United States
       Environmental Protection
       Agency
Office of Water
(WH-550)
EPA814/S-92-001
October 1992
vvEPA USE OF MICROBIAL RISK
       ASSESSMENT IN SETTING
       U.S. DRINKING WATER
       STANDARDS

-------

-------
         Use of Microbial Risk Assessment in Setting U.S. Drinking Water
                                   Standards
                         Bruce A. Macler and Stig Regli*

                     U.S. Environmental Protection Agency
                           San Francisco, California
                                     and
                            Washington, D.C., USA
This paper outlines US EPA's general strategy for using microbial risk assessment to
support the development of US National Primary Drinking Water Regulations
(NPDWRs).   It discusses specifically the use of such risk assessment in the
development of upcoming regulations for disinfection of groundwater (Groundwater
Disinfection (GWD) Rule) and for control of disinfectants and their chemical byproducts
(Disinfectant/ Disinfection Byproduct (D/DBP) Rule), and  possible amendments to the
current Surface Water Treatment Rule (SWTR).  The risk assessment and risk
management processes explicitly consider acceptable risk values for water-borne
microbial pathogens.  These values directly influence the regulatory choice of
treatment levels and methods.

The intention of the US Federal Safe Drinking Water Act is to protect the public from
unacceptable health risks arising from drinking water. The Act directs EPA to establish
drinking water Maximum Contaminant Level Goals (MCLGs) "at the level at which no
known or anticipated adverse effects on the health of persons occur and which allow
an adequate margin of safety". EPA policy requires the use of risk assessment in the
development of its regulations. MCLGs are not legally enforceable, but point EPA
towards health protective regulations. The corresponding NPDWRs are enforceable
and are required to be set as close to the MCLG as is technically and economically
feasible. They consist of either 1) a Maximum Contaminant Level (MCL) or 2) a
treatment technique, if it is not economically or technologically feasible to measure the
level of a contaminant in water. The NPDWRs are a product of risk management and
include not only risk assessment information, but considerations of analytical
capability, monitoring, available treatment technology and costs. They must be health
protective.
EPA Regulatory Development

US public health interests require that drinking water be microbiologically safe.  This
has been taken by EPA to mean not only the prevention of outbreaks of illness, but the
minimization of endemic levels of illness. EPA is in the process of developing GWD
and D/DBP Rules and re-evaluating the SWTR.  The GWD Rule is concerned with
                                     1

-------
potential health hazards from pathogenic viruses and bacteria in groundwater not
under the direct influence of surface water. The goal of the GWD Rule is to protect the
public from numerous types of water-borne viruses (e.g., hepatitis A agent, rotavirus,
Norwalk and Norwalk-Iike agents, coxsackieviruses, echovirus) and bacterial
pathogens (e.g., Salmonella,  Shigella, Campylobacter). The SWTR includes (besides
surface water sources) coverage for groundwater sources under the direct influence of
surface water. EPA distinguishes this category of groundwater as that which is
vulnerable to contamination from protozoa. The goal of the SWTR is to protect the
public from pathogenic viruses, bacteria and Giardia lamblia.

The goals of the D/DBP Rule are to ensure that drinking water remains
microbiologically safe at the limits set for disinfectants and their byproducts and that
the disinfectants and byproducts do not pose ah unacceptable risk at these limits.
EPA's approach in developing this  rule considers the constraints of simultaneously
treating for these different pathogen concerns. Considering conventional water
treatment methods, any increased  chemical disinfection to yield lower microbial risk
requires the use of more or stronger disinfectants and, depending upon the point of
application and types of byproduct  precursors present,  may produce higher levels of
byproducts, which themselves pose potential health risks. Therefore, risk comparison
and risk trade-offs must be considered. EPA has undertaken computer modeling to
estimate the relationship of microbial and chemical risks from water treatment.  This
model  examines the magnitude of these risks for a variety of source water qualities
and water treatment scenarios.

