Hospitalization due to fire-induced pollution in the Brazilian Amazon: A causal inference analysis with an assessment of policy trade-offs

Developing countries by relying on agricultural fires trade respiratory health for cheaper land preparation and greater food security of subsistence farmers. In the Brazilian Amazon, thousands of kilometres of vegetation are annually burned, releasing pollutants that impact the health of 24 million inhabitants. The paper seeks to fill two crucial informational gaps for policy planning, namely, the size of the impact on the most pollution-susceptible groups, i.e., children and the elderly, and the priority locations for health-oriented intervention on fires. A municipal-monthly panel covering ten years of the Amazonian territory was analysed by relying on exogenous and high-resolution wind direction variation to identify the effect of fires on pollution and hospitalizations. This is an unparalleled effort for applying to a large geographical area spanning five million km2 a highly refined identification strategy relying on hourly gridded wind and fire data. As the result, one extra standard deviation of “upwind” fires was estimated to increase asthma-related hospitalization of the elderly in 0.03 days/month, that is, 4 % of the bed-days commonly demanded by such kind of hospitalization, an effect whose size decayed with the distance between fires and hospitals. A policy assessment uncovered the trade-off between respiratory health of the elderly and nutritional health of fire-dependent subsistence farmers, presenting a priority map for tackling the issue with municipal-level interventions. The targeting of non-subsistence fires is advised, what could avoid 28 days of hospitalization per year. It is thus demonstrated that the trade-offs inherent to agricultural fires could be better balanced by evidence-based targeting of fire prevention policy.