Water-Quality Data for Pharmaceuticals, Hormones, and Other Organic Wastewater Contaminants in U.S. Streams, 1999-2000
By KIMBERLEE K. BARNES, DANA W. KOLPIN, MICHAEL T. MEYER, E. MICHAEL THURMAN, EDWARD T. FURLONG, STEVEN D. ZAUGG, AND LARRY B. BARBER
Open-File Report 02-94
Iowa City, Iowa
U.S. Department of the Interior
U.S. Geological Survey
Any use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government
Occurence of Organic Wastewater Contaminants
Water-quality data collected during 1999 and 2000 as part of the
first nationwide reconnaissance of the occurrence of pharmaceuticals,
hormones, and other organic wastewater contaminants (OWCs) are presented
in this report. A network of 139 streams in 30 states were sampled
and analyzed for 95 different OWCs using five new research methods
developed by the U.S. Geological Survey. Site selection was biased
toward streams more susceptible to OWC contamination because of proximity
to urban areas or livestock production. At least one OWC was detected
in 80% of the streams sampled, with 82 of the 95 analyzed OWCs determined
in this study detected in at least one sample.
Water quality is becoming more of a concern as demand for freshwater increases throughout the world. Technological advancements in industry, agriculture, medical treatment, and common household conveniences have brought increasing concerns for potential adverse human and ecological effects from chemicals present in the environment (Daughton and Ternes, 1999). Research has shown that many such compounds can enter the environment, disperse, and persist to a greater extent than first anticipated. As a result, there are a wide variety of transport pathways for many different chemicals to enter and persist in environmental waters.
Little is known about the extent of environmental occurrence, transport, and ultimate fate of many synthetic organic chemicals after their intended use, particularly pharmaceuticals and personal care products that are designed to stimulate a physiological response in humans, plants, and animals (Daughton and Ternes, 1999; Jorgensen and Halling-Sorensen, 2000). Until recently, there have been few analytical methods capable of detecting most of the target compounds at low concentrations which might be expected in the environment (Sedlak and others, 2000).
This report presents the results of data-collection activities from 139 streams in 30 states sampled during 1999-2000 (Figure 1). Sampling sites were primarily selected from streams thought to be susceptible to contamination from agriculture or urban activities. Using five newly developed analytical research methods, the U.S. Geological Survey analyzed water samples for 95 different organic wastewater contaminants (OWCs).
Little data were available on the occurrence of most of the targeted OWCs in streams at the onset of this investigation. Therefore, to determine whether the targeted OWCs are present in stream waters, the selection of sampling sites primarily focused on sites considered to be susceptible to contamination from human, industrial, and agricultural wastewater (i.e. downstream of intense urbanization and livestock production). The 139 stream sites sampled during 1999-2000 (Figure 1, Table 1) represent a wide range of geography, hydrogeology, land use, and basin size.
All samples were collected by U.S. Geological Survey personnel using consistent protocols and procedures designed to obtain a sample representative of the stream waters using standard depth and width integrating techniques (Shelton, 1994). At each site, a composite water sample was collected from about 4- 6 vertical profiles which was subsequently split into appropriate containers for shipment to the participating laboratories. For those bottles requiring filtration, water was passed through a 0.7 mm pore- size, baked, glass-fiber filter in the field where possible, or else filtration was conducted in the laboratory. Water samples for each chemical analysis were stored in amber, baked-glass bottles and collected in duplicate. The duplicate samples were used for backup purposes (in case of breakage of the primary sample) and for laboratory replicates. Following collection, samples were immediately chilled and sent to the laboratory. In addition to standard sampling procedures (Shelton, 1994), efforts were made to eliminate potential contamination of samples by contact with the collector including the avoidance of personal care items (i.e. insect repellents, colognes, perfumes), caffeinated products, and tobacco during sample collection and processing.
In general, each stream site was sampled once during the 1999-2000 study period. Select samples collected in 1999, however, were analyzed for a subset of the OWCs based on the watershed land-use characteristics. Thus, a select number of stream sites were resampled in 2000 to obtain data on OWCs not analyzed from samples collected in 1999. Stream sites sampled in 2000 were analyzed for the complete suite of OWCs. The analytical results for each stream sample are available in this report.
To determine the environmental extent of 95 OWCs in the 139 streams sampled (Table 1), five separate analytical methods were used (Tables 2-6). Target compounds analyzed by each method were selected from the large number of possible chemicals based upon estimates of usage, toxicity, potential hormonal activity, and persistence in the environment. Each method was developed independently, with somewhat different data objectives, such as identifying hormones versus identifying antibiotics in water. A number of reported concentrations were flagged with an "E" to indicate estimated values. These include all concentrations above or below the calibration curve, concentrations for compounds with average recoveries less than 60 percent, compounds routinely detected in laboratory blanks, and compounds whose reference standards were prepared from technical mixtures (Kolpin and others, 2002).
Method 1 targets 21 antibiotic compounds (Table 2 ) in filtered water samples using modifications from a previously described method (Hirsch and others, 1998). The antibiotics were extracted and analyzed by tandem solid-phase extraction (SPE) and single quadrapole, liquid chromatography/mass spectrometry positive-ion electrospray (LC/MS-ESI(+)) analysis using selected ion monitoring (SIM). The tandem SPE included an Oasis Hydrophilic-Lipophilic-Balance (HLB) cartridge (60 mg) followed by a mixed mode, HLB-cation exchange (MCX) cartridge (60 mg) (Waters Inc., Milford, MA) Additional details on this method are provided elsewhere (Kolpin and others, 2002, Meyer and others, 2000).
