Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Spatial variations and determinants of anemia among under-five children in Ethiopia, EDHS 2005–2016

Spatial variations and determinants of anemia among under-five children in Ethiopia, EDHS 2005–2016

  • Zelalem Alamrew Anteneh, 
  • Jean-Pierre Van Geertruyden
PLOS
x

Abstract

Background

Anemia has severe public health significance in sub-Saharan Africa. In Ethiopia, anemia has been increasing in the last two decades, reaching the highest national level in 2016, however, the geospatial distribution and determinants of anemia in children weren’t well explored at a national level.

Methods

We used the Ethiopian Demographic and Health Survey(EDHS) data from 2005–2016. The data consists of samples of households (HHs) obtained through a two-stage stratified sampling procedure. Our analysis included 19,699 children. Descriptive statistics, geospatial analysis, and Generalized Linear Mixed Model (GLMMs) were used.

Results

The overall prevalence of anemia was 51.5%; the spatial distribution of anemia significantly different across clusters in each survey. Children from 6 to 11 months had higher odds of anemia compared to 24–59 months (Adjusted Odds ratio (AOR) = 3.4, 95%Confidence level (CI): 2.99–3.76). Children with the first and second birth order were less likely to be anemic compared to fifth and above (AOR = 0.60, 95%CI: 0.38–0.95, and AOR = 0.83, 95%C: 0.73–0.93) respectively. Mothers’ age 15 to 24 years was associated with higher odds of anemia compared to 35 to 49 years (AOR = 1.37, 95%CI: 1.20–1.55). Children from HHs with the poorest and poorer wealth category showed a higher odds of anemia compared to the richest (AOR = 1.67, 95%CI: 1.45–1.93, and AOR = 1.25, 95%CI: 1.08–1.45) respectively. Moreover, children from HHs with one to two under-five children were less likely to be anemic compared to those three and more (AOR = 0.83, 95%CI: 0.76–0.91).

Conclusions

The geospatial distribution of anemia among children varies in Ethiopia; it was highest in the East, Northeast, and Western regions of the country. Several factors were associated with anemia; therefore, interventions targeting the hotspots areas and specific determinant factors should be implemented by the concerned bodies to reduce the consequences of anemia on the generation.

Introduction

Anemia is a condition in which the ability of red blood cells to carry oxygen is impaired, insufficient to meet the physiologic needs of the body [1, 2]. It is an indicator of both poor nutrition and health.

According to the global burden of disease report in 2016, anemia affects more than 27% of the world’s population, nearly, 1.93 billion people. Low- and middle-income countries account for more than 89% of the cases. Preschool children and reproductive-aged women are disproportionately affected by anemia [3]. In 2017, the World Health Organization (WHO) data repository showed that the global prevalence of anemia in under-five children was 41.7%. The problem is worse in the WHO Africa region, 59.3% of children under five were anemic [4]. In Ethiopia, the prevalence of anemia in children less than 5 years is persistently higher than expected for the last two decades [5]. In Sub-Saharan African countries, anemia is one of the major challenges to health affecting more than 40% of children and is considered a severe public health significance in the region [6]. This demands to investigate the component causes of anemia to implement the corresponding intervention strategies.

Iron deficiency is the major cause of anemia, but there are several other causes including folate, vitamin B12, and vitamin A deficiencies, and chronic inflammation, parasitic infections, and inherited disorders [2]. Micronutrients are often associated with specific physiological processes in the body; deficiency at an early age can affect children’s cognitive and motor development [7, 8].

Studies showed that several factors have been affecting the occurrence of anemia in children, a systematic review study indicated that the distribution of anemia is more prevalent among children from rural communities compared to urban ones [9]. Similarly, a study conducted in Bangladesh revealed that children from rural communities were at higher risk of anemia compared to children from urban communities [10]. Evidence is showing younger children are at higher risk of anemia compared to older ones; studies conducted in Uganda and Ghana indicated that children less than 24 months of age had a serious risk compared to older children [11, 12].

Studies also pointed out that maternal anemia affects the anemia status of children [13, 14]. Besides, household wealth status, family size, and the number of children in the household influenced childhood anemia [11, 15]. Regardless of the persistent higher magnitude of childhood anemia among children in Ethiopia across the waves of EDHS, spatial distributions, and determinants of anemia in children were not well explored at a national level. Studying regional variations and determinants of anemia has an important policy implication to evaluate the progress being made by the regional states in the country. In addition, studying the geospatial variations of anemia helps to identify the most hotspot areas of the problem. This helps to inform the government where to allocate its scarce resource for implementation of interventions to minimize the effect of anemia on the generation.

Methods and materials

Study design and setting

This study used data from nationally representative, cross-sectional Ethiopian Demographic, and Health Survey (EDHS) conducted from 2005 to 2016 [5, 16, 17].

Administratively, Ethiopia has nine regional states namely; Amhara, Oromia, Tigray, Benishangul-Gumuz, Somali, Afar, Harari, Southern Nations Nationalities and Peoples (SNNP), Gambella, and two city administration councils (Addis Ababa and Dire Dawa. The regions are subdivided into Zones, each Zone into Woredas, and Woreda into the lowest administrative units named Kebeles. For sampling purposes, each kebele was subdivided into census enumeration areas (EAs) [18].

Sampling strategy

EDHS consist of a sample of HHs obtained through a two-stage stratified sampling procedure. The EDHS uses the Ethiopia Population and Housing Census (PHC) sampling frame prepared by the Ethiopia Central Statistical Agency (CSA). Samples were selected using a stratified, two-stage cluster sampling design, where enumeration areas (EAs) were the primary sampling units for the first stage, & house-holds (HHs) for the second stage. Representative samples of 32,156 (9,861 in 2005, 11,654 in 2011, and 10,641 in 2016) under-five children were included in the respective surveys. The EDHS collects blood samples among all children of age 6 to 59 months included in the survey for hemoglobin tests. A total of 19,699 (3394 in 2005, 8510 in 2011, and 7795 in 2016) children gave blood specimens for hemoglobin tests; therefore, we included them in our analysis.

