Water quality for young children in Cambodia—High contamination at collection and consumption level

Abstract Unsafe drinking water is a leading cause of child morbidity, especially among young children in low‐income settings. Safe water consumption requires high‐quality water available at its source and at point of use. We examined the quality of drinking water at point of collection and point of use in 796 households in three provinces, in Cambodia. Microbiological testing for coliform and Escherichia coli contamination was conducted for samples collected. Bivariable analysis and multivariable logistic regression were used to examine associations between various factors and the deterioration in water quality (increase in the risk according coliform or E. coli concentration) between point of collection and point of use. Contamination with both coliforms and E. coli was higher at point of use than at point of collection, with contamination at point of collection to account for 76.6% (coliforms) and 46.3% (E. coli). Results suggest that child drinking water represents a considerable pathway for the ingestion of pathogens, in Cambodia. Area of residence, seasonality, type of water source, and water chlorination were associated with coliform concentration between the point of collection and point of use, whereas only seasonality was associated with E. coli contamination (OR = 1.46; 95% CI [1.05, 2.02]). Children living in rural settings were two times more likely to drink water with a deteriorating coliform concentration between the two‐time points than children living in urban settings (OR = 2.00; 95% CI [1.22, 3.30]). The increase in coliform and E. coli concentrations between point of collection and point of use indicates that water contamination mostly occurs within the household. Strengthening national legislation on water quality standards and promoting safe water management at the household are needed.


| INTRODUCTION
In Cambodia, children under 5 years of age continue to suffer high rates of death and disability from malnutrition. Annually, approximately 4,500 deaths can be attributed to malnutrition (Moench-Pfanner et al., 2016). This represents nearly one third of the overall child mortality rate in Cambodia. The adverse impact of malnutrition constitutes an economic burden that costs Cambodia an estimated 145 to 266 million USD annually (0.9-1.7% of gross domestic product; Moench-Pfanner et al., 2016). Stunting, considered the most appropriate multi-sectorial indicator for malnutrition for its nutrition sensitive and specific related, accounts for 45% of projected economic losses (Moench-Pfanner et al., 2016). Undernutrition is both a major cause and an effect in the cycle of poverty triggered by inadequate water, sanitation and hygiene (WASH) water intended for consumption by children under 5 years to better understand the extent of Cambodian children's exposure to contaminated water. In doing so, our research aims to build on existing data on main household drinking water sources (National Institute of Statistics et al., 2015) and their quality (Ministry of Rural Development & World Health Organization [WHO], 2013;WHO, 2015) in Cambodia, by (a) providing a greater age-specific focus aligned with the nutritionsensitive WASH approach proposed by Cumming and Cairncross (2016) and (b) providing a more detailed analysis of microbiological drinking water quality in line with Sustainable Development Goal 6.1 on improving access to clean water. The findings of analyses are intended to inform policies and programmes that aim to reduce water-related risks associated with child stunting and wasting as part of broader public health strategies to reduce child undernutrition, in Cambodia.

| METHODS
This study is part of a wider research project called "The Cambodian including a variety of factors that influence the "physical, mental and social well-being" of participants. These included factors related to access to health care, nutritional, access to clean water and sanitation (WASH) factors, socio-economic situation, and cognitive development.
The main interest of the collaborators of this project is child development in the first 5 years of life. This implied an observational study design with longitudinal data collection from a cohort representative of the general population selected from the six districts in three provinces (Somphos et al., 2018).

| Study sites
Interviews with mothers of children under the age of three were conducted in Phnom Penh (Russei Kaev district), Kratie province

Key messages
• The prevalence of faecal contamination in household drinking water in Cambodia is high, particularly in rural areas.
• Water contamination with coliforms and Escherichia coli is higher at point of use compared with point of collection.
• The major contributors to contaminated drinking water happen at the household level in Cambodia.
• Drinking water is a key pathway for ingestion of faecal and other pathogens among young Cambodian children.
• Strengthening national legislation on water quality standards and promoting safe water management at the household level are needed to reduce children's exposure to pathogens.
(Chitr Borie and Krong Kratie districts), and Ratanakiri province (Ou Chum, Krong Ban Lung, and Bar Kaev districts) during two rounds of survey (the first round during the wet season and the second during the dry season) as part of a project called "MyHealth, a longitudinal study." The main objective of the project was to collect health and nutrition monitoring data during 3 years in selected districts in three provinces.
Beside the urbanized district of Ban Lung, Ratanakiri is formed by rural settlements where most inhabitants are smallholder farmers who practice a subsistence agriculture supplemented by some food collection from surrounding forests and rivers. Ratanakiri has a large proportion of indigenous people with a high number of ethnic groups that have their own language and culture living in remote areas.
The clear majority of Kratie population (80%) is ethnic Khmer (Cambodian, 80%). The Mekong river provides the population in this province the opportunity to farm and produce crops, four out of five residents are employed in agriculture, and 70% of the population in the province is concentrated along the river. Phnom Penh is the capital city of the country. District selected in Phnom Penh included (a) two districts with a good offering of health services and hospitals, as well as stable food security, and (b) one peri-urban district, Russei Kaev with a typical, large proportion of poor urban population from mixed ethnic origins and with limited access to services.
Farming is not an option in Russei Kaev, but there is an availability of various jobs for different backgrounds and multiple educational opportunities.

