Inappropriate child feeding practice and primary health care as major correlates of stunting and underweight among infants and young children 6-23months of age in food insecure households of Amhara and Oromia Regions, Ethiopia. A community based cross sectional study CURRENT

Background: Child undernutrition is a major public health problem in Ethiopia. Even though the highest levels of stunting are found in food insecure areas, insufficient evidence limits effective intervention to improve the situation. Therefore, this study aimed to assess undernutrition and associated factors among infants and young children aged 6–23 mo in food insecure households of Amhara and Oromia Regions, Ethiopia. Method: A community based cross-sectional study was conducted in productive safety net targeted rural households of Amhara and Oromia Regions from April 25 to June 15, 2018.A total of 464 mother-child dyads were included in the study. Both bivariate and multivariate analyses were used to identify factors associated with stunting, wasting and underweight separately. Results: The prevalence of stunting, wasting and underweight among children 6-23mo was 42.6%, 12.4% and 27.3% respectively. Age of the child 12-23vs 6-12mo(AOR = 4.21 95% CI: 2.52, 7.05), female (AOR = 1.84 1.23, 2.75), higher number of births (AOR = 1.721.10, 2.70), lack of zinc supplement (AOR = 2.411.33, 4.38), inadequate diet diversity (AOR=1.691.02, 2.81) and lack of iodized salt in complementary food (AOR=1.551.03, 2.32) were significantly associated with stunting. Similarly, age of child 12-23 mo (AOR=1.921.14, 3.24), female gender (AOR = 1.881.22, 2.90), higher number of births (AOR= 1.671.05, 2.66), lack of zinc supplement (AOR=2.291.14, 4.61), lower maternal income (AOR= 2.341.18, 4.65), and inadequate diet diversity (AOR= 2.341.27, 4.33) showed significant association with underweight. Conclusion: The magnitude of child undernutrition was found to be very high in the study areas. Child nutrition intervention strategies should take into account nutrition education on appropriate child feeding and iodized salt utilization. In addition, improving zinc supplementation and economic empowerment of women are important to address the high prevalence child undernutrition in the study area. Evidence-based nutrition intervention focused on identified factors in early life would be expected to have a positive effect on stunting reduction.

