George M. Breuer, Ph.D., Lee A. Friell, M.S., Nelson P. Moyer, Ph.D., and Gene W. Ronald, M.S.

Executive Summary
A study was made of bottled water sold for drinking purposes in the State of Iowa. Analytes determined in each commercially manufactured water included bacteria, inorganic chemicals, and organic chemicals now regulated under the Safe Drinking Water Act and selected other analyses proposed for regulation or frequently observed. Thirty-nine samples were purchased from supermarkets in Iowa City, Des Moines, and other cities within the State of Iowa in January and February of 1990. No samples exceeded Maximum Contaminant Levels of compounds regulated under the Safe Drinking Water Act, although eleven samples had low levels of trihalomethanes, eighteen had low levels of nitrate, three had approximately 0.001 mg/L of toluene, another had 0.02 mg/L of arsenic, and a few had detectable levels of barium or other SDWA analytes. Two samples exceeded one or more secondary (aesthetic) guidelines for drinking water. Of most concern, perhaps, were significant levels of heterotrophic bacteria in a few of the samples. While not regulated under SDWA, a heterotrophic plate count above approximately 100-500 CFU/mL may be indicative of a problem with the bacterial cleanliness of the sample, although interpretation is clouded by shelf storage and lack of source information. Because many people depend on bottled water as an emergency substitute or replacement for public or private drinking water sources, the Laboratory recommends similar analytical monitoring for these waters as is done under SDWA and also recommends routine quality assurance of the product including heterotrophic plate count.

Background
Agricultural practices relying upon intensive chemical applications to increase yields have become common in Iowa. The environmental impact of these practices upon water resources has been debated by the press since the one-time-testing results of samples from Iowa public supplies were made public(1). Most of Iowa's surface water supplies occasionally have trace amounts of farm chemicals(2) and nitrate, and recent evidence points to groundwater contamination as well(3,4). Other contaminants of industrial or domestic origin which have been detected in public water include trichloroethylene and elevated trihalomethane levels(1,5).

The citizenry of Iowa is concerned about the safety of water delivered to their homes by public suppliers. Their mistrust is manifest in the resurgence of interest in point-of-use water treatment devices and the widespread practice of purchasing bottled water for drinking, especially among residents of communities with water supplies using surface water as a source.

Bottled water subject to interstate commerce is regulated by the Food and Drug Administration in 21 CFR. Part 103 of 21 CFR, "Quality Standards for Foods with No Identity Standards," establishes analytical parameters and acceptable limits for interstate distribution of bottled drinking water. These limits are essentially the same as those required for public water supplies under the Safe Drinking Water Act (SDWA), although numerical standards are listed and not referenced to the SDWA standards; e.g., a shortened list of the volatile organic chemicals (VOC) under SDWA is only now being proposed for bottled water(6). A few additional standards (color, odor and turbidity) are also included, however. Part 129, "Processing and Bottling of Bottled Drinking Water," covers plant construction and design, operations, equipment and production activities. Part 110 addresses good manufacturing practices in the manufacture, processing, packing and holding of human food including bottled water.

Water bottled in Iowa for intrastate distribution is subject to the Iowa Department of Agriculture rules contained in the Iowa Code, 159.5, which requires the Department to establish a program for inspection and regulation of water sold in sealed containers for human consumption. The Department of Agriculture has drafted rules but has not as yet promulgated these rules pending receipt of an appropriation to implement them. These proposed rules require Iowa bottlers to comply with FDA regulations contained in 21 CFR, Parts 103 and 129. Bacteriological analyses are to be required monthly and organic chemical and inorganic chemical analyses and turbidity determinations are to be required annually; radiochemical analyses are to be required every four years.

Because of the increased reliance of the public upon bottled water, governmental regulatory agencies are seeking to establish controls designed to protect consumers from unscrupulous business practices as well as ensure the safety of products marketed. The Hygienic Laboratory believes such regulations should be based upon scientific data and continued monitoring by the same procedures used for public drinking water supplies. The Safe Drinking Water Act and regulations promulgated under it by the United States Environmental Protection Agency set standards for monitoring these supplies. All analyses must be done by approved methods and must meet criteria established under the Act. Analytical work must also be performed by certified laboratories such as the State Principal Laboratory (the Hygienic Laboratory in Iowa) or a laboratory certified by the State.

