Chapter 3

ASSESSMENT OF RISK

P.R. Hunter, P. Payment, N. Ashbolt and J. Bartram

3.1

Introduction

This chapter is primarily about the role of analytical techniques in the

assessment of risk and specifically the value of water quality indicator
parameters in this process. Assessment of risk in relation to drinking water
supplies is undertaken for a number of reasons (Percival et al., 2000):

To predict the burden of waterborne disease in the community, under

outbreak and non-outbreak conditions. This is helpful in determining the
impact of improvements in water supply on health and to act as a driver
towards improvement.

To help set microbial standards for drinking water supplies that will give
tolerable levels of illness within the populations drinking that water.

To identify the most cost-effective option to reduce microbial health risks

to drinking water consumers.

To help determine the optimum treatment of water to balance microbial

risks against chemical risks from disinfection by-products.

To provide a conceptual framework to help individuals and organisations

understand the nature and risk to, and from, their water and how those
risks can be minimised.

The focus of this chapter is to review the value of indicator parameters of

water quality and other analyses in the context of three different approaches to
the assessment of risk, namely:

79

Epidemiological methods.

Quantitative microbial risk assessment (QMRA).

Qualitative risk assessment (including risk ranking).

3.2

What is risk?

Risk can be defined in the simplest form as the possibility of loss, harm or
injury. This definition includes two separate concepts; the probability of an
event and the severity of that event. These two concepts are illustrated in
Figure 3.1, and this model helps the prioritisation of risks for any risk-reduction
action. Clearly those risks that need most urgent action are high probability
high severity risks (upper right quadrant). Those that need little, if any, attention
are low probability low severity (lower left quadrant).

Severity of harm

Figure 3.1. Two-dimensional classification of risk

Low probability of severe harm

High probability of severe harm

(should be given intermediate priority

attention)

(needs most urgent attention)

Low probability of mild harm

High probability of mild harm

(can probably be ignored or given low

priority attention)

(should be given intermediate priority

attention)

Probability of occurrence

Despite the simplicity of this two-dimensional model, the processes that

allow the calculation or quantification of risk differ. Indeed, many risk-based
decisions are still subjective or semi-quantitative. Even where risk assessments
are presented in an apparently objective, numerical manner these are often
based on assumptions which are themselves subjective or semi-quantitative.
One of the major problems with all forms of assessing risk is the quality and
levels of uncertainty in much of the basic data (Macgill et al., 2001).
3.3

Types of evidence

Data used in the assessment of risk is obtained from experimental work on

animals or volunteers and from epidemiological investigations. These
epidemiological investigations may be conducted during an outbreak
investigation or be done as part of planned research to investigate the causes
and transmission of disease.
80

The most abundant source of epidemiological data on waterborne disease

comes from outbreak investigations (Chapter 7), and outbreaks provide very
valuable data for the assessment of risk. Particularly, outbreaks can provide
clear evidence that a specific pathogen can be spread by the water route.
Outbreak investigations also provide good information on what failures in the
water supply and distribution chain led to the risk to health. This enables risk
management strategies to focus on those stages in the water supply chain where
failures are likely to occur. Outbreaks can also be the setting for
epidemiological studies that provide useful information on what non-waterrelated factors affect risk of infection with the outbreak pathogen. However,
outbreak data have their limitations (Andersson and Bohan, 2001). For any
particular pathogen, it is rarely known what proportion of the burden of disease
is due to sporadic spread by the water route. Nor is it known whether those
factors responsible for failure leading to outbreaks are also those factors
responsible for sporadic disease. Consequently information reliant only on
outbreaks may not be applicable to the major proportion of waterborne disease.
Also, epidemiological investigations of water-related disease may be biased by
prior knowledge of cases and controls about the suspected cause of the outbreak
(Hunter, 2000).
Targeted epidemiological studies can provide good data on the relationship
between specific water quality parameters and disease in a population. Such
studies can identify relationships between risk factors for all waterborne disease
and not only that associated with outbreaks. Separating the waterborne fraction
of gastrointestinal disease from the numerous other routes of infection is a
challenge and the results from most epidemiological studies are presented as a
level of association between drinking water and the parameter(s) under study.
These studies are often subject to criticism as there are rarely clear-cut
conclusions, and they are potentially subject to a number of biases and
confounding factors.
Quantitative microbial risk assessment (QMRA) is an emerging field that
has applications in specific situations and is discussed in more detail below.
QMRA uses information on the distribution and concentration of specific
pathogens in the water supply along with information on the infectivity of those
pathogens to determine risk to public health.
Assessment of the quality of evidence is important yet rarely formally
addressed in the assessment of risk (Macgill et al., 2001). Requirements for
evidence related to demonstration of causality may be very different to that for
dose response. In practice the overall body of evidence may include a number of
studies each with strengths and weaknesses and employing often very different
methods and approaches (Blumenthal et al., 2001; Haas and Eisenberg, 2001).
81

When assessing the risk of disease due to drinking water it is very important to
consider the overall body of evidence, weighing each piece of evidence as to its
quality. Given the uncertainty inherent in all epidemiological studies reliance on
a single study, even an extremely well conducted one, may be misleading.
3.4

Epidemiological approaches to risk

Epidemiology is the study of the incidence and transmission of disease in

populations. Epidemiological investigations are central to the assessment of risk
(Blumenthal et al., 2001), both in providing estimates of risk and in providing
input data into risk assessment models. The epidemiological definitions of risk
are distinct from definitions used more generally, and are defined in Table 3.1.
Table 3.1. Epidemiological definitions of risk
Risk

Definition

Absolute risk

The number of new cases occurring within a certain sized

population during a specified time period, usually referred to as
incidence.

Attributable risk

The proportion of cases of a disease due to a particular risk

factor.

Relative risk

The ratio between the incidence of disease in those members of

the population exposed to a possible risk factor and those not
exposed.

Odds ratio

The ratio between the probability that someone with a disease

has experience of the potential environmental factor and the
probability that a control has experience of the same factor.
Provides an estimate of relative risk in case control studies.

Epidemiology relies on a limited range of methods and approaches to

define risk (discussed in more detail elsewhere, e.g. Gordis, 2000). Most
epidemiological studies can be classified as descriptive, analytical or
intervention. Descriptive epidemiological studies set out to describe the
distribution of cases of disease in time, place and person. Two types of
descriptive study that have been used in relation to waterborne disease are the
ecological study and the time series study. Analytical studies are generally of
the case control or cohort type, in which individuals or groups are compared.
Intervention studies are experimental studies that observe the impact of certain
interventions (such as provision of point-of-use filters) on the risk of illness.
The various types of study are described in Table 3.2.

82

Table 3.2. Types of epidemiological study that have been used in risk assessment
of waterborne disease
Study type

Description

Advantages and disadvantages

Ecological
study

Determining relationship between

disease and risk factors by
comparing the incidence of
disease in different communities
with varying exposure to risk
factors.

Relatively inexpensive to carry out

providing that disease rates and
data on risk factors are already
available. Because data is only
available for groups, it is not
known whether individuals with
disease are exposed to risk factor.
Good for generating hypotheses,
but cannot be used as evidence of
epidemiological proof.

Time series
study

Determining relationship between

disease incidence in a population
and variation in a risk factor over
time.

A type of ecological study and

subject to the same advantages
and disadvantages.