Outcome measurements

Blood specimens were collected from 6–59 months children for whom consent was obtained from their parents or caregivers responsible for them. Blood samples were drawn from a drop of blood taken from a finger prick (or a heel prick in the case of children age 6–11 months) and collected in a microcuvette. Hemoglobin analysis was carried out on-site using a battery-operated portable HemoCue analyzer. The hemoglobin values obtained using the HemoCue instrument were adjusted for altitude before classification into the level of anemia. Based on the WHO hemoglobin level cut off points, the hemoglobin from 10.0–10.9 g/dl is mild, 7.0–9.9 g/dl is moderate, and level less than 7.0 g/dl severe anemia. Therefore, the hemoglobin level less than 11 g/dl of blood is anemic, otherwise normal.

Independent variables

Potential determinant factors of anemia in children were extracted from the EDHS dataset. The factors were selected based on previous studies [10, 15, 1921], and our knowledge. These factors were categorized in the following ways:

  1. Individual-level factors. Sex, age, size at birth, stunting, wasting, underweight, birth order, fever, diarrhea, cough, child twin status, birth intervals, mother’s and father’s educational level, age of mother, marital status, mothers’ working status, religion, mother’s media exposure.
  2. Household and contextual level factors. Residence (urban, rural), source of drinking water, type of toilet facility, type of cooking fuel, wealth index, number of children U5 in the family, number of household size.

Operational definitions

Most of the determinant factors in our analysis were used as they are in their original coding from EDHS; however, some variables were created by combining two or more variables or regrouping the levels of the variables.

Sources of drinking water: the source of drinking water has several levels in the original coding, however, we regrouped into (improved and unimproved water sources), and Toilet facility: toilet facility consisted of several levels in the original coding and regrouped into (improved and unimproved toilet facility) based on WHO classifications [22]. Similarly, cooking fuel was grouped into (Electricity and gas, fossil fuels, charcoal, and agricultural products & animal dung) [23], and the presence of media exposure into “Yes” (women who listen to the radio, watch TV, or reads magazines at least once in a week), otherwise “No” [5].

Birth size.

Children whose birth weight is less than 2.5 kgs, or children reported to be ‘very small’ or ‘smaller than average’ are considered to low birth weight, otherwise normal. According to the EDHS sources, birth size (birth weight) was collected based on the presence of written records if available, or based on the mother’s report [5, 16, 17].

Stunting.

Height-for-age less than minus two standard deviations (-2SD) from the median of the reference population were regarded as moderately stunted, while below −3SD from the median of the reference population were considered severely stunted.

Underweight.

Weight-for-age less than -2SD from the median of the reference population were regarded as moderately underweight, while below −3SD from the median of the reference population were considered severely underweight.

Wasted.

Weight-for-height less than -2SD from the median of the reference population were regarded as moderately wasted, while below −3SD from the median of the reference population were considered severely wasted. These anthropometric indicators were measured based on the WHO growth standard [24].

Wealth index.

Wealth index is a measure of the socioeconomic status of the households to indicate inequalities in society. The households were given scores based on assets (television, bicycle/car, size of agricultural land, the quantity of livestock), and dwelling characteristics (sources of drinking water, sanitation facilities, and materials used for constructing houses) using principal component analysis, and the scores were compiled into five categories of wealth quintile (poorest, poorer, medium, richer and richest) each comprising 20% of the population [5, 16].

Spatial analysis

Spatial analysis was done using the application of Geographic Information System (GIS) to determine geographic variations of anemia cases among EDHS clusters for each wave from 2005 to 2016. ArcGIS software version 10.1 was used. We received the GPS points in shapefile format for each EDHS survey from the DHS office upon request. We computed the proportions of anemia cases for each cluster in each survey. We downloaded the images of administrative boundaries of Ethiopia in shapefile form from DHS website [25]. Then, we appended the proportions of anemia with the shapefile of the clusters. The high and low hotspots of anemia were visualized for each cluster in each survey. The Getis-Ord G-statistic was used to show the overall patterns of high/low clustering of anemia among children in enumeration areas.

Spatial variations of significantly high and low hotspots of anemia were computed for each cluster in each survey using the Getis-Ord G* statistic tool. The clustering of statistically high hotspots of percentages of anemia is shown by a positive z-score with a P-value of <0.05, however, clustering of statistically low spots of anemia is indicated by a negative z-score with a p-value of <0.05. A z-score near zero indicates no apparent clustering.

Statistical analysis

Descriptive statistics such as frequency distributions for respondents and children under five years were done. The prevalence of anemia in under-five children by different backgrounds and contextual characteristics of the HHs was computed. In addition, the prevalence of anemia among under-five children across the waves of EDHS 2005 to 2016 was computed for whole regions of Ethiopia. We used weighted data analysis to account for the difference in the sampling proportions to avoid distortions in our estimates. STATA version 13 and SPSS Version 25 software packages were used for analysis. Bivariate logistic regression analysis was used to identify potential candidate independent factors associated with anemia among under-five children. Variables that were associated with anemia at a p-value of 0.20 level of significance were selected to enter multilevel logistic regression models.

A generalized linear mixed model (GLMMs) was carried out to examine the effects of individual, and household, and contextual factors on childhood anemia. We fit three phases of modeling. The first (Model 1) was the null model with no individual or household level factors. It consisted of only cluster-specific random effects to model between-cluster variations in anemia. The second model (Model II) incorporated individual-level factors in addition to cluster-specific random effects. The third model (Model III) contained both individual, and household, and contextual factors in addition to cluster-specific random effects. For the fixed part of the model, the results were presented with ß-coefficients, standard errors, p-values, and OR with its 95% confidence level, however, for the random part, variance estimate with its standard error and 95% confidence level was used to present the results.

Multicollinearity tests were performed to check the presence of correlations among explanatory factors. We computed the variance inflation factor (VIF) for each predictor variable by doing a linear regression of each predictor on all the other predictors, in each case we obtained VIF within the range of recommended cut of points [26]. The intraclass correlation coefficient (ICC) was computed for each model to show the amount of variations explained at each level of modeling. Model comparisons were done using the Likelihood Ratio Test (LRT), Akaike information criteria (AIC), and Bayesian Information Criteria (BIC). The model with the lowest LRT, AIC, and BIC, was considered the best fit model.

