| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Many people living in developing countries are still reliant on water of poor quality. Figures reported in 2004 by the Joint Monitoring Program showed that, of a population of 734.6 million in sub-Saharan Africa, 56% had no access to a water supply.1 In South Africa, 34% of households did not have access to a water supply in 2000.2 For children younger than 5 years of age in South Africa, diarrhea is the third most important cause of death, after HIV/AIDS and low birth weight, representing 11% of all deaths in that age group.3 Worldwide diarrheal disease is one of the leading causes of morbidity and mortality in developing countries, accounting for 21% of all deaths in children younger than 5 years old and a total of 2.5 million deaths per year.4
The magnitude of the task of providing people without sustainable access to improved water supply with the ultimate option, piped water, in the near future has initiated studies into alternative technologies. Point of use interventions are fast becoming the preferred method for providing improved water quality5 and consequently reduction in diarrheal disease.6,7 A meta-analysis8 confirmed that point-of-use water treatment is more effective than had been previously thought. Among 37 treatment technologies reviewed for a variety of their characteristics, which included maintaining and improving microbial water quality, health impact, and costs, gravity filtration using ceramic filters was indicated as one of the five most promising technologies.9 When used appropriately, ceramic filters provide an immediate source of drinking water with reduced turbidity and up to 99.99% removal of bacteria and protozoan parasites.
To our knowledge, there have been only four field trials published that document diarrhea reduction in households using ceramic filters.7,10–12 A number of field trials using home-based chlorination methods,13,14 various combinations of flocculation and chlorination,15–17 and solar disinfection18–20 also documented the effectiveness of these water treatment methods for reduction of diarrhea incidence. However, these trials and others did not differentiate between non-bloody and bloody diarrhea.
Bloody diarrhea is often associated with shigellosis. Infections associated with Shigella spp. comprise a considerable portion of the diarrheal burden of disease. The most recent report on the global burden of Shigella infections estimated the annual number of diarrheal episodes related to Shigella spp. to be 164.7 million, of which 163.2 million were in developing countries.21 Recent studies on the Asian continent showed that incidence of Shigella diarrhea is more ubiquitous in Asian impoverished populations than previously thought22 and that the incidence is substantially underestimated in children and older people in impoverished communities.23 Shigella infections are often, but not exclusively,24 associated with bloody diarrhea and are also referred to as shigellosis or dysentery.25 Except for ad hoc outbreaks that are reported in the literature, information on the prevalence and incidence of Shigella infections on the African continent is limited.
To determine how effective point of use ceramic filtration is in preventing non-bloody diarrhea and bloody diarrhea, we conducted a controlled trial in two rural areas in southern Africa, assessing the incidence of non-bloody diarrhea and bloody diarrhea in children 24–36 months of age. Our study made use of pictorial diaries26 that enabled distinguishing between non-bloody diarrhea and bloody diarrhea and enumeration of each.
The study was conducted in two districts, Mutale in the Vhembe district, Limpopo province, South Africa, and the Zaka district, Masvingo province, Zimbabwe. The population of the Limpopo province was 5,227,432 in 2000, with a density of 43 persons/km2.27 Forty percent of the population of 69,313 residing in the South African municipality of Mutale in 2001 had no access to sanitation.28 The majority (53%) of the population used public standpipes, with a further 14% using rivers, dams, boreholes, and springs as sources of water.
The Zaka district in Zimbabwe had a population of 184,814.29 Forty-five percent of the population in the district had no access to protected water sources, and 70% had no access to any type of sanitation facility.30 More recent 2002 census data are not available. Both study sites therefore are below national averages in access to improved water supplies and sanitation.
