Water for Cambodia used biosand filters (BSFs) to provide microbiologically safe drinking water for people in Moat Khla floating village in 2010 and 2011. All 189 families use the lake, which by World Health Organization (WHO) standards is deemed unsafe for drinking water. Surveys from December 2010 to February 2011 compared 40 families using BSFs and 40 families not using BSFs. Over 92% of BSF households and 90% of non-BSF households were using high-risk lake source water (>100 colonies Escherichia coli/100 mL). Only 2.5% of BSF households had filtered water with bacteria in the high-risk range and only 5% of these 40 households showed recontamination in their storage water. Forty percent of non-BSF households had high-risk bacteria levels in their stored water, and most used no treatment. Storage water for non-BSF families showed a significant reduction in mean log10E. coli levels compared to their lake source water. Stored water for non-BSF families showed recontamination even for UV-treated water and boiled river water. Recontamination occurs in both groups but is much less for BSF households highlighting the value of proper storage containers used by BSF households and the need for water and sanitation education for floating villages in Cambodia.

INTRODUCTION

Sustainable access to improved drinking water became the proxy criterion for the original UN Millenium Development Goal 7 Target 10 guideline. Although the World Health Organization (WHO) proclaimed that the goal had been achieved in March 2012, that does not mean that the water is free of bacterial contaminants or meets WHO drinking water guidelines for E. coli (WHO/UNICEF 2012; UN News Center 2012; WHO 2013). Sustainable access for villages in developing countries using ponds, springs, or wells as a drinking water source is quite different from communities that float on their drinking water. In Cambodia, Tonle Sap Lake is a unique hydrological phenomenon increasing in size from 2,500 km2 in the dry season (November–April) to as much as 12,000 km2 in the rainy season (May through October) as a result of runoff and reversal of flow of the Tonle Sap River when the Mekong River floodwaters force their way over flooded wetlands and up into the lake. A unique fishery has developed on this lake, and, as a result, over 200 floating or stilted home communities have grown on the lake, with populations of Khmer, Vietnamese, Cham, and Chinese of over 80,000 people (TSBRS 2012). The lake itself provides the drinking water for these floating villages, as well as the receiver of all their cleaning, fish processing, and human and animal wastes.

Water source protection is an essential part of interventions to improve community health. This is challenging for floating village communities, where households literally float on their source water, making it difficult to create and protect a microbiologically safe water source. For villages like Moat Khla, the lake water is their sole drinking water source most of the year, while receiving waste water from washing clothes, cleaning fish, and human and animal waste. Such villages have nearly unlimited access to water, but the water is not microbiologically safe.

Point-of-use (POU) water treatment systems are widely used around the world to help provide microbiologically safe drinking water for villages in developing countries. Brown et al. (2007) and Liang et al. (2010) have documented that ceramic water pots (CWPs) and biosand filters (BSFs) are effective POU treatment systems in Cambodia, capable of removing over 95% of E. coli bacteria when properly used and maintained. Both studies showed that CWP and BSF usage resulted in reductions of E. coli bacteria in the filtered water between 1 and 3 log10 and averaged reductions of over 95%.

Our study evaluates BSF effectiveness in providing microbiologically safe drinking water for the Moat Khla floating village community, Siem Reap province on the Tonle Sap. Our objectives were to compare bacteria levels in lake and household storage water for families using BSFs as a water treatment and families that did not, and to compare E. coli levels between distinct geographic zones within a 1–1.5 km area of the river (Figure 1) during two periods of the dry season.

Figure 1

Map of Moat Khla floating village on the Tonle Sap and the four housing cluster zones within the community in February 2011. Arrows demarcate each zone.

Figure 1

Map of Moat Khla floating village on the Tonle Sap and the four housing cluster zones within the community in February 2011. Arrows demarcate each zone.

