Inclusion of water quality testing in the Afghanistan Living Conditions Survey and status of bacteriological contamination of drinking water in 10 provinces of Afghanistan

The UNICEF Multiple Indicator Cluster Survey module for water testing was included in the 2016/ 2017 Afghanistan Living Conditions Survey (ALCS) for 10 of the 34 provinces. The module ’ s impact on the survey implementation was assessed through interviews and focus group discussions with survey teams. To assess the level of fecal contamination, drinking water from the source and at the point of consumption was tested for Escherichia coli using on-site membrane ﬁ ltration. On-site testing of water generated signi ﬁ cant interest from community members to receive water test results and understand how to keep their drinking water safe from contamination. The inclusion of the module in the ALCS facilitated access of the enumerators to both communities and households. Only 21.0% of households used safely managed drinking water services. A majority of households (58.2%) used water sources contaminated with E. coli , while E. coli contamination at the point of consumption was found in 77.0% of households. E. coli were more frequently detected in water sources used by households with unimproved sanitation. Beside improvement and increased protection of the water supply services, water quality deterioration between source and point of consumption calls for the promotion of safe handling and storage of water at the home.


INTRODUCTION
The indicator for the SDG target 6.1 on access to the supply of safe drinking water is the proportion of the population using 'safely managed drinking water services' (WHO a, b). A safely managed drinking water service is defined as an improved drinking water source that is located on-premises, available when needed, and free of Escherichia coli bacteria, an indicator of fecal contamination and priority chemical pollutantsarsenic and fluoride. 'Improved' drinking water sources include: piped water into dwelling, yard or plot; public taps or standpipes; boreholes or tube wells; protected dug wells; protected springs; packaged water; and delivered water and rainwater (WHO a, b). Microbial contamination of drinking water is responsible for the majority of the water-related health burden, and contaminated water can transmit diseases such as diarrhoea, cholera, dysentery, typhoid, and polio (WHO ). Evidence from a systematic review suggests that 1.8 billion people use a source of drinking water contaminated with E. coli bacteria (Bain et al. ).
In many areas of developing countries and, specifically, in war-affected areas, the ability to conduct microbial testing is challenged by the availability of laboratory facilities, trained personnel, and essential infrastructure, such as electricity supply.
The WHO/UNICEF Joint Monitoring Programme (JMP) for Water Supply, Sanitation and Hygiene monitors and reports on progress toward SDG target 6.1 and 6.2.
The UNICEF-supported Multiple Indicator Cluster Survey (MICS) is one of the main global data sources to support countries to collect nationally representative data on the situation of children and women (Khan et al. ).
For monitoring access to safely managed drinking water services, the JMP and UNICEF-MICS teams have developed a water quality module for the direct testing of drinking water in household surveys (Bain et al. ). The module includes testing of E. coli bacteria, the globally recommended indicator for fecal contamination (WHO ). The main research questions of this study were the following: 1. What lessons can be learned for future surveys implementing water quality testing in Afghanistan? 2. What is the level of fecal contamination of drinking water sources and water collected at the point of consumption in Afghanistan?
The pilot also allowed for the CSO to gain experience with the module to inform a decision whether to include it in future household surveys.

METHODS
Interviews and focus group discussions to record lessons learned from implementing water quality testing in the ALCS household survey One week after completion of the survey, we conducted interviews with the enumerators responsible for the water quality module of all 11 teams and a focus group discussion (FGD) with all enumerators to determine the impact of the water quality module on the implementation of the ALCS survey. We used a standardized questionnaire for the interviews to document the interactions with the household members regarding the water quality test (interview questionnaire, Supplementary Material, Table 1). We also used a standardized questionnaire to document the enumerators' experience with the new low-cost membrane filtration kit and phase change incubator. Each interview lasted for about 25-35 min. No personal identifiers were documented.

