Green space exposure and mortality in the Florentine plain: preventable deaths at the census tract level and uncertainty analysis

Proximity to vegetation is linked to lower mortality, making urban greening a key public health strategy. This study quantified preventable deaths associated with green-space exposure in the Florentine plain (Tuscany, Italy), integrating population, deprivation, and satellite data at the census tract level, while accounting for model uncertainty. Exposure was measured in 2015 as the mean NDVI within 300 m of each tract. Deaths in 2021 were redistributed from municipalities to tracts by age–sex structure and deprivation. Mortality impacts among adults over 35 were estimated under counterfactual scenarios: NDVI thresholds (0.5, 0.7), + 0.1 absolute and + 20% relative NDVI increases, and compliance with the World Health Organization’s greenness target converted to NDVI. The exposure–response function came from a Bayesian meta-analysis. Uncertainty was propagated through Monte Carlo simulations, and a Global Sensitivity Analysis identified main sources of variance. We estimated 346 avoidable deaths for NDVI ≥ 0.5 and 877 for NDVI ≥ 0.7. Increases of + 0.1 and + 20% NDVI would have prevented 310 and 25 deaths, while meeting the WHO target would have prevented 47 deaths. Although uncertainty was high, the probability of a nonzero impact exceeded 90%. Sensitivity analysis identified the NDVI–mortality hazard ratio as the dominant source of uncertainty, while deprivation and stochastic death redistribution had minimal effects. The integrated Monte Carlo–sensitivity framework demonstrated both the potential health gains and the need for refined exposure measures and local cohorts to support equitable, evidence-based planning.