Upload
engineers-without-borders-uk
View
218
Download
0
Embed Size (px)
Citation preview
7/29/2019 Aquatest: An Affordable Water Test
1/56
Aquatest: An Affordable Water Test
by
Robert Bain (G)
Fourth-year undergraduate project
in Group D, 2007/2008
I hereby declare that, except where specifically indicated, the work submitted herein is
my own original work
Signed .. Date:
7/29/2019 Aquatest: An Affordable Water Test
2/56
Aquatest: An Affordable Water Test | Technical Abstract i
Technical Abstract
Aquatest: An Affordable Water Test
Motivation and purpose
The project was motivated by the scale and severity of problems with water supplies in
developing countries. Every year 1.8 million deaths (of these 90% are children under five)
are attributable to poor water supplies, sanitation and hygiene (WHO, 2004). Despite
considerable effort by the international community to tackle this problem, success has been
limited; a different approach is clearly needed. Community-driven development is easier to
sustain, but the technology to facilitate the shift of power has not been developed. The
monitoring of water quality is a case in point; laboratories are centralised and portable
testing kits based on Membrane Filtration are too expensive (around 1000) for most
communities to afford. The purpose of this project is two-fold:
(i) To examine water supply, sanitation and hygiene in rural areas of developingcountries and determine whether bacteriological water testing is appropriate for
this context. And if there is a need;
(ii) To develop an affordable and easy to use bacteriological water testing device thatenables widespread water quality monitoring in rural areas.
Is water testing appropriate?
An initial literature review reinforced the need for water testing:
Surveillance and verification of drinking water quality in small community water supplies is
recognised as an essential activity; however it remains potentially complex and expensiveInternational network on small community water supply management
Alice Springs, Australia July 2005
However, further exploration cast doubts on the appropriateness of water testing. A
combination of case studies, interviews and a thorough literature review were used to
establish the need for water testing, assess its limitations and gain a better understanding ofthe context.
7/29/2019 Aquatest: An Affordable Water Test
3/56
7/29/2019 Aquatest: An Affordable Water Test
4/56
Aquatest: An Affordable Water Test | Technical Abstract iii
Table of ContentsTechnical Abstract ................................................................................................................. i
List of Figures ................................................................................................................. iiv
List of Tables .................................................................................................................. iiv
List of Boxes ..................................................................................................................... v
Acronyms ......................................................................................................................... v
1 Introduction ...................................................................................................................... 1
2 Summary of the Project .................................................................................................... 3
3 Methodology ..................................................................................................................... 6
4 Part I: Exploratory Research ........................................................................................... 10
4.1 Case Studies ............................................................................................................. 10
4.2 Significance of Water Quality .................................................................................. 104.3 Alternative Measures ............................................................................................... 14
4.4 Can Testing Lead to Action? ..................................................................................... 16
4.5 Evaluation of Current Water Testing Practices ........................................................ 17
4.6 Conclusions .............................................................................................................. 22
5 Part II: Constructive Research ......................................................................................... 24
5.1 Determining Requirements ..................................................................................... 24
5.2 Critical Evaluation of the Aquatest Prototype ......................................................... 27
5.3 Experimental methods ............................................................................................. 28
5.4 Concepts and Their Evaluation ................................................................................ 39
6 Final Conclusions ............................................................................................................. 41
6.1 Findings .................................................................................................................... 41
6.2 Recommendations and Further Work ..................................................................... 42
6.3 Critical Evaluation of the Project ............................................................................. 42
7 Appendix ......................................................................................................................... 43
7/29/2019 Aquatest: An Affordable Water Test
5/56
Aquatest: An Affordable Water Test | Technical Abstract iv
List of Figures
Figure 1: Areas of Risk for Travellers Diarrhoea ................................................................. 1
Figure 2: Rotavirus distribution ........................................................................................... 1
Figure 3: Methodology Flow Diagram ................................................................................. 7
Figure 4: Faecal-Oral transmission paths ........................................................................... 11Figure 5: Categories of monitoring and interventions ...................................................... 14
Figure 6: Hydrogen Sulphide test ...................................................................................... 19
Figure 7: IDEXX Quantitray 2000 ....................................................................................... 20
Figure 8: Schematic of the relationship between indicator groups .................................. 21
Figure 9: Requirements Capture ........................................................................................ 24
Figure 10: Functional Diagram ........................................................................................... 24
Figure 11: Functional analysis of incubation ..................................................................... 25
Figure 12: Incubation temperature and time to get a result with the H2S test ................ 25
Figure 13: Model of Aquatest first device prototype ........................................................ 27
Figure 14: Likelihood plots for a single dilution with fifty wells. ....................................... 31Figure 15: Discrimination between threshold values using Monte Carlo ......................... 33
Figure 16: Heating and heat storage options that were considered ................................. 35
Figure 17: Incubation using body heat .............................................................................. 35
Figure 18: Bessel Function ................................................................................................. 38
Figure 19: Experiments to determine critical radius ......................................................... 38
Figure 20: Equivalence between empty submerged tube and suspended filled tube ...... 38
Figure 21: Final concept waterjet demonstration model .................................................. 40
Figure 22: Final design concept CAD drawing ................................................................... 40
Figure 23: Pipettes or sterile syringe ................................................................................. 40
Figure 24: Graph of Coefficient of Variation ...................................................................... 45Figure 25: Plots demonstrating the Most Probable Range ............................................... 46
Figure 26: Random number allocation .............................................................................. 46
List of TablesTable 1: Comparison of deductive and inductive research ................................................. 6
Table 2: Criteria for water testing to be considered appropriate ....................................... 9
Table 3: Summary of Case Studies ..................................................................................... 10
Table 4: Description of water testing methods ................................................................. 17
Table 5: Evaluation of testing kits against criteria ............................................................. 20
Table 6: Problem Definition ............................................................................................... 24Table 7: Requirements Specification ................................................................................. 26
Table 8: Advantages of MPNS over MF ............................................................................. 27
Table 9: Statistics Nomenclature ....................................................................................... 29
Table 10: Statistical assumptions ....................................................................................... 29
Table 11: Discrimination between example threshold values for a 50ml sample ............ 33
Table 12: Morphological Chart .......................................................................................... 39
Table 13: Case Study Findings ............................................................................................ 44
Table 14: Measures of Uncertainty ................................................................................... 45
7/29/2019 Aquatest: An Affordable Water Test
6/56
Aquatest: An Affordable Water Test | Technical Abstract v
List of Boxes
Box 1: The Aquatest Initiative
Box 2: Occupancy Theory
Acronyms
HIV/AIDS Human Immunodeficiency Virus/ Acquired Immune Deficiency Syndrome
IWA International Water Association
JMP Joint Monitoring Program
MIT Massachusetts Institute of Technology
MPN Most Probable Number
MPN Most Probable Number
MPNMD MPN-Multiple Dilution
MPNS MPN-Sample Subdivision
MTF Multiple Tube FermentationUNICEF United Nations Children's Fund
WHO World Health Organisation
7/29/2019 Aquatest: An Affordable Water Test
7/56
Aquatest: An Affordable Water Test | 1 Introduction 1
1 Introduction
1.1 Motivation
The motivation for the project is the scale and severity of disease related to problems of
poor water quality in developing countries. According to estimates 1.1 billion people in
developing countries do not have access to an improved supply of water; of these 84%
live in rural areas (JMP, 2006). Faecally contaminated water is one of the main transmission
routes of infectious diseases and poses a major threat to human health. Every year 1.8
million people are killed by diarrhoeal disease resulting from a lack of clean water,
inadequate sanitation and poor hygiene. Ninety percent of these are children under the age
of five (WHO, 2004).
These figures are likely to underestimate the true extent of the problem as a significant
proportion of the disease goes unrecorded (IWA, 2003); they also overlook the morbidity
caused by diarrhoeal disease which is aggravated by restricted access to healthcare,
malnutrition and the prevalence of HIV/AIDS. Diarrhoeal disease primarily affects those with
undeveloped or weakened immune systems: children, the elderly and HIV/AIDS suffers (and
travellers). Water-borne diseases are often perceived as an entirely developing country
problem (fig. 1). They are also affected by climate and are particularly prevalent in tropical
countries; the warmer, wetter and more humid climates tending to exacerbate problems
with water quality and sanitation (fig. 2).
