By Chang R. Lee, Ph.D., Philip Terrazas and Prosy Abarquez-Delacruz

Summary: This article expands on an article written by Larry Eils, of the National Automatic Merchandising Association (NAMA), on surveys in California in 1998 and 1999 on water quality from vending machines (“Good News is No News,” WC&P, October 1999). It’s the first public comment on a report from the State of California exonerating, to a large degree, water vending machine operators of charges in a Los Angeles County report that decried water quality delivered by their machines. But that exculpation isn’t without criticism or suggestions to improve the situation.

Water vending machines (WVM) in California are required to be licensed by the California Department of Health Services’ Food and Drug Branch (FDB) or, in an authorized county, to be “permitted” by a local health department. FDB issues a license if the applicant can show that:

  • The machine’s source water is from a state-approved and regulated source;
  • The machine is equipped with a disinfection unit to reduce the numbers of bacteria and other microorganisms, and a self-closing door to protect against contamination if the unit is located outdoors; and
  • The machine complies with the construction and performance standards established by the state with the verification of National Automatic Merchandising Association (NAMA), an independent third-party water vending authority recognized by FDB.

In 1998, a local agency in California tested 279 WVMs—9 percent of the 3,073 under its jurisdiction—for bacteria (heterotrophic plate counts [HPC], coliforms), chemicals (trihalomethanes [THMs], lead), and physical characteristics (color, odor). The agency’s report1 released in September 1998 caused considerable public alarm particularly with the microbiological quality of vended water. In the report, the agency computed an average 8 colony forming units per milliliter (cfu/ml) for 52 public drinking water samples taken from residences, compared this result with the computed 1,306 cfu/ml average for the 279 vended water samples, and concluded that the vended water is 163 times higher in HPC than public drinking water. The report also concluded that the “…disinfection process was ineffective in 26 percent of the machines tested because these units contained bacteria counts [i.e., HPC] above the level considered by the state to be disinfected [i.e., 500 cfu/ml].”

In response, FDB conducted a statewide random sampling and inspection of 794 WVMs—9 percent of the statewide total of 9,066—between Nov. 2, 1998 and Jan. 8, 1999 to determine: 1) The safety and quality of vended water, and 2) the need for additional WVM regulatory requirements. The samples were tested for heterotrophic bacteria, a broad group of bacteria, and coliform bacteria—a group of organisms used to indicate whether the water may have been exposed to pathogens. The inspection determined whether the units had self-closing doors and complied with various labeling requirements.

Materials and methods
Of the 794 WVMs selected, 731 units representing about 8 percent of the statewide total were chosen using a random table of numbers from FDB’s licensing database and the remaining 63 were selected from units reported by the study of a local agency having an HPC greater than 500 cfu/ml.1 Three 100-ml samples were collected from each machine: a first draw sample (from the first gallon of water dispensed), a second draw sample (from the third gallon of water dispensed) and a source water sample. All the sample machines used public drinking water (supplied by a municipality). A control sample of sterile water was opened at each site while taking water samples as a measure for assuring that any bacteria in the environment wasn’t accidentally introduced during the sampling procedure. All samples were obtained using containers and procedures designed to prevent accidental contamination, and they were tested for HPC and coliforms. The 794 WVMs were inspected for compliance with labeling and the condition of self-closing doors. If any were labeled as dispensing “purified water,” their samples were also tested for total dissolved solids (TDS).

Public drinking water systems and WVMs employ different disinfection methods to keep heterotrophic bacteria levels low. Public drinking water relies on maintaining a chlorine or other disinfectant residual in the water. The processes used by vending machines to reduce the levels of harmful THMs in public drinking water also remove the disinfectant chlorine residual. To ensure disinfection continues, they utilize built in ultraviolet (UV) lights to disinfect water before it travels through three-to-seven feet of plastic tubing, a connector, and a check valve to the dispensing spout. These materials are normally left in place during routine maintenance.