Case-control
study

Determining relationship between

disease and risk factors by
comparing the incidence of
disease in exposed individuals to
matched controls.

Relatively inexpensive to carry out.

Generates data on individuals
exposed to the risk factors in
comparison with healthy
individuals.

Cohort study

Comparing rate of disease in two,

or more, populations with different
levels of exposure over a specific
period of time on randomly
selected individuals.

Relatively expensive to carry out.

Generates data on the risk factors
in populations by comparing
groups of randomly selected
individuals.

Intervention
study

Comparing the rates of disease in

two or more groups (cohorts) of
randomly chosen individuals after
intervening to change the level of
exposure.

The gold standard for

epidemiological proof, but can be
time consuming and costly to carry
out.

3.5

Studies linking ill health to indicators

While many microorganisms have been implicated as causative agents in

outbreaks of various diseases, there is little epidemiological data on the endemic
level of waterborne diseases and their aetiology. The association between many
aetiological gents with a given route of exposure and their contribution to the
total burden of disease is often uncertain. Studies that have attempted to define
the burden of waterborne disease have targeted gastrointestinal illness, as it is
the most frequent and easy to measure adverse outcome associated with
drinking water (Prss et al., 2002). This frequent outcome enables researchers
to obtain information faster than with less common outcomes (e.g. hepatitis) or

83

outcomes that are less defined and are more difficult to link with specific
exposures (e.g. malignant disease). However, use of gastrointestinal disease as
an index of water-related disease impact has a number of limitations. Depending
on how gastroenteritis is measured estimates of disease burden can vary
substantially. Since the disease may be considered mild, especially amongst
adults, relatively few people seek medical attention and even if they do they
may not have faecal samples taken for laboratory investigation. Consequently,
disease burden estimates based on national surveillance systems of laboratory
reports can substantially underestimate disease burden (Wheeler et al., 1999).
This has led to the use of self-reported gastroenteritis in several studies
(discussed below). There are, however, problems with the use of self-reported
gastroenteritis as a marker of disease, as depending on how gastroenteritis is
defined rates can vary substantially. How the data is collected can also
markedly affect estimates of disease burden. Retrospective studies, where
individuals are asked whether they have had diarrhoea in the previous month
can over-estimate illness by about three times when compared to prospective
studies where volunteers maintain a daily health diary (Wheeler et al., 1999).
This overestimate may be greater in outbreak settings (Hunter and Syed, 2001).
Furthermore, since gastrointestinal disease is relatively common and may be
transmitted by various means, it may be difficult to distinguish the waterborne
contribution from the background noise.
The link between substandard drinking water and disease is relatively easy
to demonstrate. Such a demonstration becomes more difficult to make as the
quality of the water improves towards the current World Health Organization
(WHO) Guidelines (WHO, 1993; 1996; 1997). Indeed, the link between highly
treated drinking water meeting local regulations, as found in most industrialised
countries, and microbial illness has only been reported relatively recently. For
example, both waterborne Giardia and Cryptosporidium infection have clearly
been linked to drinking water meeting or exceeding current standards, thereby
challenging the value of the traditional microbial indicator parameters as well as
the efficacy of treatment procedures (Gostin et al., 2000).
3.5.1

Untreated drinking waters

In developing countries there is abundant evidence that poor quality water

containing indices of faecal pollution is the source of much disease in the
population. There is, however, little data on the exact relationship between the
two.
There is a substantial body of evidence that relates improvements in water
supply and sanitation in general and in drinking water quality in particular, to

84

specific health outcomes (most frequently reductions in diarrhoeal disease).

Many of the early studies had severe methodological flaws (Blum and Feachem,
1983), but two reviews of published studies have sought to identify betterconducted studies and assess the detected disease outcomes (Esrey et al., 1985;
1991). Most studies detected were from less-industrialized countries and a
median reduction in diarrhoeal disease of 26 27% was reported. However,
water quality was typically not assessed and in some cases opportunities for
recontamination may have cast doubt on the actual intervention tested. In some
more recent studies, far better characterisation of the intervention has been
achieved with actual water quality measurements made (e.g. Quick et al., 1999
[E. coli] and Semenza et al., 1998 [chlorine residual]). Nevertheless, the
absence of an estimate of exposure from most studies renders them unusable in
formalised risk assessment requiring description of population dose-response.
3.5.2

Substandard drinking water

In France, Collin et al. (1981) prospectively studied gastrointestinal

illnesses associated with the consumption of tap water, using reports from
physicians, pharmacists and teachers. They reported five epidemics associated
with poor quality water but they did not address the endemic level of
gastrointestinal illnesses. The same group found a relationship between faecal
streptococci and acute gastrointestinal disease (Ferley et al., 1986; Zmirou
et al., 1987) in a study of 64 villages with sub-standard water. Thermotolerant
coliforms, total coliforms and total bacteria made no independent contribution
to disease. Zmirou et al. (1995) investigated the effect of chlorination alone, on
water that did not satisfy microbiological criteria otherwise. The crude
incidence of diarrhoea was 1.4 times more frequent in children from villages
where water supplies had evidence of faecal pollution, even after chlorination.
In Israel, Fattal et al. (1988) addressed the health effects of drinking water and
did not show a relationship between health effects and total or thermotolerant
(faecal) coliforms. Beaudeau et al. (1999) reported a relationship between the
chlorine disinfection level and diarrhoeal illness in the population of Le Havre
(France).
3.5.3

Drinking water meeting current regulations

In the USA, Batik et al. (1979) attempted to use cases of hepatitis A as an

indicator of health risk, but could not establish a correlation with water quality
nor, in a later study, did they find a correlation between traditional indicator
parameters (coliforms) and the risk of waterborne outbreaks (Batik et al., 1983).

85

Craun et al. (1997) in the USA, evaluated the relationship between

coliform compliance and outbreak occurrence. They found that coliforms were
usually found in the water during an outbreak investigation but that during the
previous months, coliforms were detected in only half of the systems and
caused a violation in only a quarter of them. The violation rate was not different
between community systems that experienced an outbreak and those that did
not. In Namibia, Isackson and Sayed (1988) conducted a similar study and did
not observe an increased risk of gastrointestinal illness associated with the
consumption of recycled wastewater.
In Canada, two prospective studies have suggested that a very high
proportion of gastrointestinal illnesses could still be attributable to tap water
consumption, even when water (albeit from a degraded catchment) exceeded the
current drinking water quality guidelines (Payment et al., 1991; 1997).
Turbidity of treated drinking water has been linked to health effects in
Milwaukee (MacKenzie et al., 1994; Morris et al., 1996), in Philadelphia
(Schwartz et al., 1997; 2000) and in Le Havre (Beaudeau et al., 1999). It should
be noted, however, that these studies of turbidity and adverse health outcome
are ecological, in that they measure exposure of populations rather than of
individuals and, as such, potentially suffer from bias due to the so called
ecological fallacy (Walter, 1991). While this does not mean that these studies
are invalid, they cannot be taken as proof of an association in their own right.
3.5.4

