Stress Is in the Air: Ambient Reactive Oxygen Species and COVID-19

The paper by Stieb and colleagues (pp. 168–177) in this issue of the Journal is of interest in several domains (1). First, although still an ecological study of the potential impact of exposure to air pollution on the risk of coronavirus disease (COVID-19), it addresses some of the critiques of previously published studies (2). Instead of a comparison across regions with differing exposures that inherently includes regional differences regarding potentially confounding variables, it is a study of neighborhood differences across a single city, Toronto. Second, the authors used a novel air pollution exposure metric, estimated reactive oxygen species (ROS) in fine particulate matter (particulate matter ⩽2.5 μm in aerodynamic diameter [PM_2.5]), based on a model of ROS in human epithelial lining fluid and a land use regression model of iron and copper in PM_2.5 from multiple monitoring sites across Toronto in 2016–2017 (3). Yet another important aspect of the use of ROS as the exposure metric is the support the analysis gives to the putative oxidative stress mechanism for the PM_2.5 association with COVID-19 outcomes observed in other studies (4–6).

Although the estimated ROS exposure is an innovative method for an air pollution epidemiological study, this method would be strengthened if it were to be used effectively by other investigators in different settings to study a variety of health outcomes. Actual measurement of ROS concentrations or oxidative potential in ambient air has been advocated for air pollution health studies (7). The method used by Stieb and colleagues to estimate PM_2.5-associated ROS could be validated with comparison to ambient concentrations. Of course, this method does not account for ROS generated in human epithelial lining fluid from exposure to gaseous pollutants like ozone and NO₂. The current internationally accepted approach to regulating the outdoor air pollution mixture of particles and gases involves setting standards for ambient concentrations for single pollutants (e.g., U.S. Environmental Protection Agency National Ambient Air Quality Standards or World Health Organization Air Quality Guidelines) (8, 9). If it could be shown that ambient ROS concentrations or the oxidative potential of the pollution mixture were robustly associated with health outcomes, then air quality regulation could be targeted to an appropriately representative exposure metric.

The authors’ statistical analysis was careful to adjust for multiple likely confounders of any association of air pollution exposure and COVID-19, including socioeconomic status (SES), racialized group status, linguistic difference, use of public transportation, housing crowding, days elapsed since the first case, days since peak daily incidence of cases, case outcomes, and weekly rates of COVID-19 testing. In addition, other exposures tested were PM_2.5, NO₂, and greenness. Analyses were stratified by age, sex, and sporadic versus outbreak case status. An appropriate statistical model for count data (negative binomial) and sensitivity analyses were conducted.

A significant positive association was observed between neighborhood-level estimated ROS and COVID-19 incidence. The expected effect modification by neighborhood-level measures of racialized group membership and SES (percent unemployed, with less than high school education or with income below poverty level) was also observed (10, 11). The association with ROS was greater for men and for those under 50, perhaps because these subgroups spend more time outside of their homes during the pandemic and thus have greater opportunity for exposure to air pollution. A nonsignificant positive association with COVID-19 incidence was observed in neighborhoods where the proportion of Black residents was greater, and independent of this proportion, measures of lower SES were also positively associated with COVID-19 incidence. Individuals with lower income are more likely to be essential workers who cannot work at home and are more likely to live in crowded housing (12).

Although a measure of traffic-related air pollution, NO₂, was associated with COVID-19 incidence in a bivariate analysis, this association was attenuated when ROS was included in a joint model. Another exposure of interest, greenness as measured by the normalized difference vegetation index, was negatively associated with COVID-19 incidence in a bivariate analysis, but this association was also attenuated when ROS was included in the model. In contrast to other reports involving multiregional comparisons (4–6), Stieb and colleagues did not observe a significant positive association of COVID-19 incidence with PM_2.5 mass, likely because of a relative lack of spatial variability across Toronto.

Although this study addressed many of the limitations of other studies attempting to study the impact of exposure to air pollution on COVID-19, it remains an ecological study from which causality cannot be inferred. Studies with individual-level data for both exposures and outcomes are needed. To date, it has been difficult for investigators to obtain data from public health agencies that include both residential addresses and individual-level covariate data. Although this is understandable in terms of privacy concerns, it remains imperative to determine if exposure to air pollution is truly a risk factor for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and/or COVID-19 morbidity, especially if this exposure is a mediator of the increased risks for people of color who are at lower incomes. The COVID-19 pandemic has dramatically demonstrated the chronic health inequity that low-income communities of color experience in the United States and apparently Canada as well (13, 14). The results of the Toronto study are just another reminder of the disproportionate burden of exposure to air pollution borne by such communities (15). Environmental justice is a necessary component of dismantling the systemic racism upon which U.S. society has been built and that also may blight Canadian society.