Heterotrophic bacteria
HPC, formerly known as standard plate counts, is seldom used as an indicator for the sanitary quality of water because of its lack of specificity. Further, its measurement is difficult to accurately reproduce and is susceptible to misinterpretation because: 1) the measurements are very dependent on incubation time and temperature, and 2) heterotrophic bacteria can be easily introduced accidentally from the environment or their numbers can be inadvertently increased during sampling, transportation and testing if proper precautions aren’t taken. There’s also no federal or state (California) regulatory HPC limit for vended water, bottled water or public drinking water. The California Department of Agriculture permits an HPC of 15,000 cfu/ml in pasteurized milk. An HPC of greater than 10,000 cfu/ml in bottled water would likely trigger an inspection by the U.S. Food and Drug Administration (FDA) of the bottler for possible unsanitary conditions or inadequate treatment of contaminated source water; however, there’s no federal regulation for vended water. In 1973 and again in 1993, FDA considered and rejected a proposal to set an HPC standard of 500 cfu/ml for bottled water and concluded that the total coliform limit for bottled water is adequately protective of public health.2

Sampling results
As shown in Table 1, of first draw samples, 79.1 percent had an HPC of less than 500 cfu/ml, 20.4 percent had an HPC of 500-to-10,000 cfu/ml and the remaining 0.5 percent had HPCs of between 10,800 cfu/ml and 50,700 cfu/ml. For second draw samples, 96.2 percent had an HPC of less than 500 cfu/ml, 3.7 percent had an HPC of 500-to-10,000 cfu/ml and the remaining 0.1 percent had an HPC of 10,500 cfu/ml (corresponding to the first draw sample result of 13,300 cfu/ml). None of the WVMs with HPC levels of greater than 10,000 cfu/ml were positive for total coliforms (see coliforms, fecal coliforms and E. coli discussion in Part 2, WC&P, March 2000). For source water samples, 96 percent had an HPC of less than 500 cfu/ml, 4 percent had an HPC of 500-to-10,000 cfu/ml and none had an HPC of greater than 10,000 cfu/ml. FDB sent regulatory letters to the four operators of the units that had HPC results of greater than 10,000 cfu/ml in the first draw and second draw samples requesting the operators to take corrective actions on the machines. Three of the operators agreed to take the units out of service until new disinfection units and tubing could be installed. The fourth was cleaned and sanitized in the presence of FDB staff.

The results presented in Table 1 show, while the HPCs of second draw samples approach those of source water, the HPCs of first draw samples are higher, suggesting heterotrophic bacteria are re-growing after disinfection. Additional sampling and testing was done to learn more about this possibility and determine what actions if any could be taken to reduce HPCs. Subsequent testing confirmed the initial hypothesis that re-growth of heterotrophic bacteria after disinfection in WVMs can occur rapidly (within hours) depending on ambient temperature. Re-growth occurred mostly in the plastic tubing, suggesting: 1) this tubing needs to be regularly disinfected or replaced; or 2) the design needs to be changed by relocating the UV unit closer to the spout. The largest WVM operator has told FDB that they have begun a study to find ways to disinfect the plastic tubing during routine maintenance. FDB is working with the industry to determine if HPCs can be lowered by regular disinfection, replacement of the tubing or by a design change.

Skewed readings
The statewide testing result for first draw vended water samples was in close agreement with that reported by the local agency1 when the agency’s raw data were re-computed for a fairer evaluation. The average of 1,306 cfu/ml computed for first draw samples appears to be unfairly skewed upward because of three test results that need clarification. The agency’s report does not include raw HPC test results but the agency provided these results to FDB. The raw test results show that the agency computed 1,306 cfu/ml average by including three results of greater than 57,000 cfu/ml each. These three very high results are most likely a reporting error because HPC levels of this magnitude are approximately 10 times greater than the levels considered by most laboratories to be too numerous to count (greater than 5,700 cfu/ml in the agency’s case) and so could only be measured using serial dilutions to obtain actual numbers, but three serial dilution procedures wouldn’t all produce the same exact result. The agency didn’t report performing any serial dilutions. When the three 57,000 cfu/ml findings are excluded, the agency’s average HPC result for vended water drops from 1,306 cfu/ml to 680 cfu/ml, in close agreement with the average HPC statewide testing result for first draw vended water samples of 637 cfu/ml. (The agency took the first draw samples only for testing.) The relatively high HPCs in some first draw samples were apparently due more to the re-growth of bacteria after the disinfection by UV light—which substantially reduces the number of bacteria, but doesn’t destroy all—than to the failure of disinfection because the HPCs of second draw samples approach the HPCs of source water.

