Globally, approximately two billion people drink contaminated water. Use of household water treatment (HWT) methods, such as locally manufactured ceramic filters, reduces the diarrheal disease burden associated with unclean water. We evaluated the quality, effectiveness, and acceptability of ceramic filters in two communities in Arusha, Tanzania, by conducting: 1) baseline household surveys with 50 families; 2) filter flow rate testing; 3) filter distribution with training sessions; 4) follow-up surveys at 2, 4, and 6 weeks after distribution; and 5) project end focus group discussions. We tested Escherichia coli (E. coli) and turbidity at baseline and the first two follow-ups. We found: 1) filter quality was low, as only 46% of filters met recommended flow rate guidelines and 18% of filters broke during the 6-week study; 2) filter effectiveness was moderate, with 8% and 35% of filters effectively reducing E. coli to <1 CFU/100 mL and <10 CFU/100 mL, respectively, at follow-ups; and, 3) filter acceptability was high, with 94% overall satisfaction and 96–100% reported use in the previous day. These results highlight the importance of mixed methods research as HWT product quality, effectiveness, and acceptability all impact product efficacy, and the need for quality assurance/quality control and certification schemes for locally manufactured HWT products.

INTRODUCTION

The burden of diarrheal disease is largely borne by developing countries. Annually, inadequate water, sanitation, and hygiene are the primary causes of nearly 1.7 billion cases of diarrhea (Fischer-Walker et al. 2012). The accumulated burden of repeated diarrheal diseases also results in decreased food intake, poor nutrient absorption, malnutrition, reduced resistance to infection, impaired physical growth, and deficient cognitive development (Baqui et al. 1993; Guerrant et al. 1999).

Household water treatment (HWT) technologies can be a cost-effective method of improving drinking water quality and reducing the burden of diarrheal disease (Fewtrell & Colford 2005; Clasen et al. 2007). They are a potential solution for the 663 million people who do not have access to an improved water source (WHO/UNICEF 2015) and the 1.2 billion more whose drinking water is contaminated at the source or during collection, transport, or storage (Onda et al. 2012).

One promising HWT technology is locally manufactured ceramic water ‘pot’ filters (CMWG 2011). The filter design was conceptualized in the 1980s, and refined and distributed with support from Potters for Peace in the 1990s. Since the 1990s, independent facilities (with technical support from Potters for Peace and other advisors) have continued to manufacture and develop filters.

Filters comprise an approximately 10 L capacity filter that rests on its rim in a lidded receptacle. The receptacle serves as a safe storage container and is fitted with a tap for dispensing filtered water. Flow rate, generally recommended to be 1–3 L/hr, is the main quality control and manufacturing consistency indicator in filter factories worldwide (CMWG 2011). In the laboratory, filters effectively remove >99% of protozoan (Lantagne 2001a) and 90–99.99% of bacterial organisms from drinking water (Lantagne 2001a; Oyanedel-Craver & Smith 2008; Bielefeldt et al. 2010; Brown & Sobsey 2010; Lantagne et al. 2010). However, long-term reductions of bacteria (Bielefeldt et al. 2009) and virus removal (Brown & Sobsey 2009) remain challenging. In the field, water treated by filters is often improved to the World Health Organization's low-risk classification (Lantagne 2001b; du Preez et al. 2008) of fewer than 10 colony forming units (CFU) of Escherichia coli (E. coli)/100 mL (WHO 1997). Filter use has also been associated with a 49% reduction in diarrheal disease (Brown et al. 2008).

Given their ease of use, low cost, and simple supply chain, filters are considered one of the most sustainable HWT technologies (Sobsey et al. 2008). Filters are not without drawbacks, however. Roughly 2% of filters fall into disuse each month, mostly due to breakage of the filter element, receptacle, or tap (Brown et al. 2009), and low flow rates are frequently reported as an inhibitor of filter use (CMWG 2011).

Another drawback of filters is the difficulty in maintaining high-quality production and quality control in independent local production. Currently, there are over 50 independent filter manufacturing facilities worldwide, and continued growth is anticipated (CMWG 2011). A 2009 study found large variation, both between and within factories, in whether factories consistently used the same materials, manufacturing methods, and quality control practices (Rayner et al. 2013). Factory visits and anecdotal information suggest that these variations in manufacturing lead to filters that inconsistently meet criteria for flow rate, strength, and microbiological removal.