Two constraints have been imposed as starting conditions for the control of
disinfectant and byproduct risks and are considered independently in this analysis:  1)
minimally meeting the SWTR as written for disinfection and maintenance of a
disinfectant residual in the distribution system, and 2) meeting a potential amended
SWTR, termed here an "enhanced SWTR" (ESWTR), which would require higher
levels of treatment for Giardia to specifically ensure that the microbial risk at the first
customer is less than one infection per 10,000 people per year and that a disinfectant
residual is maintained in the distribution system (Gelderloos, et al, 1992).  The SWTR
currently only requires at least a 99.9 % and 99.99% removal and inactivation of
Giardia cysts and viruses respectively prior to the first customer, regardless of
sourcewater quality. The ESWTR would follow the EPA SWTR Guidance, which
recommends proportionally higher levels of treatment for poorer source waters to
achieve the same risk at the  first customer for all systems. Within these constraints of
disinfection, EPA considers alternatives that achieve acceptable risks from
disinfectants and byproducts.

A variety of issues for microbial risk assessment are common to these rules and will be
discussed in this paper.  These include approaches to  microbial risk assessment,
development of occurrence data, consideration of comparing microbial risks with those
from chemical contaminants, and what acceptable microbial risk levels might be.
 Microbfal pathogen risk assessment

 A number of assumptions have been used in microbial and chemical risk assessments
                                      2

-------
 that are not scientific in origin and are essentially risk management decisions.  These
 are included in the establishment of the appropriate regulatory illness endpoints of
 concern and the selection of pathogenic organisms for regulation. Additionally,
 standard, conservative "worst-case" dose-response and exposure assessments are
 not appropriate to describe this situation where treatment to decrease exposure to
 microbial pathogens in drinking water may increase exposure to chemical
 contaminants.

 Illness endpoints of concern

 The possible microbial illnesses, or "endpoints of concern", vary with the organism and
 vary markedly in their severity. In EPA's previous drinking water regulations involving
 pathogenic organisms (i.e., Total Coliform Rule and SWTR), these endpoints have
 been taken together as a broadly generalized "microbial illness" resulting from these
 organisms in total, rather than as separate defined illnesses attributable to specific
 organisms. The intention of these regulations was to minimize all microbial illnesses.
 The most common microbial illness, gastrointestinal illness or diarrhea, is generally
 considered non-life threatening in normally healthy adults. However, the US Centers
 for Disease Control (CDC) have presented data that indicate overall death rates from
 gastrointestinal illness from a variety of organisms approach 0.1% (Bennett, et al,
 1987).  In addition, studies (Glass, et al, 1991; Lew, et al, 1991) indicate that sensitive
 subpopulations, including infants and  those over 70 years old, have mortalities of
 3-5% from  diarrhea requiring hospitalization.  Additionally, specific pathogenic
 organisms  produce illness endpoints more serious than gastrointestinal illness.
 Hepatitis A infections, for example, may lead to jaundice and liver damage, as well as
 death. Death rates from hepatitis A illnesses in the U.S. have been reported at 0.6% of
 those who  are ill (CDC, 1985).  Incidence and mortality information for a variety of
 waterborne disease agents are found in Tables A and B. As a result of this, EPA is
 considering risk assessments for a variety of organisms and illness endpoints.

 Infection vs. illness

 Microbial dose-response determinations try to relate ingested levels of organisms to a
 given detection endpoint. This may be demonstrable infection or symptomatic illness
 or some other measure.  In the interest of protecting public health in a diverse
 population, EPA has focused on preventing infections and has considered defining
 acceptable risk with respect to infections avoided. For example, the goal of EPA's
 Surface Water Treatment Rule was to achieve risk reduction with respect to microbial
infection (USEPA, 1989). Generally, however, infection is not equal to illness. As an
example, in the Rendtorff (1954) study on the infectious dose for Giardia. many healthy
individuals  became infected, as shown by cysts in stool samples, but none became ill.
A survey of a waterborne outbreak of giardiasis in Berlin, NH, showed 76% of the
infections were asymptomatic.  Only 3% of those infected required hospitalization
(Lopez, et al, 1980). For Vibrio cholerae 01 (the toxigenic Latin American strain), 75%
of infections are asymptomatic. Some 20% of those infected develop mild diarrhea
and only 5% develop the severe, clinically-recognized form of the disease (CDC,
1991). It is understood, however, that sensitivity to microbial illness includes
enhanced likelihood of significant illness after infection.  To be protective of the overall
public health, EPA focuses on adverse effects to the sensitive subpopulations (in this