To provide a minimum scientific basis for rule making and to survey possible hazards to which Iowans purchasing bottled water may be exposed, the Hygienic Laboratory has evaluated the quality of representative commercial drinking waters bottled or distributed in Iowa. The results of that evaluation are presented in this report.

Previous work in the literature reported results of various surveys of bottled waters. McCurdy and Mellor(7) measured radium levels in 11 domestic bottled waters and 11 foreign waters sold commercially in the northeastern United States. Six samples, five foreign and one domestic, had radium activity statistically different from zero. Radium 226 levels in these positive samples ranged from 1.5 to 13.5 picocuries per liter (pCi/L) and radium 228 levels ranged from 0.6 to 12.8 pCi/L. The positive domestic water was below the U.S. EPA Maximum Contaminant Level (MCL) promulgated under SDWA, while all positive foreign waters would have been above the MCL.

Other workers(8) examined 41 lots of domestic purified water, 28 lots of domestic mineral water and 45 lots of imported mineral water purchased in Canada for aerobic colony count, coliforms, fecal coliforms and Escherichia coli. One lot of each of the three types of water exceeded coliform standards but none contained fecal coliforms or E. Coli. Nineteen (46%) of the domestic purified waters exceeded the Canadian bottled water standard of 100 colony forming units per milliliter (CFU/ml) for plate counts (duplicate one milliliter volumes of appropriate dilutions by the pour plate method on standard plate count agar incubated at 35ºC for 48 hr.); five each of the domestic and imported mineral waters would also have failed that standard if it were applicable to mineral waters. Hunter and Burge(9) in Wales examined 29 carbonated and 29 still (i.e., noncarbonated or noneffervescent) mineral waters and found no coliforms or Aeromonas sp. They did find high total bacterial counts using a pour plate method, however, especially in the still waters. Gonzalez, Gutierrez and Grande(10) in Spain also found a variety of bacterial species in many uncarbonated mineral drinking waters.

Allen, Halley-Henderson and Hass(11) analyzed 37 brands of domestic and imported mineral waters for pH, alkalinity, specific conductance, chloride, fluoride, nitrate, phosphate, sulfate and 23 metals including boron, sodium and potassium. Twenty-four had one or more analytes present at concentrations above United States drinking water standards. These authors and a publication by the Environmental Policy Institute(12) review other studies done by consumer organizations and state agencies. Dr. Arthur Nowak in the College of Dentistry, University of Iowa, has also performed a pilot study of fluoride levels in bottled and processed waters in the Iowa City area(13).

Experimental Design
Sampling was performed by Hygienic Laboratory staff in January and February of 1990. Several supermarkets in Iowa City and Des Moines were surveyed and all bottled waters from different manufacturers and/or brands were purchased in one gallon plastic jugs if available. Hygienic Laboratory personnel traveling on duty, as required for hospital laboratory inspections or for air quality monitoring network maintenance, purchased other samples in supermarkets across the state. Some unflavored specialty waters were purchased for comparison purposes, and it was necessary to purchase smaller containers of these products. Only waters labeled for drinking water purposes or whose labels indicated possible use for drinking were purchased. Plain distilled waters indicated to be only for other uses such as steam irons were not purchased although it should be noted that several of the purchased products were labeled as distilled. Table 1 lists the bottled water manufacturers/distributors, their addresses if given on the label, their brands as purchased for this study and the UHL sample numbers assigned to samples of their products in the course of this study. One was inadvertently sampled in duplicate. Notes are given on any label information indicating treatment. For specialty waters, the volume of the container is given (many are available in several sizes) and the country of origin if not USA.

