Progress in Water, Sanitation and Hygiene (WASH) coverage and potential contribution to the decline in diarrhea and stunting in Ethiopia

Abstract Inadequate safe water supply and poor sanitation and hygiene continue to be important risk factors for diarrhoea and stunting globally. We used data from the four rounds of the Ethiopian Demographic and Health Survey and applied the new World Health Organization (WHO)/UNICEF Joint Monitoring Program (JMP) service standards to assess progress in water, sanitation and hygiene (WASH) coverage between 2000 and 2016. We also performed an age‐disaggregated pooled linear probability regression analysis followed by a decomposition analysis to determine whether changes in WASH practices have contributed to the changing prevalence of diarrhoea and stunting in children under 5 years of age. We observed a significant increase in the coverage of safe drinking water and adequate sanitation facilities over the period. At the national level, the use of a basic water source increased from 18% in 2000 to 50% in 2016. Open defecation declined from 82% to 32% over the same period. However, in 2016, only 6% of households had access to a basic sanitation facility, and 40% of households had no handwashing facilities. The reduction in surface water use between 2000 and 2016 explained 6% of the decline in diarrhoea observed among children aged 0–5 months. In children aged 6–59 months, between 7% and 9% of the reduction in stunting were attributable to the reduction in open defecation over this period. Despite progress, improvements are still needed to increase basic WASH coverage in Ethiopia. Our findings showed that improvements in water and sanitation only modestly explained reductions in diarrhoea and stunting.


| INTRODUCTION
Despite substantial progress in reducing the burden of diarrhoeal diseases globally, diarrhoea remains one of the leading causes of mortality in children under 5 years of age, and the second leading infectious cause of mortality in this age group, behind lower respiratory infections (Troeger et al., , 2018(Troeger et al., , 2020. For example, in 2016, it was estimated that diarrhoea was responsible for approximately 9% of all deaths in children aged under 5 years (Troeger et al., 2018). In addition, diarrhoeal disease is also a major cause of disability-adjusted life years (DALYs), and in particular, it is a major contributor to the high rates of stunted growth observed in low-income settings (Danaei et al., 2016).
Inadequate safe water supply and poor sanitation and hygiene are significant risk factors for diarrhoea-related morbimortality globally . For example, it has been estimated that 94% of under-5 deaths in 2017 could have been averted by the provision of safe water, sanitation and hygiene (WASH) (Troeger et al., 2020). Inadequate water (e.g., water supplied from unprotected springs and dug wells, surface water and water tanks) and sanitation (flush toilets not connected to a sewer/septic system, pit latrines without slab, bucket latrines or open defecation) practices increase faecal-oral transmission of pathogens, which subsequently leads to diarrhoea (defined as more than three liquid stools per day) (Lutter et al., 2018). Repeated episodes of diarrhoea lead to loss of nutrients and fluids, causing overall weakness and dehydration. In addition, ingestion of faecal bacteria in large quantities leads to environmental enteric dysfunction (EED), a subclinical condition of the small intestine characterized by blunted villi, intestinal inflammation, increased intestinal permeability and consequently a reduced nutrient absorption (Humphrey, 2009) and increased likelihood of undernutrition and stunted physical and cognitive development.
Unsurprisingly therefore, it has been shown that diarrhoeal diseases disproportionately affect settings with poor access to health care, safe water and sanitation, and low-income or marginalized populations . In Ethiopia, for example, where recent estimates reveal that <15% of the population have access to safely managed water supplies and <10% using safely managed sanitation facilities and basic hygiene facilities, diarrhoeal disease remains a major public health concern (World Health Organization [WHO]/UNICEF Joint Monitoring Program [JMP], 2021). In 2016, estimates revealed a prevalence of 12%, and 13% of child deaths were attributable to diarrhoea (Central Statistical Agency [Ethiopia] and ICF, 2016). This high burden of diarrhoea is also likely contributing to the high (but decreasing) stunting burden in Ethiopia, with recent estimates revealing a stunting prevalence of 37% (Ethiopian Public Health Institute [EPHI] [Ethiopia] and ICF, 2019).
The positive effect of improved WASH practices and a reduced diarrhoea incidence in young children is well known and has been reported in several meta-analyses (Clasen et al., 2007;Fewtrell et al., 2005;Wolf et al., 2014). These findings have led to the hypothesis that, via the positive effect on diarrhoea risk, WASH may also confer benefits to linear growth. Although observational and quasi-experimental studies have shown support for an association between WASH practices and child height (Garcia et al., 2013;Lin et al., 2013;Masibo & Makoka, 2012;Merchant et al., 2003;Spears et al., 2013) (Dangour et al., 2013), two large WASH Benefits trials Null et al., 2018) and the SHINE trial  have cast doubt on this association, with all three trials observing no effect on linear growth (LAZ) at 18-24 months of age, with the authors concluding that elementary household-level WASH interventions are unlikely to improve child growth (Pickering et al., 2019).
Although the evidence regarding the efficacy of WASH interventions on reducing childhood stunting is equivocal, the benefits of promoting improved WASH practices remain important for child nutrition (e.g., wasting and micronutrient deficiencies), health and development.
The Sustainable Development Goals (SDGs) recognize WASH as central to sustainable development with SDG 6 calling for universal access to safe and adequate water, sanitation and hygiene for all by 2030 (United Nations, 2015). There is, however, a lack of evidence regarding the trends and predictors of WASH practices over time in Ethiopia and thus little indication about the progress the country is making in achieving SDG 6. The study by Beyene and colleagues, conducted in 2014, assessed the trends in sanitation practices only and revealed that more than half of the population still used unimproved sanitation facilities. A more recent cross-sectional study based on data from 2016 assessed access to and factors associated with improved drinking water sources and toilet facilities (Andualem et al., 2021). The proportion of households' access to improved drinking water sources and toilet facilities was 70% and 25%, respectively, with several factors, including wealth index, distance to water source and sex of household head, related to access. However, the cross-sectional study design meant it was not possible to assess progress in WASH trends over time.
The aim of this study therefore was to describe trends in WASH service coverage between 2000 and 2016 and investigate whether any changes in WASH practices have contributed to a decline in the prevalence of diarrhoea and stunting in children under 5 years of age in Ethiopia.

