Determinants of microbiological quality of drinking water in refugee camps and host communities in Gambella Region, Ethiopia

Inadequate improved water supply and sanitation, particularly in refugee camps contribute to the spread of infectious diseases. The study objective was to assess determinants of microbiological quality of drinking water in refugee camps and host communities in Gambella Region, Ethiopia. A cross-sectional study was conducted from September to December 2016 based on structured questionnaire-based interviews and testing household water using the portable Potatest þ water quality testing kit. Data were analyzed and P values < 0.05 with 95% con ﬁ dence interval (CI) were considered statistically signi ﬁ cant. Results showed there were signi ﬁ cant differences in fecal coliform count ( P value ¼ 0.009) and free residual chlorine concentration ( P value ¼ 0.01) between the source and stored water samples. Surface water source, water shortages in the previous month, and unavailability of free residual chlorine and caregivers without formal education were the main determinants of microbiological quality of stored water. Stored water was contaminated in many households in both the refugee and host communities. Designing and implementing appropriate community education and effective hygiene promotion programs are essential in improving community knowledge of water contamination and reducing diarrhea prevalence among under- ﬁ ve children in refugee camps and host communities in Gambella Region.

biological and chemical pollutants originating from point and non-point sources (Ali et al. ). Inadequate water supply and sanitation, particularly in conflict areas with large numbers of internally displaced people or refugees, contribute to spreading infectious diseases (Theron & Cloete ). Diarrheal diseases caused 446,000 deaths each year, contaminated drinking water accounted for 72.1%, while inadequate sanitation was associated with 56.4% deaths from diarrhea (Masquelier et al. ).
Systematic reviews indicated the risk of diarrhea in under-five children to be lower with any water quality intervention compared to no intervention (Cairncross et al. ; Clasen et al. ). Water quality improvements may significantly reduce rates of child diarrhea morbidity (Waddington et al. ). However, the water supply in many developing countries is often either inadequate or unsafe to meet basic health needs (WHO/UNICEF ). In 2015, 844 million people worldwide lived without access to improved water supply and 159 million people collected drinking water directly from surface water sources; 58% of them lived in sub-Saharan Africa (WHO/UNICEF ). Inadequate water and sanitation provision has been documented in camps of refugees and internally displaced persons in many sub-Saharan African countries (Sherlock ; Shrestha & Cronin ). In rural Ethiopia, the rate of childhood diarrhea remains high (WHO/UNICEF ) and the average domestic water consumption is much lower than the recommended amount (Kumie & Ali ). An estimated 43% of the Ethiopian population had no access to safe water and 72% had no access to basic sanitation services (FDRE-MOH ). Refugee camps with high population densities and rural communities in remote areas, including Gambella Region, may not have access to safe drinking water (Sherlock ).
Although improved sources generally delivered safe water at the point-of-supply, the quality of drinking water often deteriorates during distribution and transport to the home and subsequent storage (Gundry et al. ). Still unknown factors and substantial variation of determinants influencing water quality were observed by various studies (Clasen et al. a). Some socio-demographic characteristics such as educational level, family size, occupation, and availability of sanitation facilities may be associated with drinking water quality (McGarvey et al. ).
Behavioral factors related to variations in household water management such as type of water containers, transportation, storage conditions, hand washing, and waste disposal practices play major roles in water contamination at the household level (Bastaraud et al. ). Therefore, water kept in household containers may contain significantly more fecal bacteria than water at the source (Clasen & Cairncross ), and the high degree of contamination of drinking water in households poses a health hazard to consumers (Tallon et al. ). The presence of fecal coli- Few studies have systematically assessed determinants of water quality and reported health impacts in developing countries. Laboratory testing facilities at fixed sites are not always the most practical or readily available means to evaluate drinking water quality (WHO ). Furthermore, water sample transportation within the recommended time frame and temperature range is often impractical in rural areas in developing countries (Rice et al. ). In poorly accessible and insecure areas where there are no laboratory facilities, a more robust and portable kit is required to provide critical water quality information. These conditions necessitate detailed research on how household sociodemographic and WASH factors influence water quality in refugee and host communities using a portable water quality technique. Various commercial membrane filtration-based field test kits are available to perform testing of microbiological indicators in remote areas and refugee contexts. Therefore, this study was conducted to evaluate the microbiological quality of water at the source and household levels and its determinants in two refugee camps and three host communities using the Wagtech Potakit, one of the por-

Study area and populations
Gambella town is located 753 km west of Addis Ababa, the capital of Ethiopia. Gambella Region is subdivided into four administrative zones and one special district. In 2017, n ¼ NZ 2 * p * (1 À p)/Nd 2 þ Z 2 * p (1 À p), where n is the sample size, N the total number of surveyed households, Z at 95% CI ¼ 1.96, p the estimated population proportion of 0.5 which maximizes the sample size and d the error limit of 5% which is equal to 0.05. The computed 314 household (157 from each community) samples, 18% of the surveyed population was adequate to evaluate the microbiological quality of drinking water and identify its determining factors.

