. Water Research 2007;41(19):4413-4423. R829014 (Final)
  • Abstract from PubMed
  • Full-text: Science Direct Full Text
  • Other: Science Direct PDF
  • Journal Article Pereira VJ, Weinberg HS, Linden KG, Singer PC. UV degradation kinetics and modeling of pharmaceutical compounds in laboratory grade and surface water via direct and indirect photolysis at 254 nm. Environmental Science & Technology 2007;41(5):1682-1688. R829014 (Final)
  • Abstract from PubMed
  • Full-text: ACS Full Text
  • Other: ACS PDF
  • Journal Article Ye Z, Weinberg HS, Meyer MT. Trace analysis of trimethoprim and sulfonamide, macrolide, quinolone, and tetracycline antibiotics in chlorinated drinking water using liquid chromatography electrospray tandem mass spectrometry. Analytical Chemistry 2007;79(3):1135-1144. R829014 (Final)
  • Abstract from PubMed
  • Supplemental Keywords:

    pharmaceuticals, antibiotics, antibiotic resistance, disinfection, exposure, organics, analytical methods, ozonation, ultra-violet photolysis, chlorination byproducts, x-ray contrast media, drinking water treatment,, RFA, Scientific Discipline, Water, Ecosystem Protection/Environmental Exposure & Risk, Hydrology, Environmental Chemistry, Ecosystem/Assessment/Indicators, Fate & Transport, Ecological Effects - Environmental Exposure & Risk, Ecological Effects - Human Health, Drinking Water, Ecological Indicators, ecological effects, fate and transport, monitoring, ecological exposure, bacteria, metabolites, pharmaceuticals, vulnerability, exposure and effects, wastewater, exposure, other - risk assessment, chemical contaminants, chemical transport, treatment, water quality, organic chemicals, drinking water contaminants, water treatment, drinking water treatment

    Progress and Final Reports:

    Original Abstract
  • 2002 Progress Report
  • 2003 Progress Report
  • 2004 Progress Report
  • 2005