Household, maternal, and child related determinants of hemoglobin levels of Ethiopian children: hierarchical regression analysis

Anemia, marked by a low hemoglobin (Hb) level, continues to be a significant public health concern affecting almost a third of the world’s population. Infants and young children are of particular concern, developing anemia at a higher rate and bearing the highest burden [1]. In 2016, anemia prevalence among children under five years old in Ethiopia was 57%, rising unexpectedly from 44% in 2011 [2]. Infants and young children bear the highest burden of anemia in Ethiopia, with a 72% prevalence of anemia among those under two years of age [2]. The World Health Organization (WHO) classifies anemia prevalence above 40% as a severe public health problem [3].

Anemia is a multi-causal problem with a number of dietary and non-dietary risk factors [1, 4]. Food items with high phytate and polyphenol contents are associated with a high risk of anemia. Inadequate dietary or supplemental intake of iron, folate, and vitamin A often leads to anemia [4, 5]. While iron deficiency has long been considered the single greatest factor contributing to anemia, accounting for almost 50% of anemia globally [4, 5], recent reports suggest that iron deficiency is not as significant a culprit as was once thought [6]. Its contribution is particularly low in countries with high anemia and inflammation burdens, where it is estimated to account for 14 and 20% of the burden of anemia among preschool children, respectively [6]. Intestinal parasites, malaria, and infection are also among the main immediate causes of anemia, particularly in developing countries [4, 6]. Chronic illness or inflammatory conditions increase expression of hepcidin hormone, which reduces the absorption of iron by enterocytes and its exportation by ferroportin, thereby increasing the risk of anemia [7]. The main underlying conditions leading to anemia in developing countries are suboptimal feeding, caring and hygiene practices, coupled with poor health care. Poor socioeconomic status is one of the basic determinants of anemia [5, 8, 9].

Reducing the burden of anemia is one of the six global nutrition targets outlined by the WHO for the period 2012–2025 [10]. In Ethiopia, some interventions have been put in place to address the burden of anemia. These include distribution and promotion of the use of insecticide-treated mosquito nets, deworming and iron supplementation, and school- and community-based nutrition interventions [11]. While some studies are available on the determinants of anemia or Hb level in Ethiopia, most studies did not account for the hierarchical nature and interrelationships among the multilevel determinants [12, 13]. Their estimates were mainly based on single model regression analyses, which could be problematic. For example, the distal determinants of Hb level, like community and household factors, influence not only Hb level directly but also its underlying and proximal determinants like breastfeeding and dietary practices. Thus, including all variables in one model, a practice in most of the existing studies, may nullify or weaken the relation of the distal factors with Hb level [14]. Besides, given the recent increase in the prevalence of anemia in Ethiopia [2] and the time-varying nature of the contextual determinants, it stands timely and necessary to further investigate the determinants of Hb level. We used Hb level on a continuous scale to avoid the problem of potential statistical power loss due to dichotomization into anemic and non-anemic groups [15]. The use of Hb level on a continuous scale also enables to evaluate the relation of the determinant factors with the full spectrum of Hb level, not just with the state of anemia. Thus, in this study, we aimed to investigate the various household, maternal, and child-related dietary and non-dietary factors influencing Hb level of Ethiopian children aged 6–23 months using the latest nationally representative demographic and health survey, EDHS 2016.

In this work, we included a total of 2902 children aged 6–23 months, of which 1359 (46.83%) were boys and 1543 (53.17%) girls. The majority of study participants were from rural areas (89.22%). The mean age (± SD) was 14.01 ± 5.02 months. The majority of children were from middle- and low- income households (67.22%). The mean Hb level (± SD) was 10.00 ± 1.63 g/dL. The overall prevalence of anemia (Hb level < 11 g/dL) among the study population was 71.92%.

Table 1 shows the results of the bivariable analyses of the relation of the household and the maternal factors with Hb level. The household-related factors found significantly associated with a higher mean Hb level were living in urban areas, agrarian regions, and households of high wealth category and improved water supply. Among the maternal characteristics, age, anemia status, education level, and ANC visits were significantly associated with Hb level (P < 0.05). Toilet facility and maternal BMI were not significantly associated with Hb level during the bivariate analyses (P > 0.05).