Other inconsistencies
The local agency computed an average 8 cfu/ml HPC result for 52 public drinking water samples taken from residences. The agency compared this result with its 1,306 cfu/ml average for vended water samples and concluded that vended water is 163 times higher in HPC than public drinking water, suggesting there’s an alarming disparity between water dispensed by WVMs and by public drinking water sources.  As noted above, these machines use public drinking water as their source water and so HPC results from vended water samples must be compared with HPC results from samples of source water directly supplying the WVMs. But the agency didn’t sample any source water at WVMs. Any valid attempt to use average testing results from public drinking water samples taken from residences in a comparison with average testing results from vended water samples must be based on random sampling and must be sufficiently large in number as to be properly representative. However, the agency didn’t report using recognized random sampling methods; and 52 samples of public drinking water collected at selected residences can not reasonably be considered to be representative of the public drinking water available at the faucets of Los Angeles County’s millions of homes. In any event, wide variations in normal HPCs in public drinking water and vended water mean average HPC results have little significance for these types of comparisons. Published literature shows HPC levels found in public drinking water may vary 100 to 10,000 times depending on the chlorine level and distance from the public drinking water treatment plant to the service point.3

As indicated above, simply comparing heterotrophic bacteria levels in water dispensed from vending machines from a 1999 California statewide survey shows significant inconsistencies in the manner in which results of the earlier local Southern California study of water vending machines were evaluated. Part 2 of this article will discuss a comparison of results for coliform bacteria and TDS testing, as well as offer suggestions for improving vended water quality.


  1. Fiksdal, C., and W.W. Shindy, “Report on 1997-98 Water Vending Machine Pilot Study,” Los Angeles County Agricultural Commissioner/Weights and Measures Department, Environmental Toxicology Bureau, Aug. 28, 1998
  2. Food and Drug Administration, Federal Register, Volume 58 (No. 192), Page 52046-52047, October 6, 1993.
  3. Reasoner, D.J., Drinking Water Research Division of U.S. Environmental Protection Agency, Cincinnati, Ohio, “Monitoring Heterotrophic Bacteria in Potable Water” in Drinking Water Microbiology, Progress and Recent Developments, Gordon A. McFeters, Editor, Springer-Verlag, New York, 1990.

About the authors
Dr. Chang R. Lee is a scientist for the California Department of Health Services’ Food and Drug Branch. He coordinates the state bottled water and vended water program, and is the branch point person for issues related to recycled water and biosolids. Prior to joining FDB in 1990, he worked for Frito Lay Inc. and Stauffer Chemical Company (now part of Rhone-Poulenc Inc.) for 14 years. He earned his doctorate in food science from Ohio State University and holds five U.S. Patents and five foreign patents. He coordinated the above statewide vended water survey and was responsible for results interpretation and final report preparation. Lee can be reached at (916) 327-8041 or email: [email protected]

Phillip Terrazas, MPH, REHS, is a senior investigator for the California Department of Health Services’ Food and Drug Branch. He has worked for the agency for four years, and conducts investigations at wholesale food manufactures, water bottling operations and other related facilities in the Southern California area. He played a major role in the development, planning, implementation and oversight of the above statewide vended water survey. Terrazas holds a bachelor’s degree in biology from California State Polytechnic University, Pomona, and a master of public health (MPH) degree in environmental health from the University of California-Los Angeles School of Public Health. He can be reached at (213) 580-5745 or email: [email protected]

Prosy Abarquez-Delacruz is a regional administrator for the California Department of Health Services’ Food and Drug Branch. She provides regulatory oversight to processed foods manufacturers and distributors through a public health-law enforcement team of supervisors and investigators in Southern California. She holds both a bachelor’s degree in food technology from the University of the Philippines and a juris-doctorate (JD) degree from Whittier College School of Law in Los Angeles. She has over 20 years of experience in regulatory affairs and played a critical role in the above statewide vended water survey. Abarquez-Delacruz can be reached at (213) 580-5720 or email: [email protected]


Comments are closed.