Using structural equation modelling to understand the contributors to anaemia among young Burkinabe children

Abstract

Anaemia is a persistent problem among young Burkinabe children, yet population-specific information on its determinants is scant. We used baseline data from an evaluation of Helen Keller International's Enhanced Homestead Food Production Program (n=1210 children) to quantify household-, mother-, and child-level factors associated with anaemia in Burkinabe children aged 6-12 months. We used structural equation modelling to assess a theoretical model, which tested four categories of factors: (a) household food security and dietary diversity, (b) household sanitation and hygiene (latrine and poultry access and bednet ownership), (c) maternal factors (anaemia, stress, cleanliness, and health, hygiene and feeding knowledge and practices), and (d) child nutrition and health (iron deficiency (ID), retinol binding protein (RBP), malaria, and inflammation). The model also included household socio-economic status, size, and polygamy; maternal age and education; and child age and sex. Results showed that ID, malaria, and inflammation were the primary direct determinants of anaemia, contributing 15%, 10%, and 10%, respectively. Maternal knowledge directly explained improved child feeding practices and household bednet ownership. Household dietary diversity directly explained 18% of child feeding practices. Additionally, RBP, child age and sex, and maternal anaemia directly predicted child haemoglobin. Our findings suggest that program effectiveness could be increased by addressing the multiple, context-specific contributors of child anaemia. For young Burkinabe children, anaemia control programs that include interventions to reduce ID, malaria, and inflammation should be tested. Other potential intervention entry points suggested by our model include improving maternal knowledge of optimal health, hygiene, and nutrition practices and household dietary diversity.

Keywords: Burkina Faso; anaemia; children; determinants of anaemia; iron deficiency.

Figure 1

Conceptual model of the child‐, mother‐ and household‐level factors predicting anaemia among children 6‐12 months of age. Mother's anaemia was defined as Hb concentration <11 g/dl in pregnant mothers and mothers with missing pregnancy status and <12 g/dl in non‐pregnant mothers. Abbreviations used: Hb: haemoglobin; ID: iron deficiency; sTfR: soluble transferrin receptor; RBP: retinol binding protein; RDT: rapid diagnostic test; CRP: C‐reactive protein; AGP: α‐1‐acid glycoprotein; SRQ: self‐reported questionnaire; SES: socio‐economic status. Not depicted in the diagram are disturbance terms and their variances, the variances of exogenous variables, the covariances between exogenous variables, the scaling constant for the direct paths from disturbance terms to endogenous variables, and the covariance between the disturbance terms of “RBP (umol/l)” and “ID (sTfR>9.3 mg/l)”

Figure 2

Direct effects of child‐, mother‐, and household‐level factors predicting anaemia among children 6‐12 months of age along the hypothesized pathways. All values are standardized direct effects (SDE). P‐values were corrected for multiple comparisons. Significance level: *p<.014. Abbreviations used: Hb: haemoglobin; ID: iron deficiency; sTfR: soluble transferrin receptor; RBP: retinol binding protein; RDT: rapid diagnostic test; CRP: C‐reactive protein; AGP: α‐1‐acid glycoprotein; SRQ: self‐reported questionnaire; SES: socio‐economic status. Direct effects of child age and sex were not included in the figure. Direct effects of non‐modifiable factors were not included in the figure.