Previous studies summarized different risk factors for childhood undernutrition. Household food security and healthy household environments play important roles in preventing chronic malnutrition [3][4][5].Children under 2 years of age have high nutrient requirements because of their rapid growth and are considered the most at risk age group for undernutrition and increased vulnerability to infectious disease. Taking this age group as a window of opportunity, the World Health Organization (WHO) designed and supports the implementation of Infant and Young Child Feeding (IYCF) strategies [6].However, poor feeding and heath care in children continue as major determinants of undernutrition in developing countries [4].
Failure to wash hands, failure to dispose of feces hygienically and inadequate food hygiene were associated with a high incidence of diarrheal diseases in Ethiopia [7]. According to WHO, the global cause of death for more than half of under five children is diseases that are preventable and treatable through simple and affordable interventions. Particularly, malnourished children have a higher risk of death from common childhood illness such as diarrhea, pneumonia, and malaria. Therefore, children in sub-Saharan Africa are more than 15 times more likely to die before the age of 5 than children in high income countries [8]. Deficiencies of micronutrients like vitamin A, iron and zinc have global public health importance and represent a major threat to the health and development of populations worldwide [4]. These deficiencies are common in developing countries where a higher prevalence of undernutrition exists.
Likewise, deficiencies in key vitamins and minerals are among the major public health problems in Ethiopia [9]. A study from northern Ethiopia reported more than two thirds of school-aged children (79.5%) had at least one micronutrient deficiency and 40.5% had two or more coexisting micronutrient deficiencies. Based on biomarkers determined for 764 children, the most prevalent deficiencies were zinc, folate, vitamin A and vitamin D [10]. One of the main factors responsible for micronutrient deficiency is poor diet quality among children. Dietary diversification is one dimension of diet quality and ample evidence indicates dietary diversity is a proxy indicator of nutrient adequacy in a diet [11,12].
The period from 6 to 24 mo of age in children is one of the most critical periods for linear growth [13].The recent Ethiopian Demographic and Health Survey (EDHS) reported the national prevalence of stunting, wasting and underweight among children under five years as 38%, 10% and 24% respectively. The magnitude of undernutrition in Ethiopia shows variation from region to region with the lowest prevalence of stunting (14.6%) in Addis Ababa. On the other hand, the regional figures for stunting wasting and underweight were 46.3%, 9.8% and 28.4% respectively in Amhara region and 36.5%, 10.6% and 22.5% in Oromia region [14].
Chronic and acute food insecurity is prevalent, especially among rural populations of Ethiopia, and about 10 percent of citizens are estimated to be chronically food insecure [15,16]. This situation in Ethiopia is highly linked to severe, recurring food shortage and famine associated with recurrent drought [17]. Being one of three underlying causes of undernutrition, household food insecurity is assumed to affect the nutritional status of children by compromising quantity and quality of dietary intake [18]. But food security cannot be the only determining factor in ensuring adequate nutritional status of children. Hence, further investigation is needed to explore the most important factors behind the prevalence of undernutrition among children in Ethiopia [19].
Inadequate maternal and child health care and poor child feeding practices are among the main challenges that create a vicious cycle and contribute to long term existence of food insecurity. Often in developing countries, food insecure households are unable to produce enough food, even in times with favorable situations for agriculture, because the effect of chronic nutritional problems makes them unable to work to their full potential [20]. Nutrition program implementers often lack adequate evidence about which interventions would be most successful in reducing undernutrition in specific areas which may cause decisions to be based on assumptions and unjustified conclusions.
Inappropriate programs and interventions may be introduced only to discover the consequences after implementation [21]. Hence knowledge of the relative contribution of the predominant risk factors associated with stunting and underweight is an important prerequisite for developing nutrition intervention strategies, particularly in food insecure areas where the prevalence of undernutrition among children is high.
In these areas, identifying the most important factors contributing to undernutrition is important to guide public health planners, policy makers and implementers. Such knowledge assists to plan and design appropriate intervention strategies that will be more likely to enhance nutritional status of children and mitigate the vicious cycle of food insecurity.

Study setting and design
The study was conducted in five districts, Meket, Lasta and Sekota from Amhara and Chirro and Sirraro from Oromia where the Development Food Security Activity (DFSA) is being implemented. The districts were selected due to high levels of food insecurity. Overall, the program targeted 407,891 productive safety net program (PSNP 4 ) clients. From the five districts, kebeles (smallest administrative unit in Ethiopia) were selected randomly and a community based cross-sectional study was conducted from April 25 to June 15, 2018. Mother-child dyads were enrolled from households targeted for PSNP 4 if the child was 6-23 months old.
Sample size and sampling procedure The sample size of 494 mother-child dyads was calculated using Open-Epi analytic software and included an estimate of 5% for non-response rate. The calculated sample was proportionally allocated to the kebeles based on the total number of targeted households with 6-23 mo children in each kebele; study participants were identified by simple random sampling.
Data collection tools and procedure The primary outcome variable in this study was the anthropometric status of the index child measured as Z scores using WHO standards [22]. Socio-demographic variables, maternal and child health care and feeding practice were the independent variables. The socio-demographic variables were selected from those used by the Ethiopian Demographic and Health Survey (EDHS) [14] and collected using structured questionnaires via individual interview. Educational status was categorized according to the education levels in Ethiopia. Sixteen data collectors with BSc. degree and two community health workers were involved in the data collection process in each kebele. Two days of intensive training were provided to data collectors. Questionnaires were pretested in 25 households in non-selected kebeles and some modifications were made accordingly. Data were collected under close supervision of the principal investigator. Child weight was measured to the nearest 0.1 kg with a SECA electronic balance with graduation of 0.1 kg and a measuring range of up to25 kg. Weight was taken with light clothing and no shoes. Instrument calibration was checked before weighing each child. Furthermore, the weighing scale was tested daily against a standard weight for accuracy.
Length of the index child (aged 6-23 mo) was measured using a wooden length board in recumbent position, and read to the nearest 0.1 cm.
Standardized seven food groups were used to qualitatively assess the dietary diversity of children [6].
Mothers were asked to recall the food items consumed by the child in the24-h proceeding the day of the survey. By considering four food groups as the minimum acceptable dietary diversity, a child with less than four was classified as inadequate dietary diversity; otherwise diet diversity was considered to be acceptable. Household food security was classified as recommended by the Food and Nutrition Technical Assistance (FANTA) Guideline [23].The household wealth index was computed using a composite indicator considering selected household assets and size of agricultural land. Using principal component analysis (PCA), the factor scores were summed and ranked into quintiles. Data were checked for completeness and quality on a daily basis by the supervisors and principal investigator.

