What factors correlate with COVID-19 death rates — and which don’t? Reports have tied virus fatalities to factors ranging from air pollution to race to socioeconomic status. Now, new research from MIT Sloan explores the correlations of coronavirus death rates with common variables — and raises provocative questions.
a professor of applied economics at MIT Sloan, authored a working paper, “What Does and Does Not Correlate with COVID-19 Death Rates,” with Bora Ozaltun, a graduate research assistant with the MIT Center for Energy and Environmental Policy Research.
Using linear regression and negative binomial mixed models, the researchers analyzed daily county-level COVID-19 death rates between April 4 to May 27 in relation to a variety of factors, including patients’ race, age, and health and socioeconomic status, as well as their local climate, exposure to air pollution, and commuting patterns.
In their paper, Knittel and Ozaltun underscore that these relationships aren’t causal. However, the researchers hope that policymakers can use this work to further identify mechanisms that do affect death rates and adopt policies accordingly.
Ahead, four unexpected takeaways:
1. Higher amounts of commuting via public transportation correlate with higher death rates. This is especially important as cities reopen and people begin returning to workplaces, Knittel warned. Notably, he said that the positive correlation also exists when driving or even walking to work, not just on a crowded, stuffy subway.
“There’s a risk of just going into work, which leads to higher death rates,” Knittel said. “Then there’s an additional risk associated with being on the public transit system itself.” Given that finding, policymakers might call for extended telecommuting and/or better disinfection of and social distancing protocols on subways.
2. African Americans and elderly people are more likely to die from COVID-19 relative to Caucasians and people under 65 — but not why you think. The researchers didn’t find correlation between ICU beds per capita, obesity rates, or poverty rates and COVID-19 death rates; their research controlled for income and health insurance. This runs counter to many common narratives about higher death rates among African Americans and raises questions about whether systemic causes could be to blame.
“Because we have a lot of variables in the model, we can separate how much of the correlation that has been documented elsewhere is coming from factors in our model, such as income and insurance differences, and how much of it is coming from factors not in the model,” said Knittel.
Specifically, the researchers found evidence that African Americans still have higher death rates, even after controlling for all of the variables in their model. “This correlation is not because [African Americans] have higher diabetes rates, [and] it’s not because they have lower incomes,” Knittel said.
In light of with those findings, policymakers should focus on other explanations that might be driving the disparity. “Given the national conversation that we’re having right now, policymakers should commit resources to understanding this,” Knittel said.
He speculated that quality of insurance or quality of hospitals could explain the positive correlation. “It could be other sources of discrimination that increase the stress levels,” he said. “I would certainly encourage policymakers to investigate this and try to overcome any bias that’s inherent in the system.”
Death rates aren’t correlated with obesity rates. While those with higher body mass index are considered to be more vulnerable to COVID-19, Knittel’s research didn’t find a correlation between obesity and death.
“The obesity rate is a puzzle,” he said, which warrants further investigation. “We plan to continue to update the analysis with hope that some of these puzzles can be explained as we learn more about this virus.”
4. Counties with higher home values have higher death rates. This is counterintuitive because richer areas would likely have higher-quality hospitals and offer better access to care. This point requires deeper exploration, Knittel said.
“This is purely speculation, but we tend to think higher home values are correlated with denser homes, living in cities, and so on. So that [point] might be picking up the fact that people just live closer to each other,” he said, such as in pricier high-rise urban apartments with less opportunity for proper distancing.
Challenging the common narrative
Knittel hopes that his work helps to challenge common narratives about COVID-19; he also plans to further track how states are containing the virus.
“Trying to better understand what’s driving the trends is going to be an important question,” Knittel said. “Our study doesn’t necessarily tell you what’s driving it, but it tells you what’s not driving it, which can be just as important for policymakers.”
