1  Introduction
As of 30th of June 2020, COVID-19 has caused more than 500,000 deaths globally, with an estimated case fatality of 1–4% (Hauser et al. 2020). The UK is one of the countries most affected, with an estimated 57,300 more deaths in England and Wales than it would be expected from mid-February to end of May 2020 had the pandemic not taken place (Kontis et al. 2020). Established risk factors of COVID-19 mortality include age, sex and ethnicity (Wu et al. 2020). Previous studies have observed a correlation between pre-existing conditions such as stroke, hypertension and diabetes (Williamson et al., 2020, Yang et al., 2020). Long-term exposure to air-pollution has been hypothesised to worsen COVID-19 prognosis: either directly, as it can suppress early immune responses to the infection (E. Conticini et al. 2020), or indirectly, as it can increase the risk of stroke, hypertension and other pre-existing conditions (Giorgini et al., 2016, Scheers et al., 2015).
Previous studies suggested an effect of long-term exposure to air-pollution on COVID-19 mortality (Cole et al., 2020, Liang et al., 2020, Travaglio et al., 2020, Wu et al., 2020), however several methodological shortcomings limit their interpretability. They were based on data aggregated on large spatial units and thus suffer from ecological fallacy (grouped levels association do not reflect individual ones) (Wakefield 2008). Air pollution is characterised by high spatial variability, making the availability of mortality data at the same high spatial resolution crucial (Villeneuve and Goldberg 2020). In addition, a coarse geographical resolution might lead to inadequate adjustment for confounders, when these are available at higher resolution (Villeneuve and Goldberg 2020). Most previous studies assessed cumulative deaths until mid or end of April and thus the generalisability of their results is limited to the early stages of the epidemic (Liang et al., 2020, Travaglio et al., 2020, Wu et al., 2020). One study had data available up to June 5, 2020 (Cole et al. 2020) and another up to June 12, 2020 (Statistics 2020), capturing a proportion COVID-19 deaths attributable to the first wave.
In this nationwide study in England, we investigated the effect of long-term exposure to air pollution on COVID-19 mortality during the entire first wave of the epidemic, after accounting for confounding and spatial autocorrelation. We focused on exposure to NO2 and PM2.5 (atmospheric particulate matter that has a diameter of less than 2.5 µm). We downscaled the LTLA geographical information to the Lower Layer Super Output Area (LSOA) to alleviate the effect of ecological bias and exploit the variability of the exposure at high geographical resolution.