METHODS

Water for Cambodia (WfC) staff traveled to Moat Khla to survey drinking water use, hygiene and water quality in December 2010 and February 2011. During each visit, BSF performance was assessed by Cambodian staff for filters previously installed by WfC. Thirty households were part of a December WfC follow-up program on filter installation and February surveys targeted 40 households using BSFs and 40 households that were not yet using filters. Households were located between the river's mouth (12° 53′ 45.01 N 104° 10′ 17.11 E) and the most upstream edge of Moat Khla (12° 53′ 57.39 N 104° 11′ 8.76 E). Owing to the remote location of this village, the WfC team set up a temporary lab space at the village chief's home and lived in the community during each survey period. Each day, samples of lake water collected by the household, water from the BSF, and household storage water were collected using separate sterile Whirl pak bags. Water was checked for pH using a Mettler Toledo pH meter, and sample turbidity was measured using a Hach 2100P turbidimeter. All bacteria samples were processed for E. coli by membrane filtration and grown on modified m-Tech media (Clesceri et al. 1998). A Hach portable incubator was used to incubate plates at 44.5 °C.

Data for all comparisons were evaluated for normality using the Kolmogorov–Smirnov test in SPSS. Mean log10E. coli concentrations for lake source water samples for BSF and non-BSF households were normally distributed as well as mean turbidity for all samples and so were compared using independent-samples t-tests within each month. Mean turbidity data between the four Moat Khla zones were compared using one-way analysis of variance (ANOVA) and Dunnett C comparison of mean differences since variances were not homogeneous. Log10E. coli data for December and February for BSF-filtered water and household storage water were not normally distributed so means were compared within each month using Mann–Whitney U tests and between the four Moat Khla zones using the Kruskal–Wallis test. E. coli data for household storage water for non-BSF households were not normally distributed so median and interquartile ranges were used to describe source variability.

RESULTS

Average turbidity of the lake source water increased significantly from 2.5 ± 1.4 (n = 30) nephelometric turbidity units (NTUs) in December to 18.1 ± 6.8 (n = 40) NTUs in February (p < 0.001) due to decreasing water levels and disturbance from boat traffic. Average turbidity of the lake water used by both the BSF and non-BSF households in the February survey (18.1 ± 6.8 to 17.0 ± 7.3) was not significantly different (p > 0.506). BSFs significantly reduced mean turbidity in December (2.5 ± 0.9 to 1.4 ± 0.8) as well as February (18.1 ± 6.8 to 2.4 ± 1.4) (p < 0.001).

BSFs produced significant reductions in mean log10E. coli counts in December and February (p < 0.001, Table 1). Numbers of E. coli in the lake source water used by BSF households during December and February had medians of 415 and 725 colony-forming units (CFUs)/100 mL, respectively, while the BSF-filtered water for each month was 1 and 2 CFUs/100 mL, showing log reductions of E. coli bacterial colonies greater than 2.2 compared to the original lake source water (Table 1).

Table 1

Comparison of median and mean E. coli levels and reduction values for Moat Khla households in December 2010 and February 2011

E. coli CFUs/100 mLLog10E. coli CFUs/100 mLE. coliE. coli CFUs
BSF householdsnMedian (INTQ)Mean ± SDLRV% Red.
December 2010      
Lake water 28 415 (295) 2.56 ± 0.42   
BSF-filtered 28 1 (2.5) 0.30 ± 0.44 2.26 >99% 
Stored water 30 1 (6) 0.36 ± 0.52 2.2 >99% 
February 2011      
Lake water 40 725 (1805) 2.84 ± 0.53   
BSF-filtered 39 2 (5) 0.41 ± 0.51 2.43 >99% 
Stored water 38 2 (5.5) 0.48 ± 0.57 2.36 >99% 
E. coli CFUs/100 mLLog10E. coli CFUs/100 mLE. coliE. coli CFUs
BSF householdsnMedian (INTQ)Mean ± SDLRV% Red.
December 2010      
Lake water 28 415 (295) 2.56 ± 0.42   
BSF-filtered 28 1 (2.5) 0.30 ± 0.44 2.26 >99% 
Stored water 30 1 (6) 0.36 ± 0.52 2.2 >99% 
February 2011      
Lake water 40 725 (1805) 2.84 ± 0.53   
BSF-filtered 39 2 (5) 0.41 ± 0.51 2.43 >99% 
Stored water 38 2 (5.5) 0.48 ± 0.57 2.36 >99% 

INTQ, interquartile range; LRV, log10 reduction value based on arithmetic mean for log10E. coli for each category within a sample period.