The analytes for which analyses were performed are listed in Table 2. These are materials for which Maximum Contaminant Levels (MCL) have been set under SDWA or which must be monitored under SDWA even though MCLs have not been promulgated. The MCLs, if established, are listed in the table along with the quantitation limit achieved. The MCLs are listed in simplified form for comparison purposes and are not intended as a rigorous summation of regulations under SDWA. For example, the MCL for total trihalomethanes (listed in analyses as the four components, chloroform, bromodichloro-methane, chlorodibromomethane, and bromoform) is 100 µg/L (micrograms per liter, or ppb)--given in the list--but is in fact a yearly average value, not a single determination. Similarly, the MCL for radiochemical parameters is based on an annual composite of quarterly samples; also, the MCL for gross beta particle activity is actually expressed as a body burden of 4 mrem/yr, but the screening level of 50 pCi/L (picocuries per liter) is a reference level used. The quantitation limits will occasionally vary, depending on the background materials in the sample; perhaps the most obvious case is the radiochemical analyses where the quantitation limits will depend strongly on the volume used (which varies with dissolved solids content), instrument background, and counting time. Vinyl chloride is required to be analyzed only if a two-carbon chlorinated compound is present in the sample and is listed in the table but not in the sample analysis results.

A previous sampling of water vending machines in Iowa City had shown significant levels of heterotrophic bacteria, possibly caused by inadequate maintenance of some dispensers(14). As a consequence heterotrophic plate count and a few other analyses (such as herbicides commonly used agriculturally in Iowa, most of which are not now regulated under SDWA) were added to the list of analyses performed. These are listed in Table 3 along with a notation of existing Secondary Drinking Water Standards which have been established for a few of them. U.S. EPA has also determined(15,16) final lifetime health advisory levels (LHA) for several of the common pesticides analyzed, and these are also given in Table 3; for known or probable human carcinogens (in EPA groups A and B) EPA estimates cancer risk (the 10 ⁵ or 10⁶ risk levels are those usually considered) but does not denote them as a LHA.

A wide variety of analytical methods is required for complete analysis of samples under SDWA regulations. Those selected for this study were from routine method manuals, including the United States Environmental Protection Agency, Standard Methods, and the United States Geological Survey, for the analyses as listed below. These manuals should be consulted for details of methodology. For dissolved solids and turbidity, EPA methods 160.1 and 180.1, respectively, were used. For the anion analyses, chloride, fluoride, sulfate and nitrate, methods used were EPA 325.3, USGS 1432-7/84, EPA 275.4, and EPA 353.3, respectively. The metals aluminum, barium, iron, manganese, and sodium were all analyzed by EPA method 200.7. Arsenic, cadmium, chromium, copper, lead, mercury, selenium, silver and zinc were analyzed by EPA methods 206.2, 213.2, 218.2, 220.2, 239.2, 245.1, 270.2, 272.1, and 289.1 respectively. Volatile organics were analyzed by EPA method 502.2. The radiochemical tests were EPA method 900.0. Standard Methods 9221B multiple tube fermentation method was used for total coliform analysis, while heterotrophic plate count (pour plate method) was performed using 1 mL and 0.1 mL volumes of undiluted sample in standard plate count agar with incubation for 72 hours at 35ºC. Pesticides were analyzed using EPA National Pesticide Survey method 3 for acid herbicides (2,4-D and Silvex), EPA SW-846 method 8141 for organophosphate and carbamate insecticides and triazines and aniline-based herbicides, and EPA SW-846 method 8080 for chlorinated hydrocarbon insecticides and polychlorinated biphenyls.

Results and Discussion
Detected concentrations of contaminants in the samples taken are listed in Table 4 (see Table 1). The first column lists the log number assigned when samples were received at UHL. The second column lists the dissolved solids determined in each sample. The next 5 columns list the metals that were detected in the samples, with sodium listed under its elemental abbreviation "Na," barium under "Ba," iron under "Fe," manganese under "Mn," and arsenic under "As;" other metals listed in Tables 2 and 3 were not detected. Similarly, the next 4 columns list anions (chloride as "Cl", fluoride as "F", nitrate as "NO₃", and sulfate as "SO₄") detected. The following 2 columns list the organics detected, with concentrations of all components of total trihalomethanes (chloroform, bromodichloromethane, chlorodibromomethane, and bromoform) summed and the summed concentration reported under the heading "THM;" toluene ("tol.") was the only other organic detected in any of these samples. The next two columns present the results of the radiochemical screening tests for gross alpha ("Alpha") and cross beta ("Beta'") activity. The final two columns present the heterotrophic plate count ("HPC") and turbidity (nepholometric turbidity units, or NTU) results.

