Predictors of Underweight among Children Under-five in Ethiopia


 9013 under-five children were included for study. Descriptive Statistics and POM were employed identify socio-economic, demographic, and proximate factors of underweight to children among regions of Ethiopia. SAS version 9.4 was used. About 25.3% of children in Ethiopia were underweighted. POM showed that sex, residence, size at birth, age group, incidence of diarrhea, educational level, breastfeeding status, mothers’ employment status, and birth order were found to be significant factors of underweight. Improvement of education for caring child in appropriate age needed. Personal hygiene be improved to prevent exposures to diarrhea and environmental sanitation.


Introduction
Nutritional status of children under five years old is one of the indicators of household wellbeing and one of the determinants of child survival. Health and nutrition are important and are critical components (Smith, L.C 2000, Haddad, L, 2000. Under-nutrition was responsible for 11% of the worldwide disease burden or disability-adjusted life year (daily) loss, and it involves limitation in womb growth, stunting, wasting, underweight condition, and micro-nutrient deficiencies (Black, RE, et al., 2008). Under-nutrition is the underlying cause for about 54% of all deaths among children which is approximately 11 million worldwide. Under-nutrition also comprises 57% of mortality in Ethiopia per year (Egata, G et al., 2014). Global chronic under nutrition in children is highly prevalent and remains a big challenge. There are 11 million people undernourished in developed countries (FAO, 2014). There is a 29.9% of the total populations of Ethiopia 30.4% in rural and 25.7% of the total populations of Ethiopia in urban areas that are found to be under the poverty line (MOFED, 2013). In Ethiopia, under-nutrition is the underlying cause for about 57% of all deaths among children under five years old (Yebyo, HG, et al., 2013). Still, chronic under-nutrition causes devastating and irreversible damage among children who survived, especially if it occurs under two years old (Alemu, F, 2013).Inability to reduce the occurrence of under-nutrition among children under five years old will lead to long term impacts like having the potential of complicated delivery due to short stature among women in later life, retarded cognitive ability, and economic and productive loss of the country as whole (Berihun, M and Azizur, R, 2013). Ethiopia holds the highest level of malnutrition in children aged fifty-nine months to six years old in the world. Malnutrition has spent a significant amount of time in the history. The most pressing health problems as well as 57% of child deaths are associated with malnutrition in Ethiopia (UNICEF, 2010).
According 2016 EDHS report and based on the comparison among Ethiopian children aged under-five years with the latest World Health Organization (WHO), multi-country growth references show that 38% of children under-five years old are stunted; 18% are severely stunted; 10% are wasted; 3% are severely wasted; 24% are underweight; and 7% are severely underweight. Children in rural areas are more likely than those in urban areas. The highest percentage of underweight children is present in Afar (36%) and Benshagul (34%) (CSA, 2016).
This shows that malnutrition (underweight condition) among children is still a critical problem in Ethiopia. A study relevant to malnutrition was conducted at single district level to investigate the  2015)). Generally, the study attempts to fill the gap of considering the whole Ethiopia, and further taking account of ordering based on status of malnutrition. The core aim of this study was to explore prevalence and other proximate risk factors among children whose age is lesser than five years old in Ethiopia. Specifically, the paper aimed: To examine the prevalence of malnutrition among children who age is lower than five years old in Ethiopia; and To identify the socio-economic, demographic, and other proximate factors that affect the status of malnutrition among children who age is lesser than five years old in Ethiopia.

Significance of the Study
The study shall be beneficial for the following reasons: • It is helpful for policy making, monitoring, and evaluation activities of the government and different concerned agencies; • It is helpful for assessing the contribution and growth promotion programs for mortality reduction in the community level by improving their knowledge and practices for parents based on nutrition status of children; • it contributes a significant finding to the literature that concerns the nutritional status of children; and • It is beneficial to the literature of malnutrition that will help researchers within or within the neighboring field of study.

Data and Materials
Source of Data 2016 EDHS data is utilized in this study. The study populations were children under five years old and who were residents of Ethiopia. 9013 children whose age is less than five years old completed anthropometric measurements considered for the study.