Ethics approval and consent to participate

The data for this study was received from the DHS office upon request. The data was collected by the Ethiopian Central Statistical Agency (CSA) & the Federal Ministry of Health (FMoH) with the technical assistance of ICF through the DHS Program. The ethical clearance was provided by the Federal Democratic Republic of Ethiopia Ministry of Science and Technology and the Institutional Review Board of ICF International. Written consent to participate in the study was obtained from parents to take blood from children under five for the hemoglobin test, and the data were recorded anonymously.

Results

Sociodemographic characteristics of respondents/households

A total of 19,699 mother-child pairs were included in our analysis, 21.8% of the mothers were less than 24 years of age, and 95.1% of the women were in a marital relationship. Nearly, seven in ten women have no formal education, only 4.4% of them have a secondary and higher educational level.

About 89.9% of respondents were rural residents; 22.9% and 22.6% of the HHs were in the range of poorest and poorer wealth quantile respectively, and only 13.7% of HHs were in the richest range. Nearly, 90% of the HHs use agricultural, wood, and animal dung products as a source of cooking fuel, only 1.4% of the HHs had access to electricity and gas (Table 1).

thumbnail
Table 1. Sociodemographic characteristics of respondents/households.

https://doi.org/10.1371/journal.pone.0249412.t001

Sociodemographic, birth history and physical measurements of under-five children in Ethiopia, EDHS 2005–2016

Of the total children included in the study, 11.9% were 6 to 11 months, 21.2% were 6 to 23 months, and the remaining 66.9% were 24 to 59 months of age. About 18.1% and 8.9% of children were very small, and small at birth respectively, and 17.7% and 33.8% of children were first births, and fifth and above births orders in a family respectively. The finding revealed that 16.8%, 14.6%, and 20.2% of children were reported fever, diarrhea, and cough in the last two weeks before the survey respectively. In the anthropometric measurements, 40.1% of children were stunted, 35.3% were underweight, and 9.2% were wasted (Table 2).

thumbnail
Table 2. Sociodemographic, birth history and physical measurements of under-five children in Ethiopia, EDHS 2005–2016.

https://doi.org/10.1371/journal.pone.0249412.t002

The prevalence of anemia among under-five children by different background characteristics

The overall prevalence of anemia was 51.5% (95%CI: 50.8–52.2). Of the total 51.5% of the anemia cases, 3%, 25%, and 23% were severe, moderate, and mild cases respectively (Fig 1). The magnitude of anemia across the wave of the EDHS surveys was 55.2% (95%CI: 52.6–55.6.0) in 2005, 44.6% (95%CI: 43.6–45.6) in 2011, and 57.6% (95%CI: 56.5–58.7) in 2016.

thumbnail
Fig 1. The weighted levels of anemia among children younger than five years in Ethiopia from 2005 to 2016.

https://doi.org/10.1371/journal.pone.0249412.g001

Children from HHs utilizing agricultural products & animal dung as a source of cooking fuel showed the highest proportion of anemia (46.2%), only the remaining 5.2% of the cases were from HHs that use other sources of cooking fuels. In addition, the prevalence of anemia from HHs using unimproved water sources and unimproved toilet facilities was 37.3% and 48% respectively (Table 3).

thumbnail
Table 3. The weighted prevalence of anemia among under-five children in Ethiopia by different background characteristics, EDHS 2005–2016.

https://doi.org/10.1371/journal.pone.0249412.t003

Regional variations of anemia among under-five children in Ethiopia, EDHS 2005–2016

The findings of this study indicated that children from Somalia regional state were severely affected by anemia, where the prevalence was 86.2% in 2005, 69.6% in 2011, and 83.3% in 2016. The Afar regional state and Dire Dawa city administrations were the second most affected regions; anemia was sustainably higher than 56% across EDHS surveys. Similarly, anemia among children from Oromia, Gambella, and Hareri regions ranges between 50% to 69% across the waves of EDHS surveys.

According to results obtained, a general declining trend of anemia was observed in Tigray, Amhara, Benishangul, and SNNP regions over the years. In addition, the lowest prevalence of anemia was observed among children in Addis Ababa (40% in 2005, 33.1% in 2011, and 48.8% in 2016) (Table 4).

thumbnail
Table 4. The weighted frequency and percentage distribution of anemia among children under-five across regions in Ethiopia, EDHS 2005–2016.

https://doi.org/10.1371/journal.pone.0249412.t004

Results of the geospatial analysis

In the EDHS 2016 cluster-level (lower level) analysis, the Getis-Ord- General statistic tool indicated that the general patterns of anemia distribution among children was not similar across the clusters (Z-score of 2.76, and p-value of 0.006). The hotspots analysis using Getis-Ord- G* statistic showed that high hotspots of anemia were observed in Eastern regions (Somalia, Haregie, and Dire Dawa), Northeastern (Afar region), and Western (Gambella) and Southern parts of the country (few clusters of Oromia region). However, low hotspots were seen in most of the central regions (Addis Ababa, some Oromia zones) and Northwestern (Amhara and Benishangul) regions (Fig 2).

thumbnail
Fig 2. Maps of high and low hotspot clusters of anemia among under-five children in Ethiopia, 2016.

Reprinted from [25] under a CC BY license, with permission from [DHS], original copyright [2016].

https://doi.org/10.1371/journal.pone.0249412.g002

In the EDHS 2011 spatial analysis, the Getis-Ord- General statistic produced a Z-score of 1.87, and a corresponding p-value of 0.023 showing the overall all patterns of anemia distribution varies across the clusters. The Getis-Ord- G* hotspots analysis revealed that high hotspots of anemia were located in Eastern regions (Harergie, Dire dawa), Southern region (Oromia zones), Northeastern (Amhara and Afar regions), few clusters in North Shewa and Tigray regions, whereas low hotspots have occurred in the central regions (Fig 3).

thumbnail
Fig 3. Maps of high and low hotspot clusters of anemia among under-five children in Ethiopia, 2011.