-------
case, infants and the elderly), thus EPA believes that by controling microbial
pathogens with respect to some acceptable level of infection rather than illness
provides greater protection to all.

EPA also assumes that the susceptibility to infection of the population studied by
researchers (i.e., male prisoners, students) is representative of the U.S. population as
a whole. However,  whether an individual becomes infected depends upon pathogen
virulence and dose, as well as the health of the individual.  While an infectious unit
may represent a single virus particle (Katz and Plotkiri, 1967) or Giardia cyst,
frequently much higher doses, especially for bacterial pathogens, are required to yield
an infection. These variations are difficult, if not impossible to determine.  For risk
assessment purposes, EPA assumes for Giardia and viruses that a dose of one
infectious unit can yield an infection. Using the dose-response curve developed from
the Rendtorff data (Rose, et al, 1991), this translates to about a 2% chance for an
individual to become infected if one Giardia cyst is ingested.

Selection of appropriate pathogenic organisms for regulatory development

In the development of the SWTR, EPA selected Giardia as the representative organism
for risk assessment, regulation and treatment. Giardia was selected because data was
available for risk assessment and because it was perceived that Giardia was more
resistant to disinfection than most other known microbial pathogens in water. It was
assumed that adequate disinfection of Giardia would yield adequate disinfection for
most other microorganisms of concern.  Recent data suggest that Cryptosporidium.
because of its greater resistence to disinfection than Giardia. may be a more
appropriate target organism for defining adequate levels of treatment.

Protozoan pathogens, such as Cryptosporidium or Giardia. are not normally found in
true groundwaters not under the direct influence of surface water. The pathogens of
concern in groundwaters only include enteric viruses and bacteria.  EPA considers
viruses  as more difficult to disinfect than bacteria, thus has selected representative
viruses  for risk assessment and regulatory purposes. At issue are both general and
specific problems in defining risk from waterborne viral infection. EPA had considered
using a single virus or virus group as the basis for determining risk, but rejected this
approach  because no one virus appeared suitable. Complicating the selection of a
single virus for calculating risk is the fact that occurrence data for pathogenic viruses in
water are scant, primarily from outbreak investigations. Moreover, dose-response data
are only available for a few viruses and the relative occurrence in water for different
viruses  may vary overtime, depending on the prevalence of a particular viral disease
in nearby populations that influence the source water quality.  Additionally, sensitivities
to different disinfectants vary between viruses.  Rotaviruses, for example, are more
sensitive to chlorine than hepatitis A, but less sensitive than hepatitis A to chloramine.

Hepatitis A represents the greatest health threat in terms of severity of waterborne
illness (short of death) and is more resistant to disinfection than many other
pathogens.  Unfortunately, no practical enumeration method for hepatitis A in drinking
water and no dose-response data are yet available. This prevents a quantitative  risk
assessment based on this organism. In contrast, rotaviruses have a lower infectious
dose than most other waterborne viruses,  and dose-response data are available, but
                                      4

-------
the disease is not as severe as that from hepatitis A.

As a result of these complications, EPA is considering use of a conceptual "synthetic
virus" of combined properties for regulatory development, as described by Regli, et al
(1991), which would provide reasonable worst-case limits for any given virus. This
concept would combine the properties of several pathogenic waterborne viruses to
define a reasonable worse-case situation. EPA would use the enterovirus group
(poliovirus, echovirus, coxsackievirus) to determine waterborne occurrence, since
relatively simple quantitation methods exist and this measurement would represent
worst case occurrence for any  particular enteric virus; rotaviruses for calculating
dose-response; and hepatitis A to estimate disinfection efficiency.