| Data collection
The Multiple Indicator Cluster Survey Water Quality Module (UNICEF, 2017) was used as part of UNICEF's early childhood longitudinal study questionnaire. Two types of water samples were obtained: one collected from the source of drinking water used for children under 3 years of age and one from the main drinking vessel used by caregivers to provide water to their children aged 0-36 months (hereafter referred to as "point-of-collection" and "point-of-use" samples, respectively). For bottled water, point-of-collection samples were taken directly from the water bottle. Point-of-use samples were collected from the drinking container, or if the water was also consumed directly from the bottled water, then the point of use and point of collection were identical: at the water bottle.
Ethical approval for the study was obtained from the Cambodia National Ethics Committee for Health Research under the Ministry of Health. Participation was voluntary, and all participants provided informed consent prior to enrolment. . If total chlorine was above 0 mg L −1 , then chlorination was used as a way to improve the quality of the water.

| Methods for testing water samples
If free residual chlorine is equal or above to 0.1 mg L −1 , left over of chlorine is still available in the water to inactivate disease-causing organisms. Duplicate tests were performed for~7% of total samples for quality control, and measurement variability was lower than the predetermined 10% level of precision.

| Statistical analysis
Chi-square was used to assess bivariate relationships between independent variables and coliform and E. coli concentration dependent variables. Variables for the multivariable logistic regression model were selected through a backward stepwise conditional approach.
Variables not significant in the model (P > 0.05) were excluded. The covariates used to build the model were area of residence (urban vs. rural), province (Phnom Penh vs. North-East), source of drinking water (nonimproved vs. improved; WHO/UNICEF, 2017), seasonality (dry vs. wet), and presence of free residual chlorine (yes vs. no), and total chlorine (yes vs. no). Associations between those variables (area of residence, province, source of drinking water, seasonality, and presence of free residual chlorine and total chlorine) and the dependent variables of deterioration in microbiological water quality (increased risk of coliform and E. coli) were assessed using multivariable logistic regression models. Results are expressed as odds ratios with 95% confidence intervals. A P value of 0.05 was considered statistically significant. All analyses were performed using SPSS software version 20 (IBM Corp., Armonk, NY).

