Chapter 6

MONITORING THE QUALITY OF DRINKING WATER DURING

STORAGE AND DISTRIBUTION

W. Robertson, G. Stanfield, G. Howard and J. Bartram

6.1 Introduction

Following abstraction and treatment water becomes a vulnerable and

perishable product. It is vulnerable in that the integrity of systems used for the
storage and distribution of water can be damaged and contamination through
ingress can occur. It is perishable in that its microbial quality can deteriorate
due to the bacteria remaining after treatment growing on the residual nutrient in
the water. Water can therefore be regarded as having a finite life.

The rate at which quality deteriorates can be controlled by adding a

preservative (disinfectant residual), or by the use of advanced treatment to
remove as much biodegradable organic carbon as possible from the water.
However, these measures will be to no avail, if the distribution or storage-
systems are contaminated or permit ingress. To get the maximum degree of
protection the product (water) and the packaging (tanks and pipes) must be
clean and intact.

The importance of quality changes in distribution is based upon evidence

concerning the frequency and extent of known quality changes and their impact
upon human health. Thus, for example, a significant proportion of recognised
piped drinking water-related disease outbreaks are related to quality
deterioration in distribution (Ainsworth, 2002; Craun and Calderon, 2001).
However, most water supply world-wide is unreliable because of, for example,
intermittence and so household management is common. This may involve only
plumbed in household storage tanks or alternatives, which involve extensive

179
manual handling. Because the contamination that arises is local in character it is
unlikely to give rise to detectable outbreaks of disease but significant evidence
exists that quality changes in such circumstances may be extreme and respond
to the extent of handling (Quick et al., 1999).

6.2 Piped distribution systems

Piped distribution systems for drinking water are as important to the

quality and safety of drinking water as the treatment itself. Water entering the
distribution system must be microbiologically safe and ideally should be
biologically stable. The distribution system itself must provide a secure barrier
to post-treatment contamination as the water is transported to the user. Residual
disinfection will provide partial protection against recontamination, but may
also act to mask the presence of such contamination. On a global scale,
however, mishandling within the home is likely to be the most significant
source of fouling. This section describes the sources of contamination
progressing from treated water leaving the treatment plant, to deterioration of
water quality during storage and distribution. Non-piped systems are dealt with
in a separate section.

6.2.1 Inadequately treated water entering the distribution system

Sporadic occurrences of source water of high turbidity that overwhelm the

treatment train, or microorganism breakthrough for example resulting from sub-
optimal filtration following filter backwashing, can introduce enteric pathogens
into the distribution system. These may be in sufficient numbers to cause
detectable cases of gastrointestinal illness in the population served (MacKenzie
et al., 1994). It should be recalled that the frequency of such events in
developed and developing countries is likely to be far greater than that implied
by the number of outbreaks presently detected (see Chapter 7). Such
occurrences are often rainfall driven and may be inter-related. For
example exceptionally high pathogen loadings may arise from surface runoff
following rainfall at the same time that overall filter efficiency decreases and
the need for backwashing increases because of the concurrent high turbidity
loadings (see Chapters 4 and 5).

180
6.2.2 Integrity of the distribution system

Disease outbreaks have been traced to cross connections in spite of the fact
that water leaving the plant was deemed to be safe. In general, there are four
types of cross connections:

• Infiltration. In this situation contaminated sub-surface water is drawn

into the distribution system. In order for this to happen three conditions
must be in place. First, contaminated water must be present in the sub-
surface material surrounding the distribution system, possibly from a
leaking sanitary, storm or combined sewer. Secondly, there must be an
adjacent low-pressure zone within the system. These zones can arise
through high water usage resulting from fire fighting or other peak
demands, decreased flow arising from restrictions in the system, pump
failures or intermittent operation of the treatment plant. More recent
evidence points to the role of pressure surges, in otherwise properly
operated systems, in creating transient low pressure that may lead to the
ingress of contaminants (LeChevallier, 1999). Thirdly, there must be a
route for contaminated water to enter the system. This can occur through
pinholes caused by corrosion, cracks or outright breaks or leaking joints in
the wall of the mains. If these three situations occur simultaneously then it
is likely that contaminated water will enter the distribution system.
Leakage rates are typically high, with even well operated systems
experiencing rates of 10 – 20% (LeChevallier, 1999; WHO and UNICEF,
2000). This is likely to be associated with a relatively large number of
points of leakage and therefore present an increased risk of intrusion of
pathogenic organisms.
• Back siphonage. In this situation faecally contaminated surface water is
drawn into the distribution system or storage reservoir through a back
flow mechanism. In order for this to take place two conditions must occur
simultaneously. Firstly, there must be a reduction in line pressure as
described above. Secondly, there must be a physical link between
contaminated water and the storage and distribution system. Open taps
connected to hoses that are submerged in pools of water may provide this
link. Back flow preventers are available to stop such occurrences.
Plumbing codes should require air gaps between taps and the brim of
receptacles. The general conclusion stresses the importance of good
domestic plumbing practice.
• Open drinking water storage reservoirs. Microbial contamination can
also be introduced into the distribution system through open treated-water
storage reservoirs (Geldreich, 1996). For example, waterborne outbreaks

181
 have occurred in communities where birds contaminated the water either
because the reservoir was uncovered or because they gained access to the
reservoir through unscreened roof vents. Uncovered reservoirs can also
permit the growth of toxin-forming cyanobacteria.
• Line construction and repair. When existing mains are repaired or
replaced or when new water mains are installed strict protocols involving
disinfection and flushing must be followed to prevent the introduction of
contaminated soil or debris into the system (e.g. AWWA, 1986). These
protocols generally cover six areas of concern:
1. Protection of pipe sections at the site.

2. Restriction on the types of joint sealing materials used.

3. Preliminary flushing of pipe sections.

4. Disinfection of pipe sections.

5. Final flushing of pipe sections.

6. Bacteriological testing to confirm disinfection.

If these procedures are not adhered to pathogens may gain access to the
system. Construction materials, such as wood embedded in pipe sections, have
also been identified as a source of microbial contamination and provide an
adequate supply of nutrients to support bacterial re-growth (Martin et al., 1982).

