Bulletin of the World Health Organization

Effectiveness of emergency water treatment practices in refugee camps in South Sudan

Syed Imran Ali a, Syed Saad Ali b & Jean-Francois Fesselet c

a. University of California-Berkeley, 100 Blum Hall, MC 5570, Berkeley, CA 64720, United States of America (USA).
b. University of Toronto, Toronto, Canada.
c. Médecins Sans Frontières-Holland, Amsterdam, Netherlands.

Correspondence to Syed Imran Ali (email: imran.ali@berkeley.edu).

(Submitted: 20 September 2014 – Revised version received: 29 April 2015 – Accepted: 07 May 2015 – Published online: 01 June 2015.)

Bulletin of the World Health Organization 2015;93:550-558. doi: http://dx.doi.org/10.2471/BLT.14.147645

Introduction

The late 19th and early 20th century saw rapid declines in mortality in industrialized countries. The introduction of chlorinated piped water supplies in cities was a major contributor to this achievement.1 Today, chlorination is the most widely-used method for the treatment of piped water supplies, due to its simplicity, low cost and the residual protection it provides.2,3 Low levels of residual chlorine in water supplies limit microbial contamination during distribution and storage, reducing the risk of waterborne infectious diseases. Drawing on decades of experience with municipal piped water systems around the globe, the World Health Organization (WHO) guidelines for drinking water quality recommend a minimum concentration of 0.2 mg/L free residual chlorine at water system delivery points.4

Humanitarian agencies generally use centralized batch chlorination for water treatment in settlements for refugees and internally displaced persons.5 This treatment method entails dosing an experimentally-determined amount of chlorine solution into a known volume of clear water, and allowing adequate retention time to allow disinfection to proceed to completion. Ensuring access to adequate quantities and quality of water is essential in refugee camps as waterborne diseases are among the most significant threats facing displaced populations.69 Drawing on WHO guidelines for drinking water quality, humanitarian organizations have developed several guidelines stipulating what residual chlorine levels should be at camp water distribution points.1016 Generally speaking, guidelines recommend free residual chlorine levels should be 0.2–0.5 mg/L under normal circumstances and 0.5–1.0 mg/L during outbreaks of diarrhoeal disease, or when the water supply is especially turbid or alkaline. A balance is required between having sufficient residual protection and preventing taste and odour-driven rejection due to excessive chlorination.

WHO guidelines for drinking water quality are appropriate when users drink directly from the flowing household taps of a municipal piped water system,17 but are unlikely to provide sufficient residual chlorine protection in the fundamentally different reality of a refugee camp. In this setting, where environmental hygiene may be poor, water is collected from tapstands, transported in containers through the camp to shelters and then stored and used over 24 hours or more. Chlorine treatment based on WHO guidelines for drinking water quality may not ensure that water remains safe over its entire course in the setting of a refugee camp.

Studies in non-emergency settings in developing countries have shown recontamination of previously safe water does occur during collection and transport from distribution points, as well as during storage and drawing of water in the home,1821 representing a significant health risk to vulnerable populations.22,23 Recontamination after collection of drinking water has also been documented in refugee camps in Uganda24 and linked to the spread of diarrhoeal disease and cholera among camp populations in Malawi,25,26 Kenya27,28 and Sudan.29 Humanitarian guidelines call for facilities and practices to preserve the safe water chain including the use of covered narrow-mouthed water containers with taps and their regular cleaning, disinfection and replacement. However, recontamination after distribution in camp settings remains poorly understood and is not explicitly included in guidelines for water treatment in emergency settings.

Recent experiences in refugee camps in South Sudan brought this knowledge gap to our attention. Surveys conducted in the Jamam camp in October–November 2012 showed that 40–58% of households that collected water from chlorinated tapstand sources had no detectable residual chlorine in their stored household water.30,31 Another study carried out in Jamam and the nearby Batil camp in April 2013 found adenoviruses in stored household water, suggesting faecal contamination.32 These observations, taken in light of the prolonged hepatitis E and acute watery diarrhoea outbreaks affecting the Maban County refugee camps,33 raised pressing questions about chlorination in camp settings.

Previous work has investigated and modelled how residual chlorine decays within water distribution systems.34 However, as far as we are aware, residual chlorine decay after water leaves the tap of the distribution system has not been investigated. We sought to: (i) investigate residual chlorine decay after distribution in the refugee camp setting and (ii) identify factors that preserve or compromise the safe water chain by exploring how water quality, water handling practices and contextual factors influence residual chlorine decay. In this paper, we investigate the effectiveness of emergency water treatment practices in the field. We contribute to the evidence base on water, sanitation and hygiene in emergencies and make recommendations for best practice.35,36

Methods

Study setting

The study was carried out at Jamam, Batil and Gendrassa refugee camps in Maban County, South Sudan during March–April 2013. At the time of the study, the population at these camps was 15 500, 37 200 and 15 800, respectively. The local climate and terrain in Maban County exacerbated the crisis and impeded response. The region is part of the Nile basin floodplain and characterized by thick strata of clay-rich soil. This soil is prone to water-logging, heavy and difficult to work with and unproductive with respect to groundwater. The rainy season runs from May to October and the dry season from November to April. The rainy season of 2012 saw the camps flooding, with latrines overflowing and inundating the surface, leading to multiple waterborne infectious disease outbreaks.33 Limited water availability, poor sanitation coverage and poor environmental hygiene exacerbated the outbreaks (Table 1).