During a previous study,31 we investigated the change in water quality between source and point of use and the linkages between water quality and diarrhea among children 12–24 months of age. To recruit households to this earlier study, health centers in the study areas were ranked by rates of diarrhea in children 
5 years of age. Two villages, associated with each of the worst five health centers in each area, were selected. From each of the 10 selected villages, 12 households with children 12–24 months of age were randomly chosen. These households were drawn from lists from local birth registers in South Africa and lists drawn up by community health workers in Zimbabwe. Any households with an in-house piped water connection were excluded from this earlier study. For the intervention trial, 115 households that formed part of the earlier study were randomly selected from six villages in each of the two study areas.
The selected households were randomized into two groups by a senior researcher (RC). One half was provided with a commercially available ceramic water filter (British Berkefeld, England), whereas the other half formed the control group and continued using their usual in-house storage containers. The control group was given water filters at the end of the study. Each intervention household received four ceramic filter candles that were impregnated with silver and contained carbon for the removal of unwanted tastes. The flow rate for four candles is 80 L in 24 hours. Each household member was shown how to assemble, fill, and clean their filters. They were instructed not to open or clean the lower part of the vessel containing the filtered water. For practical reasons common to other published home water treatment interventions, neither participants nor enumerators or researchers were blinded to group assignment. This was because some source water in both our Zimbabwean and South African study areas had high turbidity levels or floating particles in the water apparent to the naked eye. The British Berkefeld filters reduced turbidity completely. Sham filters would not be able to completely reduce the turbidity, which would have made them easily identifiable, both to participating households and project staff.
Water samples were analyzed for the presence of Escherichia coli using the Colilert-18 method. Water samples were collected once at the end of the study from the storage containers in the control households and directly from the tap after filtration from intervention households. Baseline information regarding the type of household, level of education, hygiene practices, and access to sanitation, water source, and rubbish disposal methods were also documented.
All children in each household between 24 and 36 months of age were followed to determine the incidence of non-bloody diarrhea and bloody diarrhea. Adult care givers were trained to daily record episodes of diarrhea using a pictorial "smiley" diary.26 The diary recorded frequency of diarrhea episodes and distinguished between loose watery stools and stools containing blood or mucus. Trained enumerators collected the sheets at the end of each week, after which the sheets were checked by a project researcher and captured electronically. Days when a child experienced three or more loose or watery stools or one loose stool containing mucus or blood were considered as diarrhea-days, following guidance from earlier studies.32 This information was collected for the period October 2003–April 2004 in South Africa and November 2003–June 2004 in Zimbabwe. We also recorded the rate of bloody and non-bloody diarrhea for each participating child in the observation period before the trial started (average: 117 days of observation per child) to allow us to adjust for pretrial disease rates.
The trial was approved by the Research Ethics Committee of the Royal College of Surgeons in Ireland and by the appropriate regional bodies in RSA and Zimbabwe. Household participation was subject to informed consent from the head of the household. Community consent was also obtained by consulting with community elders. All participant information was translated into local languages.
Baseline demographic information and other information were analyzed with Systat version 8, using the Pearson’s 
2 test and the two-tailed Fisher exact test when appropriate.
Presence of E. coli contamination in household drinking water was modeled by Stata’s binomial regression command, which allows the calculation of relative risks as a measure of effect size.
Diarrheal data were analyzed using Poisson regression, with robust variance estimation to account for clustering of diarrhea at village and household level, using Stata/SE release 10. Poisson regression is suited to the analysis of events with low incidence rates. The use of Poisson regression allowed us to account for the differences between children in number of recorded days, giving greater weight to children with more complete follow-up. The Poisson distribution is characterized by a single parameter: the incidence rate. This makes it easily interpretable in the reporting of risk data. Groups can be compared by calculating an incidence rate ratio. To check for a differential effect of filtration in the two countries, an interaction term was added to the model. Results are expressed as predicted 30-day incidence rates, and the filter and control groups were compared using incidence rate ratios with their associated 95% confidence intervals (CIs). In each case, comparisons were adjusted for differential incidence rates in the two countries and for diarrhea incidence in the pretrial observation phase. We defined an "event" as a day on which diarrhea occurred. The 30-day incidence rates are therefore the predicted number of days of diarrhea in a 30-day period. This does not take into account clustering of days into arbitrarily defined "episodes."
