This post is a part of a series on COVID-19 and the Coronavirus Pandemic
New data compiled and analyzed by UCS shows that in counties with relatively large non-White populations, 70 percent more people test positive for COVID-19 than in predominantly White counties.
For weeks if not months, our newspaper front pages, social media feeds, and minds have been awash in statistics about the novel coronavirus, and we’ve watched with aching hearts as the death toll has risen in the U.S. and around the world. Within this ocean of data, we often see statistics on the number of tests administered at the state level and what percentage of them were positive.
But given the limited availability of test kits, particularly in the early days of the crisis, and well-documented disparities in access to health care from community to community and across racial and ethnic lines, my colleagues and I wondered how access to testing and the results of those tests varied from community to community.
County-level testing data: Spotty and scattered
Given all the data available, we naively assumed we could just find the data out there, download it, and start exploring.
The truth is that the data collection on COVID-19 testing at the community level is spotty. There’s no federal strategy, and statistics that are meticulously reported in one state are completely lacking in another. To gather county-level data on how many tests had been administered and what share of them were positive, my colleagues and I scoured department of health websites state by state and painstakingly worked data into a consistent format. (I’d say it was a labor of love, but it was more like a labor of anxiety—something to pour ourselves into during these scary times.) The data set is accurate to the best of our abilities as of April 10th.
All counties are reporting the number of positive cases of COVID-19, but not all report the total number of tests performed in that county. The number of total tests and the share of them that are positive are key pieces of information we need to understand where outbreaks are happening, how many people are infected, and who has (or lacks) access to testing. We were able to find both the total number of tests and the share of those tests that were positive for roughly 900 counties from 14 states, which represents just under one-third of all U.S. counties. Of those, about 725 reported at least one COVID-19 test and at least one confirmed case.
In all, the data provide a snapshot—from the first coronavirus cases in the U.S. through April 10th–of who has had access to testing and what the outcomes of those tests have been. It’s not a complete snapshot by any means, but to our knowledge it’s one that has not been captured elsewhere.
After gathering the testing data from as many counties as possible, we compared it with various metrics included in the Centers for Disease Control’s Social Vulnerability Index (SVI). The SVI assigns social vulnerability scores to US counties based on Census information such as race, ethnicity, income, poverty rates, and housing characteristics. We were interested in whether there were trends in the patterns of the test results that were related to measures such as a county’s overall social vulnerability score or the percent of county residents identifying as non-White or Hispanic.
In counties with more people of color, the share of COVID-19 tests that come back positive is higher
The counties reporting the total number of COVID-19 tests administered and the share of them that came back positive span a very broad range of population characteristics. The total population of the counties varies between less than 1,000 (hello, Slope County, North Dakota!) to more than 4 million (that would be Maricopa County, Arizona), and vary from having populations that are 99% White (such as Franklin County, Indiana) to more than 90% non-White or Hispanic (such as Bronx County, New York). This lets us examine patterns in the testing data across a wide diversity of demographics.
Across different metrics of race and ethnicity, socioeconomic status, and overall social vulnerability, we find that in counties with more non-White or Hispanic residents and those with higher social vulnerability, a greater percentage of COVID-19 tests are positive.
Specifically, we find that counties where more than 25% of the population is non-White or Hispanic have a positive testing rate of about 17% compared with 10% in counties where less than 25% of the population is non-White or Hispanic. That means that 70% more of the tests in counties with large communities of color come back positive than in predominantly White counties. We found similar results if we changed the population cutoffs (e.g. less than 10% minority vs. greater than 40% minority) or analyzed the CDC’s measure for overall vulnerability related to minority status and language.
Additionally, in counties with above median scores by the CDC’s measure of socioeconomic vulnerability nearly 30% more tests come back positive than in counties with below median scores from the CDC. Again, we found similar results using the CDC’s overall metric of social vulnerability for each county.
The positive testing rate did not vary significantly with total population size or with the CDC’s measures of vulnerability related to housing and transportation.
Overall, these results tell us, loud and clear, that in socially vulnerable communities and communities of color, a higher percentage of COVID-19 tests are positive.
Why do a larger share of COVID-19 tests come back positive in communities of color?