The role of index/indictor parameters in assessing risk to health

During the course of the 20th century, the absence of traditional

index/indicator parameters in drinking water was related to a significant
reduction in waterborne outbreaks. This reflected the use of these organisms to
indicate the presence of faecal contamination and through which valuable
information on effectiveness and failure of interventions was progressively
accumulated. More recently, occasional outbreaks and endemic disease have
been linked to waterborne disease in the absence of the traditional indicator
parameters. The causes are often failures in treatment or contamination of the
treated product, but the coliform parameters (total, thermotolerant or E. coli)
cannot provide information on the removal and inactivation of pathogens that
are several orders of magnitude more resistant to treatment. Hence, coliform
parameters remain useful for specific purposes described elsewhere in this
book, but future studies on waterborne disease should be targeted to additional
indicator parameters (for instance, those described in Chapter 2). There is,
however, no single direct measurement (including direct pathogen testing)
available to predict health outcomes in a population. Turbidity and faecal

86

streptococci counts are the main indicator parameters that have been shown to
have independent association with actual levels of disease in populations.
3.6

Quantitative microbial risk assessment (qmra)

The QMRA approach to risk differs from epidemiological approaches in

that the latter seeks to measure actual levels of disease in the population while
the former attempts to calculate risk from what is known, or can be inferred,
about the concentration of particular pathogens in the water supply and the
infectivity of those pathogens to humans. The relative values of QMRA and
epidemiology are strongly debated (Haas and Eisenberg, 2001).
3.6.1

The mathematical modelling of health risk

Establishing the exposure setting is the first step to the mathematical

evaluation of microbial risk. The purpose is to determine the possible pathogens
present, dose(s) consumed and the characteristics of the pathogen(s) that will
define the outcome.
The quantitative approach to microbial risk assessment is based on the
chemical risk assessment paradigm, and has been reviewed by Haas et al.
(1999). As with chemical risk assessment, this is a formalised procedure
involving four key steps (Table 3.3), each of which is briefly described below.
Table 3.3. Steps involved in quantitative microbial risk assessment
(Adapted from National Research Council, 1983)
Step

Aim

1.

Problem formalisation
To describe the overall environmental setting and
and hazard identification relevant pathogens that may cause acute or chronic
effects to human health.

2.

Dose-response analysis

To find appropriate relationship(s) between pathogen

exposure and infection or illness (from epidemiological
studies).

3.

Exposure assessment

To determine the size and nature of the populations

exposed to each identified pathogen by route, amount
and duration of the exposure.

4.

Risk characterisation

To integrate the information from exposure and doseresponse, to express public health outcomes, taking into
account the variability and uncertainty of the estimations.

87

While the conceptual framework for both chemical and microbial risk
assessments is the same, pathogens differ from toxic chemicals in several key
ways:

The variability of different strains of the one pathogen to cause disease

(differing virulence).

Virulence can evolve as the pathogen passes through various infected

individuals.

Pathogens are generally not evenly suspended in water.

Pathogens can be passed from one person to many (secondary spread),

from either healthy but infected (asymptomatic) or ill (symptomatic)
hosts.

Whether a person becomes infected or ill depends not only on the health
of the person, but also on their pre-existing immunity and pathogen dose.

3.6.2

Hazard identification (hazard assessment)

Pathogenic microorganisms are relatively well described in the scientific

literature and, apart from emerging waterborne pathogens (LeChevallier et al.,
1999a,b), data on their characteristics are generally available. Data needed for
the risk assessment process include the severity of the outcome, susceptibility
(long and short-term immunity), susceptible populations and secondary (personto-person) disease transmission. The outcomes of the exposure include noninfection, asymptomatic infection and various levels of morbidity and mortality.
Gender, age and some forms of susceptibility may also affect the outcome.
Severe morbidity or mortality resulting from waterborne exposures are
significant in developing countries, but are relatively rare in industrialised
countries.
3.6.2.1

Outbreaks

To properly conduct risk assessment, the hazard must be identified and

outbreaks provide important data in microbial risk assessment. The pathogen
responsible for the outbreak must be identified, the severity and contagiousness
of the infection can be described, the patterns of transmission in the population
can be studied and control measures can be evaluated. Waterborne disease
outbreak surveillance is key to this evaluation, and identification of the
aetiologic agent is dependent on the timely recognition of the outbreak, so that
appropriate clinical and environment samples can be obtained. The interests and
88

expertise of investigators and the routine practices of local laboratories can also
influence whether the aetiologic agent is identified (Frost et al., 1996).
Diarrhoeal stool specimens, for example, are generally examined for bacterial
pathogens, but not for viruses. In most laboratories, testing for Cryptosporidium
is only undertaken if requested and is not included in routine stool examinations
for ova and other parasites. Hence, it is not surprising that even in the USA,
with one of the most comprehensive registers of waterborne outbreaks, between
1992-1996 the causative organism was not identified in over 40% of
investigations (Levy et al., 1998).
The water quality data collected during and/or before the outbreak can be
useful in identifying the causes of the outbreak and in preventing their
reoccurrence. (Methods used for microbial water quality assessment are
discussed in Chapter 8 and their use in outbreak investigation is described in
Chapter 7). While background data on the level of faecal contamination, if not
sewage pollution in water is very valuable, care is needed in interpreting data on
finding or not finding pathogens. In particular, molecular epidemiology or
similar typing methods are necessary to confirm if the species identified from
water was also the agent present in the infected host (Chapter 7). There has been
considerable controversy over a number of species of opportunistic bacterial
pathogens with apparently non-pathogenic strains that may be found in drinking
water, versus different strains (and presumably non-water sources) causing
illness (Edberg et al., 1986; Havelaar et al., 1992; Khn et al., 1997).
3.6.2.2

Emerging pathogens

As new pathogens are being described in the population or in the

environment, their potential for being transmitted by the water route must be
evaluated. Basic characteristics that allow a pathogen to be waterborne include:

Excretion in the faeces and/or urine.

An environmentally persistent stage.

The ability to cause infection when inhaled or ingested.

Emerging pathogens include those that are increasingly being recognised

as important contributors to waterborne disease as well as those that are newly
discovered. As such, they include:

89

Viruses: new enteroviruses, human caliciviruses (including Norwalk-like

viruses), and hepatitis E.

Parasitic protozoa: Cyclospora cayetanensis, various microsporidia and

Toxoplasma gondii.

Bacteria: Mycobacterium avium complex, Helicobacter pylori, pathogenic

Escherichia coli and Campylobacter jejuni (LeChevallier et al., 1999a,b).

Toxic cyanobacteria (Chorus and Bartram, 1999).

Most faecal-oral pathogens are identified as causing acute gastrointestinal

illnesses, with the major exceptions being hepatitis A and E viruses,
Helicobacter pylori, Salmonella typhi and hookworm infection. However,
it is important to note (as mentioned in Chapter 1) that some commonly
recognised diseases (such as arthritis, type 1 diabetes mellitus, abortion,
Guillain-Barr and Miller Fisher Syndrome) have been associated with, or
are suspected to be caused by, infection with viral or bacterial pathogens
excreted by humans or animals (Duim et al., 2000; Frisk et al., 1992;
Gurgan and Diker, 1994; Havelaar et al., 2000; Maki-Ikola and Granfors,
1992; Niklasson et al., 1998).

3.6.3

Dose-response analysis

For QMRA, human dose-response studies are available for a few

pathogens and can be used to estimate the effects of low level exposure to these
microorganisms (Haas and Eisenberg, 2001). Two models of the infection
process have been proposed: the exponential model (Equation 1) and the betaPoisson model (Equation 2). These have been developed from biologically
plausible assumptions about the infection process. Models may fit available data
in a statistically acceptable sense and yet provide very different estimates for the
risk at an extrapolated low dose; a situation that has frequently caused argument
in chemical risk assessment. In QMRA, it may be possible to test the potential
appropriateness of different dose-response functions by validating with outbreak
data (Eisenberg et al., 1998).

