Spatial Heterogeneity and Factors Influencing Anemia and Severe Anemia among Under-5 Children in Ethiopia: Spatial and Multilevel Analysis

Background: Anemia is recognized as a significant public health problem in Ethiopia. Method: This secondary analysis, sampling technique and procedures done by DHS. A total of 8482 children aged 6–59 months were included in the study from EDHS 2016. Used tools for spatial heterogeneity analysis are Morans I, Local G*, Heat map, and Kulldorﬀ’s scan statistic, those carried out open source software (QGIS, GeoDa, SaTScan). Multilevel logistic regression analysis was used to identify both individual and household level factors associated with anemia and severe anemia, which measures between household variability using IHHC of the null model and generates 4 models PCV, AIC, and log-likelihood ratio used model selection for a report, which is carried out Stata 14. Result: The highest risk of both anemia and severe anemia regions are Somalia, Afar, DireDawa, and Harari. Specifically, all zones of Somalia, Afar (zone1, and 3), DireDawa, Harari, and Oromia higher risk of anemia and some of them are high risk for severe anemia. The high concentration of the disease showed in DireDawa, Harari, Jigjiga some part of Gambela and Benishangul Afar at the boundary of Amhara and Tigray and Eastern part of Tigray for anemia. Harari, DireDawa, and Afar at the border of Djibouti had high concentrated(density) severe anemia. Children (Younger age, lower preceding birth interval, stunted, underweight); mothers (younger age and anemic); households (poor and denser family number); and children from the highest and the lowest ecological zone are high risk for anemia. Additionally, children from the younger, work less, uneducated, poor, and anemic mother were high risk of a severe anemia effect of general anemia and high risk (severe) anemia in zonal level and characteristics of child, mother, and condition, and the level of associated factors in Ethiopia among under-5 children based on EDHS-2016. In this study, we investigate the burden of anemia and severe anemia of and associated factors, by level among under-5 children in Ethiopia. implementation geographical variation the main anemia causing factors. Overall anemic the highest distribution from the top are Somalia, Afar, DireDawa, Harari, and Oromia regions are highly affected their distribution is laid between the range 66.1–83.5% those are above country average distribution of childhood highest are in range 3.8–12.4%. overall than

children's mental and physical development, including learning, memory, and mental processing speed this consequently affects the latter age of school performance, working capacity [2,3]; a major cause of child and maternal mortality [4]; increases the chances of fetal deaths, abnormalities, preterm and underweight babies [5]. Short-term consequences of anemia, increased morbidity and mortality, developmental delay, and economic burden for sick children, while long-term consequences are stunted brains and stunted lives, minimize work productivity later in life, and hindering the development of entire societies [6][7][8]. Therefore, anemia affects not only the individual quality of life, but also the social and economic development of the country, particularly developing countries [7][8][9].
In 2011 WHO estimated that about 273.2 million children aged from 6 to 59 months in the world were suffering from anemia (42.6% of the total anemia), of which 9.6 million are severed anemic and above half (53.8%) anemic children found in South-East Asia and Africa Regions [1]. Anemia has a higher prevalence in sub-Saharan African (SSA) countries [11]. According to the Ethiopian Demographic and Health Survey (EDHS) report, in Ethiopia, the prevalence of anemia among children 6-59 months old was 53% in 2005, 44% in 2011, and 57% in 2016. From the sampled children in EDHS-2011, 28.6%, 21.7%, and 49.7% were above moderate (both moderate and severe), mildly anemic, and nonanemic, respectively [12]. In all years Ethiopia has severe public health problems in anemia [1]. Even though there was a study about under-5 child anemia in prevalence but not in severe, level, and spatial. There is no study showing the effect of general anemia and high risk (severe) anemia in zonal level and characteristics of child, mother, and condition, and the level of associated factors in Ethiopia among under-5 children based on EDHS-2016. In this study, we investigate the burden of anemia and severe anemia of and associated factors, by level among under-5 children in Ethiopia.