In Tanzania, ‘limited or no progress’ was reported towards meeting the Millennium Development Goal for water (to reduce the proportion of the population without access to ‘improved’ water supplies by 50% from the 1990 baseline to 2015) (WHO/UNICEF 2015). As of 2015, 77% of the urban population and 46% of the rural population has access to an improved water source. In 2007, the non-governmental organization, Safe Water Ceramics of East Africa (SWCEA), established a filter manufacturing factory in Arusha, Tanzania, with the goal of improving access to microbiologically safe water and reducing the diarrheal disease burden in Tanzania.

Shortly afterwards, SWCEA, Emory University, and the Centers for Disease Control and Prevention (CDC) collaborated to investigate the effectiveness and acceptability of these new filters in rural Tanzanian households. The results from this investigation are presented herein.

METHODS

Research design

This mixed-method, longitudinal study occurred in June and July 2008 and consisted of: 1) flow rate testing of filters to be distributed; 2) an informational session on the filters with study participants and a baseline household survey including water quality testing; 3) filter distribution and training for participants; 4) three bi-weekly follow-up household surveys including water quality testing; and 5) focus group discussions (FGDs) with participants. The Emory University and CDC Institutional Review Boards approved the study. Informed consent was obtained before initiating household visits, at training sessions, and before focus group discussions.

Sampling strategy

The communities of Sokoni and Nambala in Arusha, Tanzania, were chosen for study inclusion because they were representative of the socioeconomic status and water collection and treatment practices of the overall program target population in Arusha. A total of 50 households were invited to participate in the study, the maximum permitted by time and resource constraints. The inclusion criteria for recipient families included: 1) living within either of the two communities; 2) having at least one child five years of age or younger living in the home; and, 3) attending an informational session before filter distribution. SWCEA asked community leaders to compile a list of 25 households per community that met the inclusion criteria. Purposive sampling with snowball recruitment was used; all participants were informed through word-of-mouth, which was appropriate for the community-oriented, low-resource setting.

Flow rate testing

Before filter distribution, the 50 filters provided by SWCEA for the study were flow rate tested. Saturated filters were placed in empty receptacles and completely filled with untreated tap water. After 30 minutes, the volume of water in the receptacle was measured, recorded, and converted to obtain the flow rate in L/hr. Filters were numbered and linked to households during distribution. The correlation between filters meeting the flow criteria (1–3 L/hr) and filter breakage during the study was assessed using Fisher's Exact Test in Stata 10.0 (College Station, TX, USA).

Household surveys

Each household was visited and surveyed four times; once at baseline before filter distribution, and at follow-up visits 2, 4, and 6 weeks after receiving the filter. The baseline survey included 64 questions about household demographics and attitudes toward water use and safe storage. Follow-up surveys included 26 to 31 questions about participants’ filter usage behaviors and satisfaction with filters. Surveys employed Likert scales, multiple choice responses, open-ended questions, and enumerator observation. Enumerators were trained before conducting surveys, informed consent was obtained from all participants, and survey data were entered into Microsoft Access 2007 (Redmond, WA, USA) and analyzed using SPSS 21.0 (Chicago, IL, USA).

Focus group discussions

Six weeks after filter distribution, semi-structured focus group discussions (FGDs) were held; all study participants were invited to attend. FGDs elicited responses about participants’ perceptions of the filters’ utility and effectiveness, as well as families’, friends’, and other community members’ perceptions. Participants provided qualitative depth to quantitative survey results. FGDs were recorded with participants’ consent, then translated from Kiswahili, transcribed, and analyzed by extracting quotes representing identified themes.

Water quality testing

Microbiological indicator and turbidity tests were conducted on untreated water collected from users’ homes at the baseline survey, and on filtered water at the first two follow-up surveys. Sampling was not completed at the third follow-up survey due to resource constraints. Samples were collected in sterile WhirlPak™ bags containing sodium thiosulfate, stored in a cooler containing icepacks, and tested within 8 hours of collection using membrane filtration on a Millipore (Billerica, MA, USA) portable filtration stand. Samples were diluted appropriately with sterile buffered water, filtered aseptically through a 45-μm Millipore filter, placed in a plastic Petri dish with a pad soaked with mColiBlue24 media to test for total coliforms and E. coli, and incubated for 24 hours at 37 °C. Duplicate tests were run on 10% of the samples.