Data analysis
Statistical analyses were done using STATA 14 (Stata/SE 14) statistical package. Emergency nutrition assessment for standardized monitoring and assessment of relief and transitions (ENA for SMART) 2011 software was used to convert the anthropometric measures weight and length for age into Zscores. Descriptive statistics were calculated and bivariate analyses were done separately for the three outcome variables: stunting, wasting and underweight. Both Crude Odds Ratio (COR) and the Adjusted Odds Ratio (AOR) with a corresponding 95% Confidence Interval (CI) were computed to show the strength of the associations. All independent variables with p-value less than 0.25 during bivariate analysis were included for multivariate analysis. Multivariate logistic regressions were fitted using stepwise backwards elimination technique to identify determinants of stunting, wasting and underweight separately. In the multivariate logistic regression analysis, variables with a p-value <0.05 were considered as statistically significant. Assumptions of the regression model were found to be satisfied [24].

Results
Socio-demographic characteristics Totally 464 mother child dyads participated in the present study resulting in a response rate of 94%.
Of these, 51.7% of children were females ( Table 1). The mean (±SD) age of the children was 14.6 (±4.6) mo and mean household size was 5.0 (±1.8) persons. Only 12.7% of households had three or more under five children. Nearly 91% percent of the mothers were housewives and 64.7% had no formal education.

Obstetric characteristics of mothers
More than 90% of mothers had their first pregnancy within the age range of 15-26 years ( Table 2).
Concerning the total number of births (parity), nearly 27% of mothers had five or more live births. A limited number (20.9%) of mothers had four or more antenatal care visits for their most recent pregnancy. Only 3.5% of mothers gave birth to their most recent child at a health facility and 52.4% of mothers did not have postnatal care (PNC) even once after their most recent delivery.
Furthermore, 97.2% of mothers did not get a vitamin A supplement after their most recent birth.
Child feeding practice and health characteristics Only 52.6% of mothers reported initiating complementary food for the child at six mo of age ( Table 3).
The frequency of meals reported on the 24-h recall was four or more times for only 20. Overall, 42 % of mothers reported their children had respiratory infections, 18% had ear infection and nearly 38% experienced diarrhea in the 2 weeks preceding the survey. Treatment from a health care provider was sought for 47.6% of children with diarrhea out of which 53% of children received zinc treatment. Totally 16% of mothers reported their child had zinc supplements at least once in their life.