In each scenario, if the contaminated water contains enteric pathogens then

it is likely that consumers will be exposed to them. Even where disinfectant
residuals are employed to limit microbial occurrence it may be inadequate to
overcome the contamination or may be ineffective against some or all of the
pathogen types introduced. As a result, pathogens may occur in concentrations
that could lead to infection and illness. In many developing countries drinking
water is supplied intermittently either as a cost-saving measure or because of
water shortages. Under such conditions the resulting low water pressure will
allow the ingress of contaminated water into the system through breaks, cracks,
joints and pinholes in the walls of the system.

In the context of monitoring it should be noted that the use of residual

disinfectants to control such problems may lead to an inaccurate estimation of
their occurrence. This is because the microbial indicator parameters most
frequently used for monitoring are among the most sensitive of the organisms of
interest, leading to a potential situation where the indicator parameter is absent

182
but pathogens may be present. Thus, neither the occurrence nor the severity of
contamination will be properly identified.

6.2.3 Microbial re-growth in the distribution system

Even though treated drinking water may be free of faecal indicator

organisms and detectable enteric pathogens and therefore present a tolerable
level of risk of enteric illness, drinking water entering the distribution system
may contain free-living amoebae and environmental strains of various bacterial
species, often referred to as heterotrophic bacteria. Under favourable conditions
amoebae and heterotrophs will colonise a distribution system and form biofilms.

Many environmental strains of coliform bacteria such as Citrobacter,

Enterobacter and Klebsiella may also colonise distribution systems (Martin et
al., 1982). However, it is generally agreed that water temperatures and nutrient
concentrations are not elevated enough to support the growth of E. coli (or
enteric pathogenic bacteria) in biofilms (Geldreich and LeChevallier, 1999).
Thus the presence of E. coli should be considered as evidence of recent faecal
contamination of drinking water.

Many species of bacteria and free-living amoebae can occur in biofilms

within distribution systems including certain opportunistic pathogens. There is
inadequate evidence at present to implicate the occurrence of these
microorganisms from biofilms (excepting for example, Legionella or
Mycobacterium) with adverse health effects in humans with the possible
exception of immuno-compromised population groups (reviewed by Geldreich,
1996).

6.3 Non-piped distribution systems

Most of the population of the globe obtain their water through non-piped
systems and of the remainder the majority are supplied through systems that
require some form of storage and/or handling before use, thereby increasing the
possibility of contamination. Even where a reliable piped supply is the norm,
occasional interruptions occur (e.g. extreme weather events that cause major
line breaks and force adoption of household handling). From a public health
viewpoint it is therefore essential to respond to contamination that occurs
through the chain of supply up to the point of use and to consider all forms of
water supply used by the population.

183
 Point sources of water such as tubewells, dug wells and protected springs
represent a very significant proportion of the ’improved’ water supplies provided
to communities in developing countries (WHO and UNICEF, 2000). Such
supplies are very common in rural areas and may also represent a very
significant proportion of the water supplies available and used for domestic
purposes (including drinking) by low-income urban populations (WHO and
UNICEF, 2000; Howard et al., 1999; Ahmed and Hossain, 1997). The quality
of such sources is often very variable and they frequently show gross faecal
contamination, particularly during wet seasons (Wright, 1986; Barrett et al.,
2000). The public health consequences of consumption of contaminated water
from point sources can be severe both in relation to endemic and epidemic
disease (Pedley and Howard, 1997). The control of the quality of drinking water
from such sources is important in reducing public health risks, but requires
greater emphasis on support to community management in order to improve
operation and maintenance, and also significant user education.

As point water sources are often found in areas that also rely on on-site
sanitation, sub-surface leaching is frequently identified as being a principal
source of contamination (Melian et al., 1999; Rahman, 1996). However, there is
increasing evidence that deterioration in sanitary completion measures and
preferential flow paths are more important in causing microbial contamination
(Howard et al., 2001a; Gelinas et al., 1996; Rojas et al., 1995). Different
technologies have different vulnerability to contamination depending, largely,
on the depth of abstraction (ARGOSS, 2001).

Many people world-wide rely on water supplied by tankers or other forms

of vendor supplies (Whittington et al., 1991). The water in such supplies may
come from hydrants connected to utility supplies or may be drawn from
alternative sources. In many cases, the consumer will not be aware of the source
of the water and there may be significant concerns about the quality of water
(Lloyd et al., 1991). Where the water is supplied from hydrants via utility piped
water supplies, basic good hygiene practices, such as regular cleaning and
disinfection of the tank and sanitary storage of the connecting hoses is usually
adequate to maintain quality. Studies in Ghana, showed that the quality of water
in tanker trucks collecting water from utility hydrants was the same as the piped
water in Kumasi and only marginally less good in the capital Accra (Jabulo,
pers. comm). In the Ghana study and also studies in Uganda (Howard and
Luyima, 2000), however, it has been shown that the quality of water supplied
by other vendors, such as those selling from jerry cans or small fixed tanks, is
much worse and represents highly degraded water quality.

184
 In non-piped systems drinking water is typically either carried to the home
or is transported by truck or cart. The contamination of the water, usually as a
result of poor hygiene is common. Source contamination can then be augmented
at each point of handling through collection and transport (VanDerslice and
Briscoe, 1993). However, because such contamination is by definition local in
nature, it is unlikely to lead to large-scale single source outbreaks and therefore
unlikely to be detected or reasonably estimated.

Poor hygiene in the home is another, potentially significant source of

drinking water contamination. In many cases, contamination significantly
increases from source to household (WHO, 1997). This is of particular concern
in communities without reliable piped water to within dwellings which,
therefore, rely upon water storage containers. Evidence is accumulating for both
the impact of such contamination on human health and for the effectiveness of
interventions at this level in protecting and improving human health (Semenza
et al., 1998). Such interventions include the use of household water disinfection
technologies, including chlorination and solar disinfection (see Chapter 5;
Sobsey, 2002). They also include hygiene education interventions geared to the
maintenance of water quality during transport and storage. Such interventions
may include the testing of water and participatory approaches to health
education, which, have proven successful in both rural and urban areas (Breslin,
2000).

The selection of whether water treatment or hygiene education

interventions are most appropriate depends in part on the quality of the source
water. If it is good, then greater effort may be placed on promoting safe
handling, whereas poor source water quality greater effort may be placed on
household treatment. Where household disinfection is promoted, it is essential
that water quality programmes provide educational information on disinfecting
drinking water in the home and maintaining the quality and safety of the water
following disinfection. There is considerable evidence to suggest that this
approach is effective at reducing contamination and the spread of enteric illness
(Sobsey, 2002; Forget and Sanchez-Bain, 1999; Quick et al., 1999; Reiff et al.,
1996).