Water quality module
The water quality module was piloted in 10 of the 34 pro- Organization ). Only three households (household number 1, 5, and 9) of the 10 from each cluster were selected for water quality module. From each house, two water samples were tested for E. coli, one sample directly from the water source of the house referred to as 'point of collection' and another from a glass or pot from which households usually drink water referred to as 'point of consumption'. In case of source samples, water was first collected in sterile bags (Whirl-Pak ), and for the point of consumption, enumerators asked the household respondent for 'a glass of water that they would give to a child to drink' and directly tested water from the glass or any other vessel provided by the respondent. For quality assurance purposes, each survey team also tested a blank sample once in each cluster. The blank sample was boiled water assumed to have no E. coli provided by the supervisor to its team. These ready-to-use, chromogenic plates for the detection of E. coli give results within 22-24 h. Incubation was achieved using specially designed phase-changing incubators to maintain a temperature of 30 C even during the night. After 24 h, the number of blue colonies, signifying the presence of E. coli colony forming units (cfu), were recorded and classified into the following risk categories: low risk (<1 per 100 mL), contaminated (1-100 per 100 mL), and very high risk (>100 per 100 mL). If there were more than 100 colonies on the plate and in cases where the plate turned blue/green, this would be recorded as '101' to indicate 'too numerous to count' (MICS ). An information leaflet in Pashtu or Dari on handwashing and water treatment (English version in Figure 1) was provided to each household where water quality tests were performed. In response to requests from households and community leaders to be provided with the test results, the enumerators agreed to provide their telephone numbers for the households to call them the next day to obtain their test results. Test results were provided to all who contacted the enumerators.

Data analysis
Data analysis was conducted in Stata 16 and used svy to account for survey design. We calculated the proportion of samples in each E. coli risk level and reported 95% confidence intervals for the proportion of households where any E. coli were detected in the source of drinking water  Water quality results and estimates for safely managed drinking water services for the 10 province pilot were disaggregated by the type of residence (urban and rural or Kuchi), the type of water source (piped, boreholes, protected wells and springs, unprotected wells and spring, surface water, and other), and sanitation facility used by households (improved, unimproved) and wealth quintiles. The wealth quintile analysis followed standard procedures outlined by the Demographic and Health Surveys but excluded WASH variables in order to mitigate potential tautology between the outcome (water quality) and asset index (Rutstein & Johnson ). The CSO teams from central and regional offices conducted field monitoring visits as part of the quality assurance for the water quality module.

Ethics statement
The CSO provided ethical oversight and approved interview protocols for the ALCS survey, and the interviews conducted following the pilot and provided microdata for the secondary analysis of the water quality findings. Survey teams were informed of the voluntary nature of the interviews and FGDs following the main survey. Written informed consent was acquired from all respondents before beginning the interviews and the FGDs.

RESULTS
Lessons learned from implementing water quality testing in the ALCS household survey None of the households refused water quality testing. The enumerators reported that the water quality test was, in fact, considered an incentive by households to participate in the survey. All households asked to be provided with the test results. Enumerators (n ¼ 10) reported being asked for their telephone numberand in the absence of guidance to the contrary provided these to the survey respondents. In most cases, enumerators were called back the next day to communicate the test results. Half of the respondents (50%) asked how to solve the contamination problem if their drinking water was contaminated. Almost without exception, households asked for chlorine to disinfect their drinking water. Ten of the 11 enumerator teams responded to the questionionnaire regarding provision of chlorine tablets. Four of the 10 enumerators (40%) mentioned that it would be good to provide chlorine tablets/sachet as gifts to the households where they tested the water. The information leaflet on handwashing and water treatment was considered useful by enumerators and was well liked by the community. Most enumerators (80%) advised improving the leaflet with photos, instead of plain text, and adding information on how to protect a water source from becoming contaminateda common question that the enumerators could not readily answer. Almost all enumerators indicated that the leaflet was an essential item to provide to the households and community leaders (Malik and Imam). Many Imams indicated that they wanted to put the leaflets on the walls to remind people about ways to treat drinking water.
Often, the male enumerators tasked with conducting the water quality test were refused entry into the homes and they had to conduct the water quality testing outside of the home or even the compound on the street or sometimes in their guesthouses. The enumerators suggested that in the future, the CSO also trains female enumerators on conducting the water quality tests. Women, in Afghanistan, usually have much better access into the homes than men, and this would allow the enumerators to observe where the water for testing is actually taken from. Due to the interaction with the community members about the water quality testing, enumerators reported that they felt more at ease and safer conducting the ALCS survey than they did before the inclusion of the water quality module into the survey.
During the survey, Maliks (community leaders) and community elders usually asked for their drinking water to be tested as well. Enumerators strongly felt that they had to give in to at least some of these requests and in every cluster 1-2 water tests were conducted outside of the survey.
Similarly, prior to the survey, all enumerators also received requests for water testing from their friends, relatives, and neighbors, and most gave in to these requests.