Determination of organism concentration in finished water

Risk estimates from exposure must ultimately be based on pathogen concentrations in
water reaching consumers. It is not possible to practically measure pathogen
concentrations (at least for Giardia and viruses, since they are health concerns at very
low concentrations) in finished  water to demonstrate that acceptable risk levels are
being achieved (Regli, et al, 1991).  It is much  more practical to monitor the source
water for pathogens or to estimate such concentrations indirectly (e.g., by measuring
virus concentrations in septic tanks or sewer lines and estimating die-off and dilution in
the source water), determine the level of treatment provided and then estimate the
organism occurrence in finished water. These estimates can be used to calculate the
associated risk and determine whether the treatment in place is adequate. This
indirect approach, however, introduces the uncertainty of estimating treatment
efficiencies in addition to characterizing the occurrence in the source water. This
approach also cannot be used to quantify risk from bacterial pathogens which may
regrow in the distribution system. The assumption that these uncertainties are
acceptable,  or can be reasonably defined, is a major caveat to EPA's current risk
assessment approach. Depending on specific conditions, the uncertainties in the
quantified risks may span several orders of magnitude.
Comparing pathogenic microbiaf risk and chemical risk

As part of rule development, EPA is comparing human health risks of microbial illness
with risks from disinfectants used to minimize these microbial risks and risks from the
resulting disinfectant byproducts. This comparison is difficult in that risks from
pathogenic microorganisms are generally acute versus those from chemicals, which
are generally chronic.  Also, risks from microorganisms are not calculated in the same
manner as are those from chemical contaminants, thus they are not explicitly
comparable.  However, similarities do exist and with some care, approximate
comparisons can be made.

One difficulty is that agents and their adverse effects are considered differently for
microoganisms and chemicals. The possible microbial illnesses, or "endpoints of
concern", vary with the organism and vary markedly in their severity. As discussed
above, these endpoints have been taken together as a broadly generalized "microbial
illness" resulting from these  organisms in total, rather than as separately defined
                                      5

-------
illnesses attributable to specific organisms.  This is different from EPA's treatment of
chemical contaminants, where individual chemicals or closely related chemicals are
regulated separately. Also, each chemical is regulated to one specific endpoint of
concern. For chemicals judged to be known or probable human carcinogens, the
endpoint of concern is cancer, leading to premature mortality. For non-carcinogenic
chemicals, a specific adverse endpoint of concern is identified in the Reference Dose
determination. This endpoint is generally at the lower end of a severity progression.

Derivation of dose-response data also differs between microorganisms and chemicals.
For microbial pathogens, dose-response values are determined directly from data on
human infection or illness.  Microbial illnesses are usually rapid and acute, and thus
can be causally linked to the infecting organism.  Epidemiological data from disease
outbreaks attributed to pathogenic organisms in drinking water can in some cases be
correlated to the organisms' ambient levels. The  studies by Rendtorff (1954), which
have been used to derive dose-response values for Giardia infection, involved human
subjects given known amounts of Giardia cysts. The viral dose-response relationships
described  by Regli, et al (1991) were likewise derived from studies of human
populations (Lepow, et al, 1962; Katz and Plotkin, 1967; Minor, et al 1981; Schiff, et al,
1984; Ward, etal, 1986).

