Socio-demographic, environmental and behavioural factors of the prevalence of diarrhea among children age under-five years in Rural Ethiopia: A Cross-Sectional Study

Background : Diarrhea is one of the major contributors to deaths for under age five children in Ethiopia. Studies conducted in different countries showed that rural children are highly affected by diarrhea than urban children. Thus, the purpose of this study was to identify the socio-demographic, environmental and behavioural risk factors of the prevalence of diarrhea among children age under five years in Rural Ethiopia. Methods: Data for the study was drawn from the 2016 Ethiopia Demographic and Health Survey. A total of 8,041 under-five children were included in the study. Bivariate and multivariable binary logistic regression were employed for the analysis of the data to assess the relationships between the prevalence of diarrhea and socio-demographic, environmental and behavioural risk factors among under five children. Data was analyzed using SPSS version 23. Results : The multivariable logistic regression analysis revealed that younger mothers’ children and child’s age 6-11 months (AOR: 3.5; 95% CI: 2.58-4.87), 12-23 months (AOR: 3.1; 95% CI: 2.33-4.04) and 24-35 months (AOR: 1.7; 95% CI: 1.26-2.34) were significantly associated with diarrhea. The two week prevalence of diarrhea was also significantly associated with male children (AOR: 1.3; 95% CI: 1.05-1.58), Children in rural Afar and Gambela regions, households who shared toilet facilities with other households (AOR: 1.4; 95% CI: 1.09-1.77), higher birth order number and households with 3 or more under-five children interacted with older mothers (AOR: 4.7; 95% CI: 1.64-13.45). Conclusion: The prevalence of diarrhea among under-five children was strongly associated with younger mothers, child’s age between 6-35 months, male children, children in rural part of Afar and Gambela regions, children with higher birth order numbers, those who shared toilet facility with other households and households with 3 or more number of under-five children interacted with older mothers/primary caregivers. Thus, in order to reduce childhood diarrhea, I suggest that child’s


Background
Diarrhoeal disease is the second leading cause of death in children under five years old, and is responsible for killing around 525,000 children every year. Globally, there are nearly 1.7 billion cases of childhood diarrhoeal disease every year [1]. Each year, an estimated 2.5 billion cases of diarrhea occur among children under five years of age, and estimates suggest that overall incidence has remained relatively stable over the past two decades. More than half of these cases are in Africa and South Asia. More than 80 percent of child death due to diarrhea occur in Africa and South Asia [2].
Low and lower-middle income countries are home to 62 percent of the world's under 5 population, but account for more than 90 percent of global pneumonia and diarrhea deaths [3]. In low-income countries, children under three years old experience on average three episodes of diarrhoea every year [1]. The 2005, 2011 and 2016 Ethiopia demographic and health survey reports showed that the percentage of children under age five who had diarrhea in the 2 weeks before the survey period were 18%, 13% and 12%, respectively. Even though the magnitude of diarrhea have reduced over the past periods, diarrheal disease is still the major cause of morbidity and mortality among children in Ethiopia. Based on UNICEF report, in Ethiopia 15,500 diarrhoea deaths was occurred among children in 2015 [3].
In Ethiopia, studies conducted in different parts of the country documented many factors associated with under five diarrheal disease. Socio-demographic, economic, water, sanitation, and hygiene factors were predictors of the diarrheal illness. A study conducted in Benshangul Gumuz region, Ethiopia, revealed that low maternal education, improper child stool disposal methods, having more than two under-five children, high birth order and the age of children were the risk factors of childhood diarrhea [4].
Evidence based information is needed for child's health improvement strategy by preventing and reducing the severity of diarrheal in under-five children. Studies conducted in different countries showed that rural children are highly affected by diarrhea than urban children [11,26,27]. In rural Ethiopia, despite the high prevalence of diarrhea disease, there are limited studies available at District/Zone, but no at country level, to show the significant associated factors of diarrhea occurrence among under-five children in the rural part of the country. Thus, this study was planned to identify the socio-demographic, environmental and behavioral risk factors of the prevalence of diarrhea among children age under-five years in rural Ethiopia.

Study Design and Setting
Data for the study were drawn from the 2016 Ethiopia Demographic and Health Survey which was population based cross sectional study conducted from January 18, 2016, to June 27, 2016, based on a nationally representative sample that provides estimates at the national and regional levels and for urban and rural areas. The survey was the fourth survey in the country.

Sampling Design
The sampling frame used for the 2016 EDHS was the frame of the Population and Housing Census conducted in Ethiopia in 2007 and provided by the Central Statistical Agency (CSA). The 2016 EDHS sample is stratified and was selected in two stages. Each region was stratified into urban and rural areas, which yielded 21 sampling strata. Samples of EAs were selected independently in each stratum in two stages. In 2016 EDHS a representative sample of approximately a total of 10,641 under five children representing the number of live births born to the interviewed mothers in the period of five years preceding the date of the survey were included. Out of these children, 9,916 interviewed mothers/caregivers gave complete response about the two week prevalence of diarrhea among their under-five children in Ethiopia. Then, after a certain rearrangement and reorganization of the data 8,041 children with complete information were used as the data for this study.