The geometric mean of E. coli in each community was calculated, with E. coli values of <1 transformed to 0.5 for analysis. E. coli data were analyzed using the ‘effective use’ metric (Lantagne & Clasen 2012), calculating the percent of households that used the filter to treat contaminated source water to uncontaminated levels (at both the <1 and <10 CFU/100 mL E. coli breakpoints) (WHO 1997). The correlation between filters meeting flow criteria and having average (at follow-ups 1 and 2) filtered water E. coli of <10 CFU/100 mL was assessed using Fisher's Exact Test in Stata 10.0.

Turbidity was tested using a LaMotte 2020 turbidimeter calibrated with non-expired stock calibration solutions. Outliers were defined as points with a Cook's distance greater than 1, and were removed from analysis. A two-tailed paired t-test was used to determine whether there was a significant reduction in turbidity between baseline and the two follow-up visits.

RESULTS

Flow rate testing

The average flow rate in the filters was 3.8 L/hr (min = 0.3, max = 10.8, stdev = 2.4). Less than half (46%) of SWCEA filters provided for this study met the standard flow rate criteria of 1–3 L/hr; 4% were <1 L/hr, 28% were 3.1–6.0 L/hr, and 22% were 6.1–11.0 L/hr.

Household surveys

Baseline survey

All 50 participant households answered the baseline survey. The average respondent age was 36, most respondents were mothers (82%), most were literate (92%), and, on average, respondents had attended school for seven years (Table 1). The average household consisted of six people. All households had at least one child ≤5 years old, as per the intended inclusion criteria. The median monthly income for all participating households was 60,000 Tanzanian shillings (TSH), equivalent to 40 USD. The majority of households, 22 of 25 (88%) in each community, reported using untreated, unchlorinated public tap water for drinking. The remaining households reported using a river/creek (four households), city supply (one household), and running water in house (one household).

Table 1

Selected characteristics of study participants

 SokoniNambalaTotal
Number of participant households 25 25 50 
Number (%) respondents mothers 76 (19%) 88 (22%) 82 (41%) 
Mean age (range) 37.4 (25–70) 34.3 (20–61) 35.9 (20–70) 
Mean years of school (range) 6.8 (0–13) 7.1 (4–12) 6.9 (0–13) 
Number (%) respondents literate 21 (84%) 25 (100%) 46 (92%) 
Mean household size (range) 6.2 (3–10) 5.1 (3–8) 5.7 (3–10) 
Mean no. children ≤5 (range) 1.5 (1–3) 1.4 (1–4) 1.5 (1–4) 
Median monthly income, TSH (range) 50,000; (10,000–300,000) 65,000; (20,000–180,000) 60,000; (10,000–300,000) 
Number (%) using public tap water 22 (88%) 22 (88%) 44 (88%) 
 SokoniNambalaTotal
Number of participant households 25 25 50 
Number (%) respondents mothers 76 (19%) 88 (22%) 82 (41%) 
Mean age (range) 37.4 (25–70) 34.3 (20–61) 35.9 (20–70) 
Mean years of school (range) 6.8 (0–13) 7.1 (4–12) 6.9 (0–13) 
Number (%) respondents literate 21 (84%) 25 (100%) 46 (92%) 
Mean household size (range) 6.2 (3–10) 5.1 (3–8) 5.7 (3–10) 
Mean no. children ≤5 (range) 1.5 (1–3) 1.4 (1–4) 1.5 (1–4) 
Median monthly income, TSH (range) 50,000; (10,000–300,000) 65,000; (20,000–180,000) 60,000; (10,000–300,000) 
Number (%) using public tap water 22 (88%) 22 (88%) 44 (88%) 

About a quarter (26%) of respondents thought their drinking water supply was safe, the majority (66%) thought it was unsafe, and a few (8%) were unsure. Of the 33 participants who thought their water was unsafe, most based their assessment on visual inspection of the ‘dirtiness’ of the water (37%) and/or on the belief that their water contained microbes (31%). About a third of participants (36%) said they always treated their water, all of whom reported boiling. Four households (8%) also reported filtering water through a cloth and/or using WaterGuard-brand chlorine. Using Fisher's exact test, no significant differences in these demographic and water perception indicators were found between the communities of Sokoni and Nambala.

