COVID-19’s Disparate Impact on Low-Income Communities of Color

This research brief was written in partnership by Boston Indicators and the Economic and Public Policy Research team at the UMass Donahue Institute (UMDI). UMDI has additional data available online at the Donahue Data Dash.

COVID-19 has hit hardest many populations and communities that were already among our state’s most vulnerable. For physical health reasons, the virus has been most deadly for people with preexisting conditions, which has largely meant that the elderly are at especially high risk. But Massachusetts cities and towns with more crowding in homes, larger communities of color and larger low-income communities have also seen higher COVID case rates (places like Chelsea, Lynn and Everett). These socioeconomic factors shouldn’t make people more likely to catch a virus, but through a range of “social determinants of health,” physical health intersects with the broader social context in which people live. Many of these lower-income communities of color, for instance, have higher shares of frontline workers (e.g., nurses and grocery store clerks) who risk infection every day by going to jobs providing essential services. These workers often only have access to lower-quality health care, and some have no health insurance at all. Many of these cities are also “environmental justice” communities that for years have suffered from worse air quality, which appears to lead those who do get infected to experience more severe symptoms.

The table below summarizes COVID cases for the 100 largest Massachusetts cities and towns and shows this side-by-side with some select socioeconomic data. Click on column headers to sort.

COVID case rates by municipality can be skewed a bit by the location of large group quarters, especially assisted living facilities, which have been centers of many of our state’s largest outbreaks. At the height of the pandemic in Massachusetts, Holyoke, for instance, had the 14th-highest COVID case rate statewide, but this high ranking was in part because the Holyoke Soldier’s Home for elderly veterans is located there. The Holyoke Soldier’s Home had the highest number of COVID-related deaths of any long-term care facility in the country. If you subtract positive cases from just that one facility, Holyoke would have still ranked high, but would shift down somewhat. If you subtract just the resident cases, Holyoke would fall to 18th and if you also subtract employee cases, Holyoke would’ve declined to 28th.

It’s important to note that information in this section relies on provisional data from the Massachusetts Department of Public Health (DPH). There are large gaps in what’s been released, although the quality of data is improving. This analysis focuses on positive COVID-19 tests, which is far from a perfect measure. Recorded cases rely on tests administered. While testing has been limited everywhere, it also has varied widely across communities. While there’s no doubt Chelsea has a large outbreak of COVID-19, part of why its positive test rate so far outpaces every other city is that Mass General Hospital ramped up testing in Chelsea early and operates a clinic there, offering tests to all Chelsea residents who are symptomatic regardless of patient status, health insurance, or immigration status.

Next we look at the relationship between people who have tested positive and other socioeconomic variables through a series of scatterplots. There are high levels of correlation between each of these individual socioeconomic variables and higher COVID case rates. This can make it seem as if each variable on its own leads to higher COVID rates, when this may not actually be the case. Taken together it does appear that these socioeconomic variables help explain significant COVID variation across different types of cities and towns, but it is difficult to determine how much of the impact is driven by each individual factor.

First we look at the share of households in a given community who live in crowded homes. Researchers use different definitions of “crowding” in different contexts. For this project we look at the number of housing units with more than 1.0 occupants per room (including bedrooms, living rooms, kitchens and bathrooms), and we find a very strong correlation with higher COVID case rates (an R value of 0.76).

For a virus that is spread primarily through close physical proximity, it makes sense that we’d see worse outbreaks in more densely populated cities, and this dynamic has been a topic of news coverage as the virus has spread. But the connection between overall population density (residents per acre) and positive case rates is actually not as strong as one might expect. While New York City has had a large outbreak other dense U.S. cities have not yet had episodes nearly as bad. Many Asian cities, some with population density even higher than New York’s, have not experienced the worst outbreaks either.

Multiple people crowding into a given housing unit, on the other hand, is a more precise measure of people living in close proximity to other people, and this measure has the strongest association with higher COVID case rates of any variable we tested. Chelsea, which has the highest COVID case rate also has by far the largest share of residents living in crowded homes, at just under 10 percent of occupied housing units in the city.

Somerville and Cambridge demonstrate the importance of making a distinction between population density and residential overcrowding. They rank first and second respectively among the 100 largest cities and towns in Massachusetts for overall population density, but they shift down to 30th and 27th when using the residential crowding measure. In terms of COVID case rates, neither of them is among the top half of communities, ranking 49th (Somerville) and 83nd (Cambridge).

Our findings here are consistent with a similar analysis conducted by New York University’s Furman Center, which looked across neighborhoods within New York City, finding that “COVID-19 is more prevalent in areas where more people reside in crowded units.”

Next we look at the share of a given community that is composed of people of color, which also shows a strong relationship with higher positive case rates (an R value of 0.73). This correlation is particularly strong for communities with large Latino populations (like Chelsea) and Black populations (like Brockton).

There’s also a strong relationship between positive COVID case rates and poverty, although it’s a bit less strong than the relationship with race. Poverty rates by community are available in the table above.

Finally, below we show the share of a community’s workforce made up of frontline essential workers. These employees risk infection everyday going to work as nurses, doctors, grocery clerks and transit drivers. There’s significant overlap between these two categories because a disproportionate share of people working in frontline service jobs are people of color. The correlation between frontline workers and COVID case rates is also quite strong, although a bit less strong than the correlation with people of color.

There’s also a strong relationship between positive COVID case rates and poverty, although it’s a bit less strong than the relationship with race. Poverty rates by community are available in the table above.