6.4 Approaches to sampling in piped distribution systems

Strategies for monitoring the microbiological quality of water in supply

must be designed to allow the best possible chance of detecting reductions in
microbial quality (Ainsworth et al., 2002). It is vital, therefore, to ensure that
the samples taken are representative of the quality in the distribution network.

185
In formulating a strategy the factors outlined in the following subsections need
to be considered (ISO, 1991).

6.4.1 Choice of indicator parameter

Traditionally, microbial indicators of faecal pollution such as total

coliforms and E. coli have been the determinants used. Most national and
international microbial standards for water in supply are expressed in terms of
these organisms. Measurement of the numbers of heterotrophic bacteria provide
further valuable, but often under-utilised, additional information. Secondary
microbial parameters, for example, enterococci and clostridia, being more
resistant to disinfection, may also be used to try and gain better insight into the
source of the contamination or where particular sources of contamination are
suspected or known to cause problems. These secondary parameters are most
often used, therefore, as part of investigations where a failure of quality, in
terms of the detection of coliform bacteria, has been detected. The practice of
application of a residual disinfectant such as chlorine has a significant impact
upon the adequacy of alternative indicators as outlined above. Since the most
common microbial indicators of faecal pollution (E. coli or thermotolerant
coliforms and total coliforms) are very sensitive to chlorine their detection
implies the likelihood of either recent or substantial faecal contamination with
attendant health risks. Nevertheless, because coliform bacteria are so sensitive
to chlorine their absence provides no guarantee that pathogens such as enteric
protozoa and viruses, which are more resistant to chlorine, are also absent.

In any faecal contamination event microorganisms, whether introduced by

inadequate treatment or post-treatment contamination, are not distributed evenly
throughout the distribution system but are typically clumped (Gale, 1996a). In
these situations the probability of detecting faecal indicator bacteria in the
relatively few samples collected from the distribution system during routine
sampling is substantially reduced. Hence quantitative measurements of
contamination, such as membrane filtration (MF) or most probable number
(MPN) methods, may provide a poor estimate of the overall density of indicator
bacteria. Frequency-of-occurrence monitoring, based on the Presence/Absence
(P/A) test, can provide a better estimate of water quality in general (Pipes and
Christian, 1984; Clark, 1980). In this method the frequency of positive samples
detected during routine sampling during a reporting period is compared with
acceptable frequencies of positive samples specified in the applicable water
quality standards. Where such standards are based upon E. coli or
thermotolerant coliforms, their confirmed presence usually leads to an
immediate boil water advisory and corrective actions. Standards based upon the
presence of total coliforms specify an acceptable frequency of positive samples

186
(for example 5%) before resampling and possible corrective actions are
required. Comparative studies of the P/A and MF methods demonstrate that the
P/A method can maximise the detection of faecal indicator bacteria (Clark,
1990; Geldreich, 1996). It also allows more samples to be analysed within a
reporting period because the test is simpler, faster and less expensive than the
quantitative methods noted above (see Chapter 2.2.9). Commercial P/A kits for
faecal indicator bacteria are available.

More conservative microbial parameters (i.e. parameters of quality

deterioration that may be detected before actual faecal contamination occurs)
are logically preferred but most present practical/logistical problems. The most
frequently used alternative for distribution systems is presently the
heterotrophic plate count. Whilst taxonomically imprecise this is relatively
widely available and applicable. Its value is principally associated with changes
and trends in counts retrieved from a given system rather than comparison with
numeric reference values and it therefore requires relatively dense sampling
networks (in time and space) to provide useful information. Measuring the
numbers of heterotrophic bacteria present in a supply can be a useful indicator
of changes such as increased re-growth potential, increased biofilm activity,
extended retention times or stagnation and a breakdown of integrity of the
system. Following trends in the numbers of heterotrophs may therefore be
useful in prompting, for example, a programme of mains flushing or cleaning.

Non-microbial parameters may also be suitable for this purpose and also
require comparison on changes and trends and therefore relatively dense
sampling networks. The most frequently used examples include
conductivity/total dissolved solids, turbidity and chlorine residual (where
chlorine is applied). Where chlorine residuals are used, measurement of this
residual can often be a more rapid indication of problems than microbial
parameters. A sudden disappearance of an otherwise stable residual can indicate
ingress of contamination with a high organic loading. Alternatively, difficulties
in maintaining residuals at points in a distribution system or a gradual
disappearance of residual may indicate that the water or pipe work has a high
oxidant demand due to re-growth of bacteria in the water or biofilm growth.

Routine monitoring of distributed water for particular groups of bacteria or

specific pathogens is rarely considered worthwhile or necessary. However, in
the Netherlands guidelines for the numbers of aeromonads in final waters and in
distribution have been issued to act as an indication that better maintenance of
sand filters or that better removal of methane from anaerobic groundwater
sources is required. In this context aeromonads are acting as a sensitive
indicator of the potential for re-growth to occur within the network (see
Chapter 2.2.8).

187
6.4.2 Location of sampling points

The two principal considerations in the location of sampling points are

whether fixed or variable locations (or a mixture of these) are to be employed,
and whether sampling is to occur from normal access points (such as consumer
taps), from dedicated but otherwise normal locations or from structures
dedicated exclusively for this purpose. The purpose of monitoring and
especially whether in response to statutory requirements or for public health
investigatory purposes will have a significant impact on the choices made.

Sampling locations should be chosen to provide a means of characterising

water quality in all parts of the system. For this reason a supply system may be
divided into a series of zones on the basis of geographical area, the size of the
population served or specific areas of the pipe network. Sampling points are
then identified within each zone to try and ensure that representative samples
are taken. Sampling points will be chosen for two reasons. One will be to satisfy
statutory responsibilities and the other for strategic or operational purposes. In
the latter, a water supplier may for example be trying to obtain more
information about an area that has, in the past, yielded frequent coliform failures
or elevated counts of heterotrophs.

Fixed sampling points are frequently used, and may be chosen because of
ease of access. Often these points are located within public buildings, or in the
premises of public services, such as fire-stations. The use of fixed sampling
points alone can be regarded as unsatisfactory, since they may not give a
representative view of what is happening in all parts of the distribution system
or zones. To overcome this, additional samples may be taken from other outlets
chosen at random and these usually include the mains tap in the houses of
consumers.