Figure 1: Areas of Risk for Travellers Diarrhoea(Centre for Disease Control and Prevention, 2008)
Figure 2: Rotavirus distribution of cases. Each point represents500 out of a total of 800,000 yearly deaths (Glass, 1997)
7/29/2019 Aquatest: An Affordable Water Test
8/56
Aquatest: An Affordable Water Test | 1 Introduction 2
1.2 Approach
A new approach to monitoring is needed
Despite considerable effort and investment over the years, the international agenda to
improve water and sanitation has not seen the success expected. Furthermore, it seems
unlikely that the Millennium Development Goal for water (#7 Target 10: halving the
number of people without sustainable access to safe drinking water) will be met in many
countries the number of people without access in Sub-Saharan Africa is in fact increasing
(JMP, 2006).
As noted by the World Health Organisation the top down approach has limitations:
While global estimates of coverage will remain important, local capacity to generate and
use information will be a vital part of the monitoring effort (WHO, 2008)
The most sustainable approach would be to reduce the dependence of the poor on external
support through community-based or driven development. In many cases the infrastructure
and technology to facilitate this shift of power has not been made available; one area which
is particularly lacking is the monitoring of water supplies. This project assesses one way of
monitoring water quality: bacteriological water testing referred to as water testing below.
Water testing
Bacteriological water contamination comes from many sources much of which has little
sanitary significance. The purpose of water testing is to detect recent faecal contamination
(Hutton, 1983) and identify water that is unfit for drinking.
It is not practical to routinely test for all pathogenic micro-organisms that might be in
drinking water; instead tests should detect bacteria that are indicative of faecal pollution.
Faecal-oral disease transmission is responsible for the spread of water-borne disease and
therefore the presence of these indicator micro-organisms is thought to give a good
assessment of the health risk posed by drinking water. If indicator bacteria are not present
then the water is deemed safe to drink. The three indicators that are often used are
coliforms and more specifically thermotolerant coliforms or Escherichia coli (known as
E.coli). Bacterial densities are usually given as a number per 100ml the volume of waterusually tested.
7/29/2019 Aquatest: An Affordable Water Test
9/56
Aquatest: An Affordable Water Test | 2 Summary of the Project 3
1.4 Project Objectives
(i) To examine water supply, sanitation and hygiene in rural areas of developingcountries and determine whether water testing is appropriate in this context. And if
there is a need;
(ii) To develop a low-cost and easy to use bacteriological water testing device thatenables widespread water quality monitoring in rural areas of developing countries.
2 Summary of the Project
Familiarity with a low-cost water test developed at Massachusetts Institute of Technology
(MIT) led the author to the initial conclusion that an appropriate water test could be
produced for a fraction of the current price. Experience using this kit at a school in
Tocantins, Central Brazil reinforced the perceived need for monitoring water quality. The
boarding school (Fundao Bradesco Canuan) wanted to continue testing water as an
educational exercise in nearby communities, but was unable to because the test relies on
several consumables that are expensive and not available in rural areas (Section 6.3).
A brief literature review at the start of the project supported the need for low-cost water
testing (see quote above), however as the project progressed the benefits of water testing
were increasingly called into question. The initial proposal for the project was to further
develop the work carried out on the MIT water testing kit and to make it suitable for
community use. It quickly became clear that whilst this would result in a relatively low-cost
kit (at around 1/10 of the cost of commercially available testing kits such as DelAgua) it
would remain out of the reach of people earning less than $2 a day. A previous 4th
year
project (Gordon, 2006) had followed this path, achieving only incremental improvements to
the low-cost kit.
Surveillance and verification of drinking water quality in small community watersupplies is recognised as an essential activity; however it remains potentially complex
and expensiveInternational network on small community water supply management
Alice Springs, Australia July 2005
7/29/2019 Aquatest: An Affordable Water Test
10/56
Aquatest: An Affordable Water Test | 2 Summary of the Project 4
Part I: Is water testing appropriate for rural water supplies?
In light of the limitations of such a project, contact was made with organisations for which
an appropriate test could be developed or where the issues of water quality could be
researched. The first of these was the Karen Hilltribes Trust, a charity which builds gravity-
fed water supplies in Northern Thailand. The charity was contacted with the intention of
designing a low-cost water test that was appropriate for the Karen (one of the ethnic hill
tribes). Telephone conversations and interviews with the head of the charity led to several
interesting findings and the overall conclusion was that water testing was not appropriate
for the Karen. There were very good reasons for this including:
Mountain springs have little to no contamination and with adequate treatment(Biosand filtration) the water was known to be clean.
Volunteers had attempted to conduct water testing for the charity, but it had beenplagued by procedural difficulties and inconclusive results.
Attempts to get in contact with organisations in Northern Mozambique where water quality
problems are particularly acute (and because the author speaks Portuguese) were also
unsuccessful. It became clear that research would have to be conducted in the UK and
would have to rely primarily on secondary sources.
It was decided to build on these unexpected findings, explore case studies from secondary
sources and to conduct a thorough literature review in order to establish when water
testing is appropriate. The initial research questions were:
1 Where does water testing fit into strategies for the (improvement of) provision of cleandrinking water?
2 Is testing appropriate? If so when, and what can it achieve?3 What are the most important factors to measure? What are the best ways to do these?The case study research highlighted further concerns (technical and socio-political) and even
called into question the basic assumptions on which water testing is based ( Section 4.2.1).
Interviews and secondary sources supported these findings. Whilst water testing is seen as
inappropriate in several circumstances, it has been found that there is still a pressing need
for a low-cost device (Section 4.6.2).
7/29/2019 Aquatest: An Affordable Water Test
11/56
Aquatest: An Affordable Water Test | 5
Part II: Aquatest
In January 2008 contact was established with the Head of the Aquatest Initiative, Dr.
Stephen Gundry (Box 1). Working for Aquatest marked a significant change in the direction
of the project and greatly increased both its scope and potential. Large-scale manufacture
together with the development of a new field testing kit opened up the possibility of
producing a device at a cost that will be affordable to those in need. The research areas
agreed with Dr. Gundry were:
- Determining requirements for the device- Evaluating methods for incubating the device- Developing design concepts
With the results of the research outlined in Part I and criticisms of water testing in mind,
work began on these tasks (Section 5). And in addition, a statistical model was developed to
ascertain the number of wells and volumes needed to discriminate between a given set of
risk levels (Section 5.3.1).
Box 1: The Aquatest Initiative
Aquatest is a collaborative effort to design and disseminate a low-cost water testing kit for
developing countries. It is led by the University of Bristol and collaborators include WHO, UC
Berkeley and the University of Cape Town and Indian Participants.
With support from the Gates Foundation ($13.1 Million), they aim to develop a small single-
use device for an ambitious cost of $0.10. The test is to be used by water professionals and
communities themselves. It is hoped that within 10 years, low-cost water testing devices
will be widely used in 80% of developing countries (Aquatest website).
The first prototypes will be field tested in India and South Africa in 2009. Distribution and
Licensing of the technology will be led by PATH, an organisation that has experience with the
disposable syringe.
7/29/2019 Aquatest: An Affordable Water Test
12/56
Aquatest: An Affordable Water Test | 3 Methodology 6
3 Methodology
most people become experienced with one type of study and then stick to it, safely and
unadventurously(Hakim, 1992)
This project combines both deductive and inductive research: the former primarily in the
first half, the latter predominantly in the second. The two approaches are compared in
Table 1. The main weakness of the deductive approach, in which engineers are traditionally
trained, is that it eliminates the possibility of unanticipated findings (Eishenhardt, 1989).