90

Exponential model:
Probabilityinfection = 1 exp(-rD)

Equation 1

Where D = pathogen dose; r = fraction of pathogens that survives to

produce an infection.
Beta-Poisson model:
Probabilityinfection = 1 (1 + (D/ID50))-

Equation 2

Where D = pathogen dose; and ID50 are parameters of the betadistribution used to describe variability in survival.
Given a set of dose-response data, i.e. exposure of populations to various
doses of microorganisms and measurement of response (such as infection), the
best fitting parameters of a dose-response relationship may be computed via
standard maximum likelihood techniques. The method has been used for human
viruses, parasitic protozoa and some bacterial pathogens (Haas et al., 1999).
Confidence limits to the parameters can then be estimated, and used as a basis
for low-dose extrapolation (Kang et al., 2000). It should be noted that, in
general, dose-response studies have been conducted on healthy adults and may
not reflect the response of the general population or of more susceptible
population segments.
During an outbreak, individuals are exposed to different levels of the
pathogen(s): the volume of water ingested may be coupled with data on the
level of contamination of the water. These data can provide a dose-response
relationship confirming volunteer studies. Furthermore, information on
susceptible sub-populations (such as children and the immuno-compromised)
may also be forthcoming. For example, waterborne outbreaks of
cryptosporidiosis indicate that the general population may contract watery
diarrhoea that lasts up to several days, whereas HIV-patients may be untreatable
and die, thereby creating a much more significant health burden if the latter are
included in a risk assessment (Perz et al., 1998).
Volunteer feeding studies have provided data on the dose-response curve
for several pathogens (Haas et al., 1999). It is, however, often difficult to obtain
data on low doses as large numbers of volunteers would be needed to define the
lower bounds of the dose-response curve. It is also difficult to extrapolate from
a single strain to give a generalised model for a pathogen. Virulence differs
from one strain to another and the outcomes are often very different (e.g. E. coli
enteropathogenic versus non-enteropathogenic strains). The debate around the
human health significance of exposure to human versus animal strains of
Cryptosporidium parvum is another example. Feeding trials with three different
91

bovine strains of C. parvum have generated 50% infective doses (ID50) for
oocysts in healthy human volunteers ranging between 9 and 1 042 (Okhuysen
et al., 1999). Such a wide range is potentially problematic as the ID50 is the
parameter defining the slope of the dose-response curve in the beta-Poisson
model. A further complication is that pre-existing immunity may provide
protection from infection and illness at low oocyst doses (Chappell et al., 1999),
thereby changing the low dose-response extrapolation in a manner not
accounted for by any current model.
Relatively few data points are used to generate the curve and the degree of
uncertainty over the position of each data point is high. Each data point is a
sample mean of the probability of illness for people exposed to a set dose of
pathogen. The confidence intervals for each sample mean will be wide. It is
unlikely that all the measured points exactly correspond with the true population
means for each dose. In such circumstances it is impossible to be certain about
what dose-response model would best fit the actual curve (as opposed to the
curve of the sample means). There is, therefore, considerable uncertainty in
which model best fits the actual dose response curve and what its parameters
should be (Coleman and Marks, 1998). The impact of these uncertainties is
most marked at low doses (i.e. at the dose that will most frequently be
experienced in real life). Therefore, the predicted number of illnesses following
low dose exposure can vary by several orders of magnitude (Holcomb et al.,
1999).
3.6.4

Exposure assessment

The actual dose consumed by an individual is generally unknown and

difficult to estimate. Methods for the detection of some pathogens are not even
available, and most pathogens occur at very low levels in treated water
(generally below detection). The general level of some pathogens
(e.g. enteroviruses, Giardia, Cryptosporidium), however, are available for
sewage and untreated water. These raw water values can be used, along with the
proportion of surrogate removed by treatment, to indirectly estimate the level of
individual pathogens after treatment, thereby providing an estimate of the dose
in the water. The possible uses of surrogates and indicators are further discussed
below.
For drinking waters, the volume ingested per exposure is relatively well
defined after several studies in a number of countries (e.g. Roseberry and
Burmaster, 1992). A volume of two litres per person per day is often used to
estimate drinking water exposure, but this does not reflect the fact that only a
fraction of that volume is consumed unmodified (especially unboiled). This is
important for QMRA as microorganisms are inactivated by heat; therefore water
92

consumed in hot drinks or used in the preparation of cooked food would not be
a risk factor.
Viruses and parasites have been detected in drinking water, which was
otherwise apparently safe, without any detectable health effect being seen in the
receiving population (Bouchier, 1998). Possible reasons for this include false
positive detections, the presence of non-infective pathogens and the pathogen is
present in a concentration below that which would be expected to cause
detectable disease in the population. On the other hand, unrealistically large
volumes of drinking water would need to be sampled for example to meet the
USEPAs level of acceptable waterborne risk (<10-4 infections per annum see
1.5.1). Translating this for Cryptosporidium parvum would mean that 500
samples of 2 000 litres each would be needed to make a reasonably accurate
estimation of the allowed concentration (7 10-6 per litre) (Teunis et al., 1996).
Furthermore, depending on the detection method used, an unknown proportion
of pathogens isolated from the environment may be incapable of causing
infection. Therefore, alternative strategies are recommended to estimate
pathogen concentrations.
The applications of coliform bacteria to index the pollution of source
water, or as an indicator of water treatment efficacy or recontamination of
treated water have provided little information on health effects in developed
regions. Nonetheless, these organisms can play an important part in estimating
pathogen numbers for a screening-level or first tier of a QMRA. For example,
direct measurement of viral, parasitic protozoa and bacterial pathogens is
possible for sewage effluents, as is the estimation of pathogen prevalence data
for the faeces of some domestic animals. Hence, predictions of pathogens in
source waters can be made if the relative proportion of human and animal faecal
load is determined by, say, the analysis of faecal sterols (Leeming et al., 1998).
For environments where sewage is the primary faecal contaminant, then
pathogen dilutions in source waters can be estimated directly by the dilution of
thermotolerant coliforms (index for bacterial pathogen contamination) and
spores of Clostridium perfringens (index for the hardier viral and protozoan
pathogens) (Medema et al., 1997).
For physical treatment barriers, such as sand or membrane filters, and for
disinfection by chlorine, ozone or UV, surrogates for pathogen removal are also
generally accepted. Total aerobic spores or spores of C. perfringens are
reasonable surrogates for the cysts and oocysts of parasitic protozoa and
coliphages may also be appropriate for human enteric viruses (Facile et al.,
2000; Hijnen et al., 2000; Ndiongue et al., 2000; Owens et al., 2000). Note that
while coliphages make good models for human virus removal by physical
means, that may not be the case for mixed oxidants (Casteel et al., 2000).
93