The use of fixed or random sampling points is a topic of much continuing

debate. Geldreich (1996) reported that in a survey of 1 796 water supply utilities
in the USA, about one third used only fixed sampling points, with 50% using a
combination of random and fixed points. He recommended that some sampling
locations be varied, so that all sections of the network are monitored over time.
Burlingame and O’Donnell (1993) argue that the use of random taps increases
the detection of water contaminated within households (which is outside the
regulatory framework), since the water supplier has no control over consumers’
premises. Similarly Dufresne et al. (1997) have demonstrated that the number
of positives and the number of samples collected can be reduced by selecting
sampling points that were protected and well maintained. Similarly, a task
group, appointed by United Kingdom Water Industry (1995), concluded that
there were advantages in adopting a wholly fixed-point system. These included

188
the control and maintenance of the taps, greater compatibility of data and
reduced cost. Other studies have indicated that the type of tap, particularly
mixer taps, and material of construction may also influence water quality. These
factors need to be considered in the design of a monitoring strategy. If random
taps are to be used, it may be that identification of premises with suitable taps
may be a more important consideration than achieving true randomness (Anon,
1994).

For statutory purposes, the location and number of sampling points may be
stipulated within the appropriate regulations of that country. For example, the
UK Water Supply Regulations stipulate that at least 50% of distribution samples
must be taken from random locations (HMSO, 1989). For strategic monitoring,
locations will be selected to gain the best practicable information about the area
of the system being investigated. For this purpose, depending on the laboratory
resources available, sampling locations may be more numerous than required to
satisfy the regulations. Regardless of the reason for the monitoring it will have
the common objective of providing sufficient information to allow the water
quality in all service areas to be characterised.

6.4.3 Frequency of sampling

There is no absolute rationale for the frequency of sampling and adopted

frequencies reflect the need to balance the benefits of better information arising
from greater frequency with increasing costs (and decreasing returns) as overall
frequency increases. The guidelines and regulations of individual countries
generally prescribe the minimum sampling frequency to be achieved to meet
statutory requirements. These sampling frequencies are usually based on the
population served by the network/zone or, less frequently, on the volume of
water supplied. For example in France, a minimum of 24 samples each year
must be taken from distribution systems serving 10 000 – 20 000 inhabitants. In
the Netherlands the frequency is similar, while the UK Regulations prescribe a
more intensive programme of 48 samples per annum. In Germany the
prescribed frequency is similar to that set for treatment works at one sample per
15 000 cubic metres of water supplied.

Although jurisdictions prescribe minimum sampling frequencies it is often

necessary to collect additional samples to improve the overall picture of water
quality in the distribution system. Supplementary samples should be collected at
locations which, from historical data, are known to experience problems.
Sampling frequency should also be increased following remedial actions, for
example in response to boil water orders, or following interruptions in supply.
In most cases a failure in terms of E. coli or thermotolerant coliform detection

189
will initiate re-sampling, which may be followed by intensive investigations to
identify the source and extent of the contamination. While re-sampling is
commonly required the rationale underlying this (as a procedure prior to further
investigation and action) is unclear. Given the known temporal and spatial
variability of microbial water quality it is logical that re-sampling would often
fail to detect continued contamination without indicating that the cause of the
original contamination had been limited or controlled.

Some water suppliers carry out much more sampling within the
distribution network than required to satisfy the regulations. Strategic
monitoring programmes will be carried out to investigate specific problems
such as high incidence of taste and odour events or the need for mains
replacement or renovation. Each distribution system will have its own unique
monitoring needs and the monitoring programme should be designed to address
these using the available analytical resources. The impact of the timing of
sampling should be recalled in interpreting results. Thus, for example, most
sampling will be undertaken within the normal working week and may not
detect changes arising from abnormal patterns of demand (such as football cup
finals).

6.4.4 Volume of sample

The volume of sample collected must be sufficient to allow analysis at the

limits of detection stipulated in regulations. For coliform bacteria this is 100 ml.
Heterotrophic plate counts are usually expressed on a per ml basis. Given that
samples may have to be re-analysed, the volume of sample will therefore
generally be between 250 and 500 ml. In the Netherlands, coliform and E. coli
counts are expressed per 250 ml and so sample volumes will be greater to
satisfy these requirements.

Demonstrating the absence of indicator microorganisms in 100 ml

volumes, although adequate for statutory requirements, does not provide
information about how close to compliance water quality is. For this reason
some water suppliers routinely, or as part of special investigative studies, may
collect and analyse sample volumes of between 1 and 10 litres. Volumes of this
magnitude can be analysed by small modifications to existing analytical
methods or by multiple analyses of 100 ml to 1 l volumes.

If microorganisms were distributed randomly in distribution systems, the

use of the Poisson distribution would be appropriate in calculating the
confidence limits for occurrence of indicator bacteria (Haas and Heller, 1986).
However, it has been suggested by several workers (e.g. Pipes et al., 1977;

190
Gale, 1996a) that the distribution of bacteria tends not to fit this pattern, instead
they occur in clusters within a bulk supply. Gale (1996b) analysed the statistical
distribution of confirmed coliform densities and total heterotrophic bacteria
densities, using monitoring data from eight UK water companies. It was found
that the log value of the bacterial density varied in an approximate linear
fashion when plotted against the percentage of the data that was less than this
value: this is known as a log-normal distribution. Extrapolation of the log-
normal distribution of confirmed coliforms (1-9% of all samples) implied that
where 100 ml samples were recorded as 0/100 ml, the actual concentration may
have been as low as < 1/107 litres or as high as 1/litre. Similar clustering of a
small proportion of high counts could be caused by sporadic contamination of
the distribution system through, say, floc carryover and filter breakthrough
(Gale, 1996b). Gale (1996b) concludes that ideally a model of the risk of
exposure to pathogens would take account of the density of the organism in raw
water, the removal efficiency by subsequent treatment processes, the variation
in tendency to cluster and an estimate of water consumption.

In addition to providing some indication as to the true numbers of

microbial indicators of faecal pollution in distribution, analysis of large volume
samples increase the chance of detecting clusters or pulses of bacterial
contamination particularly if taken over a period of time such as 30-60 minutes.
An alternative to large volume grab samples is the collection of a composite
sample. This is composed of a series of small volume samples collected at time
intervals that are bulked together to form one large sample. Depending on
available analytical resources analysis of individual samples would provide an
indication of the bacterial distribution. It can be difficult to arrange for large
volume or composite samples to be collected at randomly selected locations
such as consumers’ taps, and it is best restricted to fixed sampling points.