Water quality results
The water quality module was implemented in 307 clusters.
In total, 896 samples from point of consumption and 816 samples from water source (point of collection) of households were tested for E. coli bacteria. Overall, 58.2% of drinking water sources at point of collection and 77.0% at point of consumption were found to be contaminated with E. coli (Table 1). About half (45.5%) of the households in urban areas and two out of three rural and Kuchi households (67.0%) in the 10 provinces used water from a source that was contaminated with E. coli.
Households using improved drinking water sources, including piped water and boreholes, were most likely to use a water source that was free from contamination. Fecal contamination was found in about half (48.9%) of the improved water sources. The results also indicate that access to water free of E. coli contamination is concentrated among the wealthier households and increases with increasing wealth quintiles. It is higher in urban areas than rural and Kuchi areas.
In many households, water quality was found to deteriorate between collection from the water source and at point Second column (very high risk (>100 E. coli per 100 mL)) in Table 1 is subset of the first column (contaminated (1 E. coli per 100 mL)).
of consumption within the home. Table 1 shows that the proportion of households with E. coli detected in their drinking water increased between water source (58.2%) and the glass within the home (77.0%) and likely reflects contamination that occurred during household water storage and handling.
The extent of water quality deterioration between water source and point of consumption is higher in urban households (from 45.5 to 67.1%) than in rural and Kuchi areas (67.0-84.2%). Households using improved drinking water sources, including piped water and boreholes, were most likely to use a water source that was free from contamination.
Results also highlight the likely impact of unimproved sanitation on the quality of the water source. About 71.7% of the households having E. coli in their water source were also using unimproved sanitation facilities compared with those using improved sanitation facilities where E. coli was detected in 51.3% of their water sources.
Overall, only 13.0% of households in the selected 10 pro-

Safely managed water services
Households with improved sources accessible on-premises, with sufficient quantities of water available when needed and free from E. coli contamination, meet the SDG criteria for 'safely managed' drinking water services. Table 2 shows  the proportion of the household with improved water sources located on-premises, available when needed, and free from contamination. It combines this information to provide estimates of safely managed drinking water services.
In the 10 provinces, 73.9% of households have access to improved water sources, 54% used a water service that was located within the dwelling or in the household's yard or plot ('accessible on-premises'), and 68% used a water service that was available in sufficient quantities during the last month ('available when needed'). Similar to other countries (WHO/UNICEF-JMP ), water quality was the major limiting factor for safely managed services as only 34.8% of households used a water source in which E. coli were not detected ('free from contamination'). Taken together, the findings from the ALCS pilot show that just two out of 10 households (21.0%) used safely managed services.
Households in urban areas (36.7%) were more likely to use a water service that is safely managed compared with those in rural and Kuchi areas (10.1%), due to insufficient drinking water or water found to be contaminated with E. coli. Access to a safely managed water service is higher in the richest households (40.3%) and decreases with wealth quintile as only 4.4% of the poorest households have access to safely managed water service.

DISCUSSION
This study sought to document the findings from a pilot integrating water quality testing in the ALCS in Afghanistan, examining the impact of the implementation of the survey and the findings related to levels of fecal contamination in 10 selected provinces. The study was part of a wider initiative to support countries to integrate water quality testing in household surveys and address data gaps on safely managed drinking water services.
Lessons learned from implementing water quality testing in the ALCS household survey The overall experience suggests that both survey teams and communities have shown a high level of interest in the water quality module and that its inclusion in the ALCS has facilitated access to the survey teams to enter communities which, particularly in war-torn areas, is not always a given.

Water quality results
Similar to other studies (Shields et al. ), we found that drinking water at the point of consumption was more likely to be contaminated than water collected at the source.

DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.