Reprinted from [25] under a CC BY license, with permission from [DHS], original copyright [2011].

https://doi.org/10.1371/journal.pone.0249412.g003

Similarly, in the 2005 EDHS spatial analysis, the general distribution of anemia was not similar for the clusters (Getis-Ord- General produced Z-score of 2.81 and p-value of 0.0049). The hotspots analysis of Getis-Ord- G* indicated high hotspots of anemia in the Eastern part of the country (Hareri region, Dire Dawa, and few clusters in Somalia regions), and Southern borders of the country and low spots occurred in Addis Ababa (Fig 4).

thumbnail
Fig 4. Maps of high and low hotspot clusters of anemia among under-five children in Ethiopia, 2005.

Reprinted from [25] under a CC BY license, with permission from [DHS], original copyright [2005].

https://doi.org/10.1371/journal.pone.0249412.g004

The legend, critical Z scores grouped into three classes for hotspots graphs for convenience of interpretation in the following ways; Statistically Significant coldspot (Z scores less than -1.96), not Significant (Z scores between -1.96 and +1.96) and Significant hotspot (Z scores greater than +1.96).

Determinants of anemia among under-five children in Ethiopia, EDHS 2005–2016

A univariate regression analysis was performed to have an insight into the association between anemia and predictor variables. Age, place of residence, maternal educational level, wealth index, birth order, birth intervals, fever, diarrhea, birth size, nutritional status of children, age of mother, number of under-five children in the HHs, household size, media exposure of mother, and years of the survey were showed association with anemia at 20% level of significance (Table 5).

thumbnail
Table 5. The unadjusted association between predictor variables and anemia in under-five children in Ethiopia, EDHS 2005–2016.

https://doi.org/10.1371/journal.pone.0249412.t005

Generalized linear mixed model regression analysis with an intercept only model (Model I)

In model 1, the cluster level regional variation of anemia among children was assessed without considering the effect of individual, household, and contextual level factors. There is statistically a significant variation between clusters in the prevalence of anemia (p-value <0.001). Cluster level variance was 0.253416, and the intraclass correlation coefficient (ICC) between clusters was 0.0715. This indicates that 7.15% of variations in the prevalence of anemia can be explained by clusters (higher level) factors, and the remaining 92.85% of the total variations of anemia is explained within-cluster lower-level units (Table 6).

thumbnail
Table 6. Multilevel logistic regression model without explanatory variables (model I).

https://doi.org/10.1371/journal.pone.0249412.t006

Random intercept and fixed slope GLMM regression for predictors of anemia among under-five children in Ethiopia, EDHS 2005–2016

In the random intercept model (Model II), the effect of individual-level factors on childhood anemia was assessed. The variance of a random factor was 0.2531076 with its standard error 0.0243827 and the confidence level doesn’t include 0.

The age of children was a strong predictor of anemia, children between 6 to 11 and 12 to 23 months had more than three- and two-fold times likely to be anemic compared to children over 24 months (AOR = 3.44, 95%CI: 3.07–3.85, & AOR = 2.40, 95%CI: 2.20–2.61) respectively. Birth order and birth intervals were associated with the presence of anemia. Children with first and second birth orders in a family were less likely to develop anemia compared to those with fifth and above birth orders (AOR = 0.58, 95%CI: 0.37–0.92, & AOR = 0.79, 95%C: 0.71–0.88). The odds of anemia among children whose succeeding birth interval less than 24 months was more than 20% higher compared to those with more intervals (AOR = 1.22, 95%CI: 1.09–1.36).

Children who reported fever recently had higher odds of anemia compared to those with no fever (AOR = 1.12, 95%CI: 1.02–1.22). Stunted and underweight children had more than 25% and 26% likely to develop anemia compared to their counterparts with normal anthropometric measures respectively.

In addition, children from mothers of age 15 to 24 months had more than 37% higher odds of anemia compared to those from mothers of age 35 to 49 years (AOR = 1.38, 95%CI: 1.22–1.56). Eventually, children from mothers who have no access to media exposure were more likely to report anemia compared to those from mothers with access to media (AOR = 1.32, 95%CI: 1.24–1.41 (Model II, Table 7).

thumbnail
Table 7. Random intercept, and random coefficient multilevel logistic regression model for predictors of anemia among under-five children in Ethiopia EDHS 2005 to 2016.

https://doi.org/10.1371/journal.pone.0249412.t007

Random intercept and coefficient multilevel logistic regression model for predictors of anemia among under-five children in Ethiopia, EDHS 2005–2016

The random coefficient (Model III) consists of both individual and contextual level variables in addition to cluster-specific random effects. The ICC was 0.0676 and variance was 0.2387039, with a standard error of.0235482, and 95% CI doesn’t contain 0, indicating that variation is significant.

The odds of anemia was higher among children between 6 to 11 and 12 to 23 months compared to children above two years (AOR = 3.35, 95%CI: 2.99–3.76, & AOR = 2.34, 95%CI: 2.15–2.55) respectively. Children with first and second in the birth order were 40% and 17% less likely to be anemic compared to those with 5 or above birth order (AOR = 0.60, 95%CI: 0.38–0.95, & AOR = 0.83, 95%CI: 0.73–0.93) respectively.

Children reported fever, and children with low height and weight for the age on anthropometric measurement were more likely to be anemic compared to their counterparts with no fever, and normal anthropometric measurement respectively. The finding also showed that as the age of a mother increases the odds of anemia in children decreases, and children from a mother with no access to media were more likely to be anemic compared to children from mothers with access to media.

Children from HHs with the poorest and poorer wealth quantile were more likely to be anemic compared to those the richest quantile (AOR = 1.672, 95%CI: 1.450–1.927, & AOR = 1.249, 95%CI: 1.079–1.445) respectively. Besides, the number of under-five children in the HHs has an effect on anemia, children from HHs with one to two number of children in a family had more than 17% less chance of getting anemia compared to children from HHs with three and more children (AOR = 0.830, 95%CI: 0.755–0.913) (model III, Table 7).