3.6.5

Infectious disease models and risk characterisation

As outlined in the previous sections, attempts to provide a quantitative

assessment of human health risks associated with the ingestion of waterborne
pathogens have generally focused on static models that calculate the probability
of individual infection or disease as a result of a single exposure event. They do
not address the properties that are unique to infectious disease transmission such
as secondary transmission, immunity and population dynamics (Haas and
Eisenberg, 2001). To understand the role that water plays in the transmission of
enteric pathogens and to estimate the risk of disease due to drinking water
within a defined population it is important to study the complete disease
transmission system, as illustrated in Figure 3.2. It is also important to recognise
the additional pathways that describe the natural history of enteric pathogens:
animal-to-environment-person, person-to-environment-to-person, and personto-person (Eisenberg et al., 2001).
A fundamental concept in disease transmission models is the reproduction
number, R0, which is defined as the number of infections that result from the
introduction of one index case into a population of susceptible individuals.
Therefore, R0 is a measure of the ability of a pathogen to move through a
population. An R0 >1 suggests that the pathogen is multiplying within a
community and that prevalence is increasing, whereas an R0 <1 suggests that the
disease is dying out of the population. An R0 that is on average equal to 1
suggests that the disease is endemic in the population. There are various
methods to estimate R0 for different pathogens and in different environmental
settings (Dietz, 1993). Measles, for example, is a highly infectious respiratory
transmitted disease and has been estimated to have an R0 of approximately 14.
Polio, on the other hand, a waterborne pathogen has an R0 of approximately 6.

94

Figure 3.2. Conceptual model for rotavirus infection pathways

(from Haas and Eisenberg, 2001)

S
Latency

Exposure

Incubation

SD

SC

C
CP

PC

PD

D
DP

Exposure

S: susceptible = not infectious, not symptomatic. C: carrier = infectious, not symptomatic.

D: diseased = infectious, symptomatic. P: post Infection = not infectious, not symptomatic with
short-term or partial immunity.

Summarising the previous sections, the individual daily dose of pathogenic

microorganisms via some particular product may be calculated as (Teunis et al.,
1996):

Dose = C

1
I 10 DR V
R

Equation 3

C = Concentration of pathogenic microorganisms in raw (source) materials

(or partially processed products, if data are available).
R = Recovery of the detection method.
I = Fraction of the detected pathogens that is capable of infection
(viability).
DR = Removal or inactivation efficiency of the treatment process,
expressed as its Decimal Reduction factor (DR = 0 when concentrations in the
finished product are available).
V = Daily individual consumption of the considered product.

95

In many cases, risk evaluations start from the assumption that the doseresponse relationship is approximately linear at low doses. Therefore, at very
low doses, calculation of the risk of infection simply consists of multiplying the
dose estimate with the slope of the dose-response relationship. Estimates of
daily risk may be extrapolated to yearly risk. When P1* and Pn* are the
probabilities of infection after a single (e.g. daily) exposure and after repeated
exposures (n times a daily exposure) respectively:

Pn* = 1 (1 P1* ) n n P1*

Equation 4

The latter simplification is valid as long as P1* << 1 (Haas et al., 1999).
From the above discussions it would seem that microbial data, whether
relating to indicator parameters or pathogens, have most relevance to the
exposure assessment phase of QMRA. These provide estimates of actual levels
of pathogens in water or the likelihood that water is exposed to faecal pollution.
However, caution must be exercised in assuming a direct relationship between
this level and risk to health. Despite the use of numbers and mathematical
equations, QMRA is not yet an exact science.
3.7

Qualitative risk assessment

Qualitative methods for analysing microbial hazards and managing risks

are commonplace within the food industry. They are applied as part of a
systematic process including Hazard Analysis Critical Control Points (HACCP)
(Coleman and Marks, 1999), which has recently been taken up by the water
industry (Havelaar, 1994; Barry et al., 1998; Deere and Davison,1998; Gray and
Morain, 2000; Deere et al., 2001; Dewettinck et al., 2001; Davison et al., 2002).
Although hazard identification and exposure assessment are common
issues across qualitative and quantitative methods, dose-response models and
risk characterisation steps (Table 3.3), are usually replaced with risk rankings in
qualitative assessments. These rankings are generally derived from expert
opinion summarising:

Likelihood of possible risk pathways.

Severity of outcome from each pathway.

Numbers of people that may be impacted.

96

Water agencies are now focusing on the whole system approach, as

illustrated in Figure 3.3, which includes an assessment of all types of physical,
chemical and microbiological risks. Possible ranking schemes are numerous,
but follow the generic structure indicated in Table 3.4, with Table 3.5
illustrating a simple risk scoring table.
Table 3.4. Possible qualitative risk assessment approach to
rank or scale hazardous scenarios
Step

Comment

1. Hazard scenario

Identification of hazardous scenarios, such as massive rainfallinduced contamination of source water, filter breakthrough or
loss/breakdown of chemical disinfection system (i.e. not
necessarily limited to a single pathogen).

2. Likelihood

Ranking or scaling of how likely the event is (e.g. # events per

year).

3. Consequence

Ranking or scaling of the consequence (e.g. short-term injury or

ill-health through to permanent disability or death).

4. Scale of effect

Consideration of the number of people affected by the hazard

scenario.

5. Risk score

Different weightings may be given to (2) to (3) and multiplied to

give a value for each hazard scenario.

6. Rank

Each hazard scenario is then ranked, to provide a priority list for

risk management.

Table 3.5. Simple risk scoring table for prioritising risks

(Davison et al., 2002)
Severity of consequences
Likelihood

Insignificant

Minor

Moderate

Major

Catastrophic

Almost certain

10

15

20

25

Likely

12

16

20

Moderate

12

15

Unlikely

10

Rare

The risk score for a particular hazard = likelihood severity of consequences.

An example of the descriptive terms that can be used to rate the likelihood
and severity for calculation of the risk score is given in Table 3.6.

97

98

Grazing/
Domestic
Animals

Upstream
Sewage
Treatment

Human
Pollution

Short
Circuiting

Sediments

Septic Tanks

Native
Animals

Algal Blooms

Reservoir/
River

Urban Runoff

Catchment

Pipeline
Growths

Treatment
Process
Upset

Maintenance
Contamination

Cross
Contamination

Open
Storages

Reticulation

Treatment
Effectiveness

Treatment

Cross
Contamination

Customer
Tap

(Adapted from Stevens et al., 1995)

Water
Uses

Consumption

Exposure

Infectivity/
Viability

RISK

Figure 3.3. Generic flow diagram for sources of microbial risk in a drinking water context

Table 3.6. Example descriptive terms for risk score calculation

(Davison et al., 2002)
Item

Definition

Weighting

Almost certain

Once a day

Likely

Once per week

Moderate

Once per month

Unlikely

Once per year

Rare

Once every five years

Catastrophic

Potentially lethal to large population

Major

Potentially lethal to small population

Moderate

Potentially harmful to large population

Minor

Potentially harmful to small population

Insignificant

No impact or not detectable

Compared to both epidemiological and quantitative microbial risk

assessment, this approach does not seek to determine actual levels of disease
associated with a supply. As such, criticisms cannot be made that the
conclusions are imprecise compared with reality. The other advantage that this
approach has over other methods is that out of the process itself solutions to
minimise risk will present themselves. On the other hand, reliance on expert
opinion does not always produce the correct answer as experts opinions and
models of the world are often subject to bias and inaccuracies as with any other
source of data (Hunter and Fewtrell, 2001).
3.7.1