Groundwater was pumped from boreholes in or near the camps, treated with in-line chlorination then stored in tanks before being piped to tapstands for distribution. Tapstands in each camp were provided within 500 m of shelters, in accordance with Sphere Project guidelines.16 Supply was intermittent with water delivered for several hours in the morning and afternoon. Retention times in storage tanks following chlorination varied, contributing to variation in residual chlorine at tapstands. Water systems in all camps were in a state of flux at the time of the study, with elements being added or removed as populations fluctuated. Finally, given the outbreaks affecting the camps, outbreak protocols had been adopted and agencies aimed to deliver, in this case, 0.8–1.0 mg/L free residual chlorine at distribution points.

Study design

We sought to follow the pathway of water in the camp setting from distribution at the tapstand to consumption at the household level. The study had two elements: (i) water quality analyses; and (ii) surveys of water handling practices and contextual factors. We assessed water quality at four points after distribution:

  • analysis event 1: directly from the tapstand (event 1);
  • analysis event 2: from users’ containers immediately after collection at the tapstand (event 2);
  • analysis event 3: from containers after transport to shelters (event 3); and
  • analysis event 4: from containers after several hours of household storage and use (event 4).

We analysed the following water quality parameters: free and total residual chlorine (the parameters of primary interest); pH and turbidity (both known to reduce chlorine disinfection efficiency);2,38 oxidation reduction potential (a proxy for disinfection potential);39 electrical conductivity (reflects the dissolved solids or salt content of drinking water); and air and water temperature (which affect the rate of chemical reactions).40 Quality control was achieved by calibrating analytical equipment every 1–2 days using manufacturer calibration standards.

We documented water collection, transport and storage practices, as well as other contextual factors, via spot checks and respondent self-reports at the tapstand (i.e. event 2) and in the household at follow-up (i.e. event 4). As it was crucial to follow the same unit of water, we also noted if the water had been transferred between containers, if it had been used (and, if so, how much) and whether it had been mixed with any other water at any time. The detailed study design is available from the corresponding author.

We initially set out to sample water from every borehole in each camp, however we found that boreholes were not always operational or residual chlorine was not always present at tapstands. Accordingly, we adopted a convenience sampling approach so that we could maximize data collection in the limited time available at each camp. We sought to capture spatial representativeness by: (i) visiting tapstands dispersed across the camp areas; and (ii) sampling water from tapstands attached to different boreholes. We approached whoever was collecting water at the tapstand for enrolment in the study. In total, we collected 220 unique samples roughly divided between the three camps. The study was submitted for ethics review to the Medical Director of Médecins Sans Frontières Operational Centre Amsterdam who exempted it from full ethics review as only routine operational data was being collected.

Data analysis

We adopted a pooled data approach and modelled free residual chlorine concentration in MATLAB 7.12 (MathWorks Inc., Natick, United States of America) using the general integrated rate law:(1)where Co is the initial chlorine concentration, C is the chlorine concentration at time t, N is the order of the reaction, and K is the rate constant.

(1).

We stratified by camp and by initial free residual chlorine concentration due to the nonlinearity of chlorine decay.4143 Having estimated the model parameters K and N, we calculated the initial free residual chlorine levels at the tapstand that would ensure the desired level of free residual chlorine after storage for a defined period.

Regression models

To explore associations between residual chlorine decay and water related physical and chemical parameters, handling practices and contextual factors, we used regression models in Stata 12.1 (StataCorp, College Station, USA). Regression models were run separately to investigate variables affecting water quality during three distinct stages: (i) during water collection; (ii) during transport to the household and (iii) during household storage and use (Table 2). Further details are available from the corresponding author.

Results

Summary statistics on water quality at the tapstands, in each of the three camps, are given in Table 3. Residual chlorine levels were similar in all camps. The turbidity was below the upper limit for effective chlorination at all sites (5 nephelometric turbidity units, NTU). Water temperatures averaged greater than 30 °C at tapstands in the morning – an indication of how hot March and April are in this setting. The pH of the water at all camps was below 8.0, the upper limit for effective chlorination. Oxidation reduction potential varied widely between camps, partially reflecting residual chlorine levels but also potentially undocumented factors. Electrical conductivity, a proxy for chemical quality, was statistically different across camps (P < 0.0001) suggesting that each camp’s source was unique. As discussed earlier, chlorination performance at each site was less than ideal. A histogram of free residual chlorine concentrations encountered at tapstands is presented in Fig. 1.