Table 1: Comparison of deductive and inductive research
Deductive Inductive
Starts from General Theory
Form specific hypotheses
Numerical or statistical
Starts from Observations
Identify Patterns
Textual
Narrow
Testing and confirmation
Open-ended
Explanation of reasons
Familiarity with the MIT water testing kit meant that approaching the problem with a new
perspective was of utmost importance. Inductive approaches helped to avoid coming up
with theories at the outset. In working for the Aquatest Initiative it is particularly important
to avoid advocacy; only through questioning the benefits of water testing could substantive
research be produced (Hakim, 1992). The methodology used is shown schematically in
figure 3 on the next page.
7/29/2019 Aquatest: An Affordable Water Test
13/56
Aquatest: An Affordable Water Test | 3 Methodology 7
Figure 3: Methodology Flow Diagram
7/29/2019 Aquatest: An Affordable Water Test
14/56
Aquatest: An Affordable Water Test | 3 Methodology 8
3.1 Part I: Exploratory Research
Research was conducted predominantly through the use of inductive approaches: case
studies, interviews and a thorough literature review.
Case studies were central to the first stages of the project. Theory-building case study
research is appropriate in the early stages of research on a topic or toprovide freshness in
perspective to an already researched topic (Eishenhardt, 1989). It can be seen that both of
these applied to this research project: a fresh perspective on water testing was needed and
little research had been done to establish whether tests are appropriate for use in rural
areas of developing countries. Analysis was carried out in two parts: (i) individual case-
analysis and (ii) searching for cross-case patterns. Case studies guided the literature reviewand highlighted many issues with water testing.
Interviews were informal. They ranged from telephone interviews with the head of the
Karen Hilltribes Trust (KHT) to discussions with experienced Water and Sanitation and Public
Health engineers. Several meetings were arranged with the Head of the Aquatest initiative.
TheLiterature Reviewhad several purposes: (i) to gain an understanding of the wider issues
surrounding water quality in developing countries, (ii) to validate the case-study research
findings and (iii) to pursue lines of enquiry. An appreciation of the wider context of water
testing was seen as vital to ensure that the monitoring process is adequate, appropriate and
most of all informative.
3.2 Part II: Constructive Research
The constructive research was heavily influenced by two sources:
Good Design Practice for Medical Devices and Equipment(Shefelbine, 2002) The Inclusive Design Toolkit(Clarkson, 2007).
Conventional engineering design process focuses on conceptual design: where the most
important decisions are made (Cross, 1994). The central preoccupation is seen to be not
missing good ideas. In contrast, the primary focus of the process followed in this study was
to establish the real need and ensure that the design solves a genuine problem.
7/29/2019 Aquatest: An Affordable Water Test
15/56
Aquatest: An Affordable Water Test | 3 Methodology 9
3.3 Criteria for Success: Defining Appropriate
It is interesting to note that whilst the Aquatest device will support grassroots development
it goes against much of current development thinking. It will be mass manufactured and the
technology is not easily understood by the user (they are not able to take control of the
technology). The central notion of intermediate technology (from which the term
appropriate technology was popularised) is that it addresses the need for labour intensive
rather than capital intensive work; hence lending itself to small-scale, decentralised
workplaces (Schumacher, 1973). The definition of appropriate has changed in the light of
some very successful projects that have used modern mass-manufactured goods: examples
include mobile telephones in Bangladesh and LEDs for rural lighting.
In this report, appropriate simply means that it is suited to the specific needs of the users
and the demands of the context (in this case rural areas of developing countries). The
following criteria (Table 2) were developed during the project and have been used to
determine whether water testing is appropriate in this context.
Table 2: Criteria for water testing to be considered appropriate
1. Relevant Meaningful and sensitive to the users education, aptitude andorganising skill. Relevant to their
2. Significant Of sanitary significance established link to health outcomes3. Worthwhile it must be possible to use information from the test and this
information should not be readily obtained from another source;
4. Affordable It must be low-cost to run and have a small upfront cost5. Independent It should not be reliant on external support, or materials that are not
available locally.
6. EnvironmentallySustainable
It must not generate excessive waste at the point of use.
7. Scalable It must be scalable to facilitate widespread monitoring of drinkingwater
7/29/2019 Aquatest: An Affordable Water Test
16/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 10
4 Part I: Exploratory Research
Is Water Testing Appropriate for Rural Water Supplies?
4.1 Case StudiesTable 3: Summary of Case Studies
Organisation Project Location Abbreviation
Karen Hilltribes Trust Gravity-fed water supply Northwestern Thailand CS1
Escola Bradesco
Solar water disinfection and
water testing using the low-cost
MIT testing kit
Central Brazil CS2
Student ProjectProposed water quality
assessmentTolla Islands, Indonesia
CS3
Vigyam AshramWater testing using the multiple
tube fermentation method.Pabal, India
CS4
Table 3 summarises the case studies used at the beginning of the project. The first two are
the authors own observations and experience and have been described briefly in the
summary. The last two were gathered from secondary sources and personal
communication. Conclusions from the case study research are given in theAppendix 7.2.
4.2 Significance of Water Quality
The extent of diseases related to water, sanitation and hygiene was described in the
introduction. A question that needs to be asked is: how much of the disease is attributable
to water quality? This has major implications for how diarrhoeal disease is tackled globally,
but at the community level and for this project it needs to be asked in order to justify water
quality testing which is inherently linked to interventions improving water quality.
4.2.1 The Importance of Water-borne Disease
The faecal-oral transmission route (fig. 4) is complex and there are many ways for
pathogens to be passed from one person to the next. An important question for anyone
trying to reduce diarrhoeal disease through environmental interventions is: which are the
dominant transmission paths?
7/29/2019 Aquatest: An Affordable Water Test
17/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 11
Figure 4: Faecal-Oral transmission paths (Prss, 2002)
Before examining the evidence, faecal-oral diseases need to be classified as water-borne or
water-washed(Feachem, 1978):
Water-borne diseases are those that are transmitted when contaminated water is drunk.
They can be prevented by improving water quality.Water-washeddiseases are caused by water scarcity; an inadequate supply of water inhibits
personal hygiene: washing hands and food. Water-washed diseases can be prevented by
increasing quantity of water without improving the quality.
If all of the diarrhoeal disease was transmitted via the water-washed route, as was
suggested by Prof. Cairncross in an interview, water quality interventions (and water
testing) would not be worthwhile. A causative link between water quality interventions and
diarrhoeal disease needs to be established in order to justify water testing. For this we turn
to Epidemiology, the study of the factors affecting health in a population.
7/29/2019 Aquatest: An Affordable Water Test
18/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 12
Review of the epidemiological evidence for water quality
Whilst there are many studies of water quality interventions, most have methodological
flaws and decisions cannot be based upon them. For example in the most recent meta-
analysis Clasen et al. (2007) rejected 947 out of a total of 979 studies reviewed. Thesemethodological problems have been well documented (Blum D, 1983).
Until recently, there was consensus that water quality and access were both important
means of reducing diarrhoeal disease; access was seen as more important than
bacteriological quality (Jensen, 2004). Interest in Household Water Treatment and Storage
(HWTS) has led to several new studies showing strong support for water quality
interventions in the household. In their review Fewtrell et al. (2005) found that water
quality interventions at the household gave a mean reduction of 39% in morbidity, whilst
those at the source only achieved 11%. Furthermore Clasen et al. (2007) found that water
quality interventions need not be combined with hygiene and sanitation to be effective.
Despite these encouraging findings, Prof. Cairncross is convinced that the water-washed
route is dominant water scarcity, not quality is responsible for the transmission of
diarrhoea. Of the four blinded studies reviewed in the paper he co-authored with Clasen
(2007) none showed reductions in diarrhoeal disease. Intervention studies which are not
blinded are liable to bias both because the researchers hope to show their treatment is
effective and because communities are likely to understate their disease because they are
grateful.
There are two main camps:
1. Those that advocate water quality interventions in the household;2. Those that advocate greater access to water;
In conclusion, the epidemiological evidence for water quality interventions is easily
challenged; there is no consensus as to the relative importance of these transmission
routes. The relative predominance is likely to be dependent on many factors, e.g. in drier
areas scarcity is likely to be dominant; in wetter and more populous areas waterborne
disease may be significant.