Indicators and qualitative microbial risk assessment

Microbial and other indicator analyses will be a major source of evidence

at several stages of qualitative risk assessment. The role of such information in
conducting assessments of source water quality, treatment efficacy and integrity
of the distribution system are discussed in more details in Chapters 46.
As will be seen, studies on the presence of indicator organisms frequently
provide more useful information for qualitative risk assessment than do studies
on enumeration of specific pathogens. Nevertheless, well-designed studies of
specific pathogens can also be of great value in certain situation. For example,
the detection of E. coli, faecal streptococci or sulphite-reducing Clostridia in
source water all indicate that the water is subject to contamination from human
or animal faeces. Detection and typing of Cryptosporidium in source water will

99

give a better understanding of the risk to the water supply system and the
sources of contamination.
Coliform bacteria in treated water may give an indication that water
treatment systems are not operating satisfactorily or that water is becoming
contaminated within the distribution system. However, coliform bacteria alone
are not good indicators of risk from chlorine-resistant pathogens such as
Cryptosporidium. Some indicator organisms may be naturally present in the
source water or can be deliberately seeded into the inlet to a water treatment
works and monitored at various stages in the treatment and distribution in order
to demonstrate the effectiveness of the whole system.
3.8

Summary

Microbial and other indicator parameters play an essential role in all the
models used in the assessment of risk discussed in this chapter. However, the
exact relationship between these indicator parameters and risk to health is still
far from clear. Although studies have shown that turbidity and faecal
streptococci are independent indicators of health risk there is no clear-cut
predictive relationship. Even where information on pathogens in potable water
is available, current quantitative risk assessment models have considerable
uncertainty in their calculated risk. Perhaps the real value of such indicator
parameters is in qualitative risk assessment where they can be used for
identifying where failures may occur in the water extraction, treatment and
distribution system.

100

REFERENCES

Andersson, Y. and Bohan, P. (2001) Disease surveillance and waterborne

outbreaks. In: Water Quality: Guidelines, Standards and Health.
Assessment of Risk and Risk management for Water-related Infectious
Disease. Fewtrell, L. and Bartram, J. (Eds.) IWA Publishing, London. pp.
115-133.
Barry, S.J., Atwill, E.R., Tate, K.W., Koopman, T.S., Cullor, J. and Huff, T.
(1998) Developing and implementing a HACCP-based programme to
control Cryptosporidium and other waterborne pathogens in the Alameda
Creek watershed: Case study. American Water Works Association
Annual Conference, 21-25 June 1998, Dallas. Texas Water Resources
Vol. B, 57-69.
Batik, O., Craun, G.F. and Pipes, W.O. (1983) Routine coliform monitoring and
water-borne disease outbreaks. Journal of Environmental Health 45, 227230.
Batik, O., Craun, G.F., Tuthil, R.W. and Kroemer, D.F. (1979) An
epidemiologic study of the relationship between hepatitis A and water
supply characteristics and treatment. American Journal of Public Health
70, 167-169.
Beaudeau, P., Payment, P., Bourderont, D., Mansotte, F., Boudhabay, O.,
Laubis, B. and Verdire, J. (1999) A time series study of anti-diarrheal
drug sales and tap-water quality. International Journal of Environmental
Health Research 9(4), 293-312.
Blum, D. and Feachem, R.G. (1983) Measuring the impact of water supply and
sanitation investments on diarrhoeal diseases: problems of methodology.
International Journal of Epidemiology 12(3), 357-365.
Blumenthal, U.J., Fleisher, J.M., Esrey, S.A. and Peasey, A. (2001)
Epidemiology: a tool for the assessment of risk. In: Water Quality:
Guidelines, Standards and Health. Assessment of risk and risk

101

management for water-related infectious disease. Fewtrell, L. and

Bartram, J. (Eds.). IWA Publishing, London. pp135-160.
Bouchier, I. (1998) Cryptosporidium in Water Supplies. Third Report of the
Group of Experts to Department of the Environment, Transport and the
Regions and Department of Health. November, 1998. Drinking Water
Inspectorate, London.
http://www.dwi.detr.gov.uk/pubs/bouchier/index.htm pages.
Casteel, M.J., Sobsey, M.D. and Arrowood, M.J. (2000) Inactivation of
Cryptosporidium parvum oocysts and other microbes in water and
wastewater by electrochemically generated mixed oxidants. Water Science
and Technology 41(7), 127-134.
Chappell, C.L., Okhuysen, P.C., Sterling, C.R., Wang, C., Jakubowski, W. and
DuPont, H.L. (1999) Infectivity of Cryptosporidium parvum in healthy
adults with pre-existing anti-C. parvum serum immunoglobulin G.
American Journal of Tropical Medicine and Hygiene 60(1),157-164.
Chorus, I. and Bartram, J. (Eds.) (1999) Toxic Cyanobacteria in Water. E&FN
Spon, London.
Coleman, M. and Marks, H. (1998) Topics in dose-response modelling. Journal
of Food Protection 61, 1550-1559.
Coleman, M.E. and Marks, H.M. (1999) Qualitative and quantitative risk
assessment. Food Control 10, 289-297.
Collin, J.F., Milet, J.J., Morlot, M. and Foliguet, J.M. (1981) Eau dadduction et
gastroentrites en Meurthe-et-Moselle. J. Franc. Hydrologie 12, 155-174.
Craun, G.F., Berger, P.S. and Calderon, R.L. (1997) Coliform bacteria and
waterborne disease outbreaks. Journal of the American Water Works
Association 89(3), 96-104.
Davison, A., Howard, G., Stevens, M., Callan, P., Kirby, R., Deere, D. and
Bartram, J. (2002) Water Safety Plans. WHO/SHE/WSH/02/09 World
Health Organization, Geneva, Switzerland.
Deere, D., Stevens, M., Davison, A., Helm, G. and Dufour, A. (2001)
Management Strategies. In: Water Quality: Guidelines, Standards and
Health. Assessment of risk and risk management for water-related

102

infectious disease. Fewtrell, L. and Bartram, J. (Eds.) IWA Publishing,

London. pp. 257-288.
Deere, D.A. and Davison, A.D. (1998) Safe drinking water. Are food guidelines
the answer? Water 25, 21-24.
Dewettinck, T., Van Houtte, E., Geenens, D., Van Hege, K. and Verstraete, W.
(2001) HACCP (Hazard Analysis and Critical Control Points) to
guarantee safe water reuse and drinking water production a case study.
Water Science and Technology 43(12), 31-38.
Dietz, K. (1993) The estimation of the basic reproduction number for infectious
diseases. Statistical Methods in Medical Research 2, 23-41.
Duim, B., Ang, C.W, van Belkum, A., Rigter, A., van Leeuwen, N.W.J., Endtz,
H.P. and Wagenaar, J.A. (2000) Amplified fragment length polymorphism
analysis of Campylobacter jejuni strains isolated from chickens and from
patients with gastroenteritis or Guillain-Barr or Miller Fisher Syndrome.
Applied and Environmental Microbiology 66(9), 3917-3923.
Edberg, S.L., Pisticelli, V. and Cartter, M. (1986) Phenotypic characteristics of
coliform and non coliform bacteria from a public water supply compared
with regional and national clinical species. Applied and Environmental
Microbiology 52, 474-478.
Eisenberg, J.N.S., Bartram, J. and Hunter, P.R. (2001) A public health perspective
for establishing water-related guidelines and standards. In: Water Quality:
Guidelines, Standards and Health. Assessment of risk and risk
management for water-related infectious disease. Fewtrell, L. and
Bartram, J. (Eds.) IWA Publishing, London. pp. 229-256.
Eisenberg, J.N.S., Seto, E.Y.W., Colford, J., Olivieri, A.W. and Spear, R.C.
(1998) An analysis of the Milwaukee Cryptosporidium outbreak based on a
dynamic model of disease transmission. Epidemiology 9(3), 255-263.
Esrey, S.A., Feachem, R.G. and Hughes, J.M. (1985) Interventions for the control
of diarrhoeal diseases among young children: improving water supplies and
excreta disposal facilities. Bulletin of the World Health Organization 63(4),
757-772.
Esrey, S.A., Potash, J.B., Roberts, L. and Schiff, C. (1991) Effects of improved
water supply and sanitation on ascariasis, diarrhoea, dracunculiasis,

