Air pollution and COVID-19: is there a clear relationship?

ID’s recommended reads – 22/06/2020 – Wanda Van Hemelrijck & Lucía Rodríguez Loureiro

From the very beginning of the coronavirus pandemic, several studies have tried to link air pollution and severe COVID-19 outcomes, yielding positive results. It is plausible that air pollution could exacerbate COVID-19 and increase fatality; fine particulate matter causes inflammation and pulmonary damage, and health conditions aggravated by air pollution, such as hypertension or diabetes, are important risk factors determining the prognosis of the SARS-CoV-2 infection. Besides, the virus has been found in outdoor particulate matter.

However, it should be noted that the studies mentioned were not peer-reviewed, and yet they have received a massive media coverage. Perhaps this whole good-intentioned background noise made us miss the opportunity to listen to critical voices claiming for a more parsimonious interpretation of these findings; one that allows us to reflect thoroughly on the potential limitations and biases that these studies could entail and, what is most important, how to overcome them. After all, correlation does not imply causation.

This week’s recommended reading: a dialogue between emerging research outcomes on the relationship between ambient air pollution and the COVID-19 pandemic and their methodological critique. To conclude, we include a methodological review of ecological studies including suggested guidelines to overcome biases related to this study design.

1) Exposure to air pollution and COVID-19 mortality in the United States: A nationwide cross-sectional study.
Wu, X., Nethery R C., Sabath B M., Braun D., Dominici F. (2020).

An ecological study (i.e. at an aggregated spatial level) in 3,087 counties in the United States concluded that a small increment (1µg/m3) in PM2.5 increased by 8% the COVID-19 mortality rate. The most recent version of this article has included several potential confounders at the county level, ranging from a wide set of socio-economic and demographic factors, healthcare services, environmental confounders and the date since the beginning of the lockdown measures. These findings were robust even when conducting several sensitivity analyses, for instance, excluding the New York metropolitan area or additionally adjusting by number of tests performed. Despite the ecological nature of the study, the authors argue that one of its strengths is to provide emerging evidence on the link between air pollution and COVID-19 and encourage other authors to continue this line of research.

2) Air pollution, COVID-19 and death: The perils of bypassing peer review.
Villeneuve, P. & Goldberg, MS. (2020)

What are the potential biases in the article above leading to an overestimation of the risks, and how can we overcome these limitations? In this article from The Conversation, the experts on environmental epidemiology, Paul Villeneuve and Mark S. Goldberg, enumerate key aspects that need to be taken into account in order to correctly approach the potential relationship between air pollution and the COVID-19 pandemic. To cite some of them, there is a need of using individual data, re-thinking selection of the outcome of study (due to underreported cases and deaths); taking into account the concentration of vulnerable populations in cities, such as migrants or refugees; or to account for the transmission of the disease in clusters.

3) Ecologic Studies Revisited.
Wakefield, J. (2009)

Finally, to avoid falling into methodological errors when using ecological data – data on ambient air pollution concentrations is usually accessible to the general public in an aggregated spatial level –, we present an exhaustive review on ecological studies. This paper explains the potential biases arising from ecological studies or in which cases inference might be unreliable when only analysing ecological data, but also how, when combining ecologic and individual data, a study design can become extremely powerful.