Data collection
A diarrhea survey was conducted randomly in 1,782 households with at least one under-five child prior to this study.
The systematic random sampling technique was employed to select the 314 samples from these surveyed households.
Samples were distributed over the study sites proportional to the target households. We used a questionnaire to collect information on socio-demographic characteristics and certain WASH factors which may influence water quality and child diarrheic status. The questionnaire was developed in the English language and then translated into the local Nuer and Agnuwak languages for better communication with the study subjects. Persons responsible for handling water in the household, usually the mothers or eldest daughters, were interviewed. The interviews were conducted by trained data collectors who had completed at least their secondary education and were able to write, read, and understand English. A 100 mL water sample was collected from the 306 selected households and from 51 (34%) of the direct water sources. Water samples were analyzed by the principal investigator to determine the level of fecal contamination of drinking water at the source and household levels.

Microbiological water quality test
The comprehensive Wagtech portable water testing kit allows tests to be carried out in remote areas following WHO guidelines. We used the portable water quality testing kit (WAG-WE10030 -Wagtech Potatest þ Kit) to measure fecal coliform counts and free residual chlorine (WHO ). Samples were collected using serving cups from the same containers households routinely used to collect water for drinking and by pouring the water into 100 mL uncontaminated plastic bottles.
We tightly covered the bottles with the stoppers and adhered to aseptic techniques. Each bottle was labeled with an identification number, and we recorded the time of sampling.
All samples were stored in a small cold box with ice packs immediately after collection until tested within 6 hours of collection. We filtered 100 mL water samples using 0.45 mm pore size, 47 mm diameter filter membrane and incubated the loaded petri-dish rack at 44 C for 14 hours following 1 to 4 hours resuscitation at 35 C. Then, we inspected the plates for growth of fecal coliforms systematically, column by column in the grid. We counted all the yellow colonies irrespective of colony size using a hand lens within a few minutes, as the colors are likely to change on cooling and standing. The values are recorded as number of fecal (thermo-tolerant) coliforms or CFU per 100 mL of water.

Chlorine residual testing
One hour after the addition of sodium hypochlorite solution to water there should be an average of 0.5 mg/L of free chlorine residual present. Twenty-four hours after the addition of sodium hypochlorite, the storage water should have 0.2 mg/L to 2.0 mg/L of free chlorine residual present to ensure microbiologically clean water (CDC ). Chlorine residual at direct sources and from household storage containers was tested using the Wagtech potatest þ Kit. The test allows for assessment of water quality delivered from the water utility to the home and may also help to evaluate changes in water quality at home that occurred as a result of poor hygiene practices or other reasons. Free residual chlorine (FRC) concentrations were measured in all water samples using diethyl-phenylene diamine (DPD) reagents, which are provided in tablet form for convenience and simplicity of use. We added one DPD1 tablet into each test tube, filled the test tube with sample water to the 10 mL mark and shook it vigorously to let the free chlorine react with DPD to produce coloration. The intensity of the pink color is proportional to the free residual chlorine concentration. The color intensities were measured by comparing against color standards using a Wagtech Comparator. The disk reading represents the free chlorine residual as milligrams per liter.

Data analysis
We considered water samples with <1 CFU/100 mL to be uncontaminated and samples with !1 CFU/100 mL to be contaminated. The fecal coliform count contamination levels in the drinking water samples were categorized into 0 (none), 1-10 (low risk), 11-100 (moderate risk) and >100 CFU per 100 mL (high risk) (WHO ). We reported the arithmetic mean count for sub-sets of the contaminated samples. All statistical analyses were done using STATA Version 13. Descriptive statistics of study subjects or household-level characteristics and WASH factors were reported as proportions, means, and ranges. Bivariate and multivariate models were used to assess the association of socio-demographic and WASH factors with fecal coliform count. Mann-Whitney U test was used to evaluate the relationship between presence of fecal coliform bacteria and free residual chlorine in drinking water at source and in storage vessels. P value 0.05 with 95% confidence intervals was taken as a cutoff point of statistical significance.