103

hookworm infection, schistosomiasis and trachoma. Bulletin of the World

Health Organization 69(5), 609-621.
Facile, N., Barbeau, B., Prvost, M. and Koudjonou, B. (2000) Evaluating
bacterial aerobic spores as a surrogate for Giardia and Cryptosporidium
inactivation by ozone. Water Research 34(12), 3238-3246.
Fattal, B., Guttman-Bass, N., Agursky, T. and Shuval, H.I. (1988) Evaluation of
health risk associated with drinking water quality in agricultural
communities. Water Science and Technology 20, 409-415.
Ferley, J.P., Zmirou, D., Collin, J.F. and Charrel, M. (1986) Etude longitudinale
des risques lis la consommation d' eaux non conformes aux normes
bactriologiques. Rev. Epidemiol. Sante Publique 34, 89-99.
Frisk, G., Nilsson, E., Tuvemo, T., Friman, G. and Diderholm, H. (1992) The
possible role of coxsackie A and echo viruses in the pathogenesis of type 1
diabetes mellitus studied by IgM analysis. Journal of Infection 24, 13-22.
Frost, F.J., Craun, G.F. and Calderon, R.L. (1996) Waterborne disease
surveillance. Journal of the American Water Works Association 88(9), 6675.
Gordis, L. (2000) Epidemiology 2nd Edition. W.B Saunders Company,
Philadelphia.
Gostin, L.O., Lazzarini, Z., Neslund, V.S. and Osterholm, M.T. (2000) Water
quality laws and waterborne diseases: Cryptosporidium and other emerging
pathogens. American Journal of Public Health 90(6), 847-853.
Gray, R. and Morain, M. (2000) HACCP application to Brisbane Water. Water
27, 41-42.
Gurgan, T. and Diker, K.S. (1994) Abortion associated with Campylobacter
upsaliensis. Journal of Clinical Microbiology 32, 3093-3094.
Haas, C. and Eisenberg, J, N.S. (2001) Risk assessment. In: Water Quality:
Guidelines, Standards and Health. Assessment of risk and risk
management for water-related infectious disease. Fewtrell, L. and
Bartram, J. (Eds.) IWA Publishing, London. pp. 161-183.
Haas, C.N., Rose, J.B. and Gerba, C.P. (1999) Quantitative Microbial Risk
Assessment. John Wiley, New York.

104

Havelaar, A.H. (1994) Application of HACCP to drinking water supply. Food

Control 5, 145-152.
Havelaar, A.H., de Wit, M.A.S. and van Koningsveld, R. (2000) Healthburden in
the Netherlands (1990-1995) due to infections with thermophilic
Campylobacter species. Report no. 284550 004. RIVM (National Institute
of Public Health and the Environment), Bilthoven.
Havelaar, A.H., Schets, F.M., van Silfhout, A., Jansen, W.H., Wieten, G. and van
der Kooij, D. (1992) Typing of Aeromonas strains from patients with
diarrhoea and from drinking water. Journal of Applied Bacteriology 72(5),
435-444.
Hijnen, W.A.M., Willemsen-Zwaagstra, J., Hiemstra, P., Medema, G.J. and van
der Kooij, D. (2000) Removal of sulphite-reducing clostridia spores by
full-scale water treatment processes as a surrogate for protozoan (oo)cysts
removal. Water Science and Technology 41(7), 165-171.
Holcomb, D.L., Smith, M.A., Ware, M.A., Hung, Y.-C., Brackett, R.E. and
Doyle, M.P. (1999) Comparison of six dose-response models for use with
food-borne pathogens. Risk Analysis 19, 1091-1100.
Hunter, P.R. (2000) Modelling the impact of prior immunity, case
misclassification and bias on case-control studies in the investigation of
outbreaks of cryptosporidiosis. Epidemiology and Infection 125, 713-718.
Hunter, P.R. and Fewtrell, L. (2001) Acceptable risk. In: Water Quality:
Guidelines, Standards and Health. Assessment of risk and risk
management for water-related infectious disease. Fewtrell, L. and
Bartram, J. (Eds.) IWA Publishing, London. pp.207-227.
Hunter, P.R. and Syed, Q. (2001) Community surveys of self-reported diarrhoea
can dramatically overestimate the size of outbreaks of waterborne
cryptosporidiosis. Water Science and Technology 43(12), 27-30.
Isacson, M. and Sayed, A.R. (1988) Health aspects of the use of recycled water
in Windhoek, SWA/Namibia, 1974-1983. Diarrhoeal diseases and the
consumption of reclaimed water. South Africa Medical Journal 7, 596-9.
Kang, S.H., Kodell, R.L. and Chen, J.J. (2000) Incorporating model
uncertainties along with data uncertainties in microbial risk assessment.
Regulatory Toxicology and Pharmacology 32(1), 68-72.

105

Khn, I., Albert, M.J., Ansaruzzaman, M., Bhuiyan, N.A., Alabi, S.A., Islam,
M.S., Neogi, P.K.B., Huys, G., Janssen, P., Kersters, K. and Mllby, R.
(1997) Characterization of Aeromonas spp. isolated from humans with
diarrhea, from healthy controls, and from surface water in Bangladesh.
Journal of Clinical Microbiology 35(2), 369-373.
LeChevallier, M.W., Abbaszadegan, M., Camper, A.K., Hurst, C.J., Izaguirre, G.,
Marshall, M.M., Nauovitz, D., Payment, P., Rice, E.W., Rose, J., Schaub,
S., Slifko, T.R., Smith, D.B., Smith, H.V., Sterling, C.R. and Stewart, M.
(1999a) Committee report: emerging pathogens - bacteria. Journal of the
American Water Works Association 91(9), 101-109.
LeChevallier, M.W., Abbaszadegan, M., Camper, A.K., Hurst, C.J., Izaguirre, G.,
Marshall, M.M., Nauovitz, D., Payment, P., Rice, E.W., Rose, J., Schaub,
S., Slifko, T.R., Smith, D.B., Smith, H.V., Sterling, C.R. and Stewart, M.
(1999b) Committee report: emerging pathogens - viruses, protozoa, and
algal toxins. Journal of the American Water Works Association 91(9), 110121.
Leeming, R., Bate, N., Hewlett, R. and Nichols, P.D. (1998) Discriminating
faecal pollution: a case study of stormwater entering Port Philip Bay,
Australia. Water Science and Technology 38(10), 15-22.
Levy, D.A., Bens, M.S., Craun, G.F., Calderon, R.L. and Herwaldt, B.L. (1998)
Surveillance for waterborne-disease outbreaks -- United States, 1995-1996.
Morbidity and Mortality Weekly Report 47(SS-5), 1-34.
Macgill, S., Fewtrell, L., Chudley, J. and Kay, D. (2001) Quality audit and the
assessment of waterborne risk. In: Water Quality: Guidelines, Standards
and Health. Assessment of risk and risk management for water-related
infectious disease. Fewtrell, L. and Bartram, J. (Eds.) IWA Publishing,
London. pp. 185-206.
MacKenzie, W.R., Hoxie, N.J., Proctor, M.E., Gradus, M.S., Blair, K.A.,
Peterson, D.E., Kazmierczak, J.J., Addiss, D.G., Fox, K.R., Rose, J.B.
and Davis, J.P. (1994) A massive outbreak in Milwaukee of
Cryptosporidium infection transmitted through the public water supply.
New England Journal of Medicine 331, 161-167.
Maki-Ikola, O. and Granfors, K. (1992) Salmonella-triggered reactive arthritis.
Lancet 339, 1096-1098.

