Households with unimproved water sources in Ethiopia: spatial variation and point-of-use treatment based on 2016 Demographic and Health Survey

The world population review indicates that Ethiopia has an estimated population of 114.96 million in 2020 that makes the second populous country in Africa [26]. The country has an administrative structure of nine regional states (Tigray, Afar, Amhara, Oromiya, Somali, Benishangul-Gumuz, Southern Nations Nationalities and People Region (SNNPR), Gambela, and Harari) and two city administrations (Addis Ababa and Dire Dawa) [27]. A finding on diarrhea and associated factors in the country revealed that the pooled prevalence of diarrhea among under-five children was 22%. The analysis of subgroup in this study showed that the highest prevalence was observed in the Afar region (27%), followed by the Somali and Dire Dawa regions (26%), then Addis Abeba (24%) [28]. Studies on the prevalence of childhood diarrhea with the type of water source they use show that high prevalence was observed on those households with unimproved drinking water sources [29, 30]. On the other hand, less than half (41%) of the total populations have at least basic drinking water services level, 28% have limited services (more than 30 min), 22% have unimproved services, and 9% use surface water as a source of drinking water supply in the country [31].

The EDHS data collected in 2016 was used. We obtained the data through online registration on MEASURE DHS program. A two-stage stratified sampling design based on nationally representative household surveys was implemented. Although EDHS has different data files, the household (HR) data file was used in this study.

In total, 16,650 households from 643 clusters are included and each cluster was represented by a GPS point with latitude and longitude coordinates. After initial data processing, 22 clusters without GPS points were excluded. For the final data analysis, 621 clusters and 5005 (weighted frequency of 5857) were identified to be using water from unimproved water sources based on the WHO/UNICEF category [32]. These households were used to analyze POU household water treatment practice. GPS coordinate displacement was performed on the actual locations of each cluster to produce data with displaced distances to maintain confidentiality of the surveyed respondents [33]. For urban clusters, a displacement toward a random direction by a maximum of 2 km was implemented. Rural clusters were displaced to a maximum of 5 km, with a further 1% displaced to a maximum of 10 km. A subset of clusters with GPS points were created for each wealth quintile. Since a single cluster was represented by a single GPS point and households in a single DHS cluster may fall in different quintiles, subsets of clusters were not mutually exclusive.

We have constructed a new wealth quintile by excluding drinking water as an asset of the households. The default asset scores and quintiles of DHS datasets were constructed by including drinking water supply as an input. However, drinking water is the dependant variable of this study and was excluded to construct wealth quintiles. An asset score for each household was constructed using Principal Components Analysis in Stata 14.0 [34]. All surveyed households were ranked and divided into five subsets or wealth quintiles. The first quintile included the poorest 20% of households and the fifth quintile included the wealthiest 20%. Following the approach used by Jia et al. [35] five subsets of clusters with GPS points representing five quintiles were created. Each subset included all the clusters that contained at least one household in the corresponding quintile. Since a single cluster was represented by a single GPS point and households in a single DHS cluster may fall in different quintiles, subsets of clusters were not mutually exclusive.

The raw coverage rate of unimproved water in each cluster was calculated as a proportion. The proportion was calculated as households with any of unimproved water sources to the total households in each cluster for the overall population and for each quintile. It has also accounted for the survey design and weight. The difference in raw coverage rates among the sampled clusters was statistically tested using one-way ANOVA.

A spatially smoothed rate was calculated to stabilize raw rates. To perform the smoothing, first, a Thiessen polygon which divides an area into regular sub-areas that encloses all locations closer to the central point than to any other point was created [36]. Spatial smoothing was used to produce a corresponding estimate to the raw coverage rate of each cluster from a collection of neighboring clusters enclosed by Thiessen polygon. For this study, the first order Queen Contiguity was applied as the spatial smoothing rule. Queen Contiguity spatial smoothing rule considers all neighboring polygons sharing a common edge or a common vertex with the target Thiessen polygon as neighbors. The difference between spatially smoothed and raw coverage rates for the overall population and each quintile was also calculated by subtracting the raw coverage from spatially smoothed coverage rates.

Spatial autocorrelation was performed by joining the raw and spatially smoothed coverage data to the geographic coordinates based on DHS cluster identification code. We have assumed there is a complete randomness of unimproved water distribution in the study sites. Global spatial autocorrelation was performed to analyze whether the pattern of unimproved water coverage is clustered, dispersed, or random across the study areas. The Global Moran’s I measure spatial autocorrelation based on the feature locations and attribute values. For a set of features with associated attribute, Global Moran’s I evaluate whether the pattern expressed is clustered, dispersed, or random. When the z score or p value indicates statistical significance, a positive Moran’s I index value indicates tendency toward clustering while a negative Moran’s I index value indicates tendency toward dispersion. As the global spatial autocorrelation technique provides one quantitative value for the whole dataset, it cannot identify local clusters with high or low coverage. Thus, local spatial autocorrelation analysis was applied to detect local clusters for positive global autocorrelation results. Local Moran’s I was used to calculate a test statistic for each location and to identify clusters of high and low coverage. A random permutation procedure (RPP) was used to replicate the statistics 999 times to generate reference distributions. The distribution of the test statistics was evaluated against a theoretical or random reference distribution generated. Local Moran’s I was calculated for both raw and spatially smoothed rates. Both the global and local spatial autocorrelation was calculated using GeoDa [37]. For POU water treatment, the number of households reportedly use adequate water treatment methods (chlorination, boiling, filtration, and SODIS) were considered a yes (1) and no otherwise (0 = if the household had used neither of them, i.e., households which had used either let it stand and settle, cloth straining, or never used any treatment option). Descriptive and logistic regression was used to assess the associated factors with the household POU treatment. A multivariable logistic regression was run to identify factors associated with POU water treatment practices by including variables with p value < 0.25 from bivariate analysis.