The goodness of fit of models in multilevel logistic regression

The appropriateness, adequacy, and usefulness of the models were tested using AIC, BIC, and Likely hood Ratio Test (LRT) tests. The empty model has the highest Deviance, LRT, and AIC; however, it is statistically significant indicating that a model with a random intercept is better than a model without a random intercept. The random coefficient model has the lowest AIC, BIC, and LRT values compared to the empty and random intercept models. This shows that the random coefficient model is the best fit model (Table 8).

thumbnail
Table 8. Multilevel logistic regression analysis on the prevalence of anemia among under-five children, model comparisons.

https://doi.org/10.1371/journal.pone.0249412.t008

Discussions

According to the WHO criteria, anemia is highly prevalent and considered a severe public health significance (defined as a prevalence higher than 40%) in most middle- and low-income countries [27]. Severe anemia has an increased risk of mortality and negative long-term consequences of damaging cognitive performance and motor development in children. Subsequently, this can result in impaired economic productivity and developments in the nations [2830].

The Sustainable Development Goal (SDGs) addresses anemia indirectly, the second goal is about ending hunger, aims to end all forms of malnutrition by 2030. In particular, the goal focuses to address the nutritional requirements of under-five children, adolescent girls, and mothers [31].

Despite the availability of integrated community-based child health care and various children focused prevention and intervention programs to avert nutritional disorders and childhood diseases in Ethiopia [3234], the findings of this study showed that the prevalence of anemia across the wave of the EDHS surveys was unacceptably higher.

The magnitude was 54.2% in 2005, and showed an insubstantial declining trend into 44.6% in 2011 but increased back to 57.6% in 2016. The finding also showed that there is a huge disparity in the prevalence of anemia across the country in each survey. Somalia regional state was the most affected region, the prevalence was 86.2% in 2005, 69.6% in 2011, and 83.3% in 2016. Afar regional state and Dire Dawa city administration were the second most affected regions, anemia higher than 56% in all of EDHS surveys between 2005 to 2016. The geospatial analysis supports these findings that most of the hotspots areas of anemia were located in the East, Northeast, and Western regions of the country, however, most of the low hotspot areas of anemia were located in the Central region, and South, Northwest and Northern parts of the country (Figs 24). The observed regional variability of anemia could be attributed to the regional differences in child feeding habits, infectious disease distributions, and availability and access to health care services [3537].

The age of children was a strong predictor of anemia both in the random intercept (Model 2) and random slope (Model 3). The odds of anemia decreases as the age of children increases; children whose age ranges within 6 to 11 months, and 12 to 23 months had more than three and two-fold risks of anemia compared to those with over 24 months respectively. This finding hasn’t supported the evidence that absorbed iron requirements increases with age, similar to other energy requirements in children [38], however, children above the age of one year commonly consume a variety of food sources rich in iron contents including meats, poultry, fish, cereals [39, 40]. In addition, the nutritional disorder is much higher in younger than older age children [41, 42]. Moreover, younger children are highly vulnerable to infectious diseases such as intestinal helminths as they might ingest contaminated materials into their mouths compared to older ones, particularly children living in an unsanitary environment [43, 44]. This finding is supported by similar other studies conducted across the globe that children less than two years of age were at higher risk of anemia [10, 11, 15].

In this study, children with first and second birth orders in a family were less likely to be anemic compared to those with fifth or above in the birth order. This could be an increase in the number of children associated with increased health problems due to competition for food, infections, and cross contaminations [4547]. This finding is supported by studies conducted in Uganda and Cameroon, the number of children in a family significantly associated with anemia in children [11, 48].

Birth interval showed an effect on anemia status, children with birth interval less than 24 months before their younger siblings were at higher risk of anemia compared to those with optimal intervals. This could be attributed to maternal nutritional depletion, vertical transmission of infections, suboptimal lactation due to pregnancy overlap, sibling competition for food, and transmission of infectious diseases among siblings [49]. Our finding is in line with studies conducted in African countries where birth spacing is associated with anemia in children [50, 51].

This study also showed that the odds of anemia was much higher among children who had a fever recently compared to those with no symptom. Even though the underlying cause of fever may be different, it could be due to systemic infections in the body that might affect the hemoglobin level in the blood [52, 53]. Evidence showed that fever can happen due to malaria or any other disease in a situation where both fever and anemia coexist [54, 55]. Similar studies conducted recently in Sub-Saharan African countries revealed that children who had a fever recently were more likely to be anemic compared to those with no fever [15, 56].

In this study, the odds of anemia among children with low height and weight for age on anthropometric measurements was much higher compared to those with normal anthropometric measurements. This might be due to the nutritional status of children directly affects the hemoglobin level in the blood [57, 58]; several studies conducted so far across the globe showed that stunting and underweight among children were associated with anemia [11, 5863].

Maternal age showed an effect with anemia status of their children, mothers whose age ranges between 15 to 24 years had more than 36% higher risk of anemia compared to those between 35 to 49 years. This could be due to low maternal age contributes to low birth weight (LBW) babies [6466], in turn, LBW in children might contribute to low hemoglobin level in the blood [6769].

This study also indicated that children from mothers with no media exposure had more than 30% higher odds of anemia compared to those from mothers with media exposure. This finding is supported by a study conducted in India, children from mothers with no media exposure were more likely to report anemia compared to their counterparts (63.6% vs. 53.5%) [70]. The reason could be mothers’ media exposure may affect childcare practices through enhancing the knowledge of mothers on child feeding activities, disease prevention practices, and improving health-seeking behaviors [71, 72]. Another study conducted in India showed that women who received health education were more likely to be familiar with anemia prevention practices compared to women who didn’t [73].

Socio-economic status of the households showed association with anemia, children from HHs with poorest and poorer wealth quantile had more than 67% and 45% chance of getting anemia compared to children from richest wealth quantile respectively. This could be due to HHs with higher wealth quintile are more likely to provide balanced macro and micronutrients (minerals and vitamins) to their children, and children from these HHs have more chance of accessing health care services. Several studies confirm that children from a lower economic status are vulnerable to various nutritional disorders including anemia, and at risk of easily preventable diseases [10, 15, 56, 74].

Moreover, the number of under-five children in the HHs has an effect on childhood anemia, children from HHs with two or less number of children had more than 17% less chance of getting anemia as compared to those from HHs with more number of children. This could be due to an increase in the number of children might lead to a risk of communicable disease transmission, and competition for food, consequently, nutritional deficiencies [10, 56]. This finding is supported by similar studies, that the number of children in the HHs associated with anemia [11, 15].