6.4.5 Sample collection

Sampling for microbiological analyses requires care, with observation of

the general principles of aseptic technique. All equipment used for sampling
should be thoroughly cleaned and preferably sterilised before use. The sample
containers should be sterile with wide mouths that are shielded from
contamination. They should contain sufficient sterile reagent to neutralise any
residual disinfectant in the water and samplers should be trained to ensure that
thorough mixing occurs immediately after sampling.

For some purposes and especially in relation to the performance of the

supply agency the sampling procedure used has to ensure that water from the
main is collected and not the stagnant water in the supply pipe. For other

191
purposes (especially in relation to public health investigation) interest may
focus on the water that has gone through typical processes of deterioration, as
this is what consumers will be exposed to. Collecting mains water is usually
achieved by running water to waste for a period of two-three minutes (Anon,
1994; APHA, AWWA, WEF, 1998), but given the differing lengths of service
pipes this may not always be adequate. Other methods, such as running to waste
until a constant temperature is reached (Anon, 1994), particle counts have
stabilised (Burlingame and Choi, 1998) or (if appropriate) a chlorine residual is
detected would appear to be more certain, but rather time consuming. After
reviewing the results of studies on flushing Prévost et al. (1997) suggest a
period of five-ten minutes may be necessary. However, the consumers whose
taps are being used do not like these methods because they are perceived as
being wasteful (or costly, if on a metered supply). The importance of this step in
the sampling procedure cannot be over emphasised. Since water in the service
pipe may reach relatively high temperatures, heterotrophs may grow to higher
densities than in the water main (Geldreich, 1996). In addition, the high surface
area to volume ratio of service pipes encourages the decay of chlorine residuals,
again allowing a greater potential for bacterial growth. Without an adequate
period of flushing a representative sample will not be obtained. However, not all
workers agree with this. Prevost et al. (1997) demonstrated that although
numbers of bacteria increased with distance from the treatment works,
stagnation within the service pipe did not influence numbers appreciably.
Similarly Kerneis et al. (1995) found that residence time had little influence on
the heterotrophic plate count.

The introduction of new designs of taps (mixer), the use of plastics in

construction and the provision of inserts as anti-splash devices has caused much
discussion on the need for, and the best method of, sterilising taps prior to
sampling. The traditional method of sterilisation is to heat the outside of the tap,
in the direction of spout to base. Although this method is satisfactory where
metal taps are used at premises controlled by the water supplier, they may be
considered too hazardous at consumers premises. For sampling at these
locations heat has almost completely been replaced by chemical methods of
disinfection. Sodium hypochlorite is the most widely recommended chemical
disinfectant. It is effective against a wide range of bacteria, including spores,
and viruses. Only fresh solutions should be used as the concentration of chlorine
decays with time. The consensus opinion on concentration is to use a 10%
solution of household sodium hypochlorite for swabbing, or a 1% solution
where a spray is used. It is essential that tap surfaces are clean before the use of
a chlorine-based disinfectant since they react readily with all types of organic
matter. In some cases isopropanol is used for tap disinfection as it is effective
against some of the more resistant bacteria, such as mycobacteria, it is water
miscible and evaporates at room temperature, leaving no residue. No particular

192
precautions other than simple tap flushing, have to be taken to ensure that the
disinfectant does not contaminate the sample. In common with most chemical
disinfectants, it precipitates proteins that may then form a protective layer for
bacteria. The tap body and spout must therefore be clean before application.
Isopropanol has the disadvantages that prolonged contact may result in skin
irritation, swelling of rubber and hardening of some plastics. Burlingame and
Choi (1998) suggest that the concentration used should be at least 70%, but a
slightly lower concentration 60-70% is said by Gardner and Peel (1986) to be
more effective.

6.5 Approaches to sampling in non-piped systems

In many areas where communities rely on non-piped supplies, resources to

carry out routine bacteriological monitoring are very limited. This is often
further complicated as most small water supplies are community-managed and
therefore it is difficult to ensure that the “supplier” is able undertake an
expected level of quality control monitoring and therefore there is greater
reliance on external verification. Furthermore, enforcement of action is usually
difficult as community-managed systems are often difficult to regulate directly
and this becomes impossible when water hygiene is evaluated (Howard et al.,
2001a). It is possible, however, to undertake programmes as described below.
Given that monitoring microbial quality may be expensive, when considering
the development of sampling programmes for point sources, it is essential that
the monitoring programme is designed to meet a clearly defined management
need and that the data derived will result in useful information (Bartram, 1999;
Adriaanse, 1997; Ongley, 1998).

As for all water supplies, the use of sanitary inspection is extremely

important as a means of assessing long-term risks and in analysing the causes of
contamination when this is found. Sanitary inspections are visual assessments of
the infrastructure and environment surrounding a water supply taking into
account the condition, devices, and practices in the water supply system that
pose an actual or potential danger to the health and well-being of the consumers
(WHO, 1997). The most effective way to undertake sanitary inspections is to
use a semi-quantitative standardised approach using logical questions and a
simple scoring system. There exist a number of example forms that can be used
(WHO, 1997). Sanitary inspections are complementary to water quality analysis
and there is an increase in the power of subsequent analysis when both types of
data are available.

193
6.5.1 Selection of methods and indicator organisms

For communities where point sources are commonly used, the use of on-
site testing kits is often recommended given the large distances between the
sources and laboratories and the very significant problems with sample
deterioration (Bartram and Ballance, 1996). There appears to be no significant
difference in the reliability of results obtained from such kits in comparison to
laboratory testing providing the staff using them are properly trained and
maintain an aseptic technique. Kits using MF, MPN and P/A methods are
available.

Where bacteriological testing is undertaken using test kits, it is likely that

E. coli or thermotolerant coliforms will be the indicator parameters analysed.
However, for shallow groundwater, the use of faecal streptococci may provide
more reliable results given their greater environmental resistance and because
there is evidence that thermotolerant coliforms may multiply in nutrient
enriched environmental waters (WHO, 1996; Byappanahalli and Fujioka, 1998).
Studies in Uganda using sorbitol-fermenting bifido-bacteria, which are unique
to human faeces showed a stronger relationship with faecal streptococci than
thermotolerant coliforms (Howard et al., 2001b). Most kits for thermotolerant
coliform analysis can also be used for faecal streptococci analysis, although
they must be able to sustain 48 hour incubation times.