Five subsequent wealth quintiles or subsets of clusters have been created based on the national assets score, excluding drinking water supply. Accordingly, households were assigned to one of the five wealth quintiles. Clusters categorized by the number of households each cluster denotes for each quintile are presented in Fig. 1. The number of households varies across quintiles with 26.9-41.2% of clusters had only 1-3 households, meanwhile, 11.3-48.2% of the clusters possessed 17-29 households. The spatial distribution of clusters by quintile and for the overall population (Fig. 1) showed a similar distribution of clusters in the first three quintiles, while quintile four and five had a greater number of clusters around the capital Addis Ababa. The first three quintiles had a higher number of clusters at the North (Tigray and Amhara regions) and South (SNNPR). The trend continued to the fourth quintile, except the distribution of clusters extended to the center and around the capital, Addis, Ababa. The fifth quintile had a similar pattern around the capital and clusters sparsely distributed at the periphery.

The number of clusters with varying percentage of raw coverage by wealth quintile is presented in Fig. 2. The percentage of clusters with a high number of unimproved water coverage (40-100%) declined from the poorest quintile (56.3%) to the richest (12.6%). The difference in raw coverage rates among the sampled clusters was statistically tested using one-way ANOVA. The difference in raw coverage between quintiles was statistically significant, F (3, 1697) = 21.1, p < 0.001. High percentage of unimproved water coverage was observed in the first three quintiles compared to quintile four and five which had a small number of households with unimproved water source. Spatially, clusters with a higher percentage of unimproved water coverage were located at the North (Amhara and Afar regions), South (SNNPR), and East (Somalia) regions (Fig. 3). While in the capital Addis Ababa, Dire Dawa City administration, and Gambella region, the coverage of unimproved water is low.

As 26.9-41.9% of the clusters in different quintiles had only three or less households, a spatially smoothed coverage rate was calculated for each cluster to overcome the small number issue and to identify the cluster trend better. Figure 4 showed clusters categorized by smoothed coverage in each quintile, while the distribution of spatially smoothed coverage rate for all populations and by quintile is presented in Fig. 5. The percentages of clusters with 0-20% and > 80-100% coverage has decreased in all quintiles except quintile 5, which showed a rise in the percentage of clusters with 0-20% coverage. The pattern of clusters with > 60-100% coverage of unimproved water was clearly observed at the North, South, and East parts of the country and becomes infrequent as we moved from the first quintile to the fifth.

The difference between spatially smoothed and raw coverage rates for the overall population and each quintile was calculated by subtracting the raw coverage from spatially smoothed coverage rates. Clusters with a small number of households showed a considerable change by the spatial smoothing process.

The global spatial autocorrelation was calculated using Global Moran’s I. The analysis based on feature locations and attribute values revealed a clustering pattern of unimproved water coverage across the whole clusters (Global Moran’s I = 0.174, p value < 0.0001). Following a statistically significant positive result from the global spatial autocorrelation, Local Moran’s I was also calculated for the overall population to show hot and cold spots. Unlike the global spatial autocorrelation, local spatial autocorrelation was calculated for both the raw and spatially smoothed data. The local clustering trend of high and low spots among sampled clusters was identified in both raw and spatially smoothed coverage (Fig. 6a and b). However, the clustering of spatial outliers was eliminated in the spatially smoothed coverage. The raw coverage revealed 72 high-high and 120 low-low statistically significant clusters (p < 0.05), followed by 8 high-low and 27 low-high clusters. Meanwhile, the spatially smoothed coverage showed 155 high-high and 149 low-low statistically significant clusters (p < 0.05). Clusters in the North (Amhara region and Afar), in the East (Somalia region), and in the south (SNNPR region) had statistically significant unimproved hot spots (95% confidence). Clusters in the center, South, and West had statistically significant cold spots as presented in Fig. 6a and b.

Of the total households (5005), 613 (10.46%) treat their water prior to drinking with any of the methods (use either boiling, filtration, chlorination, SODIS, let it stand and settle, and cloth straining) and 365 (6.24%) of households treat using adequate methods (use either boiling, filtration, chlorination, or SODIS). The number of households and reportedly used treatment methods are boiling 125 (2.14%), chlorination 164 (2.80%), cloth straining 204 (3.48%), filtration 105 (1.80%), let it stand and settle 41 (0.70), and SODIS 5 (0.09%). The data shows that of the adequate methods, treating with SODIS the least to be used by the households in the country (Table 1).

The logistic regression shows that the odds of treating water among household heads with the highest education level was 2.50 more (95% CI = 1.43, 4.36) compared to those who did not attend formal education. Household with the highest wealth quintile had more odds of treating water compared to the poorest (Table 1).