Environment International (Dec 2023)
Assessment of health impacts and costs attributable to air pollution in urban areas using two different approaches. A case study in the Western Balkans
Abstract
In this study, two different air quality impact assessment methodologies were adopted and combined with a sensitivity analysis to estimate the unit costs. Air pollution health impact (mortality) assessment was carried out using one methodology based on log-linear concentration response functions (CRF) and another relying on the integrated exposure response curve (IER) from the Global Burden of Disease. Morbidity impacts were estimated with the CRF approach only. To assess the inequalities between low and high income countries, an area of low-medium income countries with a critical air pollution situation, was selected. The health impact and related external costs attributable to air pollution in 2019 were assessed in 30 urban areas of the Western Balkans region, one of Europe’s air pollution hot spots. The evaluation was based on PM2.5, O3 and NO2 concentrations in background sites from official monitoring networks. In 2019, the cost of mortality attributable to PM2.5 in 26 urban areas was 7.8 and 9.0 billion Euro according to IER and CRF methodologies, respectively. The cost of O3 associated with all-cause mortality estimated with the CRF methodology in 17 urban areas was 1.0 billion Euro while the one attributable to NO2 pollution in 28 urban areas was 1.5 billion Euro. The study results suggest that the economic burden of air pollution in the Western Balkans is higher in terms of GDP than the one observed in EU27 in the same time window. The study concludes that CRF and IER methodologies are coherent, because the discrepancy in the results are explained by the differences in the assessed health outcomes. The two approaches are complementary because the combination of them makes it possible to obtain a wider range of outcomes. In addition, despite the different causes of death considered, the comparison between them is useful for cross-validation.