Chemical dose-response assessments are far less certain. For chemical
contaminants, such as the disinfectants and disinfection byproducts, which occur at
less than part per million or billion levels in drinking water,  resulting illnesses of
concern are chronic and are expected to appear only after long exposures.  Causal
linkage of  illness to these low exposures is impossible. Further, current EPA methods
to determine chemical dose-response values generally extrapolate data from high
exposures in laboratory animals to the low exposures expected for humans.  These
extrapolations use health-conservative methods that may add orders-of-magnitude
safety factors and result in considerable uncertainty. For known animal carcinogens, it
is currently assumed that no exposure threshold exists and any exposure poses a risk.
The resulting theoretical 95% upper-bound lifetime human cancer risk is estimated
such that the  real  risk is unlikely to be greater than the calculated value,  is almost
certainly lower, and may be zero. However, since in at least some instances one
infectious  unit can yield illness, microbial risk from protozoa and viruses could also be
considered to be without a threshold, thus allowing probability estimations in the same
manner as with chemical carcinogens. Comparison of microbial risks with those  from
non-carcinogenic chemical contaminants poses other problems. Endpoints of concern
are not the same and may differ substantially in their severity and the progression of
severity with increasing doses. Probability-based dose-response values  for these
contaminants cannot be calculated using current  risk methods standard at EPA.

Exposure data are generally stronger for chemical contaminants than for microbial
pathogens. Chemicals can be assayed routinely to part per billion or greater
sensitivity  and plausible chronic exposures cian be estimated from this data.  However,
it is much  more difficult to estimate exposures to pathogenic microorganisms to allow
calculation of endemic risk. Many pathogenic microorganisms lack reliable and
sensitive quantitation methods for occurrence levels typically seen in water and levels
of occurrence can vary several orders of magnitude at a given site. We desire
occurrence data to allow estimation  of endemic levels of microbial illnesses.  While

-------
 epidemiological data exist for microbial illness, reported data are from outbreaks in
 communities and do not indicate endemic levels of disease. Additionally, most
 workers in the field believe that substantial underreporting of illness outbreak occurs.

 Modeling of pathogen, disinfectant and disinfection byproduct risks

 EPA has undertaken mathematical modeling intended, in part, to produce estimates of
 pathogen  and chemical exposures and risks to individuals arising from a variety of
 source waters after various water treatments. EPA desires that these estimates
 approximate the distributions that occur nationally. These models simulate occurrence
 levels of pathogenic organisms (specifically Giardia^  in raw water, then simulate
 removal efficiencies of pathogens and production of disinfection byproducts. The
 microbial and chemical concentrations thus generated are then used to estimate
 potential health risks. By considering a variety of increasingly stringent regulatory
 options and treatment trains, relative changes in microbial and chemical risks can be
 estimated and considered. There are a number of assumptions and uncertainties in
 these models. Input occurrence data is discussed below.  Issues for dose-response
 determinations have been discussed above.

 Waterborne pathogens in surface waters include protozoa (e.g., Giardia.
 Cryptosporidium)r bacteria (e.g., Legionella. Salmonella. Campylobacter). enteric
 viruses (e.g., Norwalk and Norwalk-like agents, rotaviruses, hepatitis A agent), and
 blue-green algae. In the SWTR, EPA specified treatment to eliminate Giardia cysts,
 and assumed that treatment values were sufficiently  high to control other microbial
 contaminants in surface water.  In contrast,  the pathogens generally occurring in
 groundwater only include enteric viruses and bacteria. For groundwater, EPA's
 concern is for viruses, since Gi'ardia and Crvptosporidium are not present and viruses
 are generally more resistant to disinfection than the pathogenic bacteria. However,
 source water occurrence data for all pathogenic microorganisms but Giardia and
 Crvptosporidium are scant. Therefore, EPA has focused on surface waters and
 Giardia in  the development of the  comparative risk assessment.

 The overall purpose of this modeling effort is to determine the likely exposures to
 pathogens and chemical contaminants remaining after water treatment, for typical
 water treatment process trains, raw water quality characteristics and modeled raw
 water pathogen levels. EPA considered five treatments for surface water and five for
 groundwater as describing the majority of public water supply systems.  Depending
 on the treatment scenario, various reductions of Giardia and virus levels occur during
treatment. Assumptions for the water treatment trains are derived from the SWTR. CT
values for Gjardja and hepatitus A were used. One hundred simulations of annual city
 means for Gjardja, and disinfection byproducts (trihalomethanes and haloacetic acids)
 in finished water were generated and used in subsequent risk assessment.  Since
 EPA currently lacks appropriate virus occurrence data, only the modeling for Giardia in
surface water systems has been performed.