Study Variables
The response variable was the reported occurrence of diarrhea. The respondent/primary caregiver was asked if the child had diarrhea in the last two weeks including the last 24 hours prior to data collection time.
The independent variables included in the study were identified from literatures conducted earlier and associated with occurrence of diarrheal occurrences among children under five years of age. These are the Socio-demographic variables(respondents current age, sex of child, current age of child, child lives with whom, region, respondents educational level, respondents religion, birth order, respondents work status, household wealth index, number of children under five in the household, current marital status, current breastfeeding status, number of household members), Environmental and behavioural variables (source of drinking water, type of toilet facility, if toilet facility is shared with other households, disposal of youngest child's stools when not using toilet and main floor material).

Data Analysis Methods
After the extracted data were checked for completeness and coded, the analyses were done using both SPSS Version 23. Both descriptive and inferential data analysis methods were used in the study.
Data was described and summarized through frequency distribution tables, Cross-tabulation (Contingency tables) and charts.
To study the effect of the different predictor variables on the outcome variable, bivariate and multivariable analysis were used. In bivariate analysis, Chi-square test of association and Crude odds ratio were estimated to assess the association between each of the predictor variables and the outcome variable. Significant variables at P-value < 0.25 in the bivariate analysis were only used in the multivariable analysis.
In multivariable analysis, multilevel binary logistic regression was proposed to handle the effect of the variation of average value of the response variable among the study regions. However, during the analysis, the variation of the intercepts among the regions were not statistically significant in the random effect result meaning the regions did not differ with respect to the average value of the prevalence of diarrhea. As a result, multivariable binary logistic regression was used instead to study the overall effect of the socio-demographic, environmental and behavioural risk factors on the response variable (Prevalence of diarrhea).
Backward stepwise method was used to select variables for the best reduced model and Wald-test was used to test individual significance of the coefficients of the model. The overall goodness of the final model was checked using Hosmer-Lemeshow goodness-of-fit test. Interpretations of the strength of the associations between the risk factors and the response variable were based on significant Adjusted Odds Ratios (AOR) with 95% confidence intervals at 5% level of significance (P-value <0.05).

Descriptive Results of Socio-demographic, Environmental and behavioural characteristics
The study included 8,041 under-five children and their mothers or caregivers. Out of these 49% were females, 11.1% of children were less than 6 months old and 40.6% of them were age greater than 35 months. The majority of the children (67.9%) were currently breastfeeding at the time of the survey, very few children (1.7%) were lives with someone else. Regarding mothers/caregivers, 5,801(72.1%) had no formal education, 95.2% were married, 50.7% mothers were between 25-34 years old, 24.7% were 35 years old and above. Around 60% of the households had 6 and above family members [ Table   1].
Out of the total 8,041 households, 46.7% have used unimproved source of drinking water, 61.2% of mothers/caregivers were not properly removed youngest child's stools, there were 41% unimproved sanitation (toilet facility) and 51.6% no toilet facility in the household [ Table 2].

Magnitude of under-five Diarrhea
The overall prevalence of diarrhea among under-five children two weeks before the survey was 11.2% in rural Ethiopia. Result displayed in Tables 1 and 2 showed that the prevalence of diarrhea was the highest among children lives with respondents (11.1%), who used unimproved sanitation (10.4%), and lived in natural floor material (10.3%). The highest magnitude of diarrhea was also observed on child's still breastfeeding (8.1%), non-educated mothers (7.7%), current age of mother's/caregivers 25-34 years (5.8%) and child's age 12-23 months (3.4%).

Results of Bivariate and Multivariable Logistic Regression Analysis
In Bivariate analysis, the chi-square test result in Tables 1 and 2 and the estimated crude odds ratios in Table 3 showed that there were a significant association between prevalence of diarrhea and variables respondent's current age, sex of child, current age of child, education of mother/caregiver, region, number of under-five children, wealth index of the household, religion, Number of household members, current breastfeeding status, Type of toilet facility and toilet facilities shared with other households at 5% level of significance.
In multivariable binary logistic regression, the overall effect of the socio-demographic, environmental and behavioural risk factors on the prevalence of under-five diarrhea was assessed. The result showed that mother's current age, current age of child, sex of child, region, birth order, toilet facilities shared with other households and number of under-five children interacted with mother's current age were statistically significant risk factors of diarrheal occurrence among under five children at 5% level of significance [ Table 3]. The Hosmer-Lemeshow goodness-of-fit test result (Chi-square statistic value and their 95% confidence interval are included in Table 3. The reference categories are those indicated in brackets *Statistically significant variables at p < 0.05; **statistically significant variables at p < 0.01 COR= crude odds ratio, AOR= adjusted odds ratio, CI= confidence interval