In February, over 90% of both household groups were using high-risk lake source water with over 100 CFUs of E. coli bacteria per 100 mL (Table 2). One BSF household had high-risk water from their filter and only two had high-risk water in their household storage water (Table 2) while 40% of non-BSF household storage water had high-risk bacteria levels.

Table 2

Water from 40 BSF households and 40 non-BSF households in Moat Khla in February 2011 assessed by each of three WHO microbial risk categories for E. coli. Number of households (percent)

WHO microbial risk Range of E. coli (CFUs/100 mL)Very low <1Low 1–10Intermediate 11–100High 101–1,000Very high >1,000
BSF households      
Lake water 3 (7.5) 23 (57.5) 14 (35) 
Directly from BSF 19 (47.5) 15 (37.5) 5 (12.5) 1 (2.5) 
Household storage 17 (42.5) 15 (37.5) 6 (15) 2 (5) 
Non-BSF households      
Lake water 1 (2.5) 3 (7.5) 21 (52.5) 15 (37.5) 
Household storage 8 (20) 6 (15) 10 (25) 8 (20) 8 (20) 
WHO microbial risk Range of E. coli (CFUs/100 mL)Very low <1Low 1–10Intermediate 11–100High 101–1,000Very high >1,000
BSF households      
Lake water 3 (7.5) 23 (57.5) 14 (35) 
Directly from BSF 19 (47.5) 15 (37.5) 5 (12.5) 1 (2.5) 
Household storage 17 (42.5) 15 (37.5) 6 (15) 2 (5) 
Non-BSF households      
Lake water 1 (2.5) 3 (7.5) 21 (52.5) 15 (37.5) 
Household storage 8 (20) 6 (15) 10 (25) 8 (20) 8 (20) 

Lake water and household storage water turbidity varied greatly within and between the four housing cluster zones in February 2011, for both household groups (Figures 1 and 2). Lake water turbidity was lower for BSF households in Zone 1 nearer to the Tonle Sap Lake, but was not significantly different from either household group in the other three zones (Figure 2(a) vs. 2(d)). Household storage water turbidity was higher for non-BSF households but only significantly higher in Zone 4 than BSF households in Zones 1 and 3 (Dunnett C comparison of mean differences p < 0.05) (Figure 2(c) vs. 2(e)).

Figure 2

Mean turbidity levels by source and between household cluster zones within the Moat Khla floating village community in February 2011 for 40 households using BSFs and 40 households not using BSFs. Filtration_Source a = BSF household using lake source; b = BSF household filtered water; c = BSF household storage water; d = non-BSF household using lake source; e = non-BSF household storage water. Moat Khla Zone 1 = mouth; 2 = lower Moat Khla; 3 = mid Moat Khla; 4 = upper Moat Khla (Pagoda area). Refer to Moat Khla zone locations in Figure 2. Error bars = 95% confidence interval.

Figure 2

Mean turbidity levels by source and between household cluster zones within the Moat Khla floating village community in February 2011 for 40 households using BSFs and 40 households not using BSFs. Filtration_Source a = BSF household using lake source; b = BSF household filtered water; c = BSF household storage water; d = non-BSF household using lake source; e = non-BSF household storage water. Moat Khla Zone 1 = mouth; 2 = lower Moat Khla; 3 = mid Moat Khla; 4 = upper Moat Khla (Pagoda area). Refer to Moat Khla zone locations in Figure 2. Error bars = 95% confidence interval.

