Testing water quality is a key element of drinking water safety that has been gaining increasing attention, especially in reference to the close of the Millennium Development Goals (MDG) in 2015 [1
]. A World Health Organization (WHO) and UNICEF Task Force stated recently that it is “essential that new targets for post-2015 efforts should include water quality” [3
]. Water testing plays an important role in ensuring the correct operation of water supplies, verifying the safety of drinking-water, investigating disease outbreaks, and validating processes and preventative measures. There are significant challenges in implementing comprehensive and appropriate water quality testing, particularly in low-resource settings. As a consequence, the extent and quality of the information provided by water testing is often inadequate to support effective decision-making.
Microbial contamination is responsible for the great majority of water-related health burden [4
]. WHO recommends that the microbial quality of drinking-water be measured using faecal indicator bacteria, preferably Escherichia coli
; these bacteria are chosen to indicate the presence of faecal contamination rather than identifying pathogens directly [4
]. Conventionally, analyses take place in a laboratory environment using standard procedures, such as those described in the Standard Methods for the Examination of Water and Wastewater
], approved by the U.S. Environmental Protection Agency or set by the International Organization for Standardization. We have restricted our analysis to tests based on culturing faecal indicator bacteria as these are likely to remain the most common methods for microbial water quality monitoring in the short- to medium-term.
Conventional laboratory methods, such as membrane filtration and multiple tube fermentation, are complex and time-consuming. They require a wide range of basic laboratory equipment and skilled personnel to achieve consistent results. Sample transportation, especially within the recommended timeframe (<24 h, preferably <6 h [6
]) and temperature range (<8 °C but not frozen [5
]), is often impractical. This is particularly the case for rural and dispersed populations, for which the nearest laboratory can be at a significant distance from water supplies. Where laboratories are accessible, these may be overstretched and only able to conduct infrequent testing of a limited number of supplies. As a consequence, testing in the locations with no access to resources such as reliable mains electricity or technically trained staff may be preferable.
Several tests, most notably portable membrane filtration and the hydrogen sulphide test [7
], have sought to address the challenges of using traditional methods in remote and low-resource settings. Developments in chromogenic and fluorogenic enzyme-substrate tests have also greatly expanded the range and variety of available tests. A detailed understanding of the resource requirements and information provided by these is needed in order to select an appropriate test. However, the available information on microbial water tests is poorly consolidated and in many cases difficult to access—this is particularly the case for costs and test performance. Although previous reviews [8
] include a number of factors that are essential in the selection of tests for drinking-water, they are limited in scope, do not evaluate tests on a consistent basis or do not provide a side by side comparison. As a result, it can be difficult for practitioners to select tests for a particular setting, application and budget.
Our objectives were:
(1) To define, in reference to resource settings and the purposes of testing, important characteristics which should be considered when selecting a test for faecal indicator bacteria in drinking-water.
(2) To collate information on these characteristics for available water tests and assess their suitability based on the resources available in a given setting.
We have not carried out any microbiological assessments of the performance of the various tests. We expect users of our catalogue should satisfy themselves that the performance of the tests will meet their needs. Most manufacturers’ websites make available the findings of appropriate, objective studies. Consequently, we are able to assess suitability for resource settings, but not the fitness for purpose. Notwithstanding these limitations, we believe that the catalogue will be useful for an audience ranging from organizations responsible for monitoring national water quality and water service providers to researchers and policy makers.
There is a wide variety of characteristics within our catalogue. We recommend that users should select a short-list of tests for further consideration, based on two criteria: (i) matching tests to resources and (ii) matching tests to applications. After selecting a shortlist for further consideration, users should consult manufacturers’ websites to review the microbiological performance assessments that have been carried out to ensure that the chosen products will provide appropriate sensitivity and specificity for the target application.
