Setting Priorities for Drinking Water Contaminants (1999)

Historically, chemical and microbiological contaminants have been regulated in very different ways. Rather than regulating each type of microorganisms to a specific concentration, regulators have established a zero tolerance goal for microbiological contaminants and have used indicator organisms, particularly fecal coliforms, to show the possible presence of microbial contamination from human wastes. While this methodology has served well for indicating sewage contamination of surface waters and for controlling such diseases as cholera and typhoid fever, an increasing number of deficiencies with this approach have come to light in recent decades.

One deficiency in the current method used to regulate microbes is that, because of differences in survival and transport, viruses and protozoa can be present and viable in raw waters in which coliform organisms are inactive, so assessments of the safety of raw waters are sometimes too optimistic. A second problem is that some bacteria, many viruses, and many protozoa show greater resistance to many conventional treatment methods than do fecal coliforms, so assessments of the safety of treated water are sometimes too optimistic as well. A third limitation is that an increasing number of such pathogens as Giardia and Legionella are surfacing that can originate from sources other than human fecal material. Thus, the fecal indicator strategy is less relevant for these types of microorganisms.

In the past, the only database on microbiological contaminants has been a national database on waterborne disease outbreaks (discussed in Chapter 1). Until recently, this database was used to support the zero tolerance and fecal indicator regulatory strategy for microbial pathogens. More recently, EPA has begun to set

goals for acceptable risk, researchers have begun to publish on methods of risk assessment for microbiological contaminants, and new techniques have evolved that suggest that the development of an occurrence database for pathogens may become possible.

There are no formal schemes, such as those reviewed in Chapter 2, for chemical contaminants that might be considered for prioritizing microbial contaminants. Nonetheless, the principles of risk assessment based on exposure potential and health impacts are similar to those for chemicals. While it is likely that some adaptation will be required, the committee believes that the time is approaching when the same risk assessment principles will be applied to the management of microbiological contaminants that are applied to chemical contaminants.

The identification of microbial hazards associated with drinking water has been accomplished in the same manner since the first documented occurrence of a waterborne disease outbreak: a cholera outbreak that was associated with contamination of the Broad Street pump in London, England, in 1855. The cause of this outbreak (contaminated drinking water) was determined through an epidemiological study. Since then, epidemiology has been the major science used to study the transmission of infectious disease through drinking water.

Epidemiology is the study of occurrence and causes of diseases in populations. The field has focused on exposure to various known or suspected toxic agents and the relationship to health outcome, using statistical methods to indicate a significant association between exposure and health. To some extent epidemiologists have attempted to describe the influence of environmental factors. The field has also been described as applying a knowledge of prevention and control to health problems; in that aspect, it is tied to risk management. Epidemiological studies are sometimes referred to as risk assessments, because there is an attempt to examine both exposure (or what is often referred to as risk factors) and health outcome in humans. However, exposure in these studies is rarely specific or quantitative for microbial contaminants, and in many cases the health hazard is defined by symptomology, rather than by the specific hazard, because a more extensive investigation is required to undertake clinical or antibody tests. Regardless, epidemiological studies can help control the occurrence of future waterborne disease outbreaks, as in the Broad Street pump study.

As discussed in Chapter 1, waterborne disease outbreaks have been investigated and reported in a national database since 1920. Thus, initial efforts to control microbial pathogens focused on bacteria. At that time, typhoid, caused by a bacterium, was the waterborne illness of most concern.

Virus outbreaks began emerging in 1950 with hepatitis A being the primary

concern. From 1966 to 1970, 19 outbreaks of hepatitis A occurred. Many believe that most waterborne outbreaks caused by unidentified agents are because of viruses. Several hundred enteric viruses are possibly important agents of waterborne disease. However, information regarding the incidence of viral infections and the role of contaminated water in acquiring these is limited. Bennett et al. (1987) have reported 20 million cases of enteric viral infections and 2,010 deaths per year. Adenoviruses, which may be transmitted by the respiratory route as well, account for 10 million cases and 1,000 deaths per year, making this the most significant virus affecting U.S. populations. Rotavirus cases have been documented as the second most common viral infection and are particularly of concern for infants (MMWR, 1991).

For all virus outbreaks reported in the United States, drinking water is only one mode of transmission. Thus, epidemiological studies are needed to identify the importance of drinking water exposure. Endemic waterborne diseases of viral origin may also be important, but there is little information on the background occurrence of these diseases. Significant improvements in disease detection at endemic or low levels must occur before it is possible to assess the importance of drinking water to their occurrence.