 Preliminary work reported by Grubbs, et al (1992) examined the two levels of surface
wa^er treatment described above,  one minimally meeting the SWTR and the other
 meeting an ESWTR, where higher levels of treatment are specified for poorer quality
source waters. The results indicated that systems only minimally meeting SWTR

-------
standards of 3-log disinfection of Giardia could still produce water yielding significant
endemic levels of microbial illness, depending upon the Giardia cyst concentrations
that occur in source waters.  Increasing treatment proportionally for systems with
higher Giardia cyst concentrations to achieve approximately the same average Giardia
concentration at the first customer, reduced modeled endemic illness to de minimus
levels without substantial increases in treatment costs and without an appreciable
increase in  disinfection byproduct levels. This could reasonably be expected to apply
to disinfection  of bacteria and viruses as well, owing to their higher sensitivity to
disinfection. Comparison of appropriate model results to CDC data and to waterborne
outbreak disease data (Grubbs, et al, 1992) appears to support the validity of the
model and suggests that this model may be valuable for these analyses.
Acceptable human health risk

EPA is concerned with acceptable levels of public health risk from drinking water
pathogens.  The current approach in developing these rules considers a requirement
for an acceptable microbial risk, reflected in a promulgated MCL or treatment  ,
technique, that can feasibly be obtained. This differs from a MCLG, which is an
aspirational goal that does not take feasibility into account. MCLGs for pathogenic
organisms are generally set at zero, indicating that no risk of illness is the desired
goal.  This is similar to the approach taken for regulating chemical carcinogens, where
MCLs are set as close to the MCLG of zero, as is technically and economically feasible,
but also within an acceptable cancer risk range from 10"4 to 10'6. Whether an
acceptable risk is achieved can be determined from risk calculations. For these
purposes, the prevention of endemic illness is a concern, in addition to prevention of
illness outbreaks in a community.

For the SWTR, a risk of one infection perl 0,000 people per year was taken as the
acceptable health goal for Giardia. CDC data (Bennett, et al, 1987) indicate that
Giardia contributes about 8% (70,000 of 940,000) of all water-borne microbial illness.
Given a 70 year  lifespan, this calculates to a mean average 10% lifetime risk for
microbial infection from drinking water. Based on estimations from the maximum
likelihood analysis of GjaMa occurrence reported by Grubbs, et al (1992), the 95%
upper-bound risk would be on the order of 10-fold higher, thus yielding an estimated
lifetime risk of infection of 1. At this level of total infection, the lifetime risk of death from
waterborne microbial illness can be estimated,. Using the CDC (Bennett, al al, 1987)
ratio of approximately  0.1% for mortality resulting from  waterborne microbial illnesses
and assuming that all infections cause illness, then the estimated upper-bound lifetime
risk for this would belO"3.  If the mean risk value for infection was used along with a
plausible illness to infection ratio of 10~1, mean lifetime risk of death would be about
10-5.

Noting that significant differences exist in how risks are calculated and interpreted for
microorganisms, relative to chemical contaminants, the above developed acceptable
microbial risks can be  roughly compared to the acceptable levels for chemical
contaminants, as represented by EPA drinking water MCLs. For carcinogens, MCLs
are general set from about 1 x 10-4 to 1 x 10-° theoretical 95% upper-bound lifetime
cancer risk.
                                      8

-------
Implications for US drinking water oolicv
                                                             tsr'ijf e»Hf fes-
An immediate implication of this work is the realization that microbial pathogens
continue to dominate the comparative water-borne human health risks. Underscoring
this is recent work of Payment, et al (1991), which found that a water system meeting -
existing microbial drinking water standards could have endemic water-borne illness
rates of 25-35% per year.  This focuses more regulatory attention towards enhancing
disinfection via a GWD Rule and a possibly stricter SWTR, rather than accepting
current standards as adequate.  It also directs research towards outstanding problems
in microbial analysis, risk assessment and treatment technologies.