Key messages
• The coverage of safe and adequate drinking water and sanitation facilities significantly increased between 2000 and 2016.
• Less than 10% of households have access to basic hygiene facilities. This highlights the need for interventions that increase access to basic hygiene.
• Improvements are still needed to increase coverage of higher WASH standards and to address prowealthy and prourban inequalities in WASH coverage.
• Improvements in water and sanitation services contributed to only modest reductions in the probability of experiencing diarrhoea and stunting.

| Data source
We used data from the four rounds (2000, 2005, 2011 and 2016) of the EDHS (Central Statistical Agency [Ethiopia] and ICF, 2011ICF, , 2016 Central Statistical Agency [Ethiopia] and ORC Macro, 2001, 2006  Consequently, the highest service standards used in this analysis are 'basic drinking water facilities' and 'basic sanitation facilities'. A basic drinking water source refers to drinking water from an improved source, with collection time, not more than 30 min for a round trip. A basic sanitation facility is an improved toilet facility that is not shared with other households. The highest hygiene service standard is basic, which requires the availability of a handwashing facility with soap and water on the premises.

| Exposures
As data pertaining to hygiene practices were only available in 2011 and 2016, the analysis to assess the association between WASH practices with outcomes focused on drinking water and sanitation practices only. Specifically, the exposures of interest were the use of surface water as a drinking source and the practice of open defecation. Information on the type of water facility used was reported, whereas the type of sanitation facility used was assessed through a combination of report and observation.

| Outcomes
The primary outcome of our regression analysis was diarrhoea prevalence, the results of which are reported in the main text. Stunting prevalence was our secondary outcome, and this analysis is presented as supplementary analyses. Diarrhoea was defined as the percentage of children with diarrhoea (three or more loose stools per day) at any time in the 2 weeks preceding the survey. Stunting was defined as HAZ/LAZ below À2 SD of the median based on the WHO 2006 Child Growth Standards (de Onis et al., 2004).

| Covariates
The relationships between exposures and outcomes were adjusted for covariates, which were selected on the basis of the Lancet framework for action (Black et al., 2013) or previously reported associations with the outcome variables (Argaw et al., 2019;Danaei et al., 2016;Dearden et al., 2017;Headey & Palloni, 2019;Headey et al., 2016Headey et al., , 2017Oswald et al., 2017;Troeger et al., 2018). These included maternal and paternal education, household wealth, maternal employment, maternal height, child age, child sex, residence, region and survey round.