Study area
The study was conducted in Ethiopia, which is located in the North-Eastern part of Africa or known as the "Horn of Africa." It is bounded by the north and south Sudan on the West, Eritrea, and Djibouti on the North East, Somalia on the East and South East, and Kenya to the South. Ethiopia lies between the 3 0 N and 15 0 N Latitude and 33 0 E and 48 0 E Longitude.
The country occupies an area of approximately about 1,127,000 km 2 , which is almost twice the size of France. The Ethiopian landmass consists of a large, high elevated plateau bisected by the Rift Valley into the northwestern and the southeastern highlands, each with associated lowlands. The contrast in relief is remarkable as land elevation ranges between − 155 m of Asal Lake in the Afar depression (the lowest point in Africa) to the peak of Mt. Ras-Dejen, at 4,620 meters above sea level in the Semen Mountains [13]. There are nine regional states and two city administrations.

Study design
A population-based cross-sectional study was employed to explore geographical variation and identify the influencing factors of under-5 child's anemia and severe anemia in Ethiopia.

Data Source
The data for this study were taken from the 2016 EDHS. The 2016 EDHS is the fourth comprehensive and nationally representative survey conducted in Ethiopia as part of the worldwide Demographic and Health Surveys (DHS) project. The EDHS 2016 data were downloaded from The DHS website after being granted permission.

Dependent Variables: Anemia And Severe Anemia
The primary outcome of interest was anemia and severe anemia. Anemia is defined as a hemoglobin level of less than the 5th percentile for age and has several causes that vary by age [14]. According to WHO, children under 5 years of age with a hemoglobin level of less than 11.0 g/dl is considered anemic. The cutoff values for the various levels of severity were: < 7 g/dl for severe anemia, 7 g/dl-9.9 g/dl for moderate anemia and 10 · 0 g/dl-10 · 9 g/dl for mild anemia [15]. Based on this DHS itself puts individuals final outcome by severity.
Sample size and sampling procedures A weighted sample of 8,482 individuals from 6,460 households, out of which 4084 females and 4398 males were included in the study. All sampling procedures, data collection, and data quality control were done by the DHS team.
Data processing and analysis Spatial Statistics Analysis Statistically significant clusters defined as geographic areas where the prevalence of the disease is disproportionately higher/lower compared to neighboring areas. Tests for global clustering detect the existence of at least one cluster, but not the specific place of the cluster(s)

Mapping Cluster
The LISA cluster map shows p < 0.001 with Moran's I was 0.463 for anemia and 0.306 for severe anemia. The Moran's I indicated clustering patterns (Fig. 3.).
In both panels the red color indicates a high rate of anemia surrounded by high rates of similar cases.
Such specific areas were All Somalia region zones except Jigjiga and Shinile, In the Oromia region

Kulldoruff's Scan Statistic
Kulldoruff's scan statistic is a spatial scan statistics method for detecting and evaluating statistically significant spatial clusters risk factors for a specific disease. Final confirmatory spatial analyses were done using spatial analysis software SaTScan and QGIS. The SaTScan declares where spatially significant higher an lower rates of aggregates are found. Its output presents the hotspot areas in circular windows, indicating the areas in the windows are higher than expected distributions compared to the areas outside of the cluster windows [19][20][21][22].

Multilevel Logistic Regression
Multilevel logistic regression was utilized to assess the impact of measured lower-level and higher-level factors. Multilevel analysis was considered appropriate in order to account for the hierarchical nature of the EDHS data and to be able to estimate household-level effects on the outcome variable [23][24][25].

Model Building
The four model building and analysis were done using STATA14. Overall, four models containing variables of interest were fitted for each of the under-5 anemic and severe anemic children. The first model (M 0 ) is an empty model which was fitted without independent variables to test random variability using Intra House Hold Correlation (IHHC) [26]. The second model (M 1 ) was fitted to all lower-level (individual level) factors; the third model (M 2 ) used for all higher-level (household level) factors; and the fourth model (M 3 ) used for both lower-and higher-level factors to report. Model fitness for the report selected by the Akaike Information Criterion (AIC).

Risk Of Anemia And Severe Anemia Per 1000 Children
The left panels (anemia) of the red color indicates above 800 anemia cases per 1000 children, Those zones are in Somalia region Worder (872) in Afar region (Fig. 6, Table 1).