6.5.2 Water quality sampling approaches

The degree to which sampling of point water sources is developed is

dependent largely on the objectives of the monitoring programme being
implemented. Very few countries have fully developed programmes of rural
water supply testing. Indeed, although such approaches do provide an overall
indication of water quality and its variation, it does not necessarily yield data
from which management decisions can be easily made.

There are two main approaches (years-group approach and longitudinal

study sample) that can be recommended for sampling of point water sources,
which are outlined below. In addition, a brief section is included regarding
assessment-only approaches, which are not recommended. The recommended
approaches work on a slightly different basis and provide somewhat different
information. Common to both approaches, however, is that a sample of water
sources are tested during each sampling round.

194
 The first stage in both approaches is to compile an inventory of all the
water sources to be included in the monitoring programme (Lloyd and Bartram,
1991). A second key element in both approaches is to ensure that testing is
timed to coincide with those times when water quality may be most threatened,
usually during the wet season. However, is should be noted that some shallow
groundwater systems show very rapid response to rainfall and this may need to
be considered when designing the sampling programme (Barrett et al., 2000).
Samples should be taken from the principal outlet – handpump, spring outlet or
bucket used to take water from a well.

6.5.2.1 Year-groups approach

In this approach, all the water supplies are assigned to a particular “year-
group” and a rolling programme of visits developed (Bartram, 1999). All the
sources listed in each “year-group” would be visited, with the aim that all
supplies would receive repeated visits over a time frame of between two to five
years. On each visit, detailed surveys would be undertaken including sanitary
inspection, water quality analysis and household visits made. Key to this
approach is to ensure that stratified or cluster sampling approaches are adopted
to ensure that supplies in different parts of the country were included within
each 'year group'. If this is not done, then the results from particular “year
groups” may provide a distorted picture of the microbiological quality of water,
as it may be biased because of technology type, hydrogeology or pollution
loading specific to particular areas. The “year-group” approach provides the
monitoring body a greater breadth of information as it attempts to cover all
water sources. The data may also be used to identify actions required to improve
water quality, although the restricted numbers of samples at sources may mean
that it is more difficult to develop a full understanding of water quality
variation.

6.5.2.2 Longitudinal study sample

In this approach, a representative sample of water supplies is visited on a

regular basis. It is therefore different from the above approach in that it does not
attempt to provide data on all sources within the country, but rather addresses
developing a more in-depth understanding of the variation in quality of different
types of water sources and different parts of the country. This can be used to
assess whether microbial quality and risks vary over time and provides useful
insight into how effective community operation and maintenance is and what
improvements are required both in terms of training and design. The
longitudinal study approach aim is to use the data from the sample to inform

195
overall implementation and management of the quality of water by ensuring that
the most important factors in causing microbial contamination are identified and
preventative and remedial actions developed. These apply beyond the sample of
water sources included within the clusters and can form the basis of a broader
national strategy for water quality improvement.

6.5.2.3 Assessment only approaches

In some countries, sampling of the water sources only occurs during the
source selection stage and there are often particular levels of microbial
contamination (usually defined on the basis of thermotolerant coliforms or more
rarely E.coli) above which the source should not be used. A common value used
is 50/100ml. However, such approaches have very limited use as the results of a
single test may not provide a realistic estimate of the microbial quality of the
water (particularly as this would not typically be done during worst case
situations) and the use of an arbitrary figure of 50/100ml is unlikely to be
meaningful in terms of health risk. Clearly, such approaches also fail to provide
any indication of whether the protection works undertaken have been effective
in reducing pollution and the designs used have rarely, if ever, been properly
evaluated in terms of their ability to reduce contamination as expressed through
log-reductions in bacterial densities.

Other countries either complement or replace the testing of sources during

selection with a test on commissioning. This has the advantage that it provides a
better indication of the quality of the supply, but again may not reflect seasonal
variation in quality. Such approaches will also fail to address the greater
concerns of water quality deterioration over time in supplies where operation
and maintenance is weak.

6.5.4 Tanker trucks and vendors

These are treated separately as vending is a commercial practice and

therefore more amenable to direct control than when water supplies are
community-managed. Where trucks collect water from water supply utility
hydrants, the routine testing of the quality of water in the trucks and at the
hydrants should form part of the routine distribution testing by the water
supplier. In the first instance an assessment should be carried out covering a
sample of filling stations and tankers, followed by a lower-intensity routine
monitoring programme when the day of sampling, selection of hydrants and
selection of trucks vary in order to ensure to avoid biased results. This can be

196
linked to codes of sanitary practice both by the water supplier and the tanker
trucks (Lloyd et al., 1991).

Where vendor supplies are not taken from utility piped supplies, routine
testing and sanitary inspection is still desirable, although the type of vending
may dictate the ease with which this is implemented. For tanker trucks, a rolling
programme of random testing can be initiated, although it may be difficult to
identify source waters. For very small vendors, it may not be possible to
undertake routine monitoring, although occasional assessments would be
worthwhile. In both cases, it is critical that the monitoring will lead to some
form of action, whether this is through working with vendors to improve
practices, regulating vending practices or banning all vending.

6.5.5 Household water

The testing of water stored in households is important to ascertain the

quality of water actually being consumed. This is important because post-source
deterioration in quality may have occurred and therefore good quality water at
source may have become severely degraded by the time it is consumed and
remedial actions (for instance hygiene education programme) may be required.
The testing of household water is therefore an important component of an
integrated risk-based approach to water quality.

Household samples should be taken from the drinking water storage vessel
used by the family and containers used for collecting and transporting the water.
A water chain can be tracked from source to storage with samples taken from
source, collection vessel and water storage container. This often provides useful
information regarding where and what type of interventions (educational,
technical) are most appropriate. Generally, household testing programmes
should be linked to source water testing programmes to ensure that the
monitoring team understand whether poor quality in the home results from re-
contamination or poor source water quality.

The numbers of samples and the selection of households will depend

largely on the principal objective for the testing of household water. If the major
purpose is to simply undertake random sampling of household water (which
may be an important part of the monitoring programme whether “year group” or
longitudinal study) then a stratified random sampling approach can be adopted.
In this case, no specific intervention is being evaluated although the collection
of information about sources and the type and cleanliness of the storage
container may indicate where major problems lie.