| Statistical analysis
To account for the cluster sampling used in the EDHS, sampling weights were applied to estimate WASH service coverage, and the prevalence of stunting and diarrhoea across survey rounds. We used stacked area plots, produced at the national and regional level, to explore changes in WASH practices between 2000 and 2016.
Additionally, equity plots documenting the temporal changes in WASH practices by wealth quintile and place of residence (urban vs. rural) were produced. We computed the slope index of inequality (SII) to quantify the absolute difference (percentage point) in WASH service coverage between the wealthiest and the poorest households.
Greater SII values represent higher levels of inequality. The SII is equal to zero if there is no inequality. To account for the incidence of diarrhoea and the timing of growth faltering in childhood, we made an a priori decision to investigate the relationship between WASH practices and diarrhoea and stunting using an age-disaggregated approach. Accordingly, stratified analyses were performed in the following age ranges: 0-5, 6-11, 12-23 and 24-59 months. In addition, we made an a priori decision to include in the analysis, for households with more than one child below 59 months, only the youngest child with anthropometric data. Within the age groups listed above, we performed a regression decomposition analysis, which seeks to determine the contribution made by changes in the mean levels of exposures over time to changes in outcomes. Specifically, the decomposition analysis sought to reveal whether changes in surface water usage and open defecation practices identified in the previous step contributed to observed reductions in diarrhoea and stunting between 2000 and 2016. As has been done in other decomposition analyses, we initially examined the relationship between diarrhoea and stunting with the exposures by pooling data from all rounds of the EDHS and performing a linear probability regression analysis (adjusted for the covariates listed above and with robust standard errors to account for clustering) (Headey et al., , 2016(Headey et al., , 2017. A key assumption underpinning the use of a decomposition analysis, which is based on a pooled regression model, is that coefficients are time invariant (i.e., the magnitude and direction of coefficients are stable over time).
In order to test this assumption, we performed a series of Chow tests, which test whether coefficients differ significantly over time. We did not find support for the coefficients being time-varying, and we therefore followed the approach by  and Headey et al. (2016Headey et al. ( , 2017 and performed a simple decomposition analysis for exposures that were associated with outcomes in the regression analysis at 10% level of significance. The decomposition equation takes the form of  Table 1.  Despite this progress, in 2016, 15% of households still used a limited water source (an improved water source with a collection time of more than 30-min round trip), 24% used an unimproved water source and 11% used surface water. Although improvements in water service standards were seen across all regions, progress was uneven with some regions achieving higher coverage compared with others. Additionally, 40% had no handwashing facility on the premises.

| Trends in inequalities in basic WASH standard coverage
In general, the use of a basic water facility increased over time across all wealth quintiles. However, the change was much greater in the wealthiest quintile, increasing from less than 40% of households in The water service standards used were 'Basic': drinking water from an improved source, provided collection time is not more than 30 min for a round trip, including queuing; 'Limited': drinking water from an improved source, which collection time exceeds 30 min for a round trip, including queuing; 'Unimproved': drinking water from an unprotected dug well or unprotected spring; and 'Surface water': drinking water directly from a river, dam lake, pond, stream, canal or irrigation canal water facility increased markedly between 2000 and 2005, with little change observed thereafter. Access to basic sanitary facilities was consistently higher among the wealthiest households compared with the poorest households, though these differences were small compared with differences seen for access to basic water facilities. As expected, wealthier households and households in urban areas had more access to basic hygiene facilities compared with poorer households and households located in rural areas, respectively, and this difference increased between 2011 and 2016.

