Modelling Short-Term Health Effects in Milan Area Due to Lockdown Reduced Emissions: Combined Uncertainty Analysis from Estimated NO2 Levels and Exposure–Response Functions

Abstract

To contain the spread of the COVID-19 pandemic, several governments declared lockdowns. The reduction of human activities linked with mobility restriction caused an unprecedented drop in emissions, especially in the road transport sector. This study aims to evaluate the uncertainty of short-term health effects (i.e. avoided hospital admission (AHA) associated to NO2 ambient concentrations) derived from the change in air quality (AQ) due to lockdown. The CAMx-WRF modelling suite is applied for a series of nested domains using EMEP and Lombardy region emission inventories. The health impact analysis is focused on a 70 × 70 km domain centered on Milan metropolitan area with 1-km resolution, from February 24th 2020 to April 30th 2020. Two simulations, Business as usual (BAU) and lockdown scenario (LOCK), are carried out and results are compared with air quality monitoring data to assess the model uncertainty. Health effects for the difference between LOCK and BAU simulation are evaluated for NO2 following the WHO Health risks of air pollution in Europe (HRAPIE) project recommendation. The combined effect of both modelled concentrations and exposure–response functions (ERF) on the uncertainty of calculated AHA is then evaluated for different air quality station types. We find that the ERF is the major cause for uncertainty for Urban and suburban AQ stations, measured as the size of the 95% confidence interval (CI) of the estimated number of AHA. The AQ uncertainty of the BAU scenario has a lower impact on AHA CI than for the LOCK scenario. When comparing results according to station type, the lowest AHA uncertainty is obtained for urban background stations while the highest for rural background stations.