197
 This data can also be used to check on the use of feedback of surveillance
results on household water quality. For instance in Uganda, the simple process
of feedback of information and routine testing led to observable improvements
in water quality stored within the home (Howard and Luyima, 2000). When
such programmes are initiated it is important that different households are
visited in each sampling period to prevent a bias developing due to repeated
visits by surveillance staff. However, a cluster sampling approach may be
adopted by identifying sentinel communities believed to be at greater risk
because they have least access to direct connection or because they are more
affected by interruption in supply.

In some cases there may be other specific objectives for testing water in
the home. These may include evaluating the impact of a particular health
education programme or household water storage and treatment interventions.
In this case, a study would be designed to measure the impact between an
intervention group and a control group thus allowing an evaluation of the
impact of the intervention. Alternatively, the influence of the type of source,
frequency and duration of discontinuity, or type of storage vessel on household
water quality may be assessed in a community. In this case, a cluster sampling
approach would typically be used to keep the number of households included to
a reasonable number that allows intensive investigation.

6.6 Summary

This chapter provides a summary of possible sources of faecal

contamination in drinking water and describes recognised sampling regimes in
order to detect contamination. The small volume of samples collected during a
reporting period represents only a tiny fraction of the total quantity of water
delivered during that period. Thus, the challenge of sampling is to provide
maximum information on water quality in the distribution system using the data
from a limited number of samples. The ingress of faecal contamination into the
distribution system should trigger immediate responses and this is covered in
Chapter 7. The approaches available for sampling water quality from non-piped
systems are also covered, with indications provided for actions to improve water
quality.

198
 REFERENCES

Adriaanse, M. (1997) Tailor-made guidelines: a contradiction in terms?

European Water Pollution Control 7(4), 11-16.

Ahmed, F. and Hossain, M.D. (1997) The status of water supply and sanitation
access in urban slums and fringes of Bangladesh. Journal of Water
Supply: Research and Technology – Aqua 46, 14-19.

Ainsworth, R.A. (2002) Water quality changes in piped distribution systems.

World Health Organization.

Anon (1994) The Microbiology of Water 1994: Part 1 – Drinking water.

Reports on Public Health and Medical Subjects, No. 71. Her Majesty’s
Stationery Office, London.

APHA, AWWA, WEF (1998) Standard Methods for the Examination of Water
and Wastewaters. 20th Edition. American Public Health Association,
Washington DC.

ARGOSS (2001) Guidelines for assessing the risk to groundwater from on-site
sanitation. BGS Commissioned Report CR/01/142, British Geological
Survey, Wallingford, UK.

AWWA (1986) AWWA Standard for Disinfecting Water Mains. AWWA C651-
86, Denver CO.

Barrett, M.H., Johal, K. Howard, G., Pedley, S. and Nalubega, M. (2000)

Sources of faecal contamination in shallow groundwater in Kampala. In:
Groundwater: Past Achievements and Future Challenges. Sililo et al.
(Eds.). pp 691-696.

Bartram, J. (1999) Effective monitoring of small drinking water supplies. In:

Providing Safe Drinking Water in Small Systems: Technology, Operations
and Economics. Cotruvo, J.A., Craun, G.F. and Hearne, N. (Eds.) Lewis
Publishers, Washington, DC. USA, pp.353-366.

Bartram, J. and Ballance, R. (1996) Water Quality Monitoring. Chapman and

Hall, London, UK.

199
Breslin, E. (2000) Protecting drinking water: water quality testing and PHAST
in South Africa. In: Water, Sanitation and Health. Chorus, I., Ringelband,
U., Schlag, G. and Schmoll, O. (Eds.) IWA, London, pp 89-94.

Burlingame, G.A. and O’Donnell, L. (1993) Coliform sampling at routine and

alternate taps: problems and solutions. Proceedings of the AWWA
Water Quality Technology Conference, Miami, FL.

Burlingame, G.A. and Choi, J.J. (1998) Philadelphia’s guidelines for obtaining
representative samples from throughout drinking water systems.
Proceedings of the AWWA Water Quality Technology Conference,
November 1-4, 1998, San Diego, CA.

Byappanahalli, M.N. and Fujioka, R.S. (1998) Evidence that tropical soil can
support the growth of Escherichia coli. Water Science and Technology
38(12), 171-174.

Clark, J.A. (1990) The Presence-Absence test for monitoring drinking water
quality. In: Drinking Water Microbiology. McFeters, G.A. (Ed.)
Springer-Verlag, New York, pp 399-411.

Clark, J.A. (1980) The influence of increasing numbers of non indicator

organisms by the membrane filtration and P-A tests. Canadian Journal of
Microbiology 15, 827-832.

Craun, G.F. and Calderon, R.L. (2001) Waterborne disease outbreaks caused by
distribution system deficiencies. Journal of the American Water Works
Association 93, 64-75.

Dufresne, L., Burlingame, G., Cochrane, C., Maley, L. Shahid, S. and Toch, M.
(1997) Eliminating ‘noise’ in distribution system coliform monitoring.
Proceedings of the AWWA Water Quality Technology Conference,
Denver, CO.

Forget, G. and Sanchez-Bain, W.A. (1999) Managing the ecosystem to improve

human health: Integrated approaches to safe drinking water. International
Journal of Occupational and Environmental Health 5(1), 38-50.

Gale, P. (1996a) Coliforms in the drinking water supply: What information do

the 0/100-mL samples provide? Journal of Water Supply: Research and
Technology - Aqua 45(4), 155-161.

200
Gale, P. (1996b) Developments in microbiological risk assessment models for
drinking water: A short review. Journal of Applied Bacteriology 81, 403-
410.

Gardner, J.F. and Peel, M.M. (1986) Introduction to Sterilization and

Disinfection. Churchill Livingstone, Melbourne.

Geldreich, E.E. (1996) Microbial Quality in Water Supply Distribution Systems.

CRC Press Inc. Boca Raton, FL.

Geldreich, E.E. and LeChevallier, M. (1999) Microbiological quality control in

distribution systems. In: Water Quality and Treatment: A Handbook of
Community Water Supplies. Fifth Edition. Letterman, R.D. (Ed.)
McGraw-Hill, New York. pp.18.1-18.49.

Gelinas, Y., Randall, H., Robidoux, L. and Schmit, J-P. (1996) Well water
survey in two Districts of Conakry (republic of Guinea) and comparison
with the piped city water. Water Resources 30(9), 2017-2026.