The results of the bivariable analyses of the relationship of the dietary and non-dietary child-related factors with Hb level are shown in Table 2. The child factors found significantly associated with Hb level (P < 0.05) were sex, age, birth size, height-for-age, weight-for-height, weight-for-age, history of infection, and current breastfeeding status. Early initiation of breastfeeding, deworming medication use in the last six months, iron supplement use in the last seven days, sex, size at birth, MDD, and MMF were not found significantly associated with Hb level (P > 0.05). These estimates were, however, crude and less informative, i.e. not adjusted for any covariate factor.

The results of the hierarchical linear regression analyses between the predictor variables and Hb level are shown in Table 3. Residence in agrarian regions was associated with a significantly higher mean Hb level (aβ = 0.64, 95%CI = 0.40–0.88, P < 0.001), compared with residence in pastoral regions. Compared with children in the poor wealth category, mean Hb level was significantly higher in those in middle (aβ = 0.42, 95%CI = 0.27–0.58, P < 0.001) and rich (aβ = 0.48, 95%CI = 0.33–0.63, P < 0.001) categories. Mean Hb level was significantly higher in children of mothers with secondary and above education level (aβ = 0.69, 95%CI = 0.23–1.16, P = 0.004), compared with the value of children of illiterate mothers. Compared with children of anemic mothers, Hb level was significantly higher in children of non-anemic mothers (aβ = 0.40, 95%CI = 0.26–0.53, P < 0.001).

Being a girl was associated with a significantly higher mean Hb level (aβ = 0.16, 95%CI = 0.03–0.30, P = 0.019), compared with being a boy. Mean Hb level was significantly higher in children under 12 months of age than in those above 12 months of age (aβ = 0.72, 95%CI = 0.57–0.88, P < 0.001). Children born with an average birth size had a significantly higher mean Hb level (aβ = 0.25, 95%CI = 0.08–0.42, P = 0.003), compared with those born with small birth size. Mean Hb level was significantly higher in the non-wasted children (weight-for-age ≥ − 2) (aβ = 0.22, 95%CI = 0.02–0.42, P = 0.031) than in the wasted children. Children who had no history of infection in the last two weeks before the survey had a significantly higher mean Hb level (aβ = 0.18, 95%CI = 0.02–0.33, P = 0.025). Currently breastfeeding was associated with a significantly higher mean Hb level (aβ = 0.28, 95%CI = 0.12–0.44, P = 0.002), compared with not currently breastfeeding. Children who received vitamin A supplementation in the last six months had a significantly higher mean Hb level (aβ = 0.17, 95%CI = 0.06–0.28, P = 0.021), compared to those who did not receive the supplementation.

This study was aimed to determine the household, maternal and child factors influencing the Hb level of infants and young children in Ethiopia. We found a high level of anemia. After adjusting for covariates, the household factors found to be associated with Hb level were region of residence and household wealth category. Maternal education level and anemia status were significantly associated with Hb level of children. Sex, birth size, weight-for-age, history of infection, and duration of breastfeeding were the child factors found significantly associated with Hb level.

The high prevalence of anemia we reported was consistent with the report of EDHS 2016, which showed 78, 76, 72, and 66% of anemia prevalence among 6–8, 9–11, 12–17, and 18–23 months old children, respectively [2]. This study showed a lower mean Hb in pastoral regions, compared with the agrarian regions. This might be, in part, due to the high prevalence of malaria in pastoral regions of Ethiopia [19]. Malaria is one of the main risk factors of anemia [5]. Besides, pastoral communities depend on animal milk as a main food source. The low bioavailability of iron in milk could also account for the low Hb level in these communities. Income was a significant and independent predictor of Hb level. Children of poor households had a lower mean Hb level, compared with those of rich households. The result was consistent with previous studies that showed a higher risk of anemia in people with low socioeconomic status [8, 10] and it could be due to the fact that health-enhancing practices and options are often limited among the poor. Children of anemic mothers were more likely to have a lower mean Hb level, compared with those of non-anemic mothers. The result was in agreement with the reports of previous studies [13, 20]. Breast milk is the main source of nutrients during the early stages of life. Thus, the association of maternal anemia with the Hb level of the child could be, in part, due to the low nutrients content of the breast milk of anemic mothers [21].

