Disparity in prevalence and predictors of undernutrition in children under five among agricultural, pastoral, and agro-pastoral ecological zones of Karamoja sub-region, Uganda: a cross sectional study

BACKGROUND: Undernutrition accounts for nearly half of under-five child mortality in developing countries where household nutrition is largely dependent on agriculture. Despite the strong influence of agroecology on agriculture in those countries, limited information exists on whether undernutrition in children under five varies with agro-ecological location. METHODS: Using Karamoja sub-region of Uganda, one of the most food insecure parts of Eastern Africa as a case area, and applying a multi-stage sampling procedure, and a structured questionnaire, this study examined in a comparative manner, the prevalence and predictors of undernutrition in children under five among the agricultural, pastoral, and agro-pastoral ecological zones. Chi-square test and Kruskal-Wallis test were used to establish the disparity in prevalence of undernutrition and household contextual characteristics, respectively. Binary logistic regression was used to determine the predictors of undernutrition in children under five among the three agro-ecological zones. The level of statistical significance was set at p ≤ 0.05. RESULTS: The prevalence of underweight, stunting, and wasting ranged from 36 to 58% but varied with agroecology in terms of the peak age ranging from 6 to 37 months. Child characteristics, feeding practices, household economic factors, sanitation factors, and caregiver characteristics that predict undernutrition among children under five were identified (p ≤ 0.05). Caregiver handwashing after using latrine (p = 0.005) and diarrhoea in a fortnight (p < 0.001) increased the likelihood of stunting in pastoral agroecology only whereas cereal storage in both sacks and granary in agro-pastoral zone was associated with reduced likelihood of both underweight (p < 0.001 and p = 0.014) and stunting (p = 0.011 and p = 0.018), respectively. A male child was more likely to be underweight and stunted in pastoral (p = 0.002 and p = 0.011) and agro-pastoral (p = 0.017 and p = 0.002) agroecology, respectively. Household expenses reduced the likelihood of both underweight and wasting in pastoral (p = 0.013 and p = 0.005) and agricultural (p = 0.011 and p = 0.021) agroecology, respectively. Flour storage duration increased the stunting likelihood in pastoral (p = 0.032) and agro-pastoral (p = 0.006) agroecologies. CONCLUSION: This study has revealed that, in a food insecure developing country setting such as Karamoja sub-region of Uganda, undernutrition among children under five varies with agroecology. Thus, nutritional interventions in such locations should be agroecology specific. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12887-022-03363-6.