Determinants of stunting among under-five children in Amhara and Oromia regions, Ethiopia: the linear quantile regression analysis.

BACKGROUND: Stunting in Ethiopia slightly decreased from 38% in 2016 to 34.4% in 2022. However, with this rate the country still falls behind the targets of the Sustainable Development Goals for 2030. The underlying unobserved factors-whether constant overtime or/and evolving due to random shocks have not yet been identified. Therefore, the current study aimed to identify the determinants of stunting among under-five children by accounting for both time-invariant and time-varying random effects. METHODS: Data for this study was obtained from the International Food Policy Research Institute. It was collected through three annual follow-up surveys conducted between February 8, 2018, and April 25, 2021, targeting the children of households participating in the Productive Safety Net Programs in the Amhara and Oromia regions. A total of 3,041 households, whose children had at least two waves complete (non-missing) anthropometric measurements, were included in the analysis. The response variable, height-for-age standard score (HAZ), was used. HAZ was computed based on a child's height (in centimeters), age (in days), and sex, in accordance with the World Health Organization child growth standards. A quantile mixed hidden Markov model was employed. RESULTS: There was a significant burden of stunting among children under five in both the Amhara and Oromia regions. However, the situation was more worsening in the Amhara region. Several factors were found to significantly reduce the risk of under-five stunting, including maternal body mass index, maternal education, nutritional counseling, use of iodized salt, increased feeding frequency, and being a female child. Conversely, factors such as small size at birth, age of the child, episodes of illness, and consumption of unhealthy food were associated with increased stunting. For instance, being the female children were associated with a lower likelihood of stunting compared to males, with odds ratios of 0.263 (0.055), 0.164 (0.0544), 0.171 (0.071), and 0.358 (0.0822) at 10th, 25th, 75th and 90th quantiles respectively. Additionally, the mixture probabilities of the first component of latent process increased across quantiles, with 62% of children classified in the first hidden (stunted) state at 90th quantile. Unobserved temporal shocks were found to increase the probability of transitioning into the first hidden state. CONCLUSION: The study identified both unobserved time-constant and time-varying heterogeneities, as well as several fixed effects. These factors contribute to an increased risk of under-five stunting and raise the probability of transitioning into the stunted state. Without timely and targeted interventions, these effects may continue to exacerbate the burden of stunting.