Despite progress over the past two decades, global child undernutrition rates remain alarmingly high, particularly in South Asia and in Sub-Saharan Africa. Stunting and wasting represent the predominant forms of undernutrition: stunting, a height for age two standard deviations below the mean, associated with chronic malnutrition, was globally estimated to affect 151 million children under five years old in 2017; wasting or acute malnutrition, defined as either a weight for height two standard deviations below the mean or a mid-upper arm circumference less than 125 mm, affected 51 million children in 2017 [1
]. In Cambodia, where this study was carried out, the most recent Demographic Health Survey estimated the overall prevalence of undernutrition among children aged 0–59 months to be 32.4% for stunting and 9.8% for wasting. However, there were disparities between urban and rural provinces: 33.8% of rural children were stunted and 7.5% were wasted compared with 23.7% and 9.9% of urban children, respectively. By province, stunting prevalence was higher in Ratanakiri/Mondolkiri (39.8%) and Kratie (38.4%) than Phnom Penh (17.9%) while wasting prevalence was 8.4% in Phnom Penh, 8.2% in Ratanakiri/Mondolkiri and 6.5% in Kratie [2
Undernutrition has a complex aetiology with numerous and overlapping factors, leading to short- and long-term consequences, such as child’s and mother’s morbidity and mortality, child’s physical and neurodevelopmental impairment and reduced economic performance [3
]. Immediate causes include inadequate dietary intake and diseases; indirect factors comprise poverty, feeding and care practices, poor access to and low-quality health services and sociocultural, economic and political context [8
] and are often liked to inadequate access to water, sanitation and hygiene (WASH) services [9
]. However, most data shows associations between poor WASH practices and the incidence of diarrhoea or other infections rather than direct links to a child’s nutritional status. Diarrhoea, the main cause of mortality and morbidity among children under five years of age, has a bidirectional relationship with undernutrition: it can lead to undernutrition because it causes loss of appetite, lack of absorption and increased metabolism and, at the same time, undernutrition weakens the immune system of undernourished children, thereby increasing their susceptibility to infectious diseases [10
]. Parasitic and helminthic infections have been shown to be directly linked to poor sanitation conditions [12
], and can indirectly cause loses of nutrients, or impair absorption of nutrients. Moreover, helminth infection might affect child’s immunity and immune responses [13
]. A systematic review and meta-analysis on the direct effects of WASH interventions on child’s growth identified five cluster-randomised controlled trials (RCTs), one occurred in Cambodia, reporting that solar disinfection of drinking water, provision of soap, and improvement of water quality slightly but significantly benefit child’s linear growth [13
]. Furthermore three RCTs in Indonesia, India and Mali, showed that sanitation interventions reduced stunting or improved child height and weight [15
] and a non-randomised study reported increased height-for-age z-scores (HAZ) in villages supported with WASH interventions including water quantity, sanitation and hygiene [18
Nevertheless evidence in the literature includes discordant results. Two recent RCTs in Bangladesh and Kenya concluded that integration of water, sanitation, and handwashing with nutrition did not have any benefit on improved child’s growth [19
]. One study in India showed that improved sanitation facilities decreased open defecation but did not improve children’s growth [21
]. Furthermore three non-randomized trials targeting water quality and sanitation or hygiene promotions did not report any association with stunting [22
Limited evidence exists in the literature for Cambodia where, in 2015, 30% of the rural population lacked basic drinking water, 51% used open defecation and 40% did not have basic hygiene services [25
]. Child undernutrition has serious public health consequences for the Cambodian population and for the country’s economic development. Growth impairment in the first years of life leads to adverse health complications and to long-term economic hardships due to heavy health expenditures and decrease in productivity. In 2013–2014 the Royal Government of Cambodia, supported by UNICEF and the World Food Programme, estimated a loss of more than $
400 million due to child malnutrition [26
To understand the causes of this impoverishment UNICEF, in collaboration with the Institute of Research for Development (IRD) and Reproductive and Child Health Alliance (RACHA), performed a longitudinal study in one district in Phnom Penh, and five districts in Kratie and Ratanakiri, north-eastern provinces, to monitor the health and nutritional status of children and provide rapid feedback about the progress that can be made with enhanced health monitoring.
