Prevalence of Stunting and Its Associated Factors among Children of 6–59 Months in Arba Minch Health and Demographic Surveillance Site (HDSS), Southern Ethiopia: A Community-Based Cross-Sectional Study

Methods The community-based cross sectional study was conducted in the Arba Minch Health and Demographic Surveillance Site, Southern Ethiopia. The simple random sampling method was used to recruit 656 mother-child pairs. Height for age Z score was computed using WHO Anthro version 3.2.2 software. Multivariable logistic regression model was fitted, and adjusted odds ratio (AOR) at p value <0.05 was used to determine statistically significant association between predictors and outcome variable. Result The prevalence of stunting among children of 6–59 months in the study area was 47.9% (95% CI; 44.0–51.7). The likelihood of stunting was significantly higher among children who live in households with medium (AOR 2.20, 95% CI: 1.43–3.37) and poor (AOR 2.87, 95% CI: 1.72–4.81) wealth status. In addition, children who were not exclusively breast fed (AOR 1.55, 95% CI: 1.07–2.24), whose mothers had not participated in decision of major household purchases (AOR 2.27, 95% CI: 1.21–4.26), and whose mothers lacked decision on freedom of mobility (AOR 1.96, 95% CI: 1.05–3.66) were significantly stunted compared with counterparts. Conclusion Stunting is a severe public health problem in the area. Therefore, efforts should be taken to enhance maternal empowerment, household wealth, and infant and young child feeding practice for reducing stunting among children.


Introduction
Stunting is one of the most common markers of chronic undernutrition, which is a linear growth failure or inability to attain potential height for a particular age [1,2]. It is considered as the overall best indicator of child well-being [3]. Despite the significant achievement, the world has made towards improving nutrition and associated health burdens over recent decades, and malnutrition remains a public health threat, including stunting. Accordingly, the United Nations Nutrition report disclosed that 50.5 million children under 5 years of age are wasted and 150.8 million are stunted in 2018 [4]. Evidence shows Africa and Asia were disproportionately affected by child stunting where they accounted for more than nine out of ten of all stunted children globally [5].
In Ethiopia, stunting remains a major public health threat.
e study conducted in East Belessa district of Northwest Ethiopia reported that more than half (57.7%) of under-five children were stunted [6]. Reports showed that nearly half of the children were also stunted in Northwest, Southern, and Eastern parts of the country [7][8][9].
Stunting has a devastating and far reaching impact on individuals and nations; it ranges from diminished cognitive, low school performance, and physical development to increased risk of degenerative diseases such as diabetes and affects the countries' economy by reducing the working capacity and productivity [1,10,11]. e causes of stunting are intertwined and complexes such as socioeconomic status, inadequate infant and young child feeding practices, poor sanitation and hygiene conditions, paternal education, family size, and poor maternal health and nutrition [1,10].
ere is limited evidence regarding stunting prevalence and associated factors among under-five children in the study area. Moreover, in identifying determinants of stunting, there is also paucity of evidence regarding the association of latent variables such as maternal empowerment domains with stunting. erefore, the aim of this study was to assess the magnitude of stunting and associated factors among children of 6-59 months in Arba Minch Health and Demographic Surveillance Site (HDSS), Southern Ethiopia.

Study Setting and Design.
e community-based cross sectional study was conducted from March to April 2019 in Arba Minch HDSS. Arba Minch HDSS located in Arba Minch Zuria District by including 9 kebeles (the smallest administration unit of Ethiopia). e Kebeles were selected from Arba Minch Zuria district by stratifying them based on the climatic zone: lowland, midland, and highland.

Sample Size and Sampling
Technique. For estimating sample size, 45.6% prevalence of stunting was used, which was obtained from the study conducted in East Badawacho District, Southern Ethiopia [12]. In addition, 95% confidence interval, 4% margin of error, and 10% nonresponse were used. Accordingly, the calculated sample size was 656. All children aged 6 to 59 months in the site of Arba Minch HDSS were source population.
For the selection of the study participants, the sampling frame prepared from Arba Minch HDSS dataset, which contains child date of birth (age), kebele, household identification, household head name, marital status, child name, was used. e sample was allocated to each kebele proportionally based on the number of children from 6-59 months in the kebeles. en, study participants were randomly selected using STATA version14 from each kebele ( Figure 1).