| Trends in diarrhoea and the relationship with WASH practices
The prevalence of diarrhoea declined during this period. Reductions in the prevalence of diarrhoea between 2000 and 2016 were observed, declining from 29% to 15% in children under 5 years of age, with the largest reduction observed in children aged 12-23 months (38% in 2000 to 19% in 2016; a 19% change). Across all EDHS rounds, the prevalence of diarrhoea was lowest in the youngest infants (0-5 months). We used pooled regression analysis results (2000-2016) (see Table S4) and the per cent change in surface water use and open defecation to estimate the contribution of the reduction in these practices to the decline in diarrhoea. Among children aged 0-5 months, the use of surface water was associated with a 4% (95% confidence interval [CI]: 1%, 7%, p = 0.018) increase in the probability of diarrhoea (Table S4) (Victora & Barros, 2000). In light of the above, we observed an unexpected association with the probability of diarrhoea and the type of water source (surface water) among children aged 0-5 months. This finding is surprising because we would expect the use of surface water to increase the risk of diarrhoea once complementary feeding starts after 6 months. The lack of association between diarrhoea and surface water use in older children seen in this analysis supports findings from recent WASH trials conducted in Bangladesh , Zimbabwe  and Kenya (Null et al., 2018), which observed little effect of water treatment interventions on diarrhoea incidence in children.
In terms of sanitation practices, we observed that 7% and 8% of the estimated decline in diarrhoea among children aged 0-5 and 24-59 months, respectively, were attributed to the reduction in open defecation. This observed stronger association between sanitation and diarrhoeal incidence compared with water usage has also been reported in an analysis of data from 217 Demographic and Health Surveys (Fuller et al., 2015) as well as a difference-in-difference analysis done by Headey and Palloni (2019).
Ethiopia has made substantial progress in its coverage of safe and adequate drinking water and sanitation facilities. The first campaign for universal sanitation was implemented in the SNNPR region in 2003 (Bibby & Knapp, 2007). The campaign was reliant on a strong commitment and participation from all levels of the community, including a network of volunteer health promoters at the village level, environmental health workers at the woreda level and strong support from the regional health bureau (Bibby & Knapp, 2007). Rapid improvement was seen in just 2 years, with pit latrine ownership rising  (Novotný et al., 2018).
Despite the progress seen, improvements are still needed as the quality of WASH services is low (only 6% have basic sanitation facilities and 40% do not have handwashing facilities). Furthermore, inequalities in WASH coverage still persist, with wealthier households and those residing in urban settings more likely to have access to better water and sanitation facilities. We also reveal persisting regional Another initiative that has been established in Ethiopia with the aim of addressing the challenges faced by the WASH sector is the ONEWASH National Program (Federal Democratic Republic of Ethiopia, 2013). The overarching objective of the programme is to achieve universal, sustainable, climate resilient and equitable access to safe and affordable water for all, along with improved, low environmental impact, sanitation. To achieve this, the programme has multiple medium and longer term goals, which include the construction  Headey et al., 2016Headey et al., , 2017. Furthermore, other studies such as the Malnutrition and the Consequences for Child Health and Development (MAL-ED) birth cohort (Kosek & Mal-Ed Network Investigators, 2017) and WASH Benefit and SHINE trials (Pickering et al., 2019) have found little to no effects of improved WASH and EED and on stunting and diarrhoea. A possible reason for the lack of relationship is that neighbourhood WASH practices are more important than household practices for the reduction of environmental faecal contamination (Jung et al., 2017). However, in a supplementary analysis replacing household WASH practices with community practices, this lack of association persisted (data not shown). In light of the modest results of improvements in WASH on diarrhoea and stunting, there have been calls for research to identify interventions, labelled 'Transformative WASH', that radically reduce faecal contamination in the household environment in LMIC (Pickering et al., 2019). Proposed interventions may include high community coverage of improved sanitation facilities (Fuller & Eisenberg, 2016), complete separation of animal faeces from people's living environments (Boehm et al., 2016;Mbuya et al., 2015), continuous and convenient access to uncontaminated water (Pickering & Davis, 2012) and reductions in faecal contamination on surfaces where young children crawl and play . Given that in 2016, 40% of households had no hygiene facilities, 53% rely on unimproved sanitation facilities, and about 35% of households rely on unimproved water sources or surface water, much more work is needed to increase coverage of even basic WASH services across Ethiopia.
Our finding of an overall reduction in the stunting prevalence in children under 5 years of age in Ethiopia has been reported elsewhere (Golan et al., 2019;Takele et al., 2019). However, the agedisaggregated analysis has enabled us to also identify a changing dynamic of age-related stunting prevalence over time. For example, in 2000, the prevalence of stunting increased immediately after birth (0-5 months) and continued to increase thereafter. In 2016, however, in addition to numbers being lower overall, the time at which the stunting prevalence increased rapidly appears to be later than in 2000, with little change in stunting prevalence in infants aged 0-5 months (13%) and 6-11 months (17%), followed by a rapid rise between 12 and 23 months (39%). This delay in growth faltering may be attributable to improved breastfeeding practices in the first 6 months and improved complementary feeding practices at 6 to 12 months, which may have reduced the exposure to infections as a result of poor WASH conditions.