The potential strength of our study is the use of all available data (EDHS 2005 to 2016); this enables us to have a large sample size. In addition, the use of multilevel and geospatial analysis to handle the clustering effect of the data, and to have a better insight to locate the high and low hotspot areas of anemia across the country. Despite its strength, these findings should be interpreted considering its limitations. The temporal relationship between childhood anemia and explanatory factors can’t be established due to the cross-sectional nature of the source data. Exposure variables such as the presence of fever, and diarrhea before the survey were based on women’s self-report that could result in recall bias or varies according to the illness perception of the mother. Moreover, the birth size of children is taken from the subjective report of the mothers, which could have resulted in bias. Nevertheless, these biases are non-differential, as they are independent of the characteristics of women or children.

Conclusions

The findings of this study indicated that more than one-half of the children (51.5%) were anemic, and the prevalence of anemia sustainably higher than the expected level across the waves of the EDHS surveys from 2005 to 2016. The geospatial distribution of anemia among under-five children significantly varies across regions in Ethiopia; high hotspots of anemia were concentrated in the East, Northeast, and Western regions of the country. However, low hotspots were seen in most of the Central, South, North, and Northwest regions of the country.

The risk of anemia is highest in the first two years of life, therefore, the families, health care workers, and program planners on child health care should emphasize this critical period. Birth order and birth interval of children have a strong association with anemia, and the risk of anemia is much higher for higher-order births, and for whom birth interval is less than 24 months from their younger siblings. This demands actions by the government and concerned organizations to work further on family planning services to limit family size and integrate it with health promotion to have adequate birth spacing between siblings.

Besides, as the wealth index of the HHs improves from poorest to richest, the risk of anemia decreases significantly. Moreover, children from HHs with one or two number of under-five children had less probability of developing anemia compared to children from HHs with more children. Therefore, problem tailored interventions by government, regional health offices, and concerned organizations should work in harmony to avert the consequences of anemia in children.

Acknowledgments

We received the data from the Demographic and Health Surveys (DHS) Program. We like to thank the DHS office for grating us with the data.

We would like to acknowledge professor Annelies Van Rie & professor Joost Weyler for their unreserved constructive comments and suggestions to improve our study from the stages of development of concepts to the whole study period.