Data Collection Tools and
Procedure. Data were collected through home to home visits using a structured and pretested interviewer-administered questionnaire. e questionnaire consisted of questions that could measure socioeconomic and demographic factors, maternal empowerment status, child and mother characteristics, and environmental health condition of the household and household food insecurity. Maternal empowerment were assessed by using survey-based Women's emPowERment index, which was validated using Demographic and Health Survey data from different African countries for the estimation of the inequalities and its effects on child health [13]. Wealth index was formed by asking assets ownership based on different variables adopted from Ethiopian Demographic and Health Survey 2016 variables [14]. Household food security status was assessed by using Household Food Insecurity Accesses Scale (HFIAS), which was developed and validated by Food and Nutrition Technical Assistance (FANTA) and again validated for Ethiopian households [15,16]. Length was measured for children aged 6-23 months in a recumbent position, and height was measured for children aged 24 to 59 months in a standing-up position to the nearest 0.1 cm. A board with an upright wooden base and movable headpieces was used for measuring the height.

Data Quality Management
A questionnaire was translated into Amharic for common understanding. To check the consistency of the meaning, the Amharic version was translated back to English. Pretest was conducted on 5% of the households, which was not selected in actual sample in one kebele from HDSS. Supervision was made throughout data collection time by supervisors and principal investigator. Relative technical error of measurement (%TEM) was calculated during training among 10 under-five children to minimize random anthropometric measurement error. Both intra-and inter-observer variabilities were within the acceptable range.

Data Processing and Analysis.
Data were collected using the Open Data Kit (ODK) software via a tablet (smartphone). It was transferred to Statistical Package for Social Sciences (SPSS Inc., Chicago, Illinois, USA, version 23) for analysis. Height for age Z score was computed using WHO Anthro version 3.2.2 software [17]. e wealth status of the households was constructed by the principal component analysis based on selected household asset and grouped in to three categorizes. e dietary diversity score were assessed using seven food groups and categorized by using the acceptable four food groups, and it is classified as adequate or not adequate [18]. e empowerment status of mothers was assessed by using six domains [13,19], and the domains had shown significant internal consistency (with the minimum Cronbach's alpha � 0.74 for community group membership and maximum Cronbach's alpha � 0.84 for decision-making on major household purchase).
Binary logistic regression analysis was used to see the independent effect of predictors on stunting. ose variables with a p value of ≤0.25 during bivariable analysis were retained for multivariable analysis. e stepwise backward elimination model building procedure with likelihood ratio model comparison technique was used while building the model. Interactions and confounders were tested using beta change at cutoff point 20%. Multicollinearity was checked among predictor variables, and the variance inflation factor (VIF) was found to be less than 3. Prediction performance of predictors for stunting was checked using the ROC curve. Final model fitness status was checked by Hosmer and Lemeshow goodness of fit chi-squared test (p value = 0.77).

Demographic and Socioeconomic Characteristics.
Six-hundred fifty four child mother pairs were participated in the study with the response rate of 99.7%. e mean ± SD age of child was 33.47 ± 13.74 months. e majority (65.7%) of mothers had no formal education and protestant (70.5%) in religion. From all participated household, 34.3% were food insecure and almost equal proportion of male and female children were assessed (Table 1).

Child
Morbidity, Health Care, Feeding, and Environmental Characteristics. Ninety-four (14.4%) children had illness two weeks preceding this survey. e most common reported illness was respiratory tract infection (41.5%), followed by skin rash (27.7%) and diarrhea (26.6%). Among the total mothers participated in the study, nearly 89% had ANC visit, and more than half (56.7%) delivered their index child at home (Table 2).

Maternal Empowerment.
Almost all mothers (98%) participated in decision-making about her and her child health care either independently or jointly with their husband. Regarding the attitude towards justification of violence, overwhelming majority of mothers were (91%) justified violence (Table 3).

Prevalence of Stunting.
e mean (SD) height for age Z score of 6-59 months children was −2.01 (1.86). e prevalence of stunting among children of 6-59 months in the study area was 47.9% (95% CI; 44.0-51.7). From this, the majorities (75%) were in the age group of 24-59 months, and more than half (177) were severely stunted among those stunted children.

Factors Associated with Stunting among 6-59 Months
Children. In bivariable logistic regression analysis, paternal education, household wealth, ANC visit, place of delivery, postnatal care, breast feeding initiation time, exclusive breast feeding, mother decision on the major household purchase, the decision on freedom of mobility, source of drinking water, and waste disposal system were significantly associated with stunting. Based on multiple logistic regression analysis, the likelihood of stunting was more than 2 times higher among children who live in the households with medium (AOR � 2.20, 95% CI: 1.44-3.38) and poor (AOR � 2.87, 95% CI: 1.72-4.81) wealth status compared with children from rich households (Table 4). Similarly, the odds of stunting were higher among children who were not exclusively breast fed for the first 6 months (AOR � 1.55, 95% CI: 1.07-2.24) than those of exclusively breast-fed children. ose mothers who did not participate in decision of major household purchase had double risk to have stunted child (AOR � 2.27, 95% CI: 1. 21-4.27), and the odds of stunting were higher among children whose mothers lack decision on freedom of mobility (AOR � 1.96, 95% CI: 1.05-3.67) ( Table 4).