Mean log10E. coli levels of the four household zones showed bacterial levels increased downstream from Zone 4 toward Zone 1. Levels were significantly higher in lake water used by BSF households (Figure 3(a)) between Zone 1 and Zone 4 (Kruskal–Wallis p < 0.015) but were not significantly different for non-BSF households. Mean log10E. coli levels were not significantly different between the zones for household storage water in either BSF or non-BSF households (Figure 3(c) vs. 3(e)).

Figure 3

Comparisons of mean log10E. coli levels by source and between household cluster zones within the Moat Khla floating village community in February 2011 for 40 households using BSFs and 40 households not using BSFs. Filtration_Source a = BSF household using lake source; b = BSF household filtered water; c = BSF household storage water; d = non-BSF household using lake source; e = non-BSF household storage water. Moat Khla Zone 1 = mouth; 2 = lower Moat Khlal; 3 = mid Moat Khla; 4 = upper Moat Khla (Pagoda area). Refer to Moat Khla zone locations in Figure 2. Error bars = 95% confidence interval.

Figure 3

Comparisons of mean log10E. coli levels by source and between household cluster zones within the Moat Khla floating village community in February 2011 for 40 households using BSFs and 40 households not using BSFs. Filtration_Source a = BSF household using lake source; b = BSF household filtered water; c = BSF household storage water; d = non-BSF household using lake source; e = non-BSF household storage water. Moat Khla Zone 1 = mouth; 2 = lower Moat Khlal; 3 = mid Moat Khla; 4 = upper Moat Khla (Pagoda area). Refer to Moat Khla zone locations in Figure 2. Error bars = 95% confidence interval.

Recontamination after collection

Non-BSF households did not rely exclusively on lake source water and household storage ranged widely in bacteriological safety. Thirteen of the 40 non-BSF households used UV-treated water but stored water was no less variable in bacteriological safety than untreated river water or BSF water from their neighbor (Table 3). Recontamination occurred in the stored drinking water even for non-BSF households that boiled water or purchased BSF-filtered water from their neighbor. Although sample sizes were low, some households that used a ceramic filter or river water treated with alum had low risk levels of E. coli.

Table 3

Median number of E.coli CFUs per 100 mL in the stored household drinking water by the type of drinking water source used by Moat Khla non-BSF families in February 2011

Responses of Moat Khla non-BSF households in February 2011 on which water they use for their drinking water (dry season)nMedian E. coli CFUs per 100 mLINTQ rangeMinMax
RDB UV water system 13 470 3,000 
BSF water from neighbor 550  70 1,030 
Buy water 720    
Ceramic filter 13  26 
Korea filter 70    
River water or RDB UV water 13    
River water 13 230 1,521 2,920 
River water with boiling 514 2,020 
River water with alum    
Total 40 60.5 672 3,000 
Responses of Moat Khla non-BSF households in February 2011 on which water they use for their drinking water (dry season)nMedian E. coli CFUs per 100 mLINTQ rangeMinMax
RDB UV water system 13 470 3,000 
BSF water from neighbor 550  70 1,030 
Buy water 720    
Ceramic filter 13  26 
Korea filter 70    
River water or RDB UV water 13    
River water 13 230 1,521 2,920 
River water with boiling 514 2,020 
River water with alum    
Total 40 60.5 672 3,000 

RDB, Regional Development Bank; INTQ, interquartile range.

DISCUSSION

BSFs and ceramic filters are effective in bacterial reduction of source water from dug wells, ponds and bore wells, so drinking water has low or no risk from bacteria. The Moat Khla study supports previous findings on BSF effectiveness in bacterial removal (Duke et al. 2006; Brown et al. 2007; Liang et al. 2010) and improvement of drinking water.