Virus contamination of ground water is of great concern because of the resistant nature of the viral structure, which interferes with disinfection, and the colloidal size (20 nm) of viruses, which makes them easily transported through soil systems. Viruses can survive for months in ground waters (Yates and Yates, 1988; Gerba and Rose, 1989). National studies in the United States have found viruses in 20 percent to 30 percent of the ground waters. In these studies, coliforms were not predictive of viral contamination (LeChevallier, 1996). New detection techniques using the polymerase chain reaction have demonstrated that viral contamination of ground water is much more common than previously recognized (see Table 3-1).

Diarrhea has been one of the risks associated with many of the enteric viruses, such as the Norwalk virus; more serious chronic diseases have now been associated with viral infections, and these risks need to be better defined. Studies have reported, for example, that Coxsackie B virus is associated with myocarditis—the inflammation of cardiac muscular tissue (Klingel et al., 1992). This could be extremely significant given that 41 percent of all deaths in the elderly are associated with diseases of the heart. In a recent study of 43 cardiac patients, enteroviral RNA was detected in endomyocardial biopsies in 32 percent of the patients with dilated cardiomyopathy and in 33 percent of patients with clinical myocarditis (Kiode et al., 1992). In addition, there is emerging evidence that Coxsackie B virus is also associated with insulin-dependent diabetes (IDD), and this infection may contribute to a detectable increase in the number of IDD cases (Wagenknecht et al., 1991).

Protozoan diseases, specifically giardiasis, emerged as a concern in 1966. Prior to 1966, there had been only five outbreaks of amoebiasis. By the 1976-

1980 period, Giardia was the most identifiable cause of waterborne outbreaks in the U.S. Interestingly, outbreaks of Legionella (and several other pathogens) are not included in the waterborne disease outbreak database, although about 10,000 to 15,000 cases of Legionnaires' disease occur in the United States annually. As many as 30 percent of the respiratory diseases caused by this bacterium may be associated with tap water.

During the investigation of drinking water outbreaks, the source of the water (ground water, spring water, river water) is generally identified, along with treatment deficiency (e.g., no disinfection). More than 100 million individuals rely on ground water as a source of potable water. Only half of the community systems using ground water disinfect the water prior to distribution, while few of the noncommunity systems provide disinfection. Although ground water historically has been assumed to be safe for consumption without treatment, more than half (58 percent in 1971-94) of the reported waterborne disease outbreaks in the United States have been associated with the consumption of ground water (Craun and Calderon, 1997).

Selection of microbial contaminants for development of regulations has been based on reported waterborne disease outbreaks. Formal risk assessment methods utilizing occurrence databases and exposure assessment were not used until the 1980s. Haas (1983) was the first to look quantitatively at microbial risks associated with drinking water based on dose-response modeling. He examined mathematical models that could best estimate the probability of infection from the existing databases associated with human exposure experiments. Rose et al. (1991) then used an exponential model to evaluate daily and annual risks of Giardia infections from exposure to contaminated water after various levels of reduction through treatment. This particular study used survey data for assessing the needed treatment for polluted and pristine waters based on Giardia cyst

occurrence. This approach was used in the development of the Surface Water Treatment Rule (SWTR) to address in particular the performance-based standards required for the control of Giardia; the SWTR requires a safety goal of achieving 99.9 percent reductions of Giardia cysts through filtration and disinfection in all surface water systems. Regulators believed this level of pathogen removal would correspond to an annual risk of no more than 1 infection per 10,000 people exposed over a year from drinking water (EPA, 1989).

Because occurrence databases were not available for enteric viruses, EPA was not able to use its goal of 1 microbial infection in 10,000 exposed persons each year to specify a treatment requirement. Instead, the SWTR mandated a treatment goal of 99.9 percent removal of viruses. This goal was derived from published information on the virus removal performance that well-operated systems with filtration and disinfection can be expected to achieve (EPA, 1987).

Since that time, more work on the occurrence of enteric viruses has been conducted. For example, it has been shown that the beta-Poisson distribution best describes the probability of infection from enteric viruses. This model has been used to estimate the risk of infection, clinical disease, and mortality for hypothetical levels of viruses in drinking water (Haas et al., 1993). Meanwhile, the development and availability of new detection techniques for viruses have allowed the creation of a meaningful occurrence database that can be used in these types of risk estimates (see Table 3-1).

Although addressed in the SWTR, there are no performance-based standards (e.g., reduction requirements) for Legionella, and no risk assessment was undertaken. No occurrence databases or exposure assessments exist for this bacterium.