Treatment technology is of particular importance in terms of regulatory feasibility.  If the
risk estimations and comparisons are valid and predictive, it should be possible to
determine a minimum point for the sum of microbial and disinfection risks for a given
level of treatment.  To the limit of feasibility, this minimum could be lowered by
requiring higher levels of treatment to not only disinfect, but to reduce disinfectant
dosages and residuals and resultant byproducts. Feasibility in this case may be
limited by technology as well as by economics. For example, current treatment using
conventional filtration processes and adequate chemical disinfection contact time to
minimize microbial risks, followed by carbon filtration to remove  byproducts, may still
yield substantial disinfection byproduct risks in some systems with poor source water
quality. Use of membrane filtration to physically remove microbial pathogens
(including viruses)  produces no known byproducts of concern, and would be of
minimum risk even when followed by a residual disinfectant in the distribution system.
However, membrane filtration technology is not yet feasible for all systems and
involves high costs as well as potential water wastage. Policy implications here are
toward maintaining adequate disinfection relative to control of disinfectants and
disinfection byproducts to the limits of technology at a  given time. As technology
progresses, driven by a goal of overall drinking water risk minimization, more stringent
controls on disinfection byproducts may be possible without sacrificing disinfection.
Ultrafiltration, because it removes virtually all microbial pathogens, followed by
chloramine as a distribution system disinfectant, may be a long-term solution,
especially for bromide-containing source  waters, if health risks from chloramine
byproducts do not prove to be significant.

-------
References

Bennett, JV, SD Holmberg, MF Rogers and SL Solomon (1987). Infectious and
parasitic diseases. Am,J Prey Med 3: 102-114. Jn RW Amler and HB Dull (eds),
Closing the gap: the burden of unnecessary illness. Oxford University Press.
pp102-114.

Centers for Disease Control (1985). Hepatitis surveillance. CDC Report 49, Atlanta,
GA.

Centers for Disease Control (1991). Update:  cholera- western himisphere,
recommendations for treatment of cholera.  MMWR 40: 562-565.

Cooper, RC, AW Olivieri, RE Danielson, PG Badger, RC Spear and S Selvin (1984).
Infectious agent risk assessment water quality project.  Vol. I:  Assessment of risk
associated with water-related infectious agents.  UCB/SEEHRL Report No. 84-4,
University of California, Berkeley.

Cooper, RC, AW Olivieri, RE Danielson and PG Badger (1984).  Infectious agent risk
assessment water quality project. Vol. II: Factors associated with the transmission of
waterborne infectious disease. UCB/SEEHRL Report No. 84-5, University of
California, Berkeley.

Gelderloos, AB, GW Harrington, DM Owen, S Regli, JK Schaefer, JE Cromwell and X
Zchang (in press). Simulation of compliance choices for the regualtory impact
analysis. USEPA Publications.

Glass Rl, JF Lew, RE Gangarosa, CW LeBaron and M-S Ho (1991). Estimates of
morbidity and mortality rates for diarrhea! diseases in American children.  J. Pediatrics
118:27-33.

Grubbs WD, BA Macler and S Regli (in prejss). Simulation of microbial occurrence,
exposure and health risks after drinking water treatment processes. USEPA
Publications.

Hibler CP (1988). Analysis of municipal water samples for cysts of Giardia. In: Wallis
P and B Hammon (eds).  Advances in Giardia research. Univ of Calgary Press, pp
237-245.

Katz M and SA Plotkin (1967). Minimal infective dose of attenuated Polio virus for
man. Am J Pub Health 37:1837.

LeChevallier MW, TM Trok, MO Burns and RG Lee (1990). Comparison of the zinc
sulfate and immunofluroescence techniques for detecting Giardia and
Cryptosporidium in water. J AWWA 82:75-82.