References

  1. 1. Bandyopadhyay Sheila, B GM, Francis Richard O., et al. Iron-deficient erythropoiesis in blood donors and red blood cell recovery after transfusion: initial studies with a mouse model Blood Transfus 2017;15(2):158–64. pmid:28263174
  2. 2. World Health Organization. Anaemia 2017, [19 Dec. 2019]. https://www.who.int/topics/anaemia/en/.
  3. 3. NJ K. The Global Burden of Anemia. Hematol Oncol Clin North Am. 2016;30(2):247–308. pmid:27040955
  4. 4. World Health Organization. Global Health Observatory data repository. Prevalence of anaemia in women 2016 [cited 2018 19 Dec.]. http://apps.who.int/gho/data/node.main.ANAEMIAWOMEN?lang=en.
  5. 5. Central Statistical Agency (CSA) [Ethiopia] and ICF. Ethiopia Demographic and Health Survey. Addis Ababa, Ethiopia: CSA and ICF; 2016.
  6. 6. USAIDS. Integrated Anemia Prevention and Control Toolkit 2018 [cited 2018 19 Dec.]. https://www.k4health.org/toolkits/anemia-prevention.
  7. 7. Larson LM P K, Pasricha SR. Iron and Cognitive Development: What is the Evidence? Annals of Nutrition & Metabolism 2017;71(3):25–38. pmid:29268256
  8. 8. GM CS. The Role of Nutrition in Brain Development: The Golden Opportunity of the "First 1000 Days" Journal of Pediatrics. 2016;175:16–21. pmid:27266965
  9. 9. Aimone Ashley Mariko P N, and Donald C A systematic review of the application and utility of geographical information systems for exploring disease-disease relationships in paediatric global health research: the case of anaemia and malaria International Journal of Health Geographics. 2013;12(1). http://www.ij-healthgeographics.com/content/12/1/1. pmid:23305074
  10. 10. Khan Jahidur Rahman Aa N M F. Determinants of anemia among 6–59 months aged children in Bangladesh: evidence from nationally representative data. BMC Pediatrics. 2016;16(3). pmid:26754288
  11. 11. Kuziga Fiona Aa Y W RK. Prevalence and factors associated with anaemia among children aged 6 to 59 months in Namutumba district, Uganda: a cross- sectional study BMC Pediatric. 2017;17:25. pmid:28100200
  12. 12. Ewusie Joycelyne E A C, Beyene Joseph et al. Prevalence of anemia among under-5 children in the Ghanaian population: estimates from the Ghana demographic and health survey. BMC Public Health 2014;14:626. pmid:24946725
  13. 13. Peter A. M. Ntenda ON, Bass Paul and Senghore Thomas. Maternal anemia is a potential risk factor for anemia in children aged 6–59 months in Southern Africa: a multilevel analysis. BMC Public Health 2018;18:650. pmid:29788935
  14. 14. Prieto-Patron Alberto VdH K, Hutton Zsuzsa V. et al. Association between Anaemia in Children 6 to 23 Months Old and Child, Mother, Household and Feeding Indicators. Nutrients 2018;10(1269). pmid:30205553
  15. 15. Moschovis PP W M, Arlington L et al. Individual, maternal and household risk factors for anaemia among young children in sub-Saharan Africa: a crosssectional study. BMJ Open 2018;8:e019654. pmid:29764873
  16. 16. Central Statistical Agency (CSA) [Ethiopia] and ICF. Ethiopia Demographic and Health Survey(EDHS). Addis Ababa, Ethiopia: CSA and ICF; 2011.
  17. 17. Central Statistical Agency (CSA) [Ethiopia] and ICF. Ethiopia Demographic and Health Survey. Addis Ababa, Ethiopia: CSA and ICF; 2005.
  18. 18. Federal Republic of Ethiopia. The Regional States of Ethiopia Addis Ababa2018 [April 10, 2019]. http://www.ethiopia.gov.et/regional-states1.
  19. 19. Getaneh Z E B, Engidaye G, Seyoum M, Berhane M, Abebe Z, et al. Prevalence of anemia and associated factors among school children in Gondar town public primary schools, northwest Ethiopia: A school-based cross-sectional study. PLoS ONE. 2017;12(12). pmid:29284032
  20. 20. Gutema B A W, Asress Y, Gedefaw L. Anemia and associated factors among school-age children in Filtu Town, Somali region, Southeast Ethiopia. BMC hematology 2014;14(2014):13. pmid:25170422
  21. 21. Melku Mulugeta A KA, Terefe Betelihem and et al. Anemia severity among children aged 6–59 months in Gondar town, Ethiopia: a community-based cross-sectional study. Italian Journal of Pediatrics 2018;44(1):107. pmid:30176919
  22. 22. World Health Organization. Water sanitation hygiene 2012, [12 April, 2019]. https://www.who.int/water_sanitation_health/monitoring/jmp2012/key_terms/en/.
  23. 23. Kamal MM, Hasan MM, Davey R. Determinants of childhood morbidity in Bangladesh: evidence from the Demographic and Health Survey 2011. BMJ Open. 2015;5(10):e007538. pmid:26510724
  24. 24. World Health Organization. WHO child growth standards: WHO; 2006 [cited 2019 20 June, 2019]. https://www.who.int/publications/i/item/924154693X.
  25. 25. The DHS Program. Spatial data repository. Subnational regional boundaries. 2005–2016 [cited 2020 October 25]. http://spatialdata.dhsprogram.com/boundaries/#view=table&countryId=ET.
  26. 26. Hosmer D L S. multicolinearity diagnostics. In: Noel A.c Cressie NlFea, editor. Applied logistic regression. United States of america: Walter A Shewahrt and samuel S. Wliks; 2000.
  27. 27. World Health Organization. Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity. Vitamin and Mineral Nutrition Information System Geneva: World Health Organization; 2011.
  28. 28. Dewey ELPKG. Nutrition and brain development in early life Nutrition Reviews. 2014;72(4):267–84. pmid:24684384
  29. 29. Olness K. Effects on Brain Development Leading to Cognitive Impairment: A Worldwide Epidemic. Journal of Developmental & Behavioral Pediatrics. 2003;24(2):120–30. pmid:12692458
  30. 30. Sood JBaA. Iron Deficiency Anemia: Effect on cognitive development in children: a review. Indian Journal of Clinical Biochemistry. 2005;20(2):119–25. pmid:23105543
  31. 31. United Nations. Sustainable Development Goals (SDGs). Goal 2: End hunger, achieve food security and improved nutrition and promote sustainable agriculture 2015 [April 20, 2019]. https://unstats.un.org/sdgs/report/2016/goal-02/.
  32. 32. USAID. Maternal and child Survival program, Community-Based Care in Ethiopia November 2018 [April 23, 2019]. https://www.mcsprogram.org/resource/community-based-care-in-ethiopia/.
  33. 33. USAID. Maternal, Neonatal and child health April 22, 2019 [April 23, 2019]. https://www.usaid.gov/ethiopia/global-health/maternal-and-child-health.
  34. 34. H B. Ethiopia’s health extension program: improving health through community involvement. MEDICC Rev 2011;13(3):46–9. pmid:21778960
  35. 35. World Health Organization. Analytical summary—Health system outcomes for Ethiopia 2019 [updated 2018May 6, 2019]. http://www.aho.afro.who.int/profiles_information/index.php/Ethiopia:Analytical_summary_-_Health_system_outcomes#cite_note-ten-0.
  36. 36. Betebo B E T, Alemseged F, Massa D. Household Food Insecurity and Its Association with Nutritional Status of Children 6–59 Months of Age in East Badawacho District, South Ethiopia. J Environ Public Health. 2017;2017. pmid:28408936
  37. 37. Woldemichael A T A, Akbari Sari A, et al. Inequalities in healthcare resources and outcomes threatening sustainable health development in Ethiopia: panel data analysis. BMJ Open. 2019;9(1). pmid:30705237
  38. 38. Institute of Medicine (US) Panel on Micronutrients. Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc. 9 >Washington (DC): National Academies Press (US); 2001.