Weight-for-age was significantly associated with Hb level independent of other factors. The mean Hb of children who were not underweight (weight-for-age ≥ − 2 Z-score) was significantly higher, compared with that of underweight children. Compared with children of low birth weight (small birth size), children of average birth size demonstrated a higher mean Hb level. Previous studies also found child undernutrition associated with anemia [5, 20]. No history of infection, as measured by no cough or fever in the last two weeks prior to the survey, was associated with a significantly higher mean Hb level. The result was in concordance with other studies which reported higher risks of anemia or undernutrition to be associated with infection [5, 22‐24].

In this study, iron supplementation and deworming did not demonstrate significant associations with Hb level. At first glance, our findings might appear to conflict with public health recommendations [3, 4]. Iron is essential for the production of red blood cells [3, 21] and Hb level could be presumed to be higher in children who took iron supplements. However, some studies have shown that micronutrient interventions are less effective unless implemented in an integrated approach [25, 26]. The lack of association between these micronutrient interventions and Hb level could be due to some possible reasons like 1) in the DHS program, dose, frequency, and adherence to these interventions were not assessed, albeit they were reported to influence Hb responses to anemia prevention and control interventions [24, 25, 27]. Thus, the lack of accounting for these factors might have influenced our results, 2) the children on the micronutrient supplementations might be the ones who were already anemic and did not yet recover from the state of deficiency, 3) the high level of multiple micronutrient deficiency in Ethiopia [28, 29] could also have influenced the Hb response to iron supplementation as reported by a study in Mexico [25], or 4) it could be due to statistical power inadequacy because the data showed that only a small number of children took the iron supplement as well as the deworming medications.

Our finding was in concordance with previous reports that showed Hb level to be influenced by multiple factors, originating from individual, household, and community levels [5, 10]. Prevention of childhood anemia remains among the major public health agendas in most of the developing countries [1], including Ethiopia [11]. The WHO has outlined essential public health and nutrition interventions to reduce the burden of childhood anemia [3, 21]. Most of the WHO anemia prevention and control recommendations have been incorporated in the Ethiopians National Nutrition Programs and Strategies [11], albeit the implementation has been sub-optimal, fragmented, and mainly focused on addressing the immediate causes [2, 11]. Thus, it stands important for policy makers in Ethiopia to enhance the implementation of the existing anemia prevention and control interventions in a multi-sectoral and comprehensive approach. The determinants of Hb level, particularly the contextual factors, vary significantly across geographical regions. The relative contribution of the determinants to the burden of anemia is also contextual [6]. For example, recent reports have shown that the contribution of iron deficiency to the burden of anemia in developing countries is not as high as the widely held presumption that it accounts for almost 50% of the anemia burden [6]. Thus, it is worthy of conducting further studies considering the geospatial variations in determinants of Hb level and also design locally sensitive interventions.

One of the strengths of this study was the use of a nationally representative sample and objectively measured biomarkers: Hb level, weight, and height. The use of multilevel analysis scheme which took into account the hierarchical nature of the relationships among the various influences of Hb level might have avoided the nullifying effect of the intermediate factors on the association of the distal factors and Hb level. The inclusion of various explanatory factors (child, maternal, and household variables) might have improved the comprehensiveness of the study and enabled adjustment for various potential confounding variables. However, this study has important limitations. First, the design of the study, cross-sectional, precludes establishing temporal relations and making causal inferences. Second, the collection of data on some of the variables based on respondents’ memory of past events might have introduced recall bias. Third, the lack of data on intestinal helminths, reported to influence Hb level, could be another limitation of this study. Fourth, we did not consider hereditary anemias in the study due to the lack of data in the dataset we used. However, it is less likely that our findings were affected by the lack of inclusion of hereditary anemias because the incidence of hereditary anemias is generally low in Ethiopia [30]. Fifth, we did not account for the effects of dose, duration, and adherence of deworming, iron, and vitamin A supplementation, which might have biased our findings. Despite these limitations, we believe the findings of this study could serve as evidence basis for further studies on the determinants of Hb level of children in Ethiopia.

Hb level of infants and young children in Ethiopia was found to be influenced by various household, maternal and child related factors. Designing and scaling up comprehensive nutrition interventions, with due emphasis on the multifactorial nature of Hb, may represent a potential consideration to reduce the burden of anemia in Ethiopia.