Quality control
Training was provided by the principal investigator to ten data collectors and three supervisors a week before the commencement of the study. The questionnaire was pre-tested in households in Jawi Refugee Camp, and necessary corrections were made accordingly. We checked the data for consistency and completeness. Water samples were collected in a sterile sample bottle, stored on ice, and processed within 4 hours of collection. Escherichia coli (ATCC 25922) strain infused water was used as a positive control, and 100 mL of sterile distilled water as negative control were processed after every 20th sample to ensure that the equipment had been adequately sanitized. The standard method for testing of the drinking water quality was maintained as per World Health Organization guidelines (WHO ).

Household characteristics
The study included 306 households with an average of 6 (±SD, 2.36) person per household, and the response rate was 97.5%.
The refugee households comprised 156 (51%) households; 73 in Turkiedi and 83 in Pugnido refugee camps. One hundred and fifty (49%) of the households were from the host communities; 62 were in Tata rural Kebele and 88 in Pugnido urban Kebele. The great majority of the caregivers (292, 95.4%) were females and the mean age of the caregivers was 28.7 years (range 15-47 years), of whom 140 (45.8%) had never received formal education (Table 1). The mean age of children included in this study was 20.6 months.

Water supply and water handling in the study area
Teirkidi refugee camp is supplied with water from four borehole water sources, each of which pumped 12-13 L/second to a common collection tank. Water was centrally treated using an automatic chlorine pump. In Teirkidi, chlorine treated water is pumped to two reservoirs that distribute water to the refugee community via 43 public standpipes.
Tankers distributed the treated water in the refugee camps because it was a communal system of piping water to public taps from which the households were collecting drinking water. Similarly, Pugnido refugee camp had six boreholes and matching reservoirs with 94 public standpipes. The studied host community households were supplied with water from two boreholes piped directly to public taps, ten tube well water sources and surface water sources. No routine chlorination of water was practiced prior to distribution in the host communities.
Fifty-one water samples were collected from water sources and 306 samples from containers in selected households. In this study, 31 (60.8%) source samples were collected from refugee camps and 20 (39.2%) samples from the hosting communities. The study revealed that 224 (73.2%) of the households were using piped water, followed by surface water (48, 15.7%), and tube wells (34, 11.1%).
Water was not always available for 214 (69.6%) of the households due to interruptions in the supply. Householdlevel water treatment was uncommon, and reported by only eight (2.6%) of the households (Table 1).  (Table 2). Generally, free Witney test showed that free residual chlorine concentrations were significantly higher at the source than the stored water samples in the refugee camp (P < 0.001) and

Free residual chlorine concentrations
host communities (P < 0.001). The fecal coliform counts in the stored water samples were significantly higher than in source samples (P ¼ 0.009) ( Table 3).

Multivariate analysis of determinants of storage water quality
Logistic regression was used to estimate the odds of unsafe water quality by determining the binary outcome (potable water is equal to 1 if there is 1 or more fecal coliform colonies forming units per 100 mL water and non-potable water is 0). Multivariate logistic regression analysis was carried out to identify determinants with P values less than or equal to 0.2 in bivariate analyses. The associations between some factors and drinking water fecal contamination remained significant after further adjustment for confounders. This study shows that types of primary water sources influence storage water quality. Households which used surface

DISCUSSION
Our study shows that water quality is a major problem in both the refugee and host communities. All the households in the refugee camps and 102 (68.0%) of the households in host communities relied on improved water sources. However, water was not always available for 214 (69.6%) of the households because piped water supply was intermittent and was often available less than 8 hours a day (Mekonnen et al. ).
Nearly half (  ). These differences may be due to differences in the socio-demographic characteristics, study design, and the microbiological technique used.

Limitations of the study
This cross-sectional study did not cover the effect of seasonal variation in water quality and its determinants due to financial constraints. As the study area was remote and no standard laboratory facilities were available, the study did not employ the conventional water quality testing methods.

CONCLUSION AND RECOMMENDATION
This study found water quality in refugee camps to be better than their host communities. Nevertheless, fecal coliforms were isolated in many households with stored water in both the refugee and host communities. These findings suggest that improved water sources do not provide safe drinking water in homes. Educational level of caregivers, unimproved water sources, water supply interruptions, and lack of free residual chlorine were found to be independent predictors of fecal coliform contaminations of stored water. Additional