Prevalence of stunting wasting and underweight
In children 6-23 mo old, 42.6% were stunted, 12.4% were wasted and 27.3% were underweight ( Table   4). The children in the 12-23 mo age groups had higher prevalence of stunting and underweight when compared with the 6-11mo age groups.
Factors associated with stunting, wasting and underweight The results of the multivariate logistic regression model predicting stunting ( T h e multivariate logistic regression model ( Consistent with a previous review from Ethiopia [19], the present study showed that the likelihood of being stunted was more than fourfold higher among children aged 12-23 mo compared to infants 6-11mo of age. A plausible explanation might be the effect of malnutrition starting to manifest when the child`s nutrient requirement from complementary food increases and breast milk meets less of the needs, particularly at the age of one year and above. This is also an age at which the child is very likely to put everything to his mouth which may increase the risk of infection that negatively impacts the linear growth of the child.
We noted increased risk of stunting among children who did not meet the minimum diet diversity criteria. Previous studies have reported that dietary diversity is a proxy indicator of dietary quality and nutrient adequacy; hence, the overall nutritional quality of the child`s diet is improved with a diverse diet [36,37]. Diversity in the diet is important to meet the requirements for energy and other essential nutrients particularly for those who are at risk of nutrient deficiencies like children in food insecure households. Inadequate diet diversification among children was revealed to be a problem in the present study and lower diet diversity among children was associated with a higher likelihood of stunting. Poor diet diversification in addition to higher nutritional demand related with increased child age may increase the risk of impaired linear growth and be more obvious as these infants and young children grow older.  [41,42].
The present study also revealed higher likelihood of stunting among children who never received zinc treatment and whose caretakers didn't utilize iodized salt for complementary foods. Previous studies in developing countries have noted that multiple micronutrient deficiencies, particularly deficiencies of vitamin A, zinc, iron and iodine are evident among children [1,43].Zinc supplementation in children in endemic areas of zinc deficiency has been reported to enhance linear growth of children [44][45][46], especially when administered separately from other micronutrients. Zinc plays an important role in cellular growth, cellular differentiation and metabolism and hence has been recommended by previous researchers from Ethiopia and elsewhere to be included in national strategies to reduce stunting in children < 5 years of age in developing countries [45][46][47].
Increased demand for zinc can occur due to malnutrition and due to loss from the gut in diarrhea.
High prevalence of zinc deficiency in infants has been confirmed in Ethiopia [48], and the effect of zinc on linear growth of children was demonstrated [46]. Even though zinc supplementation is recommended protocol in diarrhea management in Ethiopia, poor compliance has been reported. The most recent EDHS reported that treatment from a health facility or provider for a child with diarrhea was sought for nearly 43% of under five children but zinc supplements were given for only33%of these children [14]. A study from northern Ethiopia confirmed poor management and inadequate treatment for children with acute diarrhea [49].
The other factor significantly associated with stunting in the present study was iodized salt utilization.
Children whose mothers did not utilize iodized salt for complementary food were 1.5 times more likely to be stunted. A low level of iodized salt utilization has been reported in previous studies in Ethiopia [50,51]. Even in places with adequate resources for child feeding, nutritional knowledge and factors such as reducing work overload of mothers are crucial to improve the nutritional status of children [53].
Hence it is important to consider a wide range of information in analyzing the association between feeding practices and child growth [54]. The determinants of stunting are varied and there is no single cause for undernutrition among children. Important pathways for effective interventions to mitigate stunting and promote healthy growth require clarification and consideration of context [55,56]. However, epidemiological studies have reported repeatedly that suboptimal breastfeeding and complementary feeding practices, recurrent infections and micronutrient deficiencies are important proximal determinants of stunting [29,57]. Considering a holistic approach, intervention against stunting during the early age of the child should not be overlooked. Also, women who were stunted as children tend to have stunted off spring. This creates an intergenerational cycle of stunting and poverty and of reduced human capital that is difficult to break [58].
We identified that as the number of births increases, so too do the odds of being stunted and underweight. Our finding is supported by another study from Ethiopia which stated that children whose mothers gave birth to more than four children were more likely to be stunted compared to children from mothers who had given birth to only one child [26]. Families with more children may face difficulty in providing the daily nutrition requirements for proper child physical development. As the number of children increases there may be scarcity of resources for the household especially for food and healthcare which ultimately may lead to stunted growth. Furthermore, parents with many children may lack adequate time to pay proper attention to the needs of each child.
The present study is one of only a few studies in Ethiopia that identify the determinants of stunting, wasting and underweight specifically in food insecure households and in one of the most vulnerable population groups, infants and young children aged 6-23mo. Despite the possibility of recall bias and measurement errors, the data revealed factors that could direct important interventions in the study area.

Conclusions
The prevalence's of stunting, wasting and underweight in the study area were higher than the national average. Undernutrition among infants and young children is a severe public health concern that needs critical attention at early ages. Improving zinc treatment, iodized salt utilization in Availability of data and materials