Water source quality differed significantly in the lake between December and February and between the four zones where homes were clustered in the mouth of the river. Turbidity and bacterial levels increased between December and February. Moreover, everyone in Zone 1 lives downstream from their neighbors (Figure 1), so they receive the cumulative impact of human and animal wastes from daily living of community members upstream. This difference was significant for BSF households. Nevertheless, BSFs performed well in turbidity reduction and bacterial removal despite the seasonal and geographic variation.

Centralized treatment systems are being implemented in Tonle Sap floating villages to provide safe water and reduce occurrences of diarrhea. RACHA (2009) has installed a solar-powered treatment station in Chong Khneas in Siem Reap province. Many of the 123 households they surveyed had portable or POU filters but many stopped using them because of low flow rates and maintenance problems. All 123 households selected for the survey had children under the age of five years. Of these, 54.1% of children under five years and 25.8% of adults had experienced diarrhea in the previous 12 months. RACHA (2009) concluded that portable household filters (POU) do not fully meet household needs for clean and safe drinking water, yet acknowledged that sanitation practice in this village was low and proper hand-washing was not practiced by most people. In Moat Khla, a floating water treatment system was installed by a regional development agency in December 2009. This centralized treatment system provided purified water to the community for 1,000 Cambodian riel (0.25 USD) per 20 L of water. A gas generator was used to power the system on an as-needed basis to pump, filter, and power UV treatment. Despite the high-quality UV treatment provided, non-BSF households purchasing this water still showed evidence of recontamination in their household storage (Table 3). Even with access to UV-treated water, villagers still requested 63 more BSFs by the time of this study despite having to pay 7 USD per filter. Villagers interviewed said they could only afford 400–600 Cambodian riel per 20 L or about 0.10 to 0.15 USD for 20 L.

Recontamination of the stored water continues to surface as a problem no matter what type of intervention is used to provide microbiologically safe drinking water. In a review of studies reporting microbiological levels in the source water and household storage, over half showed bacteriological quality declined after collection (Wright et al. 2004). Two key recontamination scenarios emerged in our study. Eighty to 95% of BSF households had very low- to intermediate-risk storage water but 5% had E. coli levels in the very high-risk range, showing a definite recontamination of stored water (Table 2). This parallels previous studies in Cambodia for ceramic filters and BSFs (Brown et al. 2007; Liang et al. 2010). In Moat Khla, 40% of the non-BSF households were using stored water with high- to very high-risk bacterial levels (Table 2). Our household survey of hygiene and water use showed over 32% of the households not using a BSF were actually using treated water from a UV floating treatment system (Table 3). Households using this water as their first water source during the dry season had E. coli ranging from 10 to 19,500 CFUs/100 mL, suggesting that either recontamination was occurring in their source water collection containers or that, in fact, they were not using the water from the community treatment system.

Providing microbiologically safe drinking water does not guarantee safe water at the point of consumption. Boiling is the commonly accepted practice for ensuring microbiologically safe water, but recontamination can negate the efforts to provide safe water. Boiling can reduce the levels of thermotolerant coliform (TTC) bacteria to the WHO standard of 0 TTCs, but storage and handling methods can negate this gain by recontamination (Oswald et al. 2007; Clasen et al. 2008; Rosa et al. 2010). Contaminated hands and improperly cleaned drinking cups were major sources of recontamination, which continued to increase from source water collection to time of consumption in a peri-urban location in Peru (Oswald et al. 2007). In this case, households were using municipal source water and not raw surface water from a lake or river. For Moat Khla village, six households not using BSFs boiled their water and had decreased mean levels of E. coli compared to 13 households using untreated lake water (Table 3). Even though they boiled their water, most were not free of E. coli. Oddly enough, this was also the case for 13 families not using a BSF who purchased water from the UV treatment filtration system (Table 3). Whether boiling municipal source water, contaminated surface water from a lake, or using a BSF or UV filtration system, storage water is subject to recontamination. Follow-up instruction on hygiene and sanitation is essential to break the bacterial transmission cycle and reduce the alarmingly high levels in the stored drinking water. Enger et al. (2013) and Peletz et al. (2013) make the point that optimal household water treatment solutions require microbiological performance, affordability, and access, but even with all of these, correct and consistent use is still required for reducing exposure to water-borne pathogens.