Follow-up surveys

At 2, 4, and 6 weeks post-distribution, 47, 49, and 47 households were surveyed, respectively. All respondents self-reported using the filters during the previous week, 96–100% reported using the filters the previous day, and 74% (n = 35), 53% (n = 26), and 54% (n = 25) reported having treated their water on the day of the survey at the three follow-up surveys, respectively. Additionally, all households at all three follow-up surveys had filtered water available in the storage container for sampling.

All respondents, at all three follow-ups, reported that filters produced enough water and were easy to use. Additionally, 100%, 98%, and 96% of respondents were satisfied or very satisfied with the time for filtration at the three follow-up surveys, respectively.

As per the study protocol, filter components were always replaced when broken; a total of nine (18%) filters broke and were replaced during the 6-week study. Additionally, leaky or broken taps were reported by five respondents over the three follow-up surveys (10%). Breakage/replacement of filters was not statistically significantly correlated with initial filter flow rates of <1.0 L/hr (p = 0.139), >3.0 L/hr (p = 1.000), or the correct flow rate of 1.0–3.0 L/hr (p = 0.277).

A total of 91%, 94%, and 100% of respondents reported the taste of filtered water was satisfactory or excellent at the three follow-up surveys, respectively; the same was true for the appearance of filtered water. However, respondents reported liking the smell of the filtered water slightly less, with 83%, 81%, and 100% satisfaction at the three follow-up surveys, respectively.

During the final follow-up survey, additional user satisfaction questions were asked. Overall, 94% of respondents were satisfied or very satisfied with the filters in general. Respondents attributed their satisfaction with their filters to improvements in the overall quality of their water (n = 37, 79%) (i.e., safety, clarity, temperature, taste, and smell) and convenience of filter usage (n = 14, 30%). All respondents (n = 47) reported that they would recommend the filter to friends and family members. When asked if they would have purchased a filter (if they had not received one for free), most respondents (n = 34, 72%) said they would not have because they would not have known about the filter or where to find it. Most respondents (n = 46, 98%) reported that they would purchase a replacement for their filters when they were no longer usable.

When asked what they might change about the filter, half (n = 24, 51%) of respondents reported that they would not make any changes. Almost a quarter of respondents (n = 11, 23%) suggested making the filters bigger. Others suggested changes to the plastic lid of the receptacle in which the filter sits, including having the lid fit completely over the receptacle and filter (n = 9, 19%); provision of spare parts to filters (i.e., extra taps, stands, strainer for larger particles) (n = 3, 6%); and, improving filters’ effectiveness to eliminate more bacteria (n = 1, 2%).

Post-implementation focus group discussions

Nine (36%) participants in Sokoni and 15 (60%) participants in Nambala attended the post-implementation FGDs. Participants were asked to explain why, before the study distribution of filters, a minority (n = 18, 36%) reported always treating their water, despite the fact that over half (n = 33, 66%) felt their water was unsafe. One participant answered:

I don't use those practices [of boiling water]. I drink the water from the tap as they are… One day I boiled the water, then [the water] smelled [like] smoke from the firewood, so even when the guests came I cannot give that water. I decided to pour [it] out… [and] continue to use the water from the tap.

Another participant explained:

I'm saying that the problem which makes us not to boil water is the case of firewood. We don't have firewood. We like so much to boil. We like to treat our water, even sometimes to use WaterGuard, [but] sometimes if you use WaterGuard, it will make our head ache and that's another problem which makes us to stop treating our water.

Some participants recommended education as a way to address the discrepancy between recognizing unsafe water and acting to treat it. As one participant explained:

Our parents were not boiling the water. There were no taps. We just used the water from the streams and wells. So the habit of treating water was not there. The way to help people [is by] giving education [on] the ways they can treat water.

Participants also provided feedback on challenges they faced when using the filter. They felt that filters were confusing to clean and suggested providing instructions to help with maintenance. Although instructed to clean filters with boiling water, participants sometimes used tap water or cleaned filters more frequently than recommended. One participant explained:

The problem is we just keep forgetting. So the people just take the easy way to wash it with tap water. Other people received the filter, and we were not there, we were at home, so we were not told the instructions on how to clean it.

Participants also discussed the need for decreasing the costs of filters, which cost approximately 25,000 TSH (16 USD), or allowing consumers to pay for filters in installments. In the words of one participant:

It is better in your organization/group to choose someone who will be there so that he can collect the little payment from the people until they finish their payment and then give them their filters for other people, they cannot afford to pay that amount of money in the right time. So to pay it slowly, slowly, slowly, slowly 25,000 TSH will be paid and then the problem is solved.