Method 2 targets eight antibiotic compounds (Table 3 ) in filtered water samples. Complete details of this method have been described previously (Lindsey and others, 2001). The antibiotics were extracted and analyzed using SPE and SIM LC/MS-ESI(+).
Method 3 targets 21 human prescription and nonprescription drugs and their select metabolites (Table 4) in filtered water samples. Compounds were extracted from water samples using SPE cartridges that contain 0.5 g of HLB. Compounds were separated and measured by high-performance liquid chromatography. Additional details on this method are provided elsewhere (Kolpin and others, 2002).
Method 4 targets a broad suite of 46 OWCs (Table 5 ) in unfiltered water samples. The whole-water samples were extracted using continuous liquid-liquid extraction (CLLE) with methylene chloride at pH 2 and analyzed by capillary gas chromatography/mass spectrometry (GC/MS). Available standards for the 4-nonylphenol compounds were composed on multiple isomers, and thus, laboratory standards for these compounds as well as octylphenol ethoxylates were prepared from technical mixtures. Additional details on this method are provided elsewhere (Barber and others, 2000; Brown and others, 1999).
Method 5 (Barber and others, 2000) targets 14 steroid and hormone compounds (Table 6 ) in unfiltered water samples. The CLLE methylene chloride extracts from Method 4 were derivatized and reanalyzed. The analysis of the steroid compounds is enhanced by derivatization to deactivate the hydroxyl and keto functional groups. The technique used in this method is the formation of the trimethylsilyl ethers of the hydroxyl groups and the oximes of the keto groups. After derivatization, the samples were analyzed by GC/MS.
At least one fortified laboratory spike and at least one laboratory blank was analyzed with each set of 10 to 16 environmental samples. Most methods had surrogate compounds added to samples prior to extraction to monitor method performance. Recoveries for method compounds spiked into reagent water and surrogate compounds in environmental samples (Kolpin and others, 2002) indicates the general proficiency of the methods. A summary of the laboratory reagent spike recoveries and laboratory reagent blanks for each compound determined from the beginning of the study through 2000 is included in this report (Table 7 ). The average mean recovery for the laboratory reagent spikes was 78.5 percent. Twenty- nine compounds were detected at least once in the laboratory reagent blanks. Only 5 compounds were detected in more than 10 percent of the laboratory reagent blanks. In general, environmental concentrations within twice the values observed in the associated laboratory reagent blanks were reported as less than the reporting level (RL).
A field quality-assurance protocol was used to determine the effect, if any, of field equipment and procedures on the concentrations of OWCs in water samples. Field blanks, made from laboratory-grade organic free water, were submitted for about 5% of the sites and analyzed for all of the 95 OWCs. Field blanks were subject to the same sample processing, handling, and equipment as the stream samples. To date, one field blank had a detection of coprostanol and testosterone, one field blank had a detection of naphthalene and tri(dichlorisopropyl)phosphate and one field blank had a detection of ethanol, 2-butoxy- phosphate; naphthalene; 4-nonylphenol; phenol; and 4-tert-octylphenol monoethoxylate. Most of these detected concentrations were near their respective reporting levels (RLs) verifying the general effectiveness of the sampling protocols used for this study. In addition, all field blanks had low level concentrations of cholesterol in Method 5 analyses (median concentration = 0.09 mg/L) documenting its ubiquitous nature in the environment. Cholesterol concentrations from 0.005 to 0.18 mg/L measured by Method 5 were set to less than the RL.
Compounds that were measured by more than one analytical method also were used to evaluate the results for this study. The presence or absence of these compounds were confirmed in 100% of the determinations for sulfamerazine and sulfathiazole; 98.8% for oxytetracycline, sulfadimethoxine, sulfamethazine, and tetracycline; 98.6% for cholesterol and coprostanol; 97.6% for chlortetracycline; 95.7% for 17b-estradiol; 94.4% for cotinine; 94.0% for trimethoprim; 89.1% for sulfamethoxazole; 86.4% for codeine; and 83.3% for caffeine. The comparisons for codeine, caffeine, and cotinine may have been affected by the differing extractions (SPE versus CLLE) as well as differing types of sample (filtered versus whole water).
Stream-sample collection began in 1999 and was completed in 2000. Results from 139 stream samples are given in Tables 2-6. Table 2 (Method 1) and Table 6 (Method 5) contain sites where samples were collected but analytical results are not yet available. One or more OWCs were found in 80% of the stream samples, with 82 of the 95 compounds detected during the study. The high overall frequency of detection is likely influenced by the design of this study, which selected stream sites that were generally considered susceptible to OWC contamination. The most frequently detected compounds (Figure 2) represent a wide variety of uses and origins. Mixtures of OWCs were prevalent during this study, with 75 percent of the streams sampled having more than one, 54 percent having more than five, 34 percent having more than 10, and 13 percent having more than 20 OWCs identified. A more complete discussion of the results of this study is provided elsewhere (Kolpin and others, 2002).
This data-collection effort could not be accomplished without the assistance of many dedicated people. The authors would like to acknowledge the USGS scientists and field technicians in 30 states who assisted in identifying candidate stream sites across the United States and in collecting and processing stream samples. This project was supported by the U.S. Geological Survey, Toxic Substances Hydrology Program.