Discussion
Chi-square test of association and bivariate logistic regression analyses were conducted to assess the relation between prevalence of under-five diarrhea and the selected variables. The bivariate analysis result showed that there were a significant association between prevalence of diarrhea with respondent's current age, sex of child, current age of child, education of mother/caregiver, region, number of under-five children, wealth index of the household, religion, number of household members, current breastfeeding status, type of toilet facility and toilet facilities shared with other households.
As we have seen from Table 3 above, in the multivariable binary logistic regression analysis, the variables mother's current age, current age of child, sex of child, region, birth order, toilet facilities shared with other households and number of under-five children interacted with mother's current age were statistically significant at 5% level of significance and then for this study they become an important risk factors for under-five diarrhea disease occurrence.The constant was also significant.
However, variables education of mother/caregiver, wealth index of the household, religion, number of household members, current breastfeeding status and type of toilet facility, which were significant in bivariate analysis, became insignificant in multivariable analysis.
The constant coefficient indicated that the odds of diarrhea occurrence in rural Ethiopia decreased by 94% (AOR: 0.059) if the effect of those factors included in the final model were zero. Current age of mothers/primary caregivers had a significant association with diarrhea. The odds of diarrhea were higher among younger mothers' children than older mothers' children. The odds of diarrhea for children whose mothers' age between 25-34 years (AOR: 0.659; 95% CI: 0.469-0.926) and 35-49 years (AOR: 0.364; 95% CI: 0.230-0.577) were 34% and 53% lower than children whose mothers' age between 15-24 years keeping the effect of other variables as constant. This result was consistent with the findings in Northern Nigeria [10].
The odds ratios of under-five diarrhea for current child's age 6-11, 12-23 and 24-35 months as compared to child's age greater than 35 months were 3.5(95% CI: 2.580-4.872), 3.1(95% CI: 2.334-4.041) and 1.7(95% CI: 1.262-2.336) respectively. These indicated that child's age group 6-11, 12-23 and 24-35 months had 3.3, 3.1 and 1.7 times more likely to be affected by diarrhea than child's age greater than 35 months respectively when adjusting the effect of other variables. In general, children age greater than 35 months had lower risk of having diarrhea than children whose age between  months. This may be due to the fact that children whose age between 6 and 23 months begin supplementary foods and also they start crawling and can touch contaminated materials in unclear environment and immediately return their hand to mouse, so it may cause them to easily vulnerable to diarrhea. This finding is in line with studies done in Benishangul region, Farta Woreda, Eastern Ethipia, Enderta woreda, and Wolitta Soddo [4,5,11,12,13].
Sex of child had significant association with diarrhea prevalence. Male children had 1.3 times more likely affected by diarrhea than female children (AOR: 1.291; 95% CI: 1.054-1.581).
The study also revealed that prevalence of diarrhea was significantly associated with region of the respondent. Children in rural Afar and Gambela regions had around 2 times more affected by diarrhea as compared to children in rural Tigray region. However, rural Somali region's children were 45% less likely affected by diarrhea than rural Tigray region's children.
Households who shared toilet facility with other households had a significant association with diarrheal disease. Children from households who shared toilet facility with other households had around 39% more risk for having diarrhea than children from households who did not shared toilet facilities (AOR: 1.387; 95% CI: 1.088-1.768).
The odds of the occurrence of diarrhea was 4.7 times more likely to be higher among household with 3 or more under-five children interacted with older mothers as compared with 2 or less under-five children of younger mothers (AOR: 4.702; 95% CI: 1.64-13.45). The 95% confidence interval also suggest that the rate (relative risk of being affected by diarrheal disease) for household with 3 or more under-five children interacted with older mothers could be as low as 1.64 and as high as 13.45. This finding is consistent with a study done in Benishangul Gumuz region [4] and in eastern Ethiopia

Conclusion
This study was intended to identify demographic, environmental and behavioural risk factors of the prevalence of diarrhea among under-five years children in rural Ethiopia based on 2016 EDHS data.
Fourteen socio-demographic factors and five environmental and behavioural factors were included in the study. The results of the study identified that current age of mother/primary caregiver, current age of child, sex of child, region, birth order number, toilet facilities shared with other households and number of under-five children interacted with mother's current age as an important risk factors for diarrhea occurrence among under-five children in rural Ethiopia.
High prevalence of diarrhea was associated with younger mothers, child's age between 6-35 months, male children, children in rural part of Afar and Gambela regions, children with higher birth order numbers, those who shared toilet facility with other households and more than or equal to 3 number of under-five children of older mothers/primary caregivers. Thus, in order to protect children from diarrhea disease, I recommend that child's health care programs, policies, plans and any other efforts in the study areas should give special attention to those categories of children who had highly affected by diarrhea. Overall prevalence of diarrhea among rural children in Ethiopia. A diagrammatic presentation using Pie-chart. The area colored red shows the proportion of diarrhea, which covered 11.24% of the total area of the chart. for top significant fixed effects, obtained in the multilevel logistic regression analysis. The line with its boundary shows the occurrence of diarrhea, on average, highly increases among children age from 6 months to 11 months, and gradually decreases to age 35 months and above.