Water quality testing

Source water turbidity was relatively low at baseline, at an average of 0.55 NTU (Table 2). There was a significant reduction in turbidity from baseline to the first follow-up (p = 0.020), and from baseline to the second follow-up (p < 0.001).

Table 2

E. coli counts, turbidity, and effective use in Sokoni and Nambala at baseline and follow-ups

 Baseline (n = 50)Follow-up 1 (n = 49)Follow-up 2 (n = 49)
E. coli geometric mean (CFU/100 mL) (range) 22.4; (<1–4,900) 2.8; ( < 1–200) 1.2; ( < 1–79) 
Turbidity mean (NTU) (min, max, stdev) 0.55; (0.01–1.93, 0.36) 0.38; (0.00–1.18, 0.30) 0.28; (0.00–1.50, 0.25) 
Effective use (<1 CFU/100 mL) – 10% (5) 24% (12) 
Effective use (<10 CFU/100 mL) – 37% (18) 39% (19) 
 Baseline (n = 50)Follow-up 1 (n = 49)Follow-up 2 (n = 49)
E. coli geometric mean (CFU/100 mL) (range) 22.4; (<1–4,900) 2.8; ( < 1–200) 1.2; ( < 1–79) 
Turbidity mean (NTU) (min, max, stdev) 0.55; (0.01–1.93, 0.36) 0.38; (0.00–1.18, 0.30) 0.28; (0.00–1.50, 0.25) 
Effective use (<1 CFU/100 mL) – 10% (5) 24% (12) 
Effective use (<10 CFU/100 mL) – 37% (18) 39% (19) 

At baseline, 50 stored household water samples were collected and analyzed for E. coli; 49 samples of filtered water from the filter container were collected at each of follow-ups 1 and 2. The geometric mean E. coli count fell from 22.4 CFU/100 mL in household water at baseline to 2.8 and 1.2 in filtered water at follow-up 1 and follow-up 2, respectively (Table 2). Before filter installation, only seven households (14%) met the <1 CFU/100 mL WHO guideline value for drinking water. An additional seven (14%) households (for a total of 14 (28%) households) met the more lenient ≤10 CFU/100 mL guideline.

In total, five (10%) and 12 (24%) households moved from higher risk to the <1 CFU/100 mL E. coli guideline at the first and second follow-up survey, respectively. A total of four (8%) households achieved effective use at this guideline at both follow-ups (Table 2). In total, 18 (37%) and 19 (39%) of households moved from higher risk to the ≤10 CFU/100 mL E. coli guideline at the first and second follow-up survey, respectively. A total of 17 (35%) households achieved effective use at the ≤10 CFU/100 mL guideline at both follow-ups. Having treated water E. coli concentrations of <10 CFU/100 mL was not statistically significantly correlated with initial filter flow rates of <1.0 L/hr (p = 0.180), >3.0 L/hr (p = 0.479), or the correct flow rate of 1.0–3.0 L/hr (p = 0.152).

DISCUSSION

We conducted a study in Arusha, Tanzania, to determine the effectiveness and acceptability of filters manufactured by a recently-established local factory. We found the filters were highly acceptable, slightly effective at removing E. coli, and, unexpectedly, that they were poorly manufactured. These results indicate that filters have the potential to be an acceptable HWT intervention, but face challenges in manufacturing that influence filter effectiveness in the household.

Overall, filter use and satisfaction were very high. All respondents surveyed had filtered water available for sampling at all three follow-up surveys, which could be, in part, because of the short study length of only 6 weeks. Respondents attributed their satisfaction with the filter technology to its improvement of water quality and convenience, and suggested making the filter bigger and having the plastic lid fit better. During FGDs, participants confirmed their acceptance of filters and suggested lower costs, improved cleaning instructions, and aesthetic changes to the filter design as approaches to reducing barriers to filter use. Participants explained that major barriers to water treatment, in general, include lack of education about water treatment methods and undesirable changes to the taste and smell of water from treatment methods, such as boiling and chlorination.