Discussion
is study had tried to assess the prevalence and associated factors of stunting among children of 6-59 months.
is variation could be due to Agroecological and sociodemographic differences.
e possible explanation could be due to sociocultural difference and age category of target population. Moreover, among nine kebeles in Arba Minch HDSS, eight of them are rural kebeles and the remaining is semiurban, so that this could affect the prevalence in the area. Due to inadequate maternal and child care, stunting is more common in rural areas when compared with urban settlement [32].
In the current study, household wealth status had a significant association with stunting among 6-59 months children. It indicated that children from poorest and medium wealth status households were two times more likely to be stunted compared with children from the richest households.
is is supported by the study conducted in Bangladesh, Yemen, and other studies conducted in Ethiopia [27,28,33]. Obviously, poverty and economic inequality hinder the financial ability for household food access, inadequate sanitation, and health care utilization, which are the major contributing factors for the development of malnutrition including stunting [34]. In contrary, household wealth had no significant association with stunting in study conducted in East Gojjam, Ethiopia [35]. Appropriate infant and young child feeding has been identified as one of key determinants of child under nutrition particularly stunting [10]. Exclusive breastfeeding has been identified as an indispensable way of providing the ideal food for the healthy growth and development of infants and children [10,36]. As shown by this study, the odds of stunting were more common among children who were not exclusively breast-fed in the first six months. ese results corroborate the findings of Fikadu et al. (2014), which found exclusive breast-feeding as a predictor of stunting in Meskan district, Southern Ethiopia [36]. is could be explained by the initiation of any type of complimentary food before six months could cause illness like diarrhea and lower respiratory tract infections, due to immature digestive and immune system of children [37].
From maternal empowerment domains, maternal decision-making on major household purchase and freedom of mobility were significantly associated with childhood stunting. Accordingly, those mothers who did not participate in financial decision of major household purchase were two times more likely to have stunted children [38]. is finding is supported by study conducted in Butajira, Ethiopia, and Benin [19,39]. is result could suggest mothers who participate in financial decision-making could give greater proportion of resources to child-favored expenditures. It could also give mothers freedom for household expenditure towards food, which could favor dietary diversity. However, this finding is inconsistent with study conducted in Tanzania [40].
In this study, mothers who lack decision on freedom of mobility had higher odds to have stunted child. is could be best explained by mothers who lack freedom of mobility could not access health care early for sick child [41]. Moreover, they may not participate in marketing and agricultural activities, which favors food security and diversity [39]. is is in agreement with studies conducted in Butajira, Benin, and analysis of sub-Saharan demographic and health survey data [13,19,39]. However, freedom of mobility had no significant association with stunting in the study conducted at Jimma, Ethiopia, and Lao PDR [42,43]. e possible reason could be due to cross cultural nature of empowerment domains from area to area [13,[44][45][46].
Community-based nature of the study gives representative sample among study subjects in Arba Minch HDSS, where it is useful for intervention strategies and actions. is study is not free from limitations, and recall and social desirability biases could be one of the limitations of the study. Furthermore, through intensive training, regular field supervision and measurement standardization the study were not free from the measurement error.  Journal of Environmental and Public Health 5

Conclusion
e current study showed stunting is a severe public health problem in Arba Minch HDSS, which is also based on the WHO cutoff point. Lack of maternal decision-making on major household purchase and freedom of mobility, lower household wealth status, and exclusive breast-feeding were independent predictors of stunting. erefore, enhancing maternal decision-making and improving household economic conditions need to be considered for reducing stunting among children. It is also crucial to give due emphasis for interventions related to infant and young child feeding with special emphasis for exclusive breast feeding.

Abbreviations
ANC: Antenatal care AOR: Adjusted odds ratio COR: Crude odds ratio HDSS: Health and demographic surveillance site WHO: World Health Organization.

Data Availability
Data will be available upon reasonable request from the corresponding author.

Ethical Approval
Ethical clearance was obtained from the Institutional Review Board of Arba Minch University. A formal letter was written to different administrative bodies and Arba Minch HDSS to obtain permission and resources to conduct the research in the settings.

Consent
Verbal informed consent was obtained from each mother after explaining the objective of the study, the risks and benefits, issues of confidentiality, and the right of participation.

Authors' Contributions
BB conceived the study, developed the tool, coordinated data collection, carried out the statistical analysis, and drafted the manuscript. BT and YC participated in the design of the study, development of the tool, statistical analysis, and reviewing of the manuscript. All authors read and approved the final manuscript.