4.1. Matching Tests to Resource Settings
When considering the resource constraints, is it valuable to consider a number of alternative testing arrangements which could include: transport of the sample to a fixed laboratory, mobile field testing laboratory, decentralized onsite testing, and sample preparation onsite followed by incubation in a laboratory. If the testing forms part of a longer term monitoring system, sampling strategies including screening and/or combining complementary tests should be considered. Decisions on where and how to conduct the testing may be equally, or more important, than the cost per test [19
]. This will be the case especially when the costs of transport, labor, and setting up, equipping and maintaining laboratories are taken into account. As such, whether testing will be taking place at the source using a portable kit, in a nearby health clinic or district laboratory warrants careful consideration.
The use of many of the tests included in this assessment in low- and middle-resource settings is limited by equipment required to conduct the tests. This is particularly the case if a decentralized approach to testing is adopted, wherein many full sets of equipment are needed. By selecting lower-cost alternatives to standard equipment, or modifying testing methods (Table 6
), significant savings on the cost of equipment may be possible. In most cases, the extent to which performance is compromised by these adaptations is not well understood. Despite these options, cold storage, safe handling and disposal, training, and temperature control during incubation or, where required, sample transport remain barriers to testing in low-resource settings. Furthermore, transport restrictions are known to apply to the consumables required for some tests, such as methanol for portable membrane filtration.
Alternatives to standard equipment and methods.
Alternatives to standard equipment and methods.
|Standard 1||Alternative 1||Advantages||Limitations|
|Laboratory Incubator ($$$)||Ambient incubation (-)||Possibly acceptable in tropical climates  or potentially indoors||Recoveries for injured bacteria may be poor; increased and poorly defined incubation time; not applicable everywhere |
|Low-cost electric incubators ($–$$) e.g., egg incubator||Good temperature control||Reliant on electricity, may not be available for higher temperatures (44.5°C for TTC)|
|Body incubation (-)||Readily available||Acceptability, health and safety issues and limited number of tests|
|Phase change incubator ($)||Good temperature control, only requires hot water||Requires heat source, can be bulky, particularly for many tests|
|De-ioniser ($$$)||Boiled water (-)||Readily available||May not inactivate all organisms; can concentrate chemical contaminants need to run blanks|
|Steam distiller ($$)||Produces very pure water||Requires high-voltage and power; requires running water; fragile|
|Autoclave (see below)||May be available or has dual use (see below)||Turbid waters may not provide suitable dilution water, especially for membrane filtration|
|Membrane Filtration assembly & vacuum ($$$)||Portable MF assembly, including hand pump or MIT D-lab kit ($–$$)||Manual, portable||Time consuming procedure|
Separate incubator required
|Autoclave ($$)||Portable autoclave ($)||Portable||Requires heat source|
|Pressure cooker ($)||Independent of electricity||Requires heat source|
|Bleach or disinfectant ($)||Readily available, good for disinfecting waste||Handling of cultures; care must be taken in reusing components to prevent false negatives due to residual disinfectant|
|Refrigerator or freezer ($$)||Storage at room temperature (-)||Independence from electricity||Shelf-life unclear for many media, particularly hydrated media; samples cannot be retained for subsequent analysis|
|Sample transport on ice||Ambient temperature transport||Simple||Potential population increase or die-off of bacteria|
|Insulated box (with cool water if available)||May be better than ambient||Change in bacterial population unknown, likely to be better than ambient|
4.2. Matching Tests to Applications
While it is relatively straight-forward to classify tests based on suitability for resource settings and most manufacturers will provide performance statistics for sensitivity and specificity, a similar, simple classification is not possible for the suitability of tests for particular applications. The purpose of testing may need to be established on a case by case basis. In general, there are three main factors in low and medium resource settings: indicator bacteria, quantitative performance and regulatory approvals.