To date there is no evidence for a significant association between WASH and child nutrition status in South East Asian countries. In this paper we aim to investigate for the first time in Cambodia the longitudinal association between water, sanitation and hygiene practices, regrouped as child-sensitive WASH composite scores, and the nutritional status of children under five years of age measured as weight-for-age z-score (WHZ) and mid-upper arm circumference (MUAC) for wasting and height-for-age z-score (HAZ) for stunting.
To our knowledge, this study is the first to investigate the longitudinal association between water, sanitation and hygiene conditions, regrouped as child-sensitive WASH composite scores, and the nutritional status of children in Cambodia. Data were collected in a longitudinal cohort study in six districts localized in Phnom Penh and Kratie and Ratanakiri, two north-eastern provinces of Cambodia. The study, started in 2016 and ending in March 2019, aims to collect in-depth data to monitor the progress on health, nutrition, WASH practices, and cognitive development of children under five years of age and to inform the government about the progress that can be made with enhanced health monitoring. Although UNICEF did not carry out interventions in the targeted districts, we cannot exclude that other development partners did. The analyses performed in this paper take into account participants’ changes over time: due to external or unknown causes, participants might have changed their main source of water or type of toilet used over time or their hygiene habits. Decreasing of wasting from baseline to follow-up 2 and increasing of stunting prevalence along the study were observed. The increase of stunting prevalence can be explained by the chronicity of the nutrition condition. Stunting is a largely irreversible condition: a child cannot recover height in the same way that they can regain weight. Multiple factors as poor maternal health and nutrition, inadequate infant and young child feeding practices and infection cause stunting [36
] which are more difficult to tackle in its entirety. Without targeted interventions, stunted children are more likely to remain stunted and the new cases explain the increased prevalence observed along the study. Actions in multiple areas are required to achieve the Global Nutrition Target 2025 of 40% reduction in the number of children under five years of age who are stunted.
The crude and adjusted—for gender, age, exclusive breastfeeding, dietary diversity score, mother education, wealth index and province of residence—analyses showed evidence for a positive association between child-sensitive WASH composite scores and the child anthropometry indicators for wasting. Regardless of the nutritional status of the child, by improving one unit of the JMP-CS composite score, the child’s WHZ increases by 0.14 SDs and the child’s MUAC by 0.18 cm. Similarly, improving one unit of the National-CS composite score, the child’s WHZ increases by 0.4 SDs and the child’s MUAC by 0.4 cm. However, the fully adjusted association between age or gender and WHZ took opposite directions compared to the fully adjusted association with MUAC. A recent study in the same cohort of Cambodian children showed that both MUAC and WHZ showed gender bias, with MUAC identifying more girls and WHZ identifying more boys with acute malnutrition. The gender bias, strongest for MUAC, diminished with older age, but remained significant up to 30 months of age [37
]. The results obtained in the cohort stressed on the importance to maintain both WHZ and MUAC for the identification of acute malnutrition in children in Cambodia.
Moreover crude and adjusted analyses showed evidence for a positive association between the JMP-CS composite score and the child anthropometry indicator for stunting. Regardless the nutritional status of the child, by improving one unit of the JMP-CS composite score, the child’s HAZ increases by 0.15 SDs. However, this association is weaker if the National-CS score is used (p = 0.08), albeit the crude analysis was strongly associated.