7/29/2019 Aquatest: An Affordable Water Test
19/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 13
4.2.2 Safe Water
There are two common principles in definitions of safe water: (i) that the water is drinkable
and (ii) that it poses little risk to health. A further principle needs to be added for a safe
water supply: access to adequate quantities. As was seen in the previous section, there is
considerable debate on the relative importance of quality and quantity.
The guideline value for bacteriological quality of drinking water is 0 Coliforms (or E.coli) per
100ml (WHO, 2006). While only a guideline value, governments often set this as a standard
to be met by all supplies despite not being affordable. This target is not achievable in most
rural areas of developing countries and does not encourage the incremental improvements
in quality and access that are needed to reduce diarrhoeal disease.
Exposure-response
Given the lack of consensus surrounding transmission paths, it is no surprise that there is
little data on exposure-response curves. Drinking water quality testing in the west has been
focused on ensuring the efficacy of treatment processes rather than measuring a health risk
thus not demanding research on the quantitative relationship between indicator density
and health outcomes. The lack of evidence is apparent when compared to the vast literature
on the health risks associated with bathing waters. It also explains the WHOs unwillingness
to suggest suitable targets for rural areas of developing countries despite much criticism.
In conclusion, there is a real need for more research before quantitative testing is
meaningful for people drinking water from sources which are not disinfected. This is also
needed for appropriate water quality standards to be set to readdress the bias towards
water quality interventions described in the next section.
7/29/2019 Aquatest: An Affordable Water Test
20/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 14
4.3 Alternative Measures
Figure 5: Categories of monitoring and interventions
4.3.1 Monitoring
Health Outcomes are the most desirable indicators because they are directly what we aim
to improve through interventions. Monitoring of the source and intervening processes is
indirect and therefore only useful if causal links between the measure taken and the desired
outcomes can be established. The main problem with measures by outcome is that they do
not usually identify actions that the community can take or locate sources of pollution.
Cholera epidemics, when all water is suspect and should be boiled are an obvious exception.
As was found in section 4.2 even with a rigorous epidemiological study, it can be hard to
isolate the impacts.
If it was always possible to know by examining the source whether water is clean, then
water testing would not be worthwhile. Unfortunately, in many cases you cannot tell
whether water from a particular source is safe to drink, let alone whether it is clean at the
point-of-use. This being said, there is much to be gained from perceptions of the quality of
7/29/2019 Aquatest: An Affordable Water Test
21/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 15
water sources, the type of source and treatment administered and the state of the water
supply. It is not surprising that the open wells tested in Brazil (CS2) were heavily
contaminated whereas the mountain springs in Northern Thailand (CS1) were clean.
Sanitary surveys are a formalized means of assessing the environmental risks they usually
constitute a series of questions with yes or no answers. A score, the number of yes
answers gives an indication of the risk. Like perceptions of quality, the current forms do not
necessarily correlate well with the actual contamination of the source. Region specific
surveys can be developed once the importance of different risk factors is understood.
Sanitary surveys are very important but require experience (Hofkes, 1983) and the user
must be literate.
One of the main advantages of a low-cost water test is that it can be used with little or no
training and could therefore be scaled up rapidly. The main weakness of water testing is
that it does not indicate possible sources of contamination or measures to improve water
supplies. Combining water testing and sanitary surveys would overcome these difficulties,
but it is not clear how this can be done without depending on expertise.
4.3.2 Interventions
Figure 5 shows that there are many interventions which can reduce the likelihood or
severity of diarrhoeal disease. Preventative measures are more ethically acceptable than
relying on treatment, especially when access to healthcare is limited (stores of Oral
Rehydration Sachets are only currently reaching 38% of 0-5 year olds in the developing
world (UNICEF, 2008)). Behavioural change in particular is often the most cost-effective way
of reducing risk, but requires trained staff and is dependent on external support.
It was generally accepted that too much emphasis has been put on water supplies
(Thompson, 2001). Feachem (1978) suggests that donors and governments both have
vested interests in water supply interventions and that there has been a degree of wishful
thinking about the benefits. With the recent findings, HWTS are becoming increasingly
popular and considerable efforts are being made in their development and dissemination.
Their development has been spurred on by very positive results in recent intervention
studies and the finding that water from improved sources is readily contaminated before
7/29/2019 Aquatest: An Affordable Water Test
22/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 16
use. There are many HWTS, these include: chlorination, solar water disinfection and ceramic
filters.
Water testing can play a role in increasing awareness of disease transmission, hygiene and
encouraging greater care for personal environment. This was its purpose in Brazil (CS2).
However, it is primarily linked to interventions to improve water quality in particular
HWTS.
4.4 Can Testing Lead to Action?
Whether water tests will lead to action to improve water quality depends on many factors.
The technology to treat water is available and in many cases affordable: it is a matter of
priorities, preferences and whether the test can be understood.
Priorities: Many people may not perceive diarrhoea as being more than an inconvenience
unless it is severe and life-threatening (Howard, 2002).Water quality is often a high priority
for women, especially mothers, but it is a much lower priority for men. Immunity means
that it is easy for adults to perceive the risk of drinking water as low. If water quality is a low
priority, it is very unlikely that a water test will be bought no matter how low-cost it is.
Interpretation:As was found in several of the case studies, many people do not share the
scientific concept of disease. Waterborne diseases are particularly prone to supernatural
beliefs. Unlike natural causes, you can drink bad water for a while before getting ill. There
is also the difficulty of blaming water when there are so many other transmission routes. If
the technology is not easily understood by the user, there is a high risk of misinterpretation.
These will need to be addressed by field testing and through careful design.
Preferences: Chlorination is heralded as a simple solution, however many people do not like
the change in taste. As noted by Zoeteman (1980) the physical characteristics of water are
often the most important for the user. More contaminated sources are drunk because of a
better taste or because they are more convenient.
7/29/2019 Aquatest: An Affordable Water Test
23/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 17
4.5 Evaluation of Current Water Testing Practices
4.5.1 Methods
The traditional method for determining faecal-contamination is multiple tube fermentation.It will be referred to as most probable number by dilution (MPND) to distinguish it from the
method used in the Aquatest Device which is MPN by subdivision (MPNS). MPND is a
laboratory based method requiring substantial quantities of reagent, glassware, distilled
water, an autoclave and large incubator (Tab. 4). Laboratory testing is particularly
problematic for remote areas of developing countries - transportation is unreliable and time
consuming making it difficult to analyse samples within the recommended 6hrs (WHO,
2006). It is also much more expensive, requires high calibre staff and the distance (physically
and organizationally) between central laboratories and rural villages means that it takes a
long time before remedial action can take place (Hutton, 1983).
The development of the Membrane Filtration (MF) and more recently chromagenic reagents
have made field testing kits feasible. The incubation requirement is greatly reduced; for MF
a single Petri dish needs incubation whereas many test tubes need incubation in the case of
MPND. MF is used by almost all commercial field testing kits, with the exception ofPresence/absence (P/A) tests which are a simplified form of MPND compromising a single
undiluted tube.
Table 4: Description of water testing methods
Membrane
Filtration
(MF)
Water is passed through a membrane (typically 45m pores) leaving the bacteria
on the filter. The filter paper is placed in a Petri dish and broth that selects bacteria
of sanitary significance is added. The Petri dish is then incubated (24hrs) to allow
these bacteria to form visible colonies. These colonies are then counted to find thenumber of indicator bacteria (Colony forming units- CFUs) that were present in the
water sample. If there are many CFUs it may become difficult to count; water that
is expected to be highly contaminated can be diluted prior to filtration in order to
lower the number of CFUs to within the counting range (around 100 CFU).
Most
Probable
Number by
Dilution
Measured volumes of the sample water are diluted and pipetted into a series of
sterile tubes containing broth. The dilution series is tailored to the expected
concentration of bacteria in the sample. Test tubes are incubated for (24-48hrs
depending on the broth used) after which the bacterial density can be inferred
7/29/2019 Aquatest: An Affordable Water Test
24/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 18
(MPND) from the tubes which show growth. The most likely number of organisms is called
the most probable number and Standard Tables are available for most dilution
series.