Model Comparison
In this study IHHC value was 16.95% and 13.35% for anemic and severely anemic respectively, which is > 5%, which indicates the level analysis is more appropriate than a binary logistic regression model. Both The AIC and BIC calculate model fitness for the report, which indicates the smallest M 3 (both individual and household level) for anemic and M 2 (only household level) for sever anemic was the best model fitness for the final report (Table 2). After controlling another individual and household level factors; When the child age increase by a month the risk of stunting decreased by 10% (AOR = 0.9, CI: 0.89-0.92). Stunted children 2.18 times higher risk of anemia (AOR = 2.18, CI: 1.58-3.00) as compared to whose counterparts. Wasted children, 63% more risk of anemia (AOR = 1.63, CI: 101-2.65). Ever-breastfed but not currently breastfeeding children were 52% (AOR = 0.48, CI: 0.33-0.69) more likely to be anemic than still breastfeeding. Underweight children were 60% more likely to risk for anemia than normal and above.
When a family household size increased by 1 person the risk of anemia in child increased by 19% (AOR = 1.19, CI: 1.08-1.31). The mother's age increased by one year the risk of anemia on the child decreased by 6% (AOR = 0.94, CI: 0.91-0.97). From poor family child 76% more risk than from rich family child (AOR = 1.76, CI: 1.15-2.69). Female household head family child 2.14 times exposed by anemia compared to their counterparts within (AOR = 2.14, CI: 1.27-3.61). Children whose mother anemia status mild, moderate, and severe were 4.18, 4.29, and 7.88 times more likely to be anemic Household who lives in subtropical was 8.45 (AOR = 8.45, CI: 2.91-24.51) and in highland 3.14 (AOR = 3.14, CI: 2.1-4.7) were times higher risk of their child with anemia than whose household who lives in temperate regions [See Additional file 1].

Multilevel Logistic Regression For Under-5 Children Severe Anemia
After controlling other household-level factors; Mother age, mother education status, household wealth, mother anemia status, mother working status, were statically significant factors for severe anemia. Mother's age increased by one year the risk of severe anemia on the child decreased by 3% (AOR = 0.97, CI: 0.915-0.99). Children born from uneducated women, 38%, (AOR = 0.62, CI: 0.42-0.92) more risk for severe anemia than a child born from primary educated mother. Children whose mother anemia status mild, moderate, and severe were 2.46, 4. Based on the WHO standard (anemia prevalence > 40% is a severe public health problem) [1], so Ethiopia is categorized as severe public health problem regions because its prevalence is 57.6%. this region anemia prevalence is high [27]; These regions are malaria-affected region, malaria was the leading factor associated with anemia [28]. Most of this region the primary, complementary food is animal milk, which is poor sources of iron or negative effect on non-heme and heme iron absorption [29][30][31]. The other tangible reason peoples who live in lowland widely dispersed and they can't get health facility and road (transport), so they can't find any intervention like iron and folic acid and have accesses advice of neither before nor after delivery. Those peoples' lives are bound up with animal-like goat, cow, and camel so they are pastoral, the first problem is this type of work the person who has no permanent residence throughout their life depend on other factors like the accesses of water and forage for their animals, so must move from place to place and the second problem this pastoral job is culturally done by females, both during pregnancy and after birth, sometimes the women will be born on work, so the mother and her child ride different burdens like a hard bit of sunlight, hanger and they are not finding rest and care. The other region is a relatively low risk of anemia; this is due to the potential of producing teff, which is iron-rich food. The lowest distribution of anemia regions is like Amhara and Benishangul Gumz but the surprising thing is these lowest regions are according to WHO lie on sever public health problem regions. So, relatively can be said high and low hot spot and cold spot but in fact all the regions of the country under severe public health problems [1].
Based on the geographical variation result, Anemia and severe anemia with respect to the specific area like zone indicates. The densest distribution of anemia with a narrow area is located in Harari and Dire Dawa and surrounded, some part of Gambela, S-W of Benishangul Gumz, North-East direction of the country especially boundary of Tigray with Afara and Amhara with Afar. And for severe anemia, high denser shows Afar region at the border of Djibouti, DireDawa and Harari and their surrounding and Jigjiga zone in Somalia region in comfort areas for treatment and other intervention for many children with a limited resource and man-power because the child on the anemia burden finds clothes each other, but the other hot spot areas the peoples are live scattered as well as the case is scattered like Somalia region so it is not easily manageable, which means they need more resource and man-power for tackling the problem.
Maternal (anemia status and age) and household wealth are show significant contribution for both anemia and severe anemia prevalence; Mother working and educational status are a significant contribution for only severe anemia, But any factor does not contradict between anemia and severe anemia. Maternal anemia directly related to both child anemia and severe anemia. This finding strongly agrees with another finding like a study done in Ethiopia DHS-2011 [32], in Togo DHS 2013-14 [33], in Malawi DHS-2010 [34]. This strong association could be due to many factors that determine anemia because many things share between mother and child such as worm infection, malaria, and other infectious diseases that affect red blood cell production and iron stores [35].
Likewise, low levels of iron, zinc, folate, and vitamin A and B12 in the breast milk of the anemic mother could also affect the Hemoglobin level of the breastfeeding child. Younger children high risk for anemia and severe anemia whose mother anemic during pregnancy [36]. The other reason may be the child finds enough iron stores in their bodies at least the first six months from his/her mother during lactation.
The other factor strong association with both anemia and severe anemia is the wealth status of the family. The child with poor family highly affected by anemia and severe anemia this output in line with other findings done in different areas [37][38][39][40]. This finding could be due to the reason for problems in production or purchasing necessary food for themselves and their children, so they can not get iron-rich foods. The other reason poor mothers might have problems in purchasing and providing good nutritious which might result in anemia. This will be a fact because, child below 12 months and their mothers share a similar sociological environment, their dietary patterns, and their quality of life similar.
Children high risk for anemia and severe anemia who are from younger mothers for anemia and severe anemia. this results in the same as others like research done Ethiopia [38], in Bangladesh [41] and in Sub Sahara Africa [42]. The reason for child feeding, hygiene, and child care older age mothers have good experienced. And the other result older women are economically developed and the child has easily gained iron-rich foods.
Child age increases the risk of anemia decrease this result similar with other findings such as a study done in Ethiopia Eastern Amhara [38], EDHS-2011 [12], and in Sub Sahara Africa [42] this could be explained by the fact that older children receive a diet that is richer and complete iron, with sufficient intake of iron it prevents the occurrence of anemia in the child.