The SWTR required that disinfection be increased to control these microorganisms, but as long as coliform standards were met, there was no way to monitor the enforcement of this rule other than requiring utilities to submit disinfectant levels and contact times. With the development of regulations to limit the levels of disinfectants and disinfectant byproducts (D/DBP), EPA recognized the possibility that efforts to reduce DBP levels could increase health risks from microbial agents. Using the Disinfection Byproducts Regulatory Analysis Model, EPA was able to examine the health and economic implications of various approaches to DBP regulation. In a direct comparison of microbial risk from Giardia infection to cancer risk for several DBP control scenarios, the predicted increases in Giardia infection were orders of magnitude higher than decreases in cancer rates. To ensure that implementation of the D/DBP rule did not increase microbial risk, the regulatory negotiating committee convened by EPA considered it necessary to review the adequacy of the existing SWTR. This revised rule, which includes regulation of Giardia and Cryptosporidium, is the Interim Enhanced Surface Water Treatment Rule (IESWTR) and is scheduled to be finalized in November 1998.

The Information Collection Rule (ICR) is the first national program to develop occurrence data in surface waters for pathogens. Since July 1997, all utilities serving more than 100,000 people have been required to collect samples from their treatment plant influents and analyze for Cryptosporidium, Giardia, and enteric viruses (as well as chemical disinfection byproducts). The monitoring is scheduled to last a total of 18 months and will end in December 1998.

On February 12, 1997, EPA established the Microbial and Disinfectants/Disinfection Byproducts Advisory Committee under the Federal Advisory Committee Act (FACA) to evaluate new information and data, as well as to build consensus on the regulatory implications of new information on DBPs and pathogens that was becoming available. The advisory committee's recommendations to EPA on the proposed changes to the D/DBP rule and the IESWTR were set forth in an Agreement in Principle document dated July 15, 1997 (EPA, 1997). Because of regulatory deadlines set by Congress under the 1996 amendments to the SDWA, however, it was not possible for EPA or the FACA committee to wait for the ICR data to be collected and analyzed before changes in the IESWTR were negotiated. Proposed changes in the IESWTR were significant and include the following:

• more stringent turbidity removal requirements to control pathogens;
• establishment of a microbial benchmarking/profiling concept;
• restoration of pre-disinfection credit;
• setting of the Cryptosporidium MCLG at zero; and
• institution of removal requirements and credits for Cryptosporidium.

The results of pathogen monitoring under the ICR will be available for the next round of negotiations, scheduled to begin with a stakeholder meeting in December 1998. Negotiations for stage two of the D/DBP rule will extend over 1999 and could lead to modifications to the final SWTR, which is not expected to be completed until after the year 2000.

Analytical methods remain a critical issue for assessment of exposure to microbiological contaminants. There has been limited development and standardization of processes, however, for laboratory approval and appropriate application of both established methods (e.g., microscopy) and newer methods (e.g., immunomagnetic capture and molecular techniques). A brief summary of the historical development and contemporary use of detection and analysis methods for waterborne pathogens is included in Chapter 5. The interpretation of analytical results has also been largely neglected. Most available detection methods may be able to address some aspect of microbial occurrence (i.e., identification, quantification, viability, virulence, source, transport), but no single analytical method can be used to address all the needs of the exposure assessment (see Table 3-2).

In order to examine microorganisms in drinking water and develop a prioritization scheme, data on both exposure and risk to human health are needed. Additive or multiplicative approaches could be used. However, many causes of waterborne disease are unknown; thus, the disease potential for microorganisms occurring in water needs to be examined carefully (see Table 3-3).

Outbreak investigations remain a significant component of the health effects assessment. This is the result of the extreme costs associated with outbreaks, not

only in medical care and days lost from work but in costs accrued in assessment of the outbreak, recall of food products, boil orders, communication efforts, remediation, and future safety efforts. The waterborne disease outbreak in Milwaukee in 1993 cost the community an estimated $25 billion, not including subsequent costs of aversion behavior because of loss of confidence in the water supply (e.g., purchase of bottled water and point-of-use devices to further treat the water). As noted in Chapter 1, investigation and reporting of waterborne disease outbreaks is not mandatory, the quality and completeness vary from state to state, and only a small proportion of the risks are identified.

Treatability of microorganisms by water processes will remain a significant part of exposure assessment. While water treatment, such as chlorination, may readily control some microbial risks, such as Shigella or Campylobacter , the reliability of treatment and the potential for growth of microbial pathogens in the water distribution system must be included in any risk assessment. Given the high

risk of violations of the coliform standard (primarily in small public water systems), if disinfection failure continues to occur in a large percentage of facilities using highly polluted water supplies, the risk could become significant. It is critical that occurrence databases are developed for microorganisms that may exhibit a high level of virulence in water in order to determine the potential effects of treatment failures. Also at issue is the question of how much treatment and how much risk reduction is appropriate and acceptable. Because zero tolerance has been maintained as the goal for microbial contaminants for so long, the idea of a non-zero maximum contaminant level has not been debated nor formalized for most microbial contaminants.