Presence/
Absence (P/A)
Presence Absence is effectively MPND reduced to a single undiluted tube. The
presence absence test does not provide a quantitative determination.
Most
Probable
Number by
Subdivision
(MPNS)
This is similar to the MPND method, but avoids dilution. Instead the sample is
mixed with the reagent and then split into small wells. The bacterial density of the
sample is inferred from the number of wells showing positive growth. Standard
tables are not always available because the configurations differ from MT-MPN.
4.5.2 Equipment
There is a wide variety of water testing kits on the market; however most are intended for
users in developed countries. Examples of commercially available field testing equipment
specifically designed for use in developing countries include the DelAgua water testing kit,
H2S test and in development the MIT water testing kit. IDEXX Quantitray is also described
as this was the inspiration behind the Aquatest approach.
DelAgua (MF)
Developed at the University of Surrey in collaboration with Oxfam, DelAgua is probably the
best known portable water testing kit. The kit was designed primarily for international aid
workers and in particular humanitarian emergencies and reflects their needs. In addition to
measuring bacteriological contamination, the testing kit can measure a variety of chemical
properties of the water including residual Chlorine. The kit costs a total of 1327 (Institute,
2008) and approximately 0.30 per test. Even with significant subsidies it can rarely be
afforded by local NGOs, let alone communities or households. The kit is cumbersome
(weighing 10kg) and, requires training (3 days), several, often imported, consumables
(methanol, filters, pads and reagents) and an electronic incubator - all barriers to its
widespread use. The DelAgua kit is only suitable for users conducting regular testing.
7/29/2019 Aquatest: An Affordable Water Test
25/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 19
Hydrogen Sulphide Test (P/A)
At a cost of around Rs. 10 (about 12p) (IWP, 2007) this test is
the most affordable water test available. It was designed not to
require incubation and has many advantages over the MF basedtesting kits. The major limitation with the H2S method is that it
only indicates whether bacteria are present or not it does not
give a quantitative estimate. This is useful for ensuring that
chlorinated supplies are working effectively because you would
expect no indicator bacteria to be present. However, most rural
water supplies are not disinfected and have some level of faecal contamination: P/A tests
condemn too many supplies (IWA, 2003). Another problem with the H2S test is that it
detects Hydrogen Sulphide producing spores, some of which do not indicate faecal
pollution. For this reason it is not recommended by WHO (2002) a more recent study
indicates that H2S only correlates with E.coli for concentrations of over 1000/100ml (Gupta,
2007).
MIT Low-cost water testing kit (MF)
The MIT low-cost water testing kit is based on MF. The filtration device is compromised of a
baby bottle and syringe. Sterilisation is avoided by using disposable sterile inserts. The need
for reliable electric supply for incubation is circumvented by a phase change incubator: a
primary alcohol is used as a heat store and releases heat at a constant temperature as it
solidifies. Experience in Brazil suggests that the sterile inserts and reagents are not readily
available. Furthermore, the reagent used in Brazil (Millipore mColiblue24) is expensive and
requires refrigeration clearly an alternative is needed if the testing kit is to be independentof electricity.
Figure 6: H2S test
(courtesy of Susan Murcott)
7/29/2019 Aquatest: An Affordable Water Test
26/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 20
IDEXX Quantitray (MPNS)
A serial dilution is equivalent to splitting a sample of water
into small wells of varying sizes. This is the basis of the
Quantitray procedure (named MPNS in this report). This is arelatively new development in water testing practices and is
more accurate and easier to use than MF. However,
Quantitray is by no means appropriate for rural areas. It
needs an electronic incubator and heat sealer.
4.5.3 Evaluation of the Testing Kits
Table 5: Evaluation of testing kits against criteria
Criteria DelAgua MIT H2S MPNS
1 Relevant no training no training yes Yes
2
Significant yes yes
not
recommended by
WHO
if E.coli or
thermotolerant
3Worthwhile yes yes
condemns too
many supplies
if counts are
within the range.
4
Affordable no no yes
can be
manufactured at
low-cost5 Independent no no yes Yes
6 Environmentally
sustainableyes
uses many
consumables
if reused or
recycled
if reused or
recycled
7 Scalable no no Yes Yes
The review of current testing equipment suggests that it will be possible to use the principle
behind Quantitray (MPNS) to develop a water test at price close to that of the H 2S test and
is able to discriminate between contaminated sources.
Figure 7: IDEXX Quantitray 2000
(www.idexx.com)
7/29/2019 Aquatest: An Affordable Water Test
27/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 21
4.5.4 Choice of Indicator Bacteria and Reagents
Water contains many bacteria, most of which are harmless. The ability to isolate particular
groups of bacteria that are indicative of a health risk is of particular importance to the
development of a water test that meets the needs of the user. In the west, it is acceptable
to use the Coliform (Total Coliforms) group because testing is used to measure the efficacy
of treatment not to gauge health risk.
An important distinction has to be made between an indicator and an index. This has
implications for the amount of information that can be derived from microbiological water
tests.Indicators are bacteria whose presence indicates faecal contamination. An index is an
indicator whose concentration is directly related to
the health risk(Gleeson, 1997).
While many alternatives have been suggested,
Escherichia coli (E.coli) is recognised as the best
indicator of faecal contamination (WHO, 2006). It
is by far the most common bacteria in the human
gut (up to 109
per gram of faeces(Gleeson, 1997))
and is found in warm-blooded animal faeces. Figure 8: Schematic of the relationship betweenindicator groups
7/29/2019 Aquatest: An Affordable Water Test
28/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 22
4.6 Conclusions
4.6.1 Discussion of the Limitations of Water Testing
It is much harder to justify quantitative water testing than was expected at the start of theproject. The main criticism of water testing is the limited evidence supporting the link
between indicator densities and health risk. To justify quantitative assessments of water
quality, the link needs to be firmly established.
Information from the water test may be used in hygiene education, but its primary use is to
ensure the bacteriological safety of water in the household. It is therefore strongly linked to
HWST and avoiding source-to-point of use contamination. Care must be taken not to further
the water quality agenda over the other preventative measures that need support: access,
hygiene education and sanitation.
Whether water testing will be a priority for individuals and communities and the value they
will put on testing is dependent on many factors these will have to be determined by field
studies. Experience in Brazil (CS2) suggests that there will be resistance where the test is not
readily understood by the community or if the scientific concept of disease is not shared.
There is a high risk of misinterpretation if the user is expected to understand the result, but
not the means by which it is found.
4.6.2 Establishing a Need for Water Testing
There is no way for households and communities in rural areas of developing countries to be
sure that the water they drink is safe. Whilst there is much to be gained from indigenous
knowledge, perceptions of quality can be misleading; they could be exposing themselves to
an unnecessary risk of illness. Water testing is attractive as one way of addressing the
question: Is this water safe to drink?
Combined with sanitary surveys, water testing would provide an effective means of
monitoring rural water supplies where contamination is expected. This would allow users to
both detect faecal contamination and identify its source. It would enable communities and
individuals to take control over their own water supply; not to rely solely on foreign aid or
7/29/2019 Aquatest: An Affordable Water Test
29/56
Aquatest: An Affordable Water Test | 4 Part I: Exploratory Research 23
the stretched government agencies which have clearly failed, in many cases, to provide
water that is suitable for consumption.
Several bacteriological water tests are commercially available, but none is suitable for
community level testing. Current tests are both too complex and expensive or they
condemn too many supplies to be of use in this context. Indeed, little has changed since
Mara (1973) noted that:
In the majority of existing water supplies in developing countries (particularly in rural areas)
facilities for the bacteriological examination of the finished water are absent.
There is a real need for widespread monitoring of water supplies in rural areas. Watertesting could offer a means in which to do this in an affordable way and with less demand
for training than other monitoring methods. There is a clear need for people who are
looking after children, the elderly or immune-compromised to assess water quality. Those
most at risk need to be able to identify ways to lower their exposure to the faecal pollution.
In the opinion of the author, this need justifies the work in the second half of the project:
working with Aquatest to develop a low cost water testing kit.