106

Medema, G.J., Bahar, M. and Schets, F.M. (1997) Survival of Cryptosporidium

parvum, Escherichia coli, faecal enterococci and Clostridium perfringens
in river water - influence of temperature and autochthonous
microorganisms. Water Science and Technology 35(11-12), 249-252.
Morris, R.D., Naumova, E.N., Levin, R. and Munasinghe, R.L. (1996)
Temporal variation in drinking water turbidity and diagnosed gastroenteritis in Milwaukee. American Journal of Public Health 86, 237-239.
National Research Council (1983) Risk Assessment in the Federal Government:
Managing the Process. National Academy Press, Washington, DC.
Ndiongue, S., Desjardins, R. and Prvost, M. (2000) Relationships between total
particle count, aerobic spore-forming bacteria and turbidity in direct
filtration. Aqua 49(2), 75-87.
Niklasson, B., Hrnfeldt, B. and Lundman, B. (1998) Could myocarditis, insulindependent diabetes mellitus and Guillain-Barr syndrome be caused by one
or more infectious agents carried by rodents? Emerging Infectious Diseases
4(2), 187-193.
Okhuysen, P.C., Chappell, C.L., Crabb, J.H., Sterling, C.R. and DuPont, H.L.
(1999) Virulence of three distinct Cryptosporidium parvum isolates for
healthy adults. Journal of Infectious Diseases 180(4), 1275-1281.
Owens, J.H., Miltner, R.J., Rice, E.W., Johnson, C.H., Dahling, D.R., Schaefer,
F.W. and Shukairy, H.M. (2000) Pilot-scale ozone inactivation of
Cryptosporidium and other microorganisms in natural water. Ozone-Sci.
Eng. 22(5), 501-517.
Payment, P., Richardson, L., Siemiatycki, J., Dewar, R., Edwardes, M. and
Franco, E. (1991) A randomized trial to evaluate the risk of
gastrointestinal disease due to the consumption of drinking water meeting
currently accepted microbiological standards. American Journal of Public
Health 81, 703-708.
Payment, P., Siemiatycki, J., Richardson, L., Renaud, G., Franco, E. and
Prvost, M. (1997) A prospective epidemiological study of
gastrointestinal health effects due to the consumption of drinking water.
International Journal of Environmental Health Research 7, 5-31.
Percival, S., Walker, J. and Hunter, P.R. (2000) Microbiological Aspects of
Biofilms in Drinking Water. CRC Press, Boca Raton.

107

Perz, J.F., Ennever, F.K. and Leblancq, S.M. (1998) Cryptosporidium in tap water
- comparison of predicted risks with observed levels of disease. American
Journal of Epidemiology 147, 289-301.
Prss, A., Kay, D., Fewtrell, L. and Bartram, J. (2002) Estimating the burden of
disease due to water, sanitation and hygiene at global level. Environmental
Health Perspectives in press.
Quick, R.E., Venczel, L.V., Mintz, E.D., Soleto, L., Aparicio, J., Gironaz, M.,
Hutwagner, L., Greene, K., Bopp, C., Maloney, K., Chavez, D., Sobsey, M.
and Tauxe, R.V. (1999) Diarrhoea prevention in Bolivia through point-ofuse water treatment and safe storage: a promising new strategy.
Epidemiology and Infection 122, 83-90.
Roseberry, A.M. and Burmaster, D.E. (1992) Lognormal distributions for water
intake by children and adults. Risk Analysis 12(1), 99-104.
Schwartz, J., Levin, R. and Goldstein, R. (2000) Drinking water turbidity and
gastrointestinal illness in the elderly of Philadelphia. Journal of
Epidemiology and Community Health 54(1), 45-51.
Schwartz, J., Levin, R. and Hodge, K. (1997) Drinking water turbidity and
pediatric hospital use for gastrointestinal illness in Philadelphia.
Epidemiology 8, 615-620.
Semenza, J.C., Roberts, L., Henderson, A., Bogan, J. and Rubin, C.H. (1998)
Water distribution system and diarrhoeal disease transmission: a case
study in Uzbekistan. American Journal of Tropical Medicine and
Hygiene 59(6), 941-946.
Stevens, M., McConnell, S., Nadebaum, P.R., Chapman, M., Ananthakumar, S.
and McNeil, J. (1995) Drinking water quality and treatment requirements:
A risk-based approach. Water 22, 12-16.
Teunis, P.F.M., van der Heijden, O.G., van der Giessen, J.W.B. and Havelaar,
A.H. (1996) The dose response relation in human volunteers for gastrointestinal pathogens. Technical report 284 550 002, RIVM (National
Institute of Public Health and the Environment), Bilthoven.
Walter, S. (1991) The ecological method in the study of environmental health.
II. Methodologic issues and feasibility. Environmental Health
Perspectives 94, 67-73.

108

Wheeler, J.G., Sethi, D., Cowden, J.M., Wall, P.G., Rodrigues, L.C., Tompkins,
D.S., Hudson, M.J., Roderick, P.J. on behalf of the Infectious Intestinal
Disease Study Executive (1999) Study of infectious intestinal disease in
England: rates in the community, presenting to general practice, and
reporting to national surveillance. British Medical Journal 318, 10461050.
WHO (1993) Guidelines for Drinking water Quality. Volume 1:
Recommendations. Second Edition. World Health Organisation, Geneva.
WHO (1996) Guidelines for Drinking water Quality. Volume 2: Health criteria
and other supporting information. Second Edition. World Health
Organisation, Geneva.
WHO (1997) Guidelines for Drinking water Quality. Volume 3: Surveillance
and control of community supplies. Second Edition. World Health
Organisation, Geneva.
Zmirou, D., Ferley, J.P., Collin, J.F., Charrel, M. and Berlin, J. (1987) A
follow-up study of gastro-intestinal diseases related to bacteriologically
substandard drinking water. American Journal of Public Health 77, 582584.
Zmirou, D., Rey, S., Courtois, X., Ferley, J.P., Blatier, J.F., Chevallier, P.,
Boudot, J., Potelon, J.L. and Mounir, R. (1995) Residual microbiological
risk after simple chlorine treatment of drinking ground water in small
community systems. European Journal of Public Health 5, 75-81.

109