In spite of this high user acceptance, we documented quality problems with the distributed filters. First, flow rates were inconsistent and, on average, above the recommended 1–3 L/hr range, demonstrating poor manufacturing consistency and quality control. These abnormally high flow rates may have contributed to over 95% of respondents reporting satisfaction with filtration time, normally considered a limitation of ceramic filtration (CMWG 2011). Flow rate was not found to be correlated with breakage or E. coli concentrations <10 CFU/100 mL, which is consistent with other data showing flow rate is a good indicator of manufacturing process control, but not filter efficacy (CMWG 2011). Second, E. coli removal was incomplete and inconsistent, as only 8% of filters effectively treated contaminated water to the <1 CFU/100 mL guideline value at both follow-up surveys, and only 35% effectively treated contaminated water to <10 CFU/100 mL. Third, 18% of filters broke over the 6 weeks of the study, equivalent to a 10% breakage rate per month. This is compared to a <2% per month disuse rate in a study in Cambodia (Brown et al. 2009) and a breakage rate of 10% over two years in a study in Sri Lanka (Casanova et al. 2012). This breakage rate indicates low filter strength, and poses a major threat to continued usage. These three problems are all indicative of manufacturing process deficiencies.

The identification of these manufacturing deficiencies was an unexpected result of the study. Filters are manufactured by pressing a mixture of locally sourced clay and a burnout material (such as sawdust or rice husk, which burns out in firing and creates the pore structure of the filter) into the filter shape (CMWG 2011). After pressing, filters are allowed to dry before being fired to a ceramic state (∼800–900 °C). The specific clay to burnout material ratio is determined during factory establishment by testing prototype filters for flow rate and microbiological efficacy. Silver is added as a bactericide, either by application to fired filters or by inclusion in the filter mixture. Although the manufacturing process is not complicated, several critical variables need to be controlled to ensure final product quality. As manufacturing processes were not investigated in this study, we are not able to determine the exact process deficiencies leading to poor filter quality.

Independently of this study, but also due to concerns about HWT product quality, the WHO has developed and launched an international certification scheme for HWT products. WHO guidance includes tiered, health-based targets which classify HWT products according to laboratory efficacy performance as: ‘***’ (4-log bacteria and protozoa reduction and 5-log virus reduction in laboratory settings), ‘**’ (2-log bacteria and protozoa reduction and 3-log virus reduction), or ‘*’ (achieving ** targets for two pathogen classes and having epidemiological evidence demonstrating disease reduction in health impact trials) (WHO 2015a, 2015b).

The results from the first round of the WHO certification scheme were released in early 2016 (WHO 2016). Of the ten HWT products evaluated, one was a ceramic filter locally manufactured in Tanzania (please note it was not the SWCEA filter, but another filter named TEMBO). According to the manufacturer, the flow rate for the TEMBO filter is intended to be 1–5 L/hr. However, during WHO testing a flow rate of <0.5 L/hr was observed. Although full laboratory testing was planned, testing was discontinued due to the low flow rate, and the TEMBO filter pot performance was found to be ‘undetermined’.

Subsequent to the study documented herein, research was conducted in coordination with the Ceramics Manufacturing Working Group (CMWG) to document manufacturing practices in over 50 independent filter production facilities and establish a certification scheme to verify manufacturing facilities. To develop a draft certification scheme, the recommendations of the CMWG Best Practice manual were translated into a factory questionnaire and evaluation protocol conducted by an external evaluator, for the quality indicators of: 1) 99% reduction of E. coli in the laboratory setting; 2) manufacturing practices suggestive of consistent production; 3) documentation of manufacturing processes; and 4) health and safety (Rayner & Lantagne 2014). Four facilities that imported filters into Haiti as part of cholera response activities were evaluated using these criteria: two of the four met the 99% reduction criteria, the documentation, and the health and safety criteria, and none met the consistent production criterion. The two factories that met the 99% reduction criteria had E. coli log reduction values of 3.5–7.1, showing from ** to *** removal according to WHO criteria.

These results – from the two of four factories that did not meet the 99% reduction criteria in the CMWG evaluation and the one factory that had ‘undetermined’ results from the WHO certification scheme – highlight that the SWCEA Tanzanian factory experience is not an isolated incidence of poor production of ceramic filters. WHO and the CMWG are currently working to determine how to incorporate locally manufactured products (such as ceramic filters, sodium hypochlorite solutions, and biosand filters) into certification processes (WHO 2015a, 2015b), either through the existing WHO scheme or through alternate mechanisms.