| RESULTS
Children selected for this analysis were in majority between 6 and 24 months old (64.2%), and 28.1% were between 25 and 36 months old. Gender distribution showed homogeneous. Among the 796 water samples collected, 52.4% were collected in rural areas; 77.4% were from an improved source, defined as piped water, public tap, tube well, protected dug well, protected spring, rainwater, and (factory) bottled water (Table 1). Almost all water samples from urban settings were from an improved source (97.8%) compared with 59.1% in rural areas (P < 0.001). The consumption of bottled water represented 30% of overall water consumption across all areas and reached 37% of water consumed in Ratanakiri. Table 2 presents the results of the microbiological tests of drinking water samples for coliform bacterial contamination. At the point of collection,~25% of water samples did not have coliform contamination (0 CFU/100 ml). However, for samples collected at point of use, coliform concentration >100 CFU/100 ml was~70%.
Higher contamination with coliform bacteria at both point of collection and point of use was observed in rural areas, in north-eastern provinces, for nonimproved sources, and in nonchlorinated water samples, with the prevalence of contamination at the point of collection to be significantly higher (Table 2).  The multivariable analysis for the deterioration in water quality due to coliform and E. coli contamination between point of collection and point of use revealed that area of residence, type of water source, and whether water was chlorine treated were associated with increased total coliform concentration between the two-time points (Table 4, Figure 1). Children consuming water from an improved As approximately 59% of water samples were found to be contaminated with coliforms at the point of collection at a level above 11 CFU/100 ml in our study, including 36% of samples from improved sources having presence of E. coli, these findings call into question the quality of drinking water obtained from what are generally considered "improved sources." Further, the increased coliform and E. coli contamination of water between point of collection and point of use suggests that an additional burden of water contamination is occurring at  (Somphos et al., 2018), and any concentration of E. coli bacteria makes water unfit for drinking, ensuring safe household drinking water constitutes a public health priority in Cambodia.
The substantial deterioration in water quality between point of collection and point of use in our study is consistent with evidence examining these effects on preventing diarrhoeal illness in children and highlights the need to reduce water microbiological contamination at household level. This is a finding in line with a study that questioned the appropriateness of interventions to improve water quality at the point of collection for preventing diarrhoea. That limited evidence of effectiveness stressed the need to focus on household water practices . Source of drinking water has shown to be a key contributor to child wasting in the north-eastern provinces  supporting the importance of water hygiene promotion in households and communities as part of a comprehensive approach for ensuring good child nutritional status.
In our study, bottled water was the most common source of household drinking water (24-37%) in the north-eastern districts, a higher prevalence than reported in a recent national survey in which   in these areas, with counts of fungi, yeast, and heterotrophic bacteria (total count plate) above the national limit in more than 75% of water samples. These data are concerning as households presumably have confidence in the safety of bottled water and, therefore, do not treat it prior to consumption.
Our multivariable analysis seems to reinforce this point, as improved sources of drinking water were two times more likely to deteriorate for coliforms and potentially for other pathogens. Of all the water sources considered in this study, water piped into the dwelling had the lowest prevalence of faecal contamination at the point of  (WHO, 2015) have shown that most urban Cambodian households store their water before use due to frequent interruptions in water supply. More than 20% of households store water in unsealed containers and dip bowls, scoops, cups, or other vessels into the water (WHO, 2015). These findings reinforce the need for education to improve storage, handling, and treatment of water in the household and reduce the risk of pathogen exposure among household members, particularly young children, who are at a higher risk of consequent undernutrition. In Indonesia, the greatest contamination and risk factors were found in the poorest households indicating the urgent need for targeted and effective interventions there (Cronin et al., 2017). In Cambodia, a significant amount of the population is at a high risk of nonchemical contamination due to poor sanitation and shallow groundwater levels. Therefore, water quality standards as well as its enforcement mechanisms are essential for the Royal Government of Cambodia through coordinated efforts. In rural areas, integrating water quality aspects as part of the overall rural water supply monitoring system is vital, whereas in urban areas, monitoring drinking water quality standards in piped systems and supporting the licensing of private operators could reduce contamination. In the meantime, behaviour change programmes that focus on household water treatment and storage practices are required with a focus on ensuring quality water for young children.
Our findings provide specific information on the extent, source, and contamination of household drinking water in distinguished locations in Cambodia. However, this research is subject to some limitations. Though data collection cycles were designed to capture humid and dry seasons, data collection for the humid season occurred during October-December 2016, with the raining season to occur between May and November, in Cambodia. Also, procedural limitations involved the use of on-site water quality testing equipment with lower levels of precision than laboratory equipment. Finally, although measurement of E. coli is regarded as the most reliable measure of public health risks associated with poor quality drinking water (Wright, Gundry, & Conroy, 2004), using faecal bacteria as an indicator precludes being able to distinguish between host sources of faecal contamination, which requires a more sophisticated molecular method (Schriewer et al., 2015). To monitor the achievements in the years to come, inclusion of water quality testing at point of use in national surveys may be recommended as it is done for salt iodization in demographic health surveys for example.

| CONCLUSION
Our study revealed high contamination of unimproved water sources at point of collection, as expected, and increased concentrations of faecal pathogens at point of household consumption, in water from both improved and nonimproved sources. The results highlight the importance of addressing unsafe drinking water as part of a comprehensive WASH programme to reduce faecal exposure and transmission and diarrhoeal illness to improve health and nutrition outcomes in Cambodian children. In Cambodia, increasing access to improved water sources, especially in rural areas, improving household water management and treatment practices through the adoption of good hygiene practices to prevent household-level contamination, and furthering national and regional water quality standards for all water sources, including bottled water, are key public health priorities for Cambodia, to improve child nutritional status and development capital.
These measures should be combined with community and household sanitation improvements to maximize reductions in child faecal exposure and prevent associated adverse consequences on health and nutritional status.