This study adds to the limited evidence on the direct association between WASH practices including water, sanitation and hygiene, and both wasting and stunting undernutrition statues. To date, mainly randomized controlled trials explored this association, showing discordant results. Three RCTs in Indonesia, India and Mali showed stunting reduction and child height and weight improvements after sanitation interventions [15
]; and similarly the systematic review and meta-analysis of Dangour et al., showed slight but significant benefits of water quality and hygiene interventions on the child’s length growth [13
]. However, two recent RCTs in Bangladesh and Kenya showed that integration of water, sanitation, and handwashing with nutrition did not have any benefit on children’s growth [19
This study brings evidence for a positive longitudinal association in the six targeted districts of Cambodia, whereby various factors of causality were accounted for, minimizing the potential residual confounding and collinearity. The large sample size (n = 5238) increases the power of the statistical analyses, but may increase study attrition and missing data. Indeed, several mothers and children did not show up at each study visit. This could be due to migration often occurring in the country, where families move across provinces seeking for seasonal jobs. The method of analysis chosen adjusts for missing data, as assuming they are missing at random. Furthermore, tablet-based data collection automatically entering information into a database, allows for data checking along the course of the collection and for recover data when they are missing. Quality data collection was ensured during the study. Anthropometry data (length, weigh and MUAC) were measured in duplicate reducing measurement bias. These data were also recorded on paper in order to calculate the z-scores in situ and detect and refer severely malnourished children for treatment. Paper-based data collection also allows having a backup in case of any mistake occurred in the tablet data entry. The WASH questions were formulated in an understandable manner supported by figures showing the different type of source of water (piped, dug well, pumping water, river…) and toilet (different type of flush/pour flush toilet, pit latrine with/without slab, bucket, field…), reducing information bias.
However, although only comparable variables among the study visits were used for the analysis, the questionnaire slightly changed after baseline, potentially biasing measurements. As shown in Supplementary Table S1
, the answers of the WASH practices questions were phrased slight differently in baseline but the values were referable to “improved” or “not improved” classification according to the guidelines. Furthermore we acknowledge weakness on the available variables used to create water, sanitation and hygiene sub-scores. Complete water quality and quantity data including indicators for accessibility (within 30 min, on premises), availability when needed and water quality, such as microbiological and chemical contaminations, were collected for only around 800 randomly selected households. These data could not be used in the longitudinal analyses because of the small proportion compared to the whole cohort and will require further analysis. The sanitation sub-scores did not include the variable indicating whether facilities were shared with other households because collected using inconsistent classifications across time. The hygiene sub-scores lack to gather information on adults’ hygiene practices after toileting which may have impacts on food preparation and other activities that impact the child. Only child-related hygiene practices were collected in the study, leading to limit the analyses on available data. Finally the indicator about handwashing with water and/or soap was not comparable between baseline and follow-ups and it was thus excluded for the creation of the hygiene sub-scores.
The longitudinal study aims to inform the government about the progress in the targeted districts underlining the importance to consider the Cambodian context in the analyses. Indeed, two CS-composite scores were independently created: the JMP-CS defined according to the worldwide accepted JMP guidelines and readapted as child-sensitive score by the authors; and the National-CS composite score, according to the national guidelines more conservative for water quality and sanitation classifications. The national drinking water classification adds an indicator about the method used for water treatment and considers bottled water as “not improved” because of the high numbers of contaminated bottles sold in the country, probably due to unsafe management (Poirot et al., submitted to Maternal and Child Nutrition). The national sanitation ladder considers buried faeces as “not improved” because of the frequent floods occurring in the country, especially in the rainy season, meaning buried stools easily re-surface. We obtained similar results for wasting highlighting the importance to integrate nutrition and WASH implementing programmes to improve the growth of children in the targeted districts. The association with stunting is significantly associated with the CS-JMP composite score but it becomes less significant when the National-CS composite score was used as independent variable. Compared to the JMP-CS composite score, the National-CS composite score included more variables leading to increased missing data. Consequently, the final models were performed on a smaller sample size decreasing the statistical power of the analyses. This may explain the differences obtained for stunting and will enforce the results obtained for wasting. Our findings are not generalizable for the whole country but rather to the selected districts where all villages were targeted.