Our study was conducted only a few months after the SWCEA factory was established, in 2008. Immediately after the study, and based on our results, SWCEA instituted a series of changes to improve manufacturing processes and quality control. Filters manufactured by SWCEA have been successfully used in subsequent studies in Tanzania, and found to have the highest adoption rate and potential for scaling-up of four distributed HWT products (Burt et al. 2014). Our work thus highlights the importance of having experienced technical assistance on-site at factory initiation to ensure filter quality, as recommended by the CMWG (2011).

Additionally, our results highlight the added value of mixed methods research, the importance of testing the laboratory efficacy of locally manufactured products before distribution, and the relationships between filter manufacturing quality, filter use over time (effectiveness), and user acceptance. Regardless of high user acceptance, HWT methods that inconsistently remove pathogens will not adequately reduce the burden of diarrheal disease (Brown & Clasen 2012). When manufactured correctly, filters are capable of providing sufficient quantities of microbiologically safe water, require a small time commitment from users, and are available and accessible to users since they can be made locally (Sobsey et al. 2008). When manufactured poorly, filters may retain some of these benefits, but – unbeknownst to the user – fail to ensure microbiologically safe water or to reduce the risk of diarrheal disease.

The small sample size and short duration of this study were among its limitations. Also, social desirability bias and recall bias may have influenced survey and FGD results; despite the fact 98% (49/50) respondents had treated water at the time of the first two follow-ups, we cannot confirm they were drinking it. In addition, E. coli was measured at baseline in untreated source water and in treated water at follow-up, but follow-up untreated source water was not directly sampled for comparison with follow-up treated water – therefore unmeasured changes in the quality of source water could have affected our evaluation of filter performance. Overall, 16% of filtered samples collected in follow-up surveys had higher E. coli than at baseline, and 7% of filtered samples collected in follow-up surveys had higher E. coli than at baseline and had E. coli values greater than 10 CFU/100 mL. Thus, while unmeasured changes may have been present, their magnitude was not large. Despite these limitations, we feel the results drawn from this study provide important insight into the challenges of maintaining the quality, effectiveness, and acceptability of locally-manufactured HWT products. We recommend further research on determining the production variables key to maintaining quality control of locally manufactured filters, and in developing certification schemes for local manufacturing facilities.

CONCLUSIONS

Filters were considered an acceptable HWT method among study participants in Arusha, Tanzania, as participants were satisfied with the amount and quality of water they produced. The main user-perceived barriers to filter usage were their cost and insufficient training about cleaning methods. However, water quality improvement was inconsistent and lower than anticipated, and key deficiencies in filter quality were identified. Our research shows that filters had desirable implementation factors for the target population, and with improved quality and effectiveness could be a promising long-term HWT option.