Baseline data for the aggregate control and intervention groups are shown in Table 1
. Forty-one of the total number of intervention and control South African households used standpipes. The remaining households used unprotected springs and unprotected wells, protected boreholes, and river and canal water. The majority of the households in Zimbabwe used unprotected springs. Only three households had access to standpipes, 14 used boreholes, and the remaining households used river water and unprotected wells.
|
Although there was no follow-up study beyond 6 months to determine sustained use of the filters, acceptance during the trial was very high. When questioned during the trial about filter use, only one household refused to use the filter. No breakage was reported during the trial. Forty-three were asked their opinion about the speed of the filtration process: 90.7% said it was "about right," 4.7% found it slightly too slow, and 4.7% found it too slow. Two of the 43 respondents had a few difficulties when cleaning the filters. The remaining 41 respondents had no difficulties. The filters were therefore well received by almost all participants.
The water was tested once for the presence of E. coli at the end of the study through an unannounced visit. Table 2 summarizes the percentage compliance of E. coli counts with the World Health Organization’s 2004 drinking water quality guidelines for control and intervention households in South Africa and Zimbabwe and for combined control and intervention groups. As shown in Table 2, we were unable to obtain water samples for 12 of 106 households, either because filters were empty or because households were absent on the day of the visit. A higher rate of compliance, 55% and 73.9%, was observed for both the control and intervention households in South Africa, respectively. In comparison, water in Zimbabwe showed compliance for 30.2% of the control households and 56.9% for the intervention households. We also examined the predictors of presence of E. coli in household drinking water using binomial regression. Presence of E. coli in source drinking water was associated with a 6-fold increase in risk of E. coli contamination in household water (relative risk [RR] = 6.1; 95% CI = 2.4–15.6). Adjusted for source contamination, there was a marginally lower rate of contamination in South African households compared with Zimbabwe (RR = 0.77; 95% CI, 0.57–1.03). Adjusted for both of these variables, households with filtered water had a lower risk of E. coli in household drinking water, with a relative risk of 0.67 (95% CI = 0.50–0.89; P = 0.006).
|
We modeled the effect of filtration using Poisson regression. We first examined the data for a potential differential effect in the two countries by using an interaction term in the regression model. The interaction effect between filtration and country was not statistically significant for total diarrhea (P = 0.77) or for non-bloody diarrhea (P = 0.90) alone or bloody diarrhea alone (P = 0.08), indicating that the effect of filtration did not differ between the two countries. Poisson regression showed a significant difference in the incidence rate of all types of diarrhea between Zimbabwe and South Africa. Adjusted for this difference, children receiving filtered water had a significantly lower incidence of bloody, non-bloody, and total diarrhea. Table 3 shows the 30-day incidence rate of all three endpoints in the filter and control groups, together with the associated incidence rate ratios.
|
The provision of ceramic water filters was associated with a lower incidence of both non-bloody and bloody diarrhea in children living in rural South Africa and Zimbabwe. During the 6-month intervention the filtered water was associated with a significant reduction in both bloody and non-bloody diarrhea and the provision of drinking water with a reduced E. coli count.
Baseline information showed no significant differences between the aggregate control and intervention households. However, significant differences between the two study sites for the type of house, level of education of carers, availability of soap for washing hands of adults and for washing children, water source type, in-house storage methods, and access to a rubbish pit were evident. This is to be expected because the sites were situated in different countries representing different circumstances and political dispensations. The South African households enjoyed a slightly higher degree of economic status as shown by better housing facilities, availability of soap, and a greater number of protected water sources compared with Zimbabwe. Political unrest in Zimbabwe was already rife when the study took place and contributed to the lower economic status of the households in Zimbabwe.
