Abstract
OBJECTIVE: This study aims to contribute to enhanced food security in Haiti through proposing targeted local interventions. Employing a spatially explicit tool, the research supports decision-making by relating undernutrition to socio-economic conditions and biophysical factors. DESIGN: Georeferenced Demographic and Health Survey (DHS) conducted in 2016–2017 combined with spatial environmental information was used for a multivariate linear regression model to identify factors associated with stunting prevalence. Missing data were imputed through kernel density regression. We converted the structural relationship estimated for the territory of Haiti into a decision support tool by adding fixed effects at communal level. Various policy scenarios were analysed. SETTING: Haiti, with spatial data across the 134 communes. PARTICIPANTS: The analysis included 5623 children under five and their mothers, sourced from DHS data. RESULTS: Approximately 22 % of all children were stunted. Implementation of the LimitedIntervention development scenario led to a 2·5 % reduction in stunting, while the ModerateIntervention and FullIntervention scenarios achieved more significant reductions of 6 % and 10 %, respectively. Areas with highest stunting incidence benefit most from interventions. CONCLUSIONS: This tool supports decisionmakers by assessing the impact of interventions at commune level and selecting areas where interventions exert the most significant effects. The study suggests to apply a strategy that starts in relatively safe communes and then scales to other areas. The flexible approach adopted in this study allows applications in other countries or regions to assess the prevalence of undernutrition among children under five.