7/29/2019 Aquatest: An Affordable Water Test
30/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 24
5 Part II: Constructive Research
Development of an affordable water test
5.1 Determining RequirementsFollowing Shefelbine (2002) there are four processes that need to be carried out before the
requirements specification can be completed (fig. 9).In the second half of this report, the
focus of the design is on what is seen as the primary user: mothers. Uses by professionals
and the trade-offs for greater coverage are also discussed.
Table 6: Problem Definition
Figure 9: Requirements Capture (Shefelbine, 2002)
Figure 10: Functional Diagram
WHO? Individuals and water professionals
(Village user, Local water engineer,
community leaders and health workers,
household).
WHAT? Water quality test that is low cost,
portable, reliable
WHY? To test bacteriological quality of water
WHERE? Rural areas of developing countries: at
the source, in the home or in a clinic
WHEN? When water quality unknown and for
regular monitoring of supplies
7/29/2019 Aquatest: An Affordable Water Test
31/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 25
5.1.1 An example of functional analysis: Incubation
Figure 11: Functional analysis of incubation
The Aquatest initiative is in the process of developing a reagent for E. coli which gives a
visual signal without the need for UV light. Since the reagent has not been developed yet
the influence of temperature on incubation time is not known. Growth models were
reviewed, but it became clear that there were too many variables. An alternative approach
is simply to use known results from the H2S test (fig. 12) as a basis for setting the
requirements. As many of the H2S bacteria are E.coli, this is seen as a good approximation.
Figure 12: Incubation temperature and time to get a result with the H2S test (J. Pillai, 1997)
Requirement: Temperature shall not drop below 22C nor rise above 44C during the 36
hours of incubation. The temperature should be close to 37C.
0
20
40
60
80
100
120
14 22 28 37 44
Time(hours)
Temperature (C)
7/29/2019 Aquatest: An Affordable Water Test
32/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 26
5.1.2 Requirements Specification
The full requirements specification is listed in Table 7. It is important to note that whenever
possible, the requirements have been relaxed so that the device is not needlessly
constrained.
Table 7: Requirements Specification
able to sample from all water supplies (accessible - at point of collection) D
able to test intrinsic water quality W
Representative encourages representative sampling W
Aseptic sample is collected without being contaminated by the user D
does not require more than one day of training. D
no training required W
Maintenance does not require maintenance by user W
Ready Signal User given signal when when the test result is ready D
InterpretableUser able to tell whether the water is suitable for drinking by (i) a normal healthy adult (ii)
children under 5, the elderly and immune-compromised/HIV.D
maintains a temperature of between 22 and 44C for a duration of at least 36 hours D
incubation close to the optimal growth temperature 37 W
users warned if the incubation has failed D
Precision able to discriminate between thresholds (to be set based on epidemiological data) D
Reliability gives the correct threshold 95% of the time when used correctly D
Sensitivity sensitive to 10/100ml D
sensitive to 5/100ml W
Significant test for E.coli. Shall test for E.coli or Thermotolerant Coliforms (WHO) D
Comparable help to prioritise the contamination problem W
Waterproof not leak contents or be affected by dampness or humidity. D
able to get the device to the regions where it will be used in a cost effective manner D
small enough to fit in a pocket W
no special storage requirements D
shelflife of at least three months; if sold in boxes (3 + 2n months). W
Independentnot reliant on sterilisation, electricity, distilled water or any other resource that is not
availableD
able to be dropped 2m onto rock without breaking D
difficult for someone to open up unintentionally or intentionally by a child W
resistant to shocks and vibrations during transport D
reusable W
not generate excessive waste D
Portable easily carried to and from the source (upto 1km by foot). D
Adaptable can be used in fixed laboratory incubators W
less than $0.30 at full scale manufacture D
$0.10 delivered to the customer, the Aquatest target price. W
costing
7/29/2019 Aquatest: An Affordable Water Test
33/56
Aquatest: An A
5.2 Critical Evaluation
It's a small, single-use devi
electricity or skilled techni
target manuf
Figure 13: Mod
1. The design is complicreaction (red) surrou
efficient way of incub
2. It is not certain wheth3. The choice of MPNS
change needed to ma
the most important o
Table 8: Advantages of MPNS ove
Lower Skill
It is easy to cont
laboratory worke
individual practice
1983). This is som
Robens Institute
Affordability MPSS is more affo
ReliabilityThe assay is more
growth of injured
Range
It is possible to t
supplies this m
dilution.
fordable Water Test | 5 Part II: Constructive
of the Aquatest Prototype
e for testing water quality. It can be used in
ians. It is being designed for use in developi
ctured cost of 10 US cents.(Aquatest Web
l of Aquatest first device prototype (Davis & Gundry, 200
ated and it is unlikely to meet the cost tar
ded by an insulating layer (yellow in fig. 1
ting the sample.
er this configuration is adequate.
is appropriate to the requirements and r
ke water testing affordable. It has many ad
which are:
r MF
minate the samples when using MF and it requir
should receive a weeks training and have an o
before he undertakes membrane filter analyses (M
ewhat of an exaggeration, but MF is not easy to u
ffer a 3 day training courses for their DelAgua kit.
dable than MF
reliable because suspended growth inherently allow
acteria (Fujioka, 1997).
ailor the device to the concentrations that are ex
ans it can detect higher concentrations than MF (
Research 27
the field, without
g countries at a
ite)
7)
et. An exothermic
) is clearly not an
presents the step
vantages over MF,
s training a skilled
pportunity for further
junkin 1976, in Hutton
se: by comparison the
s the resuscitation and
pected in rural water
ax 200CFUs) without
7/29/2019 Aquatest: An Affordable Water Test
34/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 28
5.3 Experimental methods
5.3.1 Statistics for the Device
Progress was hindered without results from the Aquatest Initiative describing theconfiguration of wells that were needed in the device. In order to develop the concepts and
preliminary designs an appreciation of the how the design would affect the operational
characteristics of the test was required.
Two first observations can be drawn about MPNS:
1. More information can be gained about the concentration for a given volume if thenumber of wells is increased.
2. If there is at least one bacterium in the sample the device will show growth in atleast one of the wells. The sensitivity of the test is therefore related to the total
volume of water in the sample.
5.3.1.1 Relevance of the Statistics and Ease of Interpretation
The test must be easily interpreted and the result relevant to the user. An MPN table: a
chart which tabulates the most probable number (an estimate of the bacterial density) for
each configuration of positive wells is clearly not appropriate.The user is not interested in a
count rate, but rather an indication of the level of risk from which a decision will need to be
made. Often, this will be a drink/dont drink decision.
Both Quantitray 2000 and the Aquatest prototype have used wells of different sizes. This
may be confusing as it is not intuitive that the largest wells are the least important. It is
suggested that the wells ought to be the same size this will make interpretation easier and
increase the information gained by the test.
5.3.1.2 Literature Review
There is an extensive literature surrounding the most probable number. At first efforts were
focused on determining the most probable or most likely number, but have since turned to
the more difficult matter of defining a measure of the uncertainty of bacterial counts using
the method. It is a Standard Method and has been used widely in microbiology and
medicine since 191 (McBride, 2005). However, approximate MPN calculations, rounding
7/29/2019 Aquatest: An Affordable Water Test
35/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 29
conventions and different methods of calculating the confidence limits lead to considerable
variability in the Standard Tables (McBride G. , 2003). The literature describes serial
dilutions as used in MPNMD; this is equivalent to the method used in this project as long as
the organisms are assumed to be randomly distributed.
It is helpful to introduce some notation and review the assumptions on which the calculation
of the MPN is based before describing the methods used:
Table 9: Statistics Nomenclature
R Number of wells
r Number of positive wells
s Number of sterile wells
n Number of E. coli in sample
V Total volume of sample
v Volume of a single alquilot
L (D|H) Likelihood of the data given the hypothesis
Table 10: Statistical assumptions
Assumption Description Comments
Detects one
or more
organisms.
If there is at least one organism
there is an obvious change which
can be clearly and consistently
identified.
There are many ways in which a test may fail
to detect a viable organism. If the medium is poor
it can require several bacteria and therefore
underestimate contamination.