No primary (health-based) drinking water standards were exceeded in any sample. Detection of arsenic in one of the specialty waters was unusual, but the level of 0.02 mg/L (milligrams per liter, or ppm) is below the MCL of 0.05 mg/L. Nitrates were detected in 18 of 39 samples, but the highest concentrations were only about half the MCL of 45 mg/L nitrate as nitrate (10 mg/L nitrate as nitrate-nitrogen). Similarly, 11 of 39 samples had detectable levels of the trihalomethanes, but all were at or below 15 µg/L while the MCL is 100 µg/L as a yearly average. The traces of toluene noted in 3 samples at extremely low levels (1.1 µg/L or less) were unexpected, but may be due to trace contamination by gasoline fumes of these samples during storage or shipment in their plastic bottles. Interestingly, the sample of Perrier water collected in this study did not contain the benzene reported widely in the media in this time frame; benzene is also a component of gasoline, but the source of the Perrier contamination is reportedly the source spring itself(17), not chance contamination in shipping the product to market or storing it. The radiochemical screenings were all within guidelines, and sequential radium (the 226 and 228 isotopes) done on some of the samples with higher gross alpha levels as a check did not indicate levels above the MCL. One sample did show turbidity at 2 NTU. Turbidity is a parameter usually measured on-site at water treatment works because if high it may interfere with disinfection of the water; in this case the value observed is unlikely to be of consequence.

However, some secondary standards-nonenforceable guidelines related to taste, odor or other aesthetic aspects-were exceeded in a few cases. A secondary standard of 500 mg/L is used for total dissolved solids, and two samples at 530 and 960 were over that level. A secondary standard for sulfate of 250 mg/L was exceeded by one sample at 480 mg/L (the sample with 960 mg/L dissolved solids), and this could result in taste and laxative effects. This same sample also exceeded the secondary standard for manganese of 0.05 mg/L by a factor of 8, but manganese is largely a problem of taste and staining of laundry and the latter at least sould be of concern for bottled water.

Interpretation of the heterotrophic plate count (HPC) figures is clouded by the lack of microbiological information about the source water being bottled or the containers being used. Five of 31 bottled water samples in one gallon containers had HPCs greater than 200 CFU/mL. Method of treatment was listed for 4 of 5 of these samples and there was no correlation between source or treatment and the number of bacteria present. Unusually high plate counts (>10⁴ CFU/mL) were obtained from two samples, one treated by distillation and the other treated by reverse osmosis and ozonation. A third sample purified by reverse osmosis with a plate count >10³ CFU/mL had a label recommendation that the water was suitable for contact lens care. Because HPC did not correlate with source or treatment, the bottles themselves and the bottling process may have contributed to subsequent growth of bacteria during storage and transit. Four of eight specialty waters exceeded 200 CFU/mL by HPC and all of these samples were bottled from natural springs. A high plate count may result from non-pathogenic bacteria present in the water initially followed by growth during the significant amount of time elapsing in bottling, storage at the plant, shipping, and storage on the supermarket shelf. If a disinfected water (e.g., a public drinking water supply) was used as a source, it may also be an indicator of unsanitary conditions in processing or bottling the water. The best procedure is probably to use HPC as a quality control check or regulatory check on the source water and immediately after the bottling process itself, using a control value on the order of the European standard of approximately 100 CFU/mL. Additional quality control checks are highly recommended for containers to prevent bacterial growth during storage and transit.

Additional testing may be desirable to completely characterize contaminants in bottled water. For example, this sampling in January and February will probably minimize any detections of the common herbicides (e.g., atrazine or alachlor). Further testing of bottles using surface water sources could be done to determine the impact of spring and summer application of pesticides on the bottled product.