Although we did not find an association between surface water use and stunting, open defecation was associated with an increased probability of stunting across the age groups. A lower risk of stunting with improved sanitation but not improved water has been reported for Ethiopian children elsewhere (Dearden et al., 2017). The mixed findings between WASH practices and stunting observed in our study are mirrored in the mixed evidence reported in the literature, with observational studies reporting positive associations between WASH practices and linear growth in childhood (Garcia et al., 2013;Lin et al., 2013;Masibo & Makoka, 2012;Merchant et al., 2003), whereas results from recent RCTs suggest no effect of WASH interventions on LAZ-scores at 18-24 months of age Luby et al., 2018;Null et al., 2018). Our findings showed that the observed reduction in open defecation contributed to 9% and 7% of the decline in stunting seen among children aged 6-11 and 12-59 months, respectively. Because one proposed pathway between poor WASH practices and stunting is via increase in the incidence of diarrhoea (and EED), our findings of increased probability of diarrhoea and stunting with open defecation are expected.
We have utilized nationally representative data to identify the contributions made by WASH practices to reductions observed in diarrhoea and stunting. We constructed age-disaggregated regression models that have several benefits. First, such an approach has the benefit of being able to identify differential contributions made by variables in specific age periods and thus periods in which intervening on a particular variable may be more advantageous. For example, infants under 2 years represent an age group of particular interest, as most growth faltering takes places in the first 1000 days of life. Second, because age-disaggregated models respect the changing age dynamics of growth faltering observed across childhood (Alderman & Headey, 2018), they are less likely to underestimate the effects of any factor on stunting. In terms of limitations, due to the cross-sectional nature of the data, causal inferences cannot be made about the relationship between exposure and outcome variables. A second limitation is that we were unable to assess the contribution of the use of higher quality WASH services to the decline in diarrhoea and stunting.
The EDHS relies predominantly on self-reports or proxy reports, which are therefore subject to recall bias and which may affect older children to a greater extent than younger ones. Reliance on the recall may have also led to the underestimation of diarrhoea prevalence if, for example, respondents omit to mention episodes of diarrhoea that did not result in the utilization of medical care or medicines. Lastly, the lack of data on additional predictors of diarrhoea and stunting meant that our models only explained a small portion of the change in outcome variables.
Given the persistence of inequalities in WASH coverage, particularly regional disparities, we advocate a reconsideration of the CLTSH to improve suitability across all regions, whether this would be greater investment in the HEP in these regions or at a more basic level, ensuring that CLTSH materials are in an accessible format. We do acknowledge however that due to the nomadic/semipastoralist nature of some of the communities in these regions, actioning change in these settings may be a challenge. Given the greater time spent by females performing WASH-related activities, it is perhaps unsurprising that female sex and female empowerment have been associated with improved household WASH practices in Ethiopia (Bogale, 2021) and elsewhere (Dickin et al., 2021). The continued empowerment and involvement of females in WASH initiatives should therefore be promoted. Finally, continued promotion of the Sanitation Marketing Initiative and engagement of the private sector will increase financial investment in the WASH sector and improve access to affordable and improved basic sanitation and hygiene products. Given that costbenefit analyses show that for each $1 invested in WASH results in a return of between $2 and $9 via the benefits conferred by WASH interventions (WHO, 2012) on health, social and environmental wellbeing, investment is likely to result in substantial long-term economic benefit to the country.
In conclusion, the current analysis has highlighted the progress made in WASH practices between 2000 and 2016 in Ethiopia and their contribution to the reduction in diarrhoea and stunting in children under 5 years of age. Although progress has been made, improvements are still needed to increase the WASH standards and to address prowealthy and prourban inequalities in WASH coverage.
Our findings showed that improvements in water and sanitation only modestly explained reductions in diarrhoea and stunting. More research is needed to identify other unexplored drivers of diarrhoea and stunting in Ethiopia.

ACKNOWLEDGMENTS
The authors acknowledge Kalle Hirvonen (IFPRI) for his substantial comments on the analysis. Financial support for the National Informa-

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest. The opinions and statements in this article are those of the authors and may not reflect official policies or opinions of the organizations they belong to.