To augment existing literature with more diverse methodologies, we took a more holistic approach than prior reviews of stunting reduction in Ethiopia, triangulating across literature reviews, quantitative analyses, and qualitative inquiry.

Literature review

We initiated our research process with a systematic literature review of contextual factors, national and subnational interventions, policies, strategies, programs, and other initiatives that may have theoretically contributed to reductions in child stunting over time. Broad searches, followed by de-duplication and predefined exclusions, abstract and full-text reviews resulted in a total of 102 records with information pertinent to child stunting in Ethiopia. Targeted additional searches were completed for key topics and a total of 402 additional documents spanning grey literature and published peer-reviewed reports were collated and summarized.

Quantitative analysis

Quantitative methods involved (1) descriptive analyses to provide contextual understanding of the stunting decline across geographic, socioeconomic, gender, and age segments and (2) hierarchical multivariable regression and regression-based decomposition analyses to understand the major predictors of the stunting decline, as well as their relative importance to Senegal’s progress. Ethiopia’s series of Demographic and Health Surveys (DHS) (2000, 2005, 2011, 2016) were the primary quantitative datasets used in this study. These nationally-representative household surveys have comparable, standardized methods with data for a wide range of indicators in the areas of population, health, and nutrition. Ethiopia had standard DHS completed in each of the years studied, and we felt confident that the data collected did not suffer from anthropometry quality concerns as has been the case for other surveys. Thus, we used all available Ethiopia DHS for our study.

We organized available variables for analysis into a conceptual framework adapted from the UNICEF / Lancet nutrition series [link to framework below] undernutrition conceptual framework, which groups determinants of stunting into their causal hierarchical levels.

UNICEF / Lancet nutrition series undernutrition conceptual framework

Framework for the relations between poverty, food security, and other underlying and immediate causes to maternal and child undernutrition and its short-term and long-term consequences.
Framework for the relations between poverty, food security, and other underlying and immediate causes to maternal and child undernutrition and its short-term and long-term consequences

Descriptive analysis

To explore geospatial within-region stunting variation, we utilized 5x5 km area stunting estimates modeled by the Institute of Health Metrics and Evaluation (IHME). IHME used Ethiopia DHS datasets incorporated into Bayesian spatial models to generate posterior predicted prevalence of stunting. The model draws strength from covariables, years, and locations where data is available.

To examine inequalities across population subgroups, standardized and well-established methods were used to estimate stunting in Ethiopia across important subnational dimensions: wealth quintile, level of maternal education, area of residence (urban or rural), child gender, and geographic region (administrative and broader regions).1  2  3

Finally, to examine the growth faltering process from birth to five years of age, we estimated child growth curves also known as Victora curves. Smoothed polynomial curves were used to plot predicted child height-for-age Z-scores against child age to assess growth trajectories and gain an understanding of how stunting risk changes with age.

Linear multivariable regression (difference-in-difference analysis)

The linear multivariable regression is based on a difference-in-difference analysis framework to evaluate if a change in a proposed predictor of HAZ leads to a change in HAZ over the studied time period. We analyzed individual-level data from four rounds of Ethiopia’s DHSs: 2000, 2005, 2011, 2016 for children under five years of age.

To examine the association between HAZ and various indicators, we conducted a series of step-wise linear regression models. A hierarchical modelling approach using distal, intermediate and proximal level variables was executed as suggested by Victora 1997 to generate the final multivariable models. Variables within each level were selected from our general conceptual framework.

Step 1 was a series of bivariate regressions to determine crude associations between indicators in our conceptual framework and HAZ outcome. Step 2 was to use all candidate variables for multivariable model building (i.e., with p-value =0.20) irrespective of their direction to move forward for multivariable modeling. Bivariate correlations estimate the absolute crude associations between the covariable and the outcome, and they highlight the total (unadjusted) effect of the factor on HAZ. In multivariable analysis, the final multivariable regression coefficient is adjusted for child age, sex and region (control variables) and all confounders in preceding levels.

Oaxaca-Blinder decomposition

Our Oaxaca-Blinder decomposition analysis (link with data and analysis page) is based largely on individual and household-level data, with larger sample sizes and higher statistical power. We focused on index mother-child pairs (i.e., the youngest child and youngest mother in any given household). This standard approach simplifies modeling and interpretation with minimal loss in data. Again, we analyzed the 2000-2016 time period and applied a similar hierarchical modeling approach as used for the linear mixed effects model.

We selected height-for-age Z-score (HAZ) rather than child stunting (i.e., yes or no) as the outcome of interest for greater statistical efficiency. The analysis was stratified by child age group (this time, just the under-five and 24-59 month age groups). For the under-five group, we also looked at the time period 1992/93-2005.

Importantly, the Oaxaca-Blinder decomposition allowed us to identify the relative contribution of each predictive factor to HAZ change. We did this by using a linear least square regression model, accounting for survey design and weights, to assess associations between HAZ, time, control variables (i.e., child age and sex), and any trend effects.

Qualitative analysis

Focus group discussion and in-depth interviews

To investigate a diverse range of stakeholder perspectives (e.g. national experts, mothers, childcare workers, etc.), we conducted a series of in-depth interviews with national informants and regional resource personnel, followed by a set of focus group discussions with mothers.

A total of 11 key national informants were interviewed in Addis Ababa with 12 regional personnel interviews conducted in two districts of SNNP region (Bonga and Yeki) and two districts of Somali region (Harshin and Aware). These interviews targeted resource persons with several functions: teachers and health staff (health extension workers, maternal, newborn and child health care workers, district health surveillance focal persons, and senior health centre staff). Respondents were asked to describe key nutrition-specific and -sensitive events in Ethiopia as well as comment on key trends and contextual factors that impacted child undernutrition over time.

12 focus group discussions were also conducted in three groups per district, across the two regions. In each region, three focus groups were held with mothers who gave birth between 1987-1991, three focus groups with mothers who gave birth between 2003-2007, and three focus groups with younger women who gave birth between 2011-2015, commenting on topics including, but not limited to pregnancy and breastfeeding practices, hygiene and child nutrition.

Policy and program review

To understand implementation of policies, programs, and strategies, we conducted additional research and corroborated our findings with country experts. A similar multi-pronged data collection and corroboration exercise was undertaken to track financial data linked to the nutrition policy and program timeline.

  1. 1
    Hosseinpoor AR, Bergen N, Barros AJD, et al. Monitoring subnational regional inequalities in health: measurement approaches and challenges. Int J Equity Health.
  2. 2
    Restrepo-Méndez MC, Barros AJ, Black RE, Victora CG. Time trends in socio-economic inequalities in stunting prevalence: analyses of repeated international surveys. Public Health Nutr. 2014;18:2097-104
  3. 3
    Fink G, Victora CG, Harttgen K, et al. Measuring Socioeconomic Inequalities with Predicted Absolute Incomes Rather Than Wealth Quintiles: A Comparative Assessment Using Child Stunting Data from National Surveys. Am J Public Hea. 2017;107:550-5.

Resources