  39. 39. Queensland Government. Diet and Eating. Iron for toddlers and children 1995–2020 [updated 25/09/2017 cited 23 october, 2020]. http://conditions.health.qld.gov.au/HealthCondition/condition/8/78/418/iron-for-toddlers-and-children.
  40. 40. UNICEF. The State of the World’s Children 2019. Children, Food and Nutrition: Growing well in a changing world. 2019.
  41. 41. Shaza O H Kanan MOS. Prevalence and outcome of severe malnutrition in children less than ive-year- old in Omdurman Paediatric Hospital Sudan Sudanese Journal of Paediatrics. 2016;16(1).
  42. 42. WHO. World health Organization fact sheet. Infant and young child feeding 16 February 2018 [April 1, 2019]. https://www.who.int/news-room/fact-sheets/detail/infant-and-young-child-feeding.
  43. 43. Jill E. Weatherhead PJH. Worm Infections in Children. Pediatrics in Review. 2015;36(8).
  44. 44. World Health Organization. Deworming in children 2017 [updated 11 February 2019; cited 2019 April 20]. https://www.who.int/elena/titles/deworming/en/.
  45. 45. Feleke BE. Nutritional Status and Intestinal Parasite in School Age Children: A Comparative Cross-Sectional Study. Int J Pediatr. 2016;2016. pmid:27656219
  46. 46. House T K M. Household structure and infectious disease transmission. Epidemiol Infect. 2008;137(5):654–61. pmid:18840319
  47. 47. Nesti MMM, & Goldbaum Moisés. Infectious diseases and daycare and preschool education. Jornal de Pediatria. 2007;83(4):299–312. pmid:17632670
  48. 48. Kana Sop Marie Modestine M M-J, Tetanye Ekoe and Gouado Inocent. Risk factors of anemia among young children in rural Cameroon. International Journal of Current Microbiology and Applied Sciences. 2015;4(3):925–35.
  49. 49. Conde-Agudelo A ea. Effects of Birth Spacing on Maternal, Perinatal, Infant, and Child Health: A Systematic Review of Causal Mechanisms 2012 [April 2, 2019]. https://www.k4health.org/toolkits/anemia-prevention/effects-birth-spacing-maternal-perinatal-infant-and-child-health.
  50. 50. Afeworki R SJ , Tolboom J, van der Ven A. Positive Effect of Large Birth Intervals on Early Childhood Hemoglobin Levels in Africa Is Limited to Girls: Cross-Sectional DHS Study. PLoS One. 2015;10(6). pmid:26121362
  51. 51. Rana MJ, Goli S. Family Planning and Its Association with Nutritional Status of Women: Investigation in Select South Asian Countries. Indian Journal of Human Development. 2017;11(1):56–75.
  52. 52. World Health Organization. Humanitarian Health Action. Communicable diseases fact sheet 2019 [April 21, 2019]. https://www.who.int/hac/techguidance/ems/flood_cds/en/.
  53. 53. Centers for Disease Control and Prevention. Malaria, disease 2019 [April 24, 2019]. https://www.cdc.gov/malaria/about/disease.html.
  54. 54. MB V. Anemia and infection: a complex relationship. Rev Bras Hematol Hemoter. 2011;33(2):90–2. pmid:23284251
  55. 55. Charles Patrick Davis. Anemia Symptoms and Signs, Types, Treatment and Causes. What Is Anemia? 2017 [April 24, 2019]. https://www.onhealth.com/content/1/anemia_causes_treatments.
  56. 56. Habyarimana Faustin Za T R S. Structured Additive Quantile Regression for Assessing the Determinants of Childhood Anemia in Rwanda. Int J Environ Res Public Health 2017;14:652. www.mdpi.com/journal/ijerph. pmid:28629151
  57. 57. Osazuwa F AO. Contribution of malnutrition and malaria to anemia in children in rural communities N Am J Med Sci 2010;2(11). pmid:22558561
  58. 58. Yang W L X, Li Y, et al. Anemia, malnutrition and their correlations with socio-demographic characteristics and feeding practices among infants aged 0–18 months in rural areas of Shaanxi province in northwestern China: a cross-sectional study. BMC Public Health. 2012;12:1127. pmid:23273099
  59. 59. da Silva LLS F W, Cardoso MA; ENFAC Working Group. Factors associated with anemia in young children in Brazil. PLoS One. 2018;13(9). pmid:30252898
  60. 60. Gebreegziabiher G E B, Niggusie D. Gebreegziabiher G, Etana B, Niggusie D. Determinants of Anemia among Children Aged 6–59 Months Living in Kilte Awulaelo Woreda, Northern Ethiopia. Anemia. 2014;2014. pmid:25302116
  61. 61. Huynh G H Q, Nguyen NHT, Do QT, Khanh Tran V. Malnutrition among 6-59-Month-Old Children at District 2 Hospital, Ho Chi Minh City, Vietnam: Prevalence and Associated Factors. Biomed Res Int. 2019;2019. pmid:30868070
  62. 62. Malako BG, Asamoah BO, Tadesse M, Hussen R, Gebre MT. Stunting and anemia among children 6–23 months old in Damot Sore district, Southern Ethiopia. BMC Nutrition. 2019;5(1):3. pmid:32153918
  63. 63. Woldie H, Kebede Y., & Tariku A. Factors Associated with Anemia among Children Aged 6–23 Months Attending Growth Monitoring at Tsitsika Health Center, Wag-Himra Zone, Northeast Ethiopia. Journal of Nutrition and Metabolism. 2015;2015:9. pmid:26106486
  64. 64. Aras R. Is maternal age risk factor for low birth weight? Archives of Medicine and Health Sciences. 2013;1(1):33–7.
  65. 65. Dennis JA M S. Young maternal age and low birth weight risk: An exploration of racial/ethnic disparities in the birth outcomes of mothers in the United States. Soc Sci J. 2013;50(4):625–34. pmid:25328275
  66. 66. Restrepo-Méndez MC L D, Horta BL, et al. The association of maternal age with birthweight and gestational age: a cross-cohort comparison. Paediatr Perinat Epidemiol. 2014;29(1):31–40. pmid:25405673
  67. 67. Ferri C, Procianoy RS, Silveira RC. Prevalence and Risk Factors for Iron-Deficiency Anemia in Very-Low-Birth-Weight Preterm Infants at 1 Year of Corrected Age. Journal of Tropical Pediatrics. 2013;60(1):53–60. pmid:24044971
  68. 68. Kejo D P P, Martin H, Kimanya ME, Mosha TC. Prevalence and predictors of anemia among children under 5 years of age in Arusha District, Tanzania Pediatric Health Med Ther. 2018;9(1):9–15. pmid:29443328
  69. 69. Mohammed SH, Habtewold TD, Esmaillzadeh A. Household, maternal, and child related determinants of hemoglobin levels of Ethiopian children: hierarchical regression analysis. BMC Pediatrics. 2019;19(1):113. pmid:30987632
  70. 70. Baranwal Annu Ba A R N. Association of household environment and prevalence of anemia among children under-5 in India. FrontiersinPublicHealth. 2014;2(1). pmid:25368862
  71. 71. Tassew AA T D, Belachew AB, Adhena BM. Factors affecting feeding 6–23 months age children according to minimum acceptable diet in Ethiopia: A multilevel analysis of the Ethiopian Demographic Health Survey. PLOS ONE. 2019;14(2):e0203098. pmid:30789922
  72. 72. Menberu Molla T E, and Girma Nega. Associated Factors among Mothers Having Children 6–23 Months of Age, Lasta District, Amhara Region, Northeast Ethiopia Advances in Public Health. 2017;2017.
  73. 73. Pernilla Ny E D-K, Giggi Udén & Ted Greiner. Health Education to Prevent Anemia Among Women of Reproductive Age in Southern India. Journal Health Care for Women International. 2006;27(2):131–44. pmid:16484158
  74. 74. Iglesias Lucía Vázquez EV, Marcela Villalobos et al. Prevalence of Anemia in Children from Latin America and the Caribbean and Effectiveness of Nutritional Interventions: Systematic Review and Meta–Analysis. Nutrients 2019;11(183). pmid:30654514