The choice of indicator bacteria will be influenced by the application; a distinction can be drawn between cases where presence of the indicator is evidence of faecal pollution, and therefore potential health risk, or an assessment of the efficacy of a treatment process [21
]. The former requires that the indicator be ubiquitous in faeces but must not occur naturally. As some total coliforms occur naturally in the environment E. coli
, or alternatively thermotolerant coliforms, are recommended by the WHO [4
]. This is reflected by the sanitary significance column of Table 4
and Table 5
. E. coli
are also used for treatment assessment purposes, but in this context total coliforms are generally recommended [22
]. Both indicators suffer from being more sensitive to disinfection processes than some pathogens [4
]. Tests that detect the presence of H2
S-producing bacteria are frequently used, particularly where resources are limited; however there is ongoing debate about their sanitary significance [23
Quantitative tests are generally more expensive and require more resources. If the purpose of testing is to ascertain whether water meets national regulations (or the WHO Guidelines [4
]), PA tests may be entirely adequate as long as the volume is sufficient and the test has the necessary validation and approvals. PA tests are also valuable when monitoring water supplies that are usually free of contamination. The resulting string of “non-detects” or infrequent positives gives more confidence than a single quantitative test [25
]. However, if there is a need for relative prioritization (e.g., source selection) or if monitoring changes over time and there is a reasonable risk of contamination, a quantitative test will generally add value. Furthermore, the cost per analysis increases if a wide range of contamination levels are to be measured with high precision. As such, the range of a test, its lower and upper detection limits, needs careful consideration. For operational monitoring this decision should be based on an understanding of the likely levels of contamination in the sources being assessed. Guidance on the volumes which should be assessed using membrane filtration and multiple tube fermentation are available elsewhere [6
]. It should be noted that the ranges are likely to be strongly influenced by both indicator bacteria and source type. For surveillance monitoring the testing of volumes lower than the WHO Guideline of 100 mL should only be considered if the majority of supplies are known to be contaminated. The range and precision should also be chosen with thought given to data analysis, decision-making, responsibilities and integration with existing data.
Regulatory approvals are required for compliance and, usually, surveillance monitoring. Furthermore they provide additional reassurance of tests’ performance. We have not conducted a review of international regulations; instead, we compiled information on whether tests have obtained U.S. EPA approval or are featured in the Standard Methods [5
] or standards published by the International Standards Organization. These approvals are the basis of the standards in a number of countries.
There are number of limitation to this assessment. Firstly, the full cost of testing will include a number of factors which we have not been able to take into account in this catalogue. This includes variability in the per test and equipment costs resulting from shipping and distribution. In most cases, a significant element of the overall cost of testing will be related to the resources such as labor, transport and infrastructure [19
]. We have listed many of the resource requirements, but we do not calculate their associated costs; clearly this will vary considerably depending on the circumstances. Secondly, a number of characteristics were not included in this catalogue. Test performance in terms of false positive and false negative rates (or specificity and sensitivity) was not included as this information is not available for all tests and, unless a comparative study (e.g., [26
]) is undertaken, these cannot be compared on a consistent basis. A review of the validations and national regulatory approvals each test has obtained was beyond the scope of this assessment. The precision of tests varies depending on the concentration of indicator bacteria. The availability of tests and equipment will vary both within and between countries; this would need to be established for a particular setting and factored into test selection. Thirdly, we do not include all microbiological growth media or tests based on the detection time for which sufficient information was not made available. Finally, we have not assessed the suitability of individual or combinations of tests for particular applications. This is because the information required depends on a number of variables (such as source type and regulatory standards) and is context specific. Moreover, ongoing debate about the sanitary significance and applicability of indicator bacteria, particularly H2
], limits the guidance that can be provided.
Important characteristics to consider when choosing a microbial drinking-water test have been identified and this information has been compiled for 44 tests. The tabulated information should assist users in short listing tests for their particular requirements and setting. The identified tests include both presence/absence and quantitative tests for E. coli, total and thermotolerant coliforms, and H2S. The results are provided in tabular form to facilitate comparisons between tests.
The cost per test was found to range from $0.60 to $5.00 for a presence/absence device and from $0.50 to $7.50 for a quantitative test. Although the costs of tests themselves are important, in fact they are likely to be a small component of the overall costs of testing, when the infrastructure and human resources are considered. The ability of tests to support alternative testing arrangements that reduce reliance on these resources and consequently the overall cost of testing is a key issue.
Few of the identified tests are ideal for low-resource settings if implemented according to their standard protocols. This is especially the case for quantitative tests. A number of alternative procedures which would greatly simplify testing in low-resource environments have been identified. We encourage further work to evaluate these and establish guidelines for their application.