Study Design
Cross sectional design was applied. Variables are collected for several sample units at the same points in time (one-time shoot. The study only utilized the data collected from the respondents directly in a particular time (2016).

Sampling Technique
The 2016 EDHS data was used based on multistage stratified cluster sampling. The structure of data in the population is hierarchical, and a sample from the population can be viewed as a multistage sample.

Study Variables
For this study, the potential determinant factors are expected to have a significant effect on the malnutrition among children under five years old.

Outcome Variable
The interest variable for this study was the status of underweight according to Z-score (WHO, 2006
Where, αj represents the threshold (intercept), Xi represents the explanatory variables, and β represents the regression coefficient. The present study estimates the underweight status ordered by a categorical response of variable Yik having three ordered categories (not underweight, moderate underweight, and severe underweight). Ordered model estimates the cumulative probability γk or cumulative log odds log ( γk 1−γk ) up to K th category where, K=1, 2, 3. One category (last) of this taken as reference category and cumulative probability for reference category is always equal to one. Cumulative logit probability model takes the form log � γ1

Proportional Odds Assumption
Score test for the proportional odds assumption for SAS procedure produced 2 =45.924 with 32 df and their insignificant p-value=0.0528. The result revealed that assumption has appropriate for study to the evidence the data. However, we cannot reject the null hypothesis at 5% significant level. The model satisfies proportional odds assumption. Hence, it is not necessary to go for another model

Result of Proportional Odds Model
In this study, the idea behind fitting Proportional Odds Model was applied. POM was applied for this analysis of data. The coefficients of the explanatory variables in the model is interpreted as ratio of the odds of status of the underweight condition. -2.0242 showed that for comparison of moderately underweight to severely underweight and not underweight, -3.4177 is also used to compare the moderately underweight category, severely underweight category, and the not underweight category.  4.198 ] times more likely compared to children who had < 6 months respectively for each, assuming others variables were constant. The probability of underweight status among children whose age ranges from 48-59 months, children whose age ranges from 6-11 months, children whose age ranges from 12-23 months, and children whose age ranges from 24-47 months were significantly at 5% level of significance.

Discussions
The main core objective for this study was to identify the determining factors of underweight better model to analyze considered study data for underweight status as test of proportionality is not significant at 5% with p-value=0.0528, indicating that proportional odds assumption is valid/satisfied. Generally speaking, we can say that POM is well done fitted due to parallel line assumption [proportionality assumption]. Based on the findings of previous results, this study made a few comparative discussions as follow. POM was fitted in terms of examining the socioeconomic, demographic, and health characteristics for underweight status among children whose age is less than five years old. This was tested through PROC LOGISTIC statement in SAS.
Result of proportional odds model revealed that sex had statistically significant effect on status of underweight children. This finding shows that female were 0.269 times to be underweighted were 0.199 times less likely to be underweight than children whose father had not undergone a school. Also, children whose father had higher levels of education [β=-0.2604, 95% CI: 0.607, 0.979] were 0.771times less likely to be underweight compared to children whose father had not undergone a school. This finding is similar with the one conducted by Chowdhury et al. (2018).

Conclusions:
The study revealed that about 25.3% of children under five years old in Ethiopia were underweighted. The results proportional odds model revealed that factors like sex, place of residence, mothers' educational level, husbands' educational level, diarrhea status in the last present in two weeks, employment status of mothers, age-group of child in months, birth status, birth order, status of breastfeeding, size at birth, and consumption of fortified baby food had significant effects on underweight among children under five years old at 5% significance level.
However, the place of delivery, marital status, wealth index, and number of household members did not have a significant effect on the underweight status of children under five years old.

Recommendations
Based on this study finding, we recommend the following: Initiate programs related to improvement of mothers' and husbands' education for caring children under five years of age, monitor personal hygiene to prevent exposures to diarrhea, other diseases, and problems with environmental sanitation, strengthen interventions aimed at improving the practice of parents on ways of feeding their children and initiate safety net programs to focus on the poorest segment of societies; and Provide nutrition education on dietary diversity and family planning, and special attention should be given to adult literacy program in order to promote adequate feeding practice and to curb chronic nutritional problems.