Invited Perspective: Ambient Air Pollution and SARS-CoV-2: Research Challenges and Public Health Implications

Approximately five million people are known to have died from coronavirus disease 2019 (COVID-19) since the pandemic was declared in March 2020 (Johns Hopkins University 2021); however, the actual number of such deaths has been estimated to be as much as 3–4 times higher (Farge and Revill 2021; Spinney 2021; Economist 2021). At an early stage of the pandemic, it was suggested that ambient air pollution was a modifiable risk factor for both infection and severity of COVID-19 on the basis of spatial correlation (Pansini and Fornacca 2021) (Ogen 2020), ecological (Liu and Li 2020; Wu et al. 2020), and time-series (Zhu et al. 2020) analyses. As outlined by Villeneuve and Goldberg (2020), these studies were susceptible to important biases relating to case ascertainment, differences in timing on the pandemic curve, and control of confounding owing to the adoption of preventive public health measures.

In this issue of Environmental Health Perspectives, Kogevinas et al. (2021) publish findings from a carefully conducted analysis within a cohort study in Spain that avoids many of the limitations of past studies. A distinct advantage over past studies was the identification of prior severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection using antibodies. The authors found that residential air pollution concentrations estimated for 2018–2019 were not significantly associated with COVID-19 infections, with risk ratios (RRs) of 1.07 [95% confidence interval (CI): 0.97, 1.18] for nitrogen dioxide (${\mathrm{NO}}_2$) and 1.04 (95% CI: 0.94, 1.14) for fine particulate matter [PM with an aerodynamic diameter of $\le 2.5\mathrm{\mu }\mathrm{m}$ (${\mathrm{PM}}_{2.5}$)] per interquartile range (IQR). In contrast, they found significant associations with the extent of COVID-19 disease based on hospitalizations and self-reported symptomology with adjusted RRs per IQR of 1.14 (95% CI: 1.00, 1.29) for ${\mathrm{NO}}_2$ and 1.17 (95% CI: 1.03, 1.32) for ${\mathrm{PM}}_{2.5}$. Associations were stronger for more severe disease than for mild disease.

This study had several key strengths: a) the individual-level design, which can avoid the potential for clustering of cases; b) data for an extensive series of confounding variables; c) a relatively small geographical area with good surveillance data; d) a short time scale prior to emergence of more infectious variants; e) data collection prior to vaccination that otherwise could have complicated the modeling of transmission and case severity; f) use of established well-evaluated European air pollution models at high (100-m) resolution; and g) careful ascertainment of outcomes, particularly by using serology measures. Important limitations of the study included a modest participation rate (62%). Thus, some cases and deaths from COVID-19 were likely missed.

Sensitive antibody tests were employed in this study that could detect current and past infections in all participants, not just those with symptoms or in contact with a case. Notably, 40% of infections in their analyses were asymptomatic, suggesting that this approach to case ascertainment reduces possible bias inherent in studies reliant on test surveillance data, administrative health records, or both. These latter approaches have been used in previous individual-level studies in the United States (Bowe et al. 2021; Mendy et al. 2021), Mexico (López-Feldman et al. 2021), and the United Kingdom (Elliott et al. 2021; Zhang et al. 2021), but they may result in collider bias—those coming forward for SARS-CoV-2 testing can be highly selected on other factors that may bias the exposure–response relationship (Griffith et al. 2020). An additional issue in the UK studies was the use of air pollution exposure estimates at the place of residence $>10\mathrm{y}$ prior to the pandemic.

What, then, should be made of the latest paper by Kogevinas et al. (2021)? Ambient air pollution is widely recognized to increase the risk of many diseases, including chronic cardiorespiratory disease (Thurston et al. 2017) and, potentially, diabetes (McAlexander et al. 2021). These health conditions are important risk factors for more severe COVID-19 (Mazucanti and Egan 2020; Ssentongo et al. 2020). The finding in the paper of a stronger association between air pollution and more severe COVID-19 is consistent with this knowledge. Further, there are other plausible mechanisms by which air pollution impacts case severity, including effects on the immune system (Glencross et al. 2020) and inflammation (Pope et al. 2016). However, it should be noted that the associations between air pollution and COVID-19 disease in the present well-designed study are much lower than many earlier studies.

It is perhaps surprising that there was no significant association with infection given suggestions from animal studies that air pollution increases the expression of both the angiotensin-converting enzyme 2 receptor (Sagawa et al. 2021) and transmembrane protease serine 2 (Sagawa et al. 2021; Vo et al. 2020), both of which are involved in SARS-CoV-2 virus entry into target cells (Bourgonje et al. 2020; Jackson et al. 2021). One interpretation is that air pollution does not affect infection risk. Alternatively, it may be that the relevant impacts of air pollution on infection are short term and that ambient air pollution exposures in 2018–2019 are a poor proxy for exposures in 2020. More mechanistic studies are needed to examine this hypothesis.

Studies of air pollution and COVID-19 require substantially different methods to characterize risks than studies of chronic disease. Future studies would benefit from additional data that reliably capture the source, time, and location of infection, as well as information about predictors of indoor air quality such as ventilation rates. Preliminary analyses suggest that the spread of COVID-19 in outdoor settings is many orders of magnitude lower than indoors (McGreevy 2021; Qian et al. 2021). Therefore, great care must be taken in risk communication messaging so as to avoid giving individuals the false impression that their risks of COVID-19 are reduced by avoiding outdoor activities on days with poor air quality.

In conclusion, we cannot help but remark on the progress made over the last year in the methodology to evaluate links between air pollution and SARS-CoV-2. Yet, evidence to date, including from the study by Kogevinas et al. (2021), does not support moving reducing air pollution to the front line during the pandemic as a mitigation measure for COVID-19. From a public health perspective, policies such as vaccination, face mask wearing, and employee leave benefits (to encourage test and vaccination uptake, or to provide the means for infectious individuals to stay home) represent more effective and powerful tools to help us emerge from the pandemic.

Refers to https://doi.org/10.1289/EHP9726