Randomly
distributed
Bacteria are separate, not
clustered and they do not repel
each other
Supposed to thoroughly mix the sample. It is
possible to take a sub-sample that contains no
bacteria, especially if concentrations are low.
Poisson theory
or
Binomially
distributed
The volume examined is a
relatively small portion of a large
sample which had been
thoroughly mixed such that the
volume of each alquilot is small
compared to the sample volume
(v/V
7/29/2019 Aquatest: An Affordable Water Test
36/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 30
5.3.1.3 Methods
Calculating the MPN Using Occupancy TheoryEstimates for the Most Probable Number (Thomas 1942, Cochran 1950) are not suitable for
large R or multiple dilutions (McBride, 2005). While Standard Tables, simple estimates and
even an MPN Calculator (Blogett, 2003) are available the exact MPN had to be calculated.
Only with the likelihood function could measures of uncertainty be determined.
The likelihood functions were evaluated directly from Occupancy Theory (box 2) using
Matlab. It is relatively computationally intensive and the code had to be sped up in order to
calculate the likelihood functions within a reasonable time. This was achieved by creating
lookup tables thus reducing the number of times slow functions (e.g. binomial[], ln[]) were
called. Most Probable Number lookup tables were created to in order to evaluate the
operational characteristics described in section 5.3.1.6. The lookup tables were compared to
Standard tables to verify the technique.
| , = 1
Box 2: Occupancy Theory(David, 1962)
If n bacteria are randomly distributed among R test tubes of equal volume, the likelihoodthat (R-r) tubes will not receive any of these bacteria and therefore remain sterile is:
The Most Probable Number is the value of n that maximizes this likelihood.
7/29/2019 Aquatest: An Affordable Water Test
37/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 31
Figure 14:Likelihood plots for a single dilution with fifty wells (top graph has seven positive wells, lower graph showing
forty). The MPN is found by selecting the largest conditional likelihood. The plots demonstrate that the test is better at
discriminating between small numbers of organisms - when many wells show growth the curve is broader.
5.3.1.4 Measures of Uncertainty
As demonstrated infig. 14, it is difficult to discriminate between MPNs when the majority of
the wells show growth. An appreciation of the uncertainty introduced by the MPN
technique is needed to design the test. This is much harder than determining the MPN and
there is no exact way to calculate the uncertainty in the MPN result. In fact the confidence
or credible intervals are subjective (McBride, 2003). One must make prior assumptions
about the likelihood of contamination in order to determine a measure of uncertainty.
7/29/2019 Aquatest: An Affordable Water Test
38/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 32
Typically a diffuse prior (uniform distribution of mean bacterial densities) is used. Counter to
experience and intuition, this assumption implies that high concentrations are more likely to
be found than low concentrations.
McBride (2003) suggests that water testing lends itself to the Bayesian statistics. He
assumes the experimentalist is concerned with the probability of a particular result being
within a range. While this may be the case for scientific studies, when tests are to be
compared with risk assessments they are firmly in the classical statistical philosophy. The
device is looking for performance in the long run rather than being specific about the
current result.
5.3.1.6 Calculating Uncertainty
The most traditional means of expressing uncertainty, the confidence interval, is not strictly
appropriate for likelihoods and three alternatives were evaluated (Appendix 7.3). The
Monte Carlo Method was suggested by Prof. Spiegelhalter as a means to determine the
uncertainty of the measurement.
Operational Characteristics: Monte Carlo
From a risk profile, threshold values that the test should be able to discriminate between
can be chosen. The values chosen are (based on examples in 2 nd edition WHO Guidelines
Vol.3 (1993) pg. 78) are arbitrary: low risk (50/100ml) and high
risk (>100/100ml).
A Monte Carlo method was developed to examine the operational characteristics of
different device designs. It is described in the Appendix. This also facilitated the examination
of thresholds in line with the chosen risk levels: to establish how many wells are needed to
distinguish between chosen E.coli concentrations.
7/29/2019 Aquatest: An Affordable Water Test
39/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 33
Figure 15: Discrimination between threshold values using Monte Carlo a comparison of 25 and 50 wells with fixed total
sample volume of 50ml
Table 11: Discrimination between example threshold values for a 50ml sample
Number
of Wells
Discrimination between 10
and 100/100ml10 and 50/100ml 50 and 100/100ml
5 Good Poor None
10 Good Adequate Poor
20 Good Good Poor
25 Good Good Adequate
50 Good Good Good
7/29/2019 Aquatest: An Affordable Water Test
40/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 34
Findings and implications for the design
1. The device is looking for assured quality in a large number of different water suppliesand thus requires a frequentist approach. We are not interested in the scientific
measurement of the exact value of the MPN. The user is only interested in the
extent of contamination of the water and the resultant health risk of drinking it.
2. The MPN method is not a precise technique. When many wells show growth, there ispoor discrimination.
3. There is a trade-off between complexity of test (both design and interpretation) andprecision. It is easier to interpret the device if all of the wells are the same size.
4. The first prototype does not appear to be adequate for the purposes required.5. The high bacterial concentrations expected in source selection (often >1000/100ml)
are a challenge for a device using MPN: the well sizes would have to be very small. It
may be necessary to design two devices: one for assessing drinking water and
treated supplies and the second for more contaminated supplies such as surface
waters.
5.3.2 Energy Strategy
Purpose: to ensure bacterial growth and reliable results within a reasonable time.
In Section 5.1.1 it was found that the temperature must remain within the range 23 to 44C.
Growth is less sensitive than the precision of electronic incubators first led the author to
believe. To maintain temperatures close to the optimal and within this range, several
approaches were considered. These can broadly be termed low-tech and high-tech. The
options for storing and accepting heat that were considered in this project are described
below (fig. 16). Temperatures were recorded using an Omega 4-channel Datalogger in an
environment fluctuating between 16 and 20C. Whilst these do not reflect tropical
temperatures or diurnal variations, the results give a good indication of the approaches that
are likely to be successful.
7/29/2019 Aquatest: An Affordable Water Test
41/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 35
Figure 16: Heating and heat storage options that were considered
5.3.2.2 Low-tech: Body Heat
Body surface temperature is typically between 32 and 35C, higher on some areas of the
body such as the armpits. It was soon realized that placing the device on the body gave
sufficient temperature regulation, a finding validated by experiments (fig.16). Body heat has
been shown to be a very effective solution.
Figure 17: Incubation using body heat (50ml sample)
Whilst acceptability can only be established through field trials, reports from South Africa
suggest that individuals and households will be happy to carry the device with them for the
duration of incubation (Gundry, 2008). For these users, where there is a choice between
speed of the assay and cost, the design should invariably aim for the lowest cost.
Store Energy
Chemical EnergyExothermic
Reaction
Inorganic
PCMSalt Hydrates
Sensible HeatInsulated water
bath
Latent Heat
Liquid-Gas
Vapourisation
Solid-Liquid
Melting
Organic PhaseChange Materials
Lauric Acid
T=42-44
1-Tetradecanol
T=38Solid-Gas
Sublimation
Solid-Solid
Accept energy
Solar
Biomass
Body Heat
0
5
10
15
20
25
30
35
0 0.5 1 1.5 2 2.5 3
T
emperature(C)
Time (hours)
Water Temperature
Environment Temperature
7/29/2019 Aquatest: An Affordable Water Test
42/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 36
The use of body heat however is unlikely to be accepted by water professionals. They will
not want to have several water samples attached to their bodies all day and night. It is
suggested that high-tech solutions may have a role for incubating several samples at a time
or during transport to a location where laboratory incubation is available.
5.3.2.3 A Reusable Incubator?
A simple solution which could also solve the temperature requirement for professional
users is to heat water to just above the optimal temperature and warm the devices in this
water. The correct temperature could be attained by using a thermometer, or mixing three
parts boiling water and one part source water. The samples could then be stored in an
insulated container (e.g. vacuum flask) with a fraction of the warm water. This approach was
evaluated using a 0.5l thermos flask. After 33 hours the temperature had dropped from
42C to 24C, within the required limits. The main problem with this approach is the need to
boil and transport more water.