What could be causing these disparities? In thinking through some possibilities, it’s helpful to think of the fraction that we’re evaluating (positive tests/total tests). Looking at this fraction, we can see that if the total tests term increases and the positive tests term stays the same, the overall percentage would decrease. So that begs the question of whether predominantly white communities have more access to COVID-19 tests.
The answer, based on this county-level data set, is probably not. If we look again at counties above or below the 25% people of color mark, the number of tests per 1,000 residents is slightly higher in counties where more than 25% of residents are people of color than in counties where less than 25% of residents are people of color—6 per 1,000 residents vs. 5 per 1,000 residents, respectively. The difference is just barely significant statistically. And if we slice the minority population percentages differently, there’s no significant difference.
If the total number of tests being administered per 1,000 residents is not correlated with the county’s non-White or Hispanic population, we’re left to explore why communities of color have larger percentages of positive tests. There are reasons to think that people of color may be experiencing more severe cases of COVID-19 that would lead them to seek tests and treatment more often than Whites. Some of the underlying factors could include:
- On the whole, Black and Hispanic people in the U.S. have relatively high rates of underlying health conditions, such as hypertension, obesity, and diabetes, which the CDC lists as being risk factors for severe COVID-19 illness. In many cases this is a direct result of long-standing disparities in exposure to environmental pollution and economic opportunities, health care access,
- These same groups have lower rates of health insurance than Whites, which translates to having less preventative care and, in the long-term, potentially poorer health outcomes that increase risk to falling ill to COVID-19.
- Non-White groups in the U.S. tend to have higher exposure to air pollution from industrial or vehicle sources. Recent research has shown that the COVID-19 death rate is higher in counties with certain types of air pollution.
It’s important to note that what’s missing from this data set could change this story or could be indicative of even broader issues. What does it mean, for example, that several Southeastern states home to large African American populations aren’t reporting the total number of tests administered at the county level?
If you’re interested in exploring this data set or contributing to it, you can find it on github.
Why is a climate scientist looking at this at all?
I am not a public health expert or an epidemiologist, so I can only draw from what has previously been studied and reported to make sense, from a scientific perspective, of what is happening with respect to community-level race and COVID-19 testing. Undoubtedly, it will take the work of experts in these specific scientific fields to truly understand the factors that drive the disparities in test results I report here.
What strikes me about these results, though, is that many of the negative forces that have led to communities of color being disproportionately impacted by the coronavirus may be the same forces that leave these communities more exposed to and impacted by climate-related risks.
For example, for many years, policies like redlining limited Black Americans’ housing options. While redlining is no longer legal, it’s legacy casts a long shadow such that today, people in communities of color are less likely to have access to healthy food options than other communities. That lack of access may be contributing to higher rates of heart disease among Black Americans.
Heart disease may increase the risk of experiencing a severe case of COVID-19. But heart disease can also be exacerbated by extreme heat, meaning that the increasingly frequent extreme heat expected in the coming decades due to climate change could have an outsized impact on communities of color.
One can make the same argument for factors such as lack of access to affordable housing (overcrowding increases the risk of virus transmission; the urban heat island effect increases the temperature of cities during heat waves), or health care (underlying conditions left untreated can affect the ability to recover from COVID-19; they can also exacerbate the risk of heat-related illness).
As Juan Declet-Barreto points out in his recent blog post, many of the communities most vulnerable during the COVID-19 crisis are the same communities that stand to be hit hardest by climate change.
Solutions are complex, but data collection is a start
Addressing and preventing disparities in the impacts of crises like pandemics or climate change on different demographic groups will require us to rip at the stitches that have held the fabric of America’s racism together for hundreds of years. Equitable solutions to the current pandemic must be developed with the full engagement of the people and communities most affected and with the recognition that we are all in this together.
In the near term, we need to deliver clear, science-based information on who is being affected by COVID-19 and where, including the number of tests administered at the community-level. With only a select number of states reporting the race of COVID-19 cases and race data missing for 65% of cases reported to the CDC, the need for data is acute. Over the longer term, we must address environmental injustice and work to, among other issues, reduce the pollution that puts low-income communities and communities of color at a disadvantage, rather than allowing the Trump Administration’s EPA to do the exact opposite. The policies we put in place during this crisis for protecting the most vulnerable among us could very well alleviate the impacts of the next one.
Support from UCS members make work like this possible. Will you join us? Help UCS advance independent science for a healthy environment and a safer world.