New Harvard Study Links COVID Deaths and Air Pollution: An Interview with Study Author Dr. Francesca Dominici
By Gretchen Goldman, Union of Concerned Scientists
A new study was made public yesterday that sheds light on the connection between COVID-19 health impacts and air pollution. I sat down (virtually of course) with Dr. Francesca Dominici, author and Director of the Data Science Initiative at Harvard’s T.H. Chan School of Public Health, to learn more about the study’s findings, which focused on fine particulate matter (PM2.5) and what researchers plan to do next to further our understanding of air pollution’s impact on coronavirus health outcomes.
Dr. Gretchen Goldman: These are groundbreaking new findings on the relationship between PM2.5 air pollution and COVID-19. What are the key findings?
Dr. Francesca Dominici: To our knowledge, this is the first study that quantifies on a national scale the potential increase in COVID mortality rate that is associated with long-term exposure to air pollution. We use data from counties that account for 90% of the COVID-19 deaths in the US as of April 4, 2020.
The most important result is that we found that people living in counties in the US that have experienced a higher level of air pollution over the past 15-17 years have a substantially higher COVID-19 mortality rate. To quantify, we found that a one unit increase in long-term average exposure to fine particulate matter is associated with a 15% increase in COVID-19 mortality rate on average in the analysis. This increase accounts for adjustments for any systematic differences between county level characteristics [such as population density or smoking rates].
We also wanted to put into perspective this 15% increase in mortality rate for COVID with all the other evidence we have produced in the past for the long-term effect of fine particulate matter on all-cause mortality. This is important to report: A one unit increase in long-term exposure to PM2.5leads to a 15% increase in COVID-19 mortality rate with a magnitude that is 20 times that of PM2.5and all-cause mortality (which is 0.7%). So in the Medicare study [Di et al, 2017], we report that a one unit increase in long-term exposure to PM2.5 is associated with a 0.7% increase in all-cause mortality. In this study, we have that one unit associated with a 15% increase in COVID-19 mortality so the relationship between PM2.5 and COVID mortality is 20 times stronger than the relationship between PM2.5 and all-cause mortality.
GG: These are notable findings. Did they surprise you?
FD: I was expecting a statistically significant association. At the beginning, I was surprised at how strong the association is but then as I was thinking more about it, I was less surprised because as we are learning more and more about COVID-19, we are also learning that all the diseases that are affected by fine particulate matter are all of the disease that make the outcomes for COVID-19 much worse.
So basically, this is like adding gasoline to the fire. People that have been breathing polluted air for a long time, we know that long-term exposure to fine particulate matter increases inflammation in the lungs and potentially in the cardiovascular system. If on top of that, these individuals are affected by COVID-19, then it’s not surprising given that they’ve already been impacted by fine particulate matter, that they might respond with much worse health outcomes than someone who lives in a clean air county.
GG: We know that some groups (such as the elderly and those with lung diseases) are at higher risk of adverse health effects from PM2.5 exposure, and that some populations (such as environmental justice communities) have higher burdens of air pollution. How do these new findings inform how we think about who is most vulnerable to severe outcomes from COVID-19?
FD: To be rigorous, we have to consider that this is early data. For now, we have just looked at the relationship between county level exposure to PM2.5 and COVID-19 mortality. Unfortunately, the mortality data is not currently available by age and race and so on. From the data in this particular study, I don’t think we can characterize vulnerability and susceptibility, but I do think it is not too much of an extrapolation from the data to expect that people that have been experiencing and breathing and living in very polluted areas, whether they are young or old, and they have pneumonia or asthma or any other chronic lung disease, and also people that are generally socially disadvantaged (e.g. African Americans, low socioeconomic status) [may face worse COVID outcomes].
These are all the communities that will experience higher COVID deaths. But I do think we need a little more refined data to be able to pinpoint this type of vulnerability profile more precisely. The unfortunate thing is that more deaths will arrive. In the next few months, we will be able to characterize vulnerability a little bit more precisely based on what we know.
GG: What additional research is needed to understand links between air pollution and COVID-19?
FD: There is an enormous amount of work that needs to be done. This is a first look at the data. First of all, this is county-level data; we need to look at the relationship between air pollution and COVID health outcomes at a much smaller spatial resolution. In some states, for example New York, they are now making available zip code-level data. It will be really important to repeat the analysis at a spatial resolution much smaller than the county.
Second, we have only looked at deaths. As we have testing done in a more systematic fashion (though the number of cases is always a little controversial outcome because it depends on practices), as we have hospitalization data, as we have data on the outcome after you’ve been hospitalized, there will be important analyses and questions to explore, regarding the potential interaction between exposure to fine particulate matter, race, and socioeconomic status with respect to COVID deaths. We know, as of today, it has been reported that there is a much higher burden of COVID deaths among African Americans so I would expect there is going to be a high interaction with exposure to fine particulate matter and race in terms of that outcome.
Third, we also need to look at many other pollutants. We are only looking at fine particulate matter but we can also look at traffic pollutants, we can look at NO2, we can also look at ozone. I wanted to get the first data out there because I wanted to make sure that the general public and the government will start really paying attention to the communities with higher levels of particulate matter because they have potential to be more affected by the virus.
GG: What ways could public health officials and political leaders use these findings to best protect the public?
FD: There is a very concrete action which is to implement stricter social distancing measures and to make sure that there is adequate access to health care resources for the most polluted areas in the US.
It is pretty clear from the data that these are the counties in the US that have been affected and have high levels of pollution in the last few years, even if we haven’t seen it yet. In some of them, we are already seeing very high levels of deaths. But in ones where we haven’t yet seen higher numbers of deaths, we need to pay attention because they are at higher risk of much worse health outcomes for COVID. So we have to give them higher priority and really look closely at these counties because this is where getting COVID could get much worse in terms of outcomes than places where people are breathing cleaner air.
GG: The EPA is in the process of reviewing the National Ambient Air Quality Standards for particulate matter. How do these findings inform the EPA’s decision on how to protect public health from particulate pollution? What is at stake here?
FD: It seems pretty clear to me that we are now living in a new world with this COVID virus. So not paying enough attention and weakening the National Ambient Air Quality Standards standards, I actually see it as a very unwise decision, I would go as far as saying irresponsible decision because we now know that exposure to fine particulate matter puts American people at risk to die from COVID, in addition to everything else we know about the harmful effects of fine particulate matter. So I will call it unwise and irresponsible.
GG: Anything else people should know about this new research?
FD: This study is completely open-sourced. You can go on the website, you can download the data, and you can run all of our code. There is absolutely no question that this is a fully reproducible and fully publicly available study. We will continue to update the analysis as unfortunately more deaths occur and more data will come in.
Now that we have developed the platform, if for example hospitalization data becomes available, we will definitely continue to analyze data in a way that we can protect and inform public health in the best ways possible.