The viability of Phase Change Material (PCM) to incubate several water samples (taken as
0.5l of water roughly ten) at a time was assessed through a combination of experiments and
(lumped mass) heat transfer modelling. This began with a test of Portatherm - the incubator
used in the MIT kit. It was unable to maintain a steady temperature despite containing only
500ml of water and over 2kg of PCM: either the insulation or the mass of PCM would need
to be increased. Improving insulation is expected to be more cost effective and would result
in a lighter incubator. 1-Tetradecanol, the PCM used in Portatherm, should not be ingested
and its use is not recommended for safety reasons (note: a leading cause of poisoning in
some developing countries is ingestion of kerosene). An alternative, Lauric Acid (Tm=42C),
was found by searching through two reviews of PCMs (Farid (2003), Zalba (2002)) and the
CRC handbook.
7/29/2019 Aquatest: An Affordable Water Test
43/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 37
Findings and Implications for the Design
(i) Body heat is by far the most appropriate solution. The device ought to bedesigned so that it can be kept near the body comfortably for over 36 hours.
(ii)
Phase change materials are not a viable option for a single sample and even lessfor a single-use device.
(iii) Placing the device against the skin may not protect it from extremely hightemperatures. In this case the device could be submersed in more water from
the source and left in the shade.
(iv) A reusable incubator would be desirable for professional users who need toundertake frequent monitoring. This incubator could use phase change in an
insulated box Lauric Acid is recommended for this purpose.
5.3.3 Bubbles and Stability
Purpose: response to problem during prototyping of narrow tubes 3mm I/D. Narrow blocked
tubes would not fill when submerged in source water.
When a narrow tube blocked at one end is submerged, air cannot escape and a bubble
chamber forms. Surface tension forces dominate and water is stable above air. As the radius
is increased the gravitational forces become more important and at a certain value, the
bubble is no longer stable and the tube fills with water. This radius, , had to bedetermined because it would impact the form of potential designs, if not the minimum
volume of water that could be separated. A stability analysis can be used to describe this
phenomenon, the result of which gave an upper bound on the critical radius as just less than
5mm. Mestrel (2008) gives a derivation of the critical radius, leading to Equ. 2:
> Equ. 2
Where is the smallest value given by the Bessel function which satisfies the rigidwall boundary condition Jm(kr) = 0. Figure 17 demonstrates that this is on the (blue) J1
(curve); it has a value of 18.41. This predicts an antisymmetric perturbation as is to be
expected: air must go up one way as the water goes down the other.
7/29/2019 Aquatest: An Affordable Water Test
44/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 38
The numerical result in these lecture
notes (r = 2cm) was in fact correct. The
answer above was confirmed by a second
source (Joseph, 1970) and experimentswere conducted to verify this model.
Kitchen experiments
This was a very simple experiment: holes of different diameters were drilled into an acrylic
block. The block was then carefully submerged in dyed water. Holes larger than the critical
radius filled with the dyed water: the critical radius was between the smallest tube that
filled and the largest tube that remained empty (fig 19).
Spurious initial results could be accounted for by residual soap in the water (as confirmed by
later experiments). It is useful to note that impurities (e.g. turbidity) will always weaken the
surface tension, which supports the use of the surface tension of pure water in determining
the upper bound.
Eq. 1 predicts that the stability depends only on density and surface tension. Results
supported this theory for deep wells (40 mm), but nearly all of the 10-15mm deep wells
filled. This suggests that the stability also depends on the depth of the well. The theory was
Figure 18: Bessel Function
Figure 20: Experiments to determine critical
radius
Figure 19: Equivalence between empty submerged tube and
suspended filled tube
7/29/2019 Aquatest: An Affordable Water Test
45/56
Aquatest: An Affordable Water Test | 5 Part II: Constructive Research 39
also verified by an equivalence it implies (fig.20) between this problem and the emptying of
narrow tubes filled with water (then blocked at one end).
Findings and Implications for the Design
The trapping of air in thin wells is a limit on the smallest volume of water that can bedivided by pouring water into or submerging the device. This is therefore a limit on
the range of the MPNS method. One way around this is to pipette the water into the
device.
The diameter should be at least 10mm to ensure that the wells can be filled bysubmersion in water so that bubbles will not be stable.
By inspection, the smallest wells in the first Aquatest prototype will suffer from thisproblem.
5.4 Concepts and Their Evaluation
Table 12: Morphological Chart
Sample Reagent Incubate Make safe Dispose
Directly into
device e.g. BailerPill packet Body Heat
Solar water
disinfectionSterilise in lab
Samplecup/whirlbag
Powered Organic PCM +insulate
Chlorinedisinfection
Sterilise in field
Transfer using
disposable pipetteSaturated
paper
Body heat +
insulateBoil Generate waste
HydratedExothermic
reactionRecycle
Mixing boiling
waterBiodegradable
A morphological chart (Tab. 12) was used to explore different combinations of functions.
Having established that body heat would adequately incubate the water sample, the critical
function was taking the sample. With indicator bacteria present in the environment, it
would need to be very simple in order to make sure it is carried out aseptically. The most
promising concept is described on the next page; other ideas are given in the Appendix 7.4.
7/29/2019 Aquatest: An Affordable Water Test
46/56
Aquatest: An A
Final Concept: pipette transf
The Aquatest prototype wou
gloves and the device would
when opening the bag would
pipette to transfer water fro
the water, hands can be kep
stability limit described in sec
A screw-on lid with neoprene
The neoprene could be white
(shown in green) would need
due to solar disinfection, but
overheating. The screw-on li
that it is not easy to contami
and would preferably be tran
The choice of materials will b
and if so where the re-sterili
not to be encouraged both
this would impose the weste
may be difficult to impleme
further dependence on centr
Figure 23: Pipettes or stsyringe
fordable Water Test | 5 Part II: Constructive
r into the device
ld be complicated to use and relatively wa
be packaged in a sterile bag. The risk of c
be high. A much simpler solution is to use
the source to the device. As pipettes are o
clear from the device. Using a pipette also
tion 5.3.3.
under it is suggested to stop cross-contami
as shown in the figure to facilitate reading
to be opaque to prevent unintentional den
it is suggested that the colour is not dark
d should be deep, entirely encasing the si
ate the interior. The base (shown in grey)
sparent.
e determined on whether the device is exp
sation will take place. Generating waste at
ecause the collecting systems are not ava
n disposable culture. However, the autho
nt a collection scheme and furthermore t
lised laboratories.
Figure 22: Final ConceptWaterjet demonstration model Figure 22: FinaldrFigure 21: Final concept waterjetdemonstration modelrile
Research 40
teful. It would use
ntaminating these
sterile disposable
perated away from
avoids the bubble
ation of the wells.
the results. The lid
turisation of E.coli
s this may lead to
es of the base, so
ust be translucent
cted to be reused,
the point of use is
ilable and because
r recognises that it
is may result in a
design concept CADawing
7/29/2019 Aquatest: An Affordable Water Test
47/56
Aquatest: An Affordable Water Test | 6 Final Conclusions 41
If the device is to be reused, the lid could be made from polypropylene and the base from
polystyrene, both of which are resistant to UV and can be autoclaved. A disposable device
would need to have a seal to show that the device has not been used.
6 Final Conclusions
6.1 Findings
Water testing
1. Without stronger epidemiological evidence quantitative water quality testing is onlya measure of the extent of faecal contamination: it does not measure health risk.
2. An affordable water test can be produced using the MPNS method. This will be muchmore suited to the needs of the rural poor than current equipment.
3. The significance of water quality and in particular the health benefits of improvingthe quality of water supplies have often been overstated. In spite of water quality
being important it is also essential to have an integrated approach to solving health
problems.
Design of the Aquatest device
1. Field testing kits were designed to strict specifications many of these are notjustified in this context and add unnecessary expense. Minimum requirements for
the device have been proposed.
2. Aquatest ought to pursue low-tech incubation methods: the use of body heat isdeemed appropriate.
3. Statistical models suggest that ten wells may not provide enough information todistinguish between sources. A statistical model has been developed which will
enable the configuration to be dete