Although this study’s focus was not to test the performance of the filter for the removal of bacteria, a one-off measurement for the presence of E. coli indicated that the ceramic filters reduced the risk of having E. coli present in the drinking water of the intervention households. Much effort has gone into the provision of greater quantities and better microbiological quality of water over the last 10 years in both rural and peri-urban areas of South Africa. Piped water, in the households, on site, or from a communal tap, was available in 78% of the households in the Limpopo Province.33 The majority of households taking part in the South African study had access to standpipes. The higher compliance rate to the 2004 World Health Organization’s drinking water quality guidelines for E. coli counts, observed in both storage and filtered water in South Africa may be concomitant to their access to standpipes.
There was no follow-up study to determine sustained use of the filters, but acceptance during the trial was very high. Only one household refused to use the filter. Generally the filters were perceived as prized items by the study participants and non-participants. This agrees with previous work, such as the high acceptance and continued use of ceramic filters showed in a study conducted in response to severe flooding in the Dominican Republic.34
Household water filtration reduced the incidence of both bloody and non-bloody diarrhea by 80%. Other studies using ceramic filters at point-of-use showed a reduction of 70%,11 45.3%,7 60%,10 and 46%.12 This is somewhat higher than the median reduction of 42% that was observed for an analysis of 21 studies consisting of water quality interventions at the point-of-use regardless of the nature of the intervention.6
The large reduction we observed relative to previous studies may in part be attributed to the high quality of the British Berkefeld ceramic filters used in the study. They have been independently tested and have absolute filtration ratings of 0.9 µm. The filters used in the study were impregnated with silver and contained carbon. They effectively remove Salmonella typhi, Shigella spp., Vibrio cholerae, Klebsiella, E. coli, and protozoan parasites such as Cryptosporidium and Giardia. Viral pathogens such as rotavirus, hepatitis A and E, and poliovirus are, however, not removed. If used appropriately, filtered water of good microbial quality can be obtained. Observations in the field suggested that the study had an additional effect of heightening awareness with regard to water contamination, bacteria, and hygiene practices such as hand washing. This may have also contributed to the large reduction observed in diarrhea incidence. Because safe water and sanitation provision are particularly poor in the two study sites relative to the national averages, the findings of this trial may not therefore be generalizable to the wider population in the two countries.
The statistically significant reduction of bloody diarrhea in this study, which is predictive of the presence of Shigella infection,35 although not exclusively,24 shows that ceramic filters can reduce the devastating effects of Shigella infections. Diarrheal infections caused by Shigella in particular affect malnourished children younger than 5 years old in impoverished areas and account for proportionately greater morbidity and mortality than non-bloody diarrhea.22 The possibility of outbreaks caused by one of a growing number of antibiotic-resistant strains23,24 and costly vaccination campaigns can be minimized or averted by interventions with the advantages of ceramic filters. The filter can easily be moved to new locations if necessary, introduce no chemicals into the water or environment, improve the water aesthetically, remove taste and odors, and is easy to use and readily accepted.
Dilute hypochlorite solution and solar disinfection, a zero cost technology, are the least expensive methods for treatment of water at the point of use. Approximate cost for chlorination has been estimated at $1.60 to $8 for initial cost of hardware (per capita; per household) and $0.60 to $3.00 for annual operating cost per capita per household.9 Both dilute hypochlorite solution16,36 and solar disinfection19,20 are effective in reducing the incidence of diarrhea. Ceramic filters treat water at the point of use, immediately providing water free of bacteria and parasites and consequently reduce diarrhea more effectively than the former methods. Providing people with ceramic filtration, however, is costly at present. The ceramic filter used for this study costs approximately $60 (
46), one candle costs approximately $6 (4.60), and replacement parts are not readily available in the areas the study was conducted. Most rural households will not have the financial means for the initial purchasing costs of a water filter. Although increased demand and local manufacture may drive down unit costs, it is possible that the technology will remain too expensive for some communities in poorer regions. However, in regions where financial security is increasing, promotion of manufacture of ceramic filtration systems may be warranted. Given the easy acceptance, easy use, and low maintenance of the filters and the possible provision of affordable replacement parts, long-term sustainability is perhaps not unattainable.