Fig. 1. Histogram of free residual chlorine concentrations at the tapstand in refugee camps, South Sudan, March–April 2013
Fig. 1. Histogram of free residual chlorine concentrations at the tapstand in refugee camps, South Sudan, March–April 2013
Note: Free residual chlorine concentrations ranged from just detectable (i.e. 0.01 mg/L) to excessive (i.e. 5.2 mg/L), reflecting poor chlorination performance of camp water systems.

Modelling residual chlorine

As seen in Table 4, modelling suggested that residual chlorine decay was a second-order process.4143 The average model R2 was 0.76 (range: 0.57–0.95). Examining modelling graphs and residual plots offers additional insight regarding goodness of fit. The pattern of the residuals was consistent across models, so the most general model (including data for all camps and all initial free residual chlorine data) is shown in Fig. 2.

Fig. 2. Decay model for free residual chlorine in refugee camps, South Sudan, March–April 2013
Fig. 2. Decay model for free residual chlorine in refugee camps, South Sudan, March–April 2013

We calculated the initial free residual chlorine concentrations required to ensure a desired level of free residual chlorine at a designated time after distribution (Table 5). A primary target of 0.2 mg/L at 24 hours after distribution was selected,44 with a secondary target of 0.2 mg/L at 12 hours. Table 5 reports projections using modelling outputs from the 0.5–1.5 mg/L initial free residual chlorine strata, as this is the range most relevant to field practice.

According to our model results, the primary target of 0.2 mg/L at 24 hours after distribution could not be achieved. For the secondary target, 0.2 mg/L at 12 hours after distribution, the initial concentration of free residual chlorine averaged 1.1 mg/L (range 0.37–2.19). If free residual chlorine at the tapstand was set to 1.0 mg/L, the residual concentration remained above 0.2 mg/L for an average of 13.8 hours (range 10.4–19.7; Table 5).

Linear regression models were used to explore relationships between free residual chlorine decay and water quality, water handling practices and contextual factors. These variables accounted for only about 25% of the variance in the data, suggesting that other unknown or undocumented factors are also important. Ambient air temperature during water collection was positively associated with free residual chlorine decay, suggesting that where ambient temperatures are high we can expect decay to be accelerated.40 There was some evidence of a direct relationship between decay and electrical conductivity, a proxy measure for dissolved metals and salts content of water, raising the possibility that complete oxidation of dissolved metals and other compounds, as required for effective chlorination and maintenance of residual, was not being achieved before distribution. Covering household water storage containers had a protective effect, while there was inconsistent evidence for other hygienic water handling practices (e.g. container cleanliness and method of drawing water). Detailed regression results are available from the corresponding authors.

Discussion

According to our models, under the conditions in the South Sudan camps, it is not possible to ensure 0.2 mg/L of free residual chlorine 24 hours after distribution. Therefore, we must accept a lower level of protection than is ideal, or consider improvements to current practice. Centralized batch chlorination may be less appropriate where populations are dispersed across rural areas or within existing urban settlements; among populations with low chlorine taste or odour acceptance; or during the transitional phase from an acute to a stabilized emergency.45 In some situations, chlorine decay is so rapid that alternatives or adjuncts such as point-of-use water treatment may need to be considered.46

Regarding the safe water chain, covering household water storage containers was confirmed to be protective while there was weak or inconsistent evidence on the effect of other important hygienic water handling practices. The fact that an effect was not observed in the present study does not imply that these practices are necessarily ineffective; promotion of hygienic water handling practices remains an essential component of emergency safe water supply. Although there was some evidence concerning the effect of container and tap type, the study was not sufficiently powered to draw strong conclusions in this regard.

Current guidelines for free residual chlorine in emergency water supplies are not based on field evidence and offer inadequate protection after distribution in refugee camps in South Sudan. We recommend that the free residual chlorine guideline be increased to 1.0 mg/L in all situations, irrespective of disease outbreak, pH, or turbidity conditions. This is a tentative recommendation because the degree to which these findings can be generalized to other camps in different settings is unknown. According to our findings, an initial concentration of 1.0 mg/L will provide 0.2 mg/L free residual chlorine protection for at least 10 hours after distribution. This is consistent with the recommended concentration for point-of-use water chlorination in emergency and non-emergency settings44,47 and is within the limits generally considered to be acceptable to users (2.0 mg/L).44 Further studies are required in diverse climatic and environmental settings to expand the evidence base.


Acknowledgements

We thank Simon Juma Choul, Thomas Bashir, Alfaki Yusif, Abdalbagi Madani, Mark Onna, Sebit Khalil, Issa Wallah, Biserka Pop-Stefanija, Annick Lenglet, Matt Arnold, Edward McBean, Cono Ariti, Kristin George and Ivan Gayton.

Competing interests:

None declared.

References