CONCLUSIONS

  • Biosand filters effectively reduced turbidity and E. coli levels and provided an improved low bacteria risk source of drinking water to at least 80% of households in Moat Khla floating village despite seasonal changes in lake source water.

  • Location within the floating village impacts the level of contamination of the lake source water used by families.

  • E. coli levels varied widely in household storage water for these non-BSF families no matter what source or treatment used and highlights why water, sanitation, and hygiene education must be combined with treatment interventions to provide safe water.

  • An effort to provide clean water can be negated by recontamination, thus a multi-barrier approach is necessary that includes hygiene and sanitation education.

REFERENCES

Brown
J.
Sobsey
M.
Proun
S.
2007
Use of Ceramic Water Filters in Cambodia
.
WSP – World Bank Field Notes
,
Washington DC
.
Clasen
T.
McLaughlin
C.
Nayaar
N.
Boisson
S.
Gupta
R.
Desai
D.
Shah
N.
2008
Microbiological effectiveness and cost of disinfecting water by boiling in semi-urban India
.
Am. J. Trop. Med. Hyg
.
79
(
3
),
407
413
.
Clesceri
L. S.
Greenberg
A. E.
Eaton
A. D.
(eds).
1998
Standard Methods for the Examination of Water and Wastewater
, 20th edn.
APHA, AWHA, WEF
,
Baltimore, MD
.
Duke
W. F.
Nordin
R. N.
Baker
D.
Mazumder
A.
2006
The use and performance of biosand filters in the Artibonite Valley of Haiti: a field study of 107 households
.
Rural Remote Health
6
,
570
.
Liang
K.
Sobsey
M.
Stauber
C.
2010
Use of BioSand filters in Cambodia
.
WSP World Bank Field Note May 2010
,
Washington DC
,
40
.
Oswald
W. E.
Lescano
A. G.
Bern
C.
Calderon
M. M.
Cabrera
L.
Gilman
R. H.
2007
Fecal contamination of drinking water within peri-urban households, Lima, Peru
.
Am. J. Trop. Med. Hyg.
77
(
4
),
699
704
.
Peletz
R.
Mahin
T.
Elliott
M.
Montgomery
M.
Clasen
T.
2013
Preventing cryptosporidiosis: the need for safe drinking water
.
Bull. World Health Org
.
91
,
238
.
http://dx.doi.org/10.2471/BLT.13.119990 (accessed 20 August 2013)
.
RACHA (Reproductive and Child Health Alliance)
2009
Water and sanitation baseline survey in Chong Khneas, Siem Reap Province, Cambodia. December 2009. Phnom Penh. http://rc.racha.org.kh/download.asp?file=/resources/documents/301-400/341-360/358/WATER%20AND%20SANITATION%20BASELINE%20SURVEY.pdf (accessed 18 August 2012)
.
TSBRS (Tonle Sap Biosphere Reserve Secretariat)
.
2012
Tonle Sap Biosphere Reserve: The Heart of Cambodia
.
Phnom Penh
,
Cambodia
, p.
12
. .
United Nations News Centre
.
2012
World meets goal of boosting access to clean water but lags on better sanitation – UN. http://www.un.org/apps/news/story.asp?NewsID=41465&Cr=MDGs&Cr1 (accessed 13 August 2012)
.
WHO
.
2013
Water Sanitation and Health: Health through safe drinking water and basic sanitation
.
World Health Organization
,
Geneva
,
Switzerland
. .
WHO/UNICEF
.
2012
Progress on drinking water and sanitation: 2012 update
.
WHO/UNICEF Joint Monitoring Programme for Water Supply and Sanitation
. .