Historically, microbial contaminants in drinking water have not been individually prioritized for regulation. Rather, microbial contaminants have been controlled by specifying treatment methods for various types of source water and by monitoring for fecal coliform bacteria, which indicate possible presence of contamination but are not in themselves pathogenic. This system for regulating microbial contaminants has been relatively effective, but emerging new pathogens have raised concerns about whether the system is sufficient (Craun et al., 1997). Emerging waterborne pathogens of concern include protozoans (primarily Giardia and Cryptosporidium), Legionella, and several viruses (including various enteric viruses and adenoviruses). Limitations in data on health effects of these organisms and levels of human exposure make it difficult to establish specific priorities for their future regulation. While a tremendous amount of resources have been devoted to the control of chemical hazards in the environment (including drinking water contaminants), related expertise for controlling microbial contaminants is far less developed, even though the majority of reported waterborne illness outbreaks are known or thought to be caused by microorganisms.

Bennett, J. V., S. D. Homberg, and M. F. Rogers. 1987. Infectious and parasitic diseases. Pp. 102-114 in Closing the Gap: The Burden of Unnecessary Illness , R. W. Amber and H. B. Dull, eds. New York: Oxford University Press.

Craun, G. F., and R. Calderon. 1997. Microbial risks in groundwater systems epidemiology of waterborne outbreaks. Pp. 9-20 in Under the Microscope: Examining Microbes in Groundwater. Denver, Colo.: American Water Works Association.

Craun, G. F., P. S. Berger, and R. L. Calderon. 1997. Coliform bacteria and waterborne disease outbreaks. Journal of the American Water Works Association 89(3):96-104.

EPA (Environmental Protection Agency). 1987. National Primary Drinking Water Regulation; Filtration and Disinfection; Turbidity, Giardia Lamblia, Viruses Legionella, and Heterotrophic Bacteria; Proposed Rule. Federal Register 52(212):42194.

EPA. 1989. National Primary Drinking Water Regulations; Filtration and Disinfection; Turbidity; Giardia lamblia, Viruses, Legionella, and Heterotrophic Bacteria. Federal Register 54(124):27486-27541.

EPA. 1997. National Primary Drinking Water Regulations: Interim Enhanced Surface Water Treatment Rule Notice of Data Availability; Proposed Rule. Federal Register 62(212):59485-59557.

Gerba, C. P., and J. B. Rose. 1989. Viruses in source and drinking water . Ch. 19 in Advances in Drinking Water Microbiology Research, G. A. McFeters, ed. Madison, Wis: Science Technology.

Haas, C. N. 1983. Estimation of risk due to low doses of microorganisms: A comparison of alternate methodologies. American Journal of Epidemiology 118:573-582.

Haas, C. N, J. B. Rose, C. Gerba, and S. Regli. 1993. Risk assessment of virus in drinking water. Risk Analysis 13:545-552.

Haas, C. N., J. B. Rose, and C. P. Gerba, eds. 1998. Quantitative Microbial Risk Assessment. New York: John Wiley and Sons.

Kiode, H., Y. Kitaura, H. Deguchi, A. Ukimura, K. Kawamura, and K. Hirai. 1992. Genomic detection of enteroviruses in the myocardium studies on animal hearts with Coxsackievirus B3 myocarditis and endomyocardial biopsies from patients with myocarditis and dilated cardiomyopathy. Japanese Circulation Journal 56:1081-1093.

Klingel, K., C. Hohenadl, A. Canu, M. Albrecht, M. Seemann, G. Mall and R. Kandolf. 1992. Ongoing enterovirus-induced myocarditis is associated with persistent heart muscle infection: Quantitative analysis of virus replication, tissue damage and inflammation . Proceedings of the National Academy of Sciences 89:314-318.

LeChevallier, M. L. 1996. What do studies of public water system groundwater sources tell us? Under the Microscope: Examining Microbes in Groundwater, Sept. 5-6. Lincoln, Nebr.: The Groundwater Foundation.

MMWR (Morbidity and Mortality Weekly Report). 1991. Rotavirus surveillance--United States, 1989-1990. Morbidity and Mortality Weekly Report 40(5):80-81, 87.

Rose, J., C. N. Haas, and S. Regli. 1991. Risk assessment and control of waterborne giardiasis. American Journal Public Health 1:709-713.

Wagenknecht, L. E., J. M. Roseman, and W. H. Herman. 1991. Increased incidence of insulin-dependent diabetes mellitus following an epidemic of Coxsackievirus B5. American Journal of Epidemiology 133:1024-1031.

Yates, M. V., and S. R. Yates. 1988. Modeling microbial fate in the subsurface environment. Critical Reviews in Environmental Control 17:307-343.