Conclusion
This finding identifies based on different dimensions which are based on sample populations (prevalence, hotspot, association with neighbors), based on specific area density (accumulation), based on child characteristics, and maternal characteristics with some community characteristics. So anybody easily understands the situation with other characteristics supporting or reducing this situation and can be intervention easily based on this finding. Specialty planners and decision makers easily understand the disease situation and to identify the type of intervention based on place, characteristics, severity, accumulation. And additionally, researchers can be further analysis this identifies the place and group why/how are high risks and how to minimize this problem. Generally, anemia is high, especially rift value areas highly affected by both anemia and severe anemia especially Somalia regions and its neighbors. Those regions are not productive areas, lack of health facility, malaria region, backward for any access. So the concerned body must be tackled to minimize this childhood series problem. Before starting any intervention first must be prioritized according to risk, concentration, and characteristics. Declarations Ethics approval and consent to participate Ethical clearance with written consent (AuthLetter_130669) was obtained from Measure DHS International Program which authorized the data sets. All the data which used in this study are publicly available, aggregated secondary data with not having any personal identifying information that can be linked to particular individuals, communities, or study participants. Confidentiality of data maintained anonymously.

Consent for publication
Not Applicable Prevalence of anemia by severity and region in Ethiopia, 2016. The pie-chart of the black color indicates severe (< 7 g/dl), red color for moderate(7 g/dl-9.9g/dl), yellow color for mild(10 • 0 g/dl-10 • 9 g/dl), and the green color for not anemia(≥11g/dl). To carry out this analysis used QGIS 3.4.3, which available at https://www.npackd.org/p/qgis/3.4.3.  The hotspots are public health importance. To conduct this analysis use GeoDa version 1.14, which is available at: https://geodacenter.github.io/download_windows.html.

Figure 4
Hotspot map of anemia and severe anemia in Ethiopia, 2016 The left side of the map indicates anemia and the right side of severe anemia. Each polygon on the map represents a single zone area with a burden of anemia and severe anemia. High (red color) means high rates (hot spot) of anemia and severe anemia. Low (green color) shows a low rate (cold spot) of severe anemia and undefined (black color) indicates lakes and zones (did not sample taken in this area). To conduct this analysis use GeoDa version 1.14, which is available at: https://geodacenter.github.io/download_windows.html.  Anemia and severe anemia hotspot clusters identified using SaTScan spatial analysis tool, in Ethiopia 2016. The red color circle in both panels indicates hotspot windows. And the dot indicates enumeration cluster. To conduct this analysis use GeoDa version 1.14, which is available at: SaTScan v9.6, which available at https://www.satscan.org/download.html and QGIS 3.4.3.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.