Conclusion
On the basis of the data obtained in this study, the quality of typical bottled waters sold in Iowa appears to be neither much better nor much worse than typical drinking water from public drinking water supplies in the state. However, this study represents only one brief time period in the distribution of bottled waters. Although these data are not remarkably different from Iowa public water supplies, we must caution that bottled waters may not be tested as frequently or for as many contaminants as public water supplies.

Acknowledgments
The authors are grateful to Randy R. Hudachek and Robert D. Jensen for collection of samples at sites throughout the state. Although the personnel involved are too numerous to mention, the work reported here could not have been done without the expertise and assistance of the analytical staff of the University (State) Hygienic Laboratory at both the Iowa City and Des Moines locations, and the authors express their heartfelt thanks to them. We are also grateful to the Laboratory Director, Dr. William J. Hausler, Jr., who requested that this study be undertaken and supported it through to completion.

References

1. K. L. Cherryholmes, G. M. Breuer, and W. J. Hausler, Jr., "One Time Testing of Iowa's Regulated Drinking Water Supplies," University (State) Hygienic Laboratory, Iowa City, March 1989.
2. M. Wnuk, R. Kelley, G. Breuer, and L. Johnson, "Pesticides in Water Supplies Using Surface Water Sources," Iowa Department of Natural Resources, Des Moines, September 1987.
3. M. G. Detroy and R. L. Kuzniar, "Occurrence and Distribution of Nitrate and Herbicides in the Iowa River Alluvial Aquifer, Iowa--May 1984 to November 1985," Water-Resources Investigations Report 88-4117, United States Geological Survey, Iowa City, 1988.
4. M. G. Detroy, P. K. B. Hunt and M. A. Holub, "Ground-Water-Quality-Monitoring Program in Iowa: Nitrate and Pesticides in Shallow Aquifers," Water-Resources Investigations Report 88- 4123, United States Geological Survey, Iowa City, 1988.
5. United States Environmental Protection Agency data on the Des Moines TCE (trichloroethylene) Superfund site.
6. Food and Drug Administration, Proposed Rule, Fed. Reg., 55, 27831-27835, July 6, 1990.
7. D. E. McCurdy and R. A. Mellor, "The Concentration of ²²⁶Ra and ²²⁸ Ra in Domestic and Imported Bottled Waters," Health Physics, 40, 250-253 (1981).
8. D. W. Warburton, P. I. Peterkin, K. F. Weiss and M. A. Johnston, "Microbiological Quality of Bottled Water Sold in Canada," Can. J. Microbiol., 32, 891-893 (1986).
9. P. R. Hunter and S. H. Burge, "The Bacteriological Quality of Bottled Natural Mineral Waters," Epidem. Inf., 99, 439-443 (1987).
10. C. Gonzalez, C. Gutierrez and T. Grande, "Bacterial Flora in Bottled Uncarbonated Mineral Drinking Water," Can. J. Microbiol., 33, 1120-1125 (1987).
11. H. E. Allen, M. A. Halley-Henderson and C. N. Hass, "Chemical Composition of Bottled Mineral Water," Arch. Environ. Health, 44, 102-116 (1989).
12. S. Marquardt, V. Smith, J. Bell and J. Dinne, "Bottled Water: Sparkling Hype at a Premium Price," The Environmental Policy Institute, Washington, D.C., January 1989.
13. A. J. Nowak, "Fluoride Content of Bottled/Processed Waters," Memorandum to Dentists and Physicians of Johnson County, July 17, 1989.
14. N. Hall and N. Moyer, unpublished results, 1988.
15. United States Environmental Protection Agency, "Health Advisories for 16 Pesticides," NTIS PB87-200176, March 1987.
16. United States Environmental Protection Agency, "Health Advisories for 50 Pesticides," NTIS PB88-245931, August 1988.
17. E. S. Browning, A. M. Freedman and T. R. King, "Perrier Expands North American Recall to Rest of Globe," The Wall Street Journal, Thursday, February 15, 1990, p. B1.