DISCLAIMER

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

REFERENCES

REFERENCES
Baqui
A. H.
Black
R. E.
Sack
R. B.
Chowdhury
H. R.
Yunus
M.
Siddique
A. K.
1993
Malnutrition, cell-mediated immune deficiency, and diarrhea: a community-based longitudinal study in rural Bangladeshi children
.
Am. J. Epidemiol.
137
(
3
),
355
365
.
Bielefeldt
A. R.
Kowalski
K.
Summers
R. S.
2009
Bacterial treatment effectiveness of point-of-use ceramic water filters
.
Water Res.
43
(
14
),
3559
3565
.
Bielefeldt
A. R.
Kowalski
K.
Schilling
C.
Schreier
S.
Kohler
A.
Summers
R. S.
2010
Removal of virus to protozoan sized particles in point-of-use ceramic water filters
.
Water Res.
44
(
5
),
1482
1488
.
Brown
J.
Sobsey
M. D.
Loomis
D.
2008
Local drinking water filters reduce diarrheal disease in Cambodia: a randomized, controlled trial of the ceramic water purifier
.
Am. J. Trop. Med. Hyg.
79
(
3
),
394
400
.
Burt
Z.
Mussa
M.
Mbatia
Y.
Msimbe
V.
Brown
J.
Clasen
T.
Malebo
H.
Ray
I.
2014
User Preferences and Willingness to pay for Household Water Treatment in Rural Tanzania
.
UNC Water and Health: Where Science Meets Policy
,
Chapel Hill, NC
,
USA
.
Casanova
L. M.
Walters
A.
Naghawatte
A.
Sobsey
M. D.
2012
Factors affecting continued use of ceramic water purifiers distributed to tsunami-affected communities in Sri Lanka
.
Trop. Med. Int. Health
17
(
11
),
1361
1368
.
CMWG
2011
Best Practice Recommendations for Local Manufacturing of Ceramic Pot Filters for Household Water Treatment
. 1st Edn.
Ceramic Manufacturing Working Group, Centers for Disease Control and Prevention
,
Atlanta, GA
,
USA
.
du Preez
M.
Conroy
R. M.
Wright
J. A.
Moyo
S.
Potgieter
N.
Gundry
S. W.
2008
Use of ceramic water filtration in the prevention of diarrheal disease: a randomized controlled trial in rural South Africa and Zimbabwe
.
Am. J. Trop. Med. Hyg.
79
(
5
),
696
701
.
Fewtrell
L.
Colford
J. M.
Jr.
2005
Water, sanitation and hygiene in developing countries: interventions and diarrhoea--a review
.
Water Sci. Technol.
52
(
8
),
133
142
.
Fischer-Walker
C. L.
Perin
J.
Aryee
M. J.
Boschi-Pinto
C.
Black
R. E.
2012
Diarrhea incidence in low- and middle-income countries in 1990 and 2010: a systematic review
.
BMC Public Health
12
,
220
.
Guerrant
D. I.
Moore
S. R.
Lima
A. A.
Patrick
P. D.
Schorling
J. B.
Guerrant
R. L.
1999
Association of early childhood diarrhea and cryptosporidiosis with impaired physical fitness and cognitive function four-seven years later in a poor urban community in northeast Brazil
.
Am. J. Trop. Med. Hyg.
61
(
5
),
707
713
.
Lantagne
D.
2001a
Investigations of the Potters for Peace Colloidal Silver Impregnated Ceramic Filter. Report 1: Intrinsic Effectiveness
.
Alethia Environmental
,
Allston, MA
,
USA
.
Lantagne
D.
2001b
Investigations of the Potters for Peace Colloidal Silver Impregnated Ceramic Filter. Report 2: Field Investigations
.
Alethia Environmental
,
Allston, MA
,
USA
.
Lantagne
D.
Klarman
M.
Mayer
A.
Preston
K.
Napotnik
J.
Jellison
K.
2010
Effect of production variables on microbiological removal in locally-produced ceramic filters for household water treatment
.
Int. J. Environ. Health Res.
20
(
3
),
171
187
.
Onda
K.
LoBuglio
J.
Bartram
J.
2012
Global access to safe water: accounting for water quality and the resulting impact on MDG progress
.
Int. J. Environ. Res. Public Health
9
(
3
),
880
894
.
Oyanedel-Craver
V. A.
Smith
J. A.
2008
Sustainable colloidal-silver-impregnated ceramic filter for point-of-use water treatment
.
Environ. Sci. Technol.
42
(
3
),
927
933
.
Rayner
J.
Lantagne
D.
2014
Investigations on Household Filtration in Haiti: Final Report
.
Tufts University for the Centers for Disease Control and Prevention
,
Medford, MA
.
Available from corresponding author
.
WHO
1997
Guidelines for Drinking-Water Quality, 2nd Edition: Volume 3; Surveillance and Control of Community Supplies
.
World Health Organization
,
Geneva
,
Switzerland
.
WHO
2015a
Strategic Meeting of the WHO International Scheme to Evaluate Household Water Treatment Technologies: Nieuwegein, the Netherlands, 23–24 March, 2015
.
World Health Organization
,
Geneva
,
Switzerland
.
WHO
2015b
WHO International Scheme to Evaluate Household Water Treatment Technologies
.
World Health Organization
,
Geneva
,
Switzerland
. (accessed 26 May 2016).
WHO
2016
Results of Round 1 of the WHO International Scheme to Evaluate Household Water Treatment Technologies
.
World Health Organization
,
Geneva
,
Switzerland
.
WHO/UNICEF
2015
Progress on Drinking Water and Sanitation: 2015 Update and MDG Assessment
.
World Health Organization and UNICEF
,
Geneva
,
Switzerland
and New York City, NY, USA. Available at: http://www.unicef.org/publications/files/Progress_on_Sanitation_and_Drinking_Water_2015_Update_.pdf (accessed 26 May 2016).