LeChevallier MW, WD Norton and RG Lee (1991). Occurrence of Giardia and
Cryptosporidium spp. in surface water supplies.  Appl Env Microb 57: 2610-2616.
                                     10

-------
 Lepow ML, RJ Warren and VG Ingram (1962).  Sabin Type I oral poliomyelitis vaccine:
 effect of dose upon response of new-borne infants.  Am JDis Child
 Lew JF, R! Glass, RE Gangarosa, IP Cohen, C Bern and CL Moe (1991). Diarrheal
 deaths in the United States, 1979 through 1987. A special problem for the elderly
 JAMA 265: 3280-3284.

 Lopez, et al (1980). Waterborne giardiasis: a communitywide outbreak of disease and
 a high rate of asymptomatic infection. Am J Epid 112: 495-507.

 Minor, TE, Cl Allen and AA Tsiatis (1 981 ).  Human infective dose determinations for
 oral Poliovirus Type I vaccine in infants. J Clin Microbiol 13: 388.

 Payment, P, L Richardson, J Siemiatycki, R Dewar, M Edwardes and E Franco (1991).
 A randomized trial to evaluate the risk of gastrointestinal disease due to consumption
 of drinking water meeting current microbiological standards. Am J Pub Health 81 •
 703-708.

 Regli S, JB Rose, CN Haas and CP Gerba (1 991 ).  Modeling the risk from Giardia and
 viruses in drinking water. J AWWA 83: 76-84.

 Rendtorff RC (1954). The experimental transmission of human intestinal protozoan
 parasites.  II. Giardia lamblia cysts given in capsules.  Am J Hyg 59: 209-220.

 Rendtorff RC and CJ Holt (1954). The experimental transmission of human intestinal
 protozoan paracites.  IV. Attempts to trasmit  Endamoeba coli and Giardia lamblia by
 water. Am J Hyg 60: 327-328.

 Rose JB (unpublished,1988). Crvptosporidium in water:  risk of protozoan waterborne
 transmission.  Report prepared for Office of Drinking Water, U.S. EPA.

 Rose JB, CN Haas and S Regli (1991). Risk  assessment and control of waterborne
 giardiasis. Am J Pub Health 81 : 709-713.

 Schiff GM, et al (1984). Studies of Echovirus-12 in volunteers: determination of
 minimal infectious dose and the effect of previous infection on infectius dose  J Infect
 Dis150: 858.

 Sobsey MD, T Fuji and RM Hall (1 991). Inactivation of cell-associated and dispersed
 Hepatitis A virus in water. J AWWA 83: 64-67.

 USEPA (1989). National Primary Drinking Water Regulations: Filtration Disinfection;
Turbidity, Giardia lamblia. Viruses, Legionella and Heterotrophic Bacteria. Final Rule,
40 CFR parts 141 and 142. Federal Register, June 29, 1989, 54: 27486-27541.

Ward RL, et al (1986). Human Rotavirus studies in volunteers: determinations of
infectious dose and serological rsponse to infection.  J Infect Dis 154: 871.
                                     11

-------
Table A  Water-borne Domestic Hicrobial infections, 1985
Disease or agent
Total Water-borne
Campylobacteriosis
E. coli
Misc. enteric
Salmonella, nontyphi.
Shigella
Typhoid
Vibrio (excl. Cholera)
Yersiniosis (excl. plague)
Norwalk
Giardia
Incidence
940,000
320,000
150,000
10,000
60,000
30,000
60
1,000
1,800
300,000
70,000
Fatality/case (%)
0.1
0.1
0.2
1.0
0.1
0.2
6.0
4.0
0.05
0.0001
0.0001
Data calculated from Bennett, et al (1987)

-------
Table B  Suspected Water-borne Diseases and Organisms
Disease or agent
.Fatality/  case  (%)
Cholera
Legioneliosis
Enteroviral disease (excl. polio)
Hepatitis A
Poliomyelitis
Rotavirus
Coxsackieviruses
Echovirus
Reovirus
      1.0
      15
      o.ooi
      0.3
      10
      0.01
Data from Bennett, et al (1987)

-------

-------