A limitation of the study was that it was not blinded, a shortcoming common to many intervention studies of home water filtration and storage. A recent systematic review,37 for example, found only four double-blinded intervention studies of home water treatment among 33 published reports. Because none of the double-blinded trials identified in this review found significant protective effects, it is possible that the observed improvements in waterborne disease are because of a placebo effect caused by lack of blinding. However, the impact of filtration on the presence of E. coli contamination in household drinking water is not explainable as response bias and would argue that the reported reduction in diarrhea rates is attributable to improved water quality. The extent to which the provision of filters may have changed household health behaviors, and thus had an indirect effect on diarrhea through, for example, an increased awareness of the importance of hygiene practices, is unknown, but would provide a suitable area for future research.
Ceramic filters were effective in reducing incidence of bloody and non-bloody diarrhea in this study. However, providing people with ceramic filtration is costly at present and out of reach for those living in some of the poorer regions. A higher percentage reduction in diarrheal disease was recorded using the ceramic filters than reported using solar disinfection, but the very low cost of solar disinfection still makes it a better choice in very poor regions. Ceramic gravity filters effectively improve water quality and reduces diarrheal disease, which make them an attractive alternative for households without safe water supply and in emergency conditions. Ceramic filters can provide safe water at the point of use within minutes. They are durable and easy to use, only minimum training is necessary in their use, and they are easily maintained. They effectively remove diarrheal-causing bacteria and parasites from unsafe water and consequently contribute to the overall health of people exposed to unsafe water.
Received September 1, 2007. Accepted for publication June 25, 2008.
Acknowledgments: The authors acknowledge the assistance of final year students at the University of Venda for Science and Technology, the Department of Health in the Limpopo Province, South. Africa, and the local authorities in Zaka District. We acknowledge the cooperation of the study participants and community leaders in both study sites, without whose assistance this work would not have been possible.
Financial support: This work was funded by the European Union under the INCO-DEV: International Co-operation with Developing Countries Programme (Contract ICA4-CT-2000-30039: Title: The Policy Implications of Contamination of Rural Water Between Source and Point-of-Use in Kenya, South Africa and Zimbabwe—AQUAPOL; www.bristol.ac.uk/aquapol/).
* Address correspondence to Martella du Preez, Natural Resources and the Environment, CSIR, PO Box 395, Pretoria 0001, South Africa. E-mail: mdupreez{at}csir.co.za 
Authors’ addresses: Martella du Preez, Natural Resources and the Environment, CSIR, PO Box 395, Pretoria 0001, South Africa, Tel: 27-12-8413950, Fax: 27-12-8413954, E-mail: mdupreez{at}csir.co.za. Ronán M. Conroy, Department of Epidemiology and Preventive Medicine, Royal College of Surgeons in Ireland, Mercer Building, Dublin 2, Ireland, E-mail: rconroy{at}rcsi.ie. James A. Wright, Department of Geography, University of Southampton, Highfield, Southampton SO17 1BJ, UK, E-mail: J.A.Wright{at}soton.ac.uk. Sibonginkosi Moyo, Research and Technical Services, Institute of Water and Sanitation Development, PO Box MP 422 Mount Pleasant, Harare, Zimbabwe, E-mail: smoyo{at}iwsd.co.zw. Natasha Potgieter, Department of Microbiology, University of Venda for Science and Technology, Thohoyandou, Venda, South Africa, E-mail: Natasha.potgieter{at}univen.ac.za. Stephen W. Gundry, Water and Environmental Management Research Centre, University of Bristol, 83 Woodland Road, Bristol BS8 1US, UK, E-mail: stephen.gundry{at}bristol.ac.uk.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