Haas, C.N. and Heller, B. (1986) Statistics of enumerating total coliforms in

water samples by membrane filter procedures. Water Research 20(4),
525-530.

HMSO (1989) The Water Supply (Water Quality) Regulation 1989. Statutory
Instrument No. 1147.

Howard, G., Bartram, J.K. and Luyima, P.G. (1999) Small water supplies in urban
areas of developing countries. In: Providing Safe Drinking Water in Small
Systems:Technology, Operations and Economics. Cotruvo, J.A., Craun,
G.F. and Hearne, N. (Eds.) Lewis Publishers, Washington, DC. USA.
pp.83-93.

Howard, G. and Luyima, P.G. (2000) Report on water supply surveillance

activities in 10 selected urban areas of Uganda. Report published for the
Ministry of Health, Uganda and available at www.lboro.ac./watermark

Howard, G., Bartram, J., Schuab, S., Deere, D. and Waite M. (2001a)
Regulation of microbiological quality in water cycle. In: Water Quality:
Standards, Guidelines and Health. Assessment of risk and risk
management for water-related infectious disease. Fewtrell, L. and
Bartram, J. (Eds.) IWA Publishing, London, pp. 377-393.

201
Howard, G., Barrett, M., Pedley, S., Johal, K. and Nalubega, M. (2001b)
Distinguishing human and animal faecal contamination in shallow
groundwater. Unpublished technical report for DFID, Robens, Guildford.

ISO (1991) Sampling – Part 5: Guidance on sampling drinking water and water
used for food and beverage preparation. ISO 5667-5. International
Organization for Standardization, Geneva, Switzerland.

Kerneis, A., Nakache, F., Deguin, A. and Feinberg, M. (1995) The effects of
water residence time on the biological quality in a distribution network.
Water Research 29(7), 1719-1727.

LeChevallier, M. (1999) The case for maintaining a disinfectant residual.

Journal of the American Water Works Association 91(1), 86-94.

Lloyd, B. and Bartram, J. (1991) Surveillance solutions to microbiological

problems in water quality control in developing countries. Water Science
and Technology 24(2), 61-75.

Lloyd, B., Bartram, J., Rojas, R., Pardon, M., Wheeler, D. and Wedgewood, K.
(1991) Surveillance and improvement of Peruvian drinking water
supplies. University of Surrey, Guildford, UK.

MacKenzie, W.R., Hoxie, N.J., Proctor, M.E., Gradus, M.S., Blair, K.A.,
Peterson, D.E., Kazmierczak, J.J., Addis, D.G., Fox, K.R., Rose, J.B. and
Davis, J.P. (1994) Massive waterborne outbreak of Cryptosporidium
infection associated with a filtered public water supply, Milwaukee,
Wisconsin, March and April 1993. New England Journal of Medicine
331(3), 161-167.

Martin, R.S., Gates, W.H., Tobin, R.S. et al. (1982) Factors affecting coliform
bacteria growth in distribution systems. Journal of the American Water
Works Association 74, 34-37.

Melian, R., Myrlian, N., Gouriev, A., Moraru, C. and Radstake, F. (1999)
Groundwater quality and rural drinking water supplies in the Republic of
Moldova. Hydrogeology Journal 7, 188-196.

Ongley, E. (1998) Modernisation of water quality programmes in developing

countries: issues of relevancy and cost efficiency. Water Quality
International September/October, 37-42.

202
Pedley, S. and Howard, G. (1997) The public health implication of groundwater
microbiology. Quarterly Journal of Engineering Geology 30(2), 179-188.

Pipes, W.O., Ward, P. and Ahn, S.H. (1977) Frequency distributions for
coliform bacteria in water. Journal of the American Water Works
Association 69, 664-668.

Pipes, W.O. and Christian, R.R. (1984) Estimating mean coliform densities of
water distribution systems. Journal of the American Water Works
Association 76, 60-64.

Prévost, M., Rompré, A., Baribeau, H., Callier, J. and Lafrance, P. (1997)
Service lines: their effect on microbiological quality. Journal of the
American Water Works Association 89(7), 78-91.

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-of-use water treatment and safe storage: a promising new strategy.
Epidemiology and Infection 122, 83-90.

Rahman, A. (1996) Groundwater as a source of contamination in rapidly

growing megacities of Asia: Case of Karachi, Pakistan. Water Science
and Technology 34(7-8), 385-292.

Reiff, F.M., Roses, M., Venczel, L., Quick, R. and Witt, V.M. (1996) Low-cost
safe water for the world: A practical interim solution. Journal of Public
Health 17(4), 389-408.

Rojas, R., Howard, G. and Bartram, J. (1995) Groundwater quality and water
supply in Lima, Peru. In: Groundwater Quality. Nash, H. and McCall,
G.J.H. (Eds.) AGID Report No. 16. Chapman and Hall, London. pp. 159-
167.

Semenza, J.C., Roberts, L., Henderson, A., Bogan, J. and Rubin, C.H. (1998)
Water distribution system and diarrheal disease transmission: a case study
in Uzbekistan. American Journal of Tropical Medicine and Hygiene
59(6), 941-946.

Sobsey, M.D. (2002) Household water treatment and storage as appropriate

technology for the developing world. World Health Organization.

203
UK Water Industry Research (1995) Microbiological growth in domestic
pipework and fittings: Cambridge Workshop. Report DW 04/BUKWIR,
1995.

VanDerslice, J. and Briscoe, J. (1993) All coliforms are not created equal: a
comparison of the effects of water source and in-house water
contamination on infantile diarrheal disease. Water Resources Research
29(7), 1983-1995.

Whittington, D., Lauria, D.T. and Mu, X. (1991) A study of water vending and
willingness to pay for water in Onitsha, Nigeria. World Development
19,179-198.

WHO (1996) Guidelines for Drinking water Quality, Volume 2: Health and
supporting criteria. Second Edition. World Health Organization, Geneva.

WHO (1997) Guidelines for Drinking water Quality. Volume 3: Surveillance

and control of community supplies. Second Edition. World Health
Organization, Geneva.

WHO and UNICEF (2000) Global Water Supply and Sanitation Assessment
2000 Report. WHO/UNICEF, USA.

Wright, R.C. (1986) The seasonality of bacterial quality of water in a tropical

developing country (Sierra Leone). Journal of Hygiene 96(1), 75-82.

204