Across Two Waves: COVID-19 Disparities in Massachusetts
By Mark Melnik and Abby Raisz, Economic & Public Policy Research (EPPR) group at the UMass Donahue Institute
Jessica Pearlman, Institute for Social Sciences Research at UMass Amherst
With support provided by Ian Dinnie and Ellen Aron, Research Associates at EPPR
December 18, 2020
As we approach the end of 2020, we stand at an interesting time in the COVID-19 pandemic. It has been nine months since Massachusetts—and many other states—declared a state of emergency. Over that time, public policy makers have put in place a wide array of measures and policies aimed at stopping the spread of the virus and helping relieve pressure on the public health infrastructure.
As is well-known by residents of this state, the outbreak in Massachusetts was particularly severe early in the pandemic, when several northeastern states (including Massachusetts, New York and New Jersey) suffered extremely high COVID-19 caseloads. More recently, we’ve seen dramatic increases in testing across the state, and lower positivity rates, cases per capita and death rates. Despite these positive indicators, broad concerns remain as Massachusetts and its most vulnerable populations struggle through the second wave of COVID-19 infections.
Massachusetts, race and COVID-19 deaths
The White population has accounted for roughly 76 percent of total COVID-19 related deaths in Massachusetts, a disproportionately high share given its share of total COVID-19 cases (32 percent). This pattern is contrary for the Black and Hispanic/Latinx populations, whose shares of total deaths are lower than their shares of total cases.
While the White population comprises the largest share of overall deaths, this crude measure does not account for age differences across racial groups. Massachusetts’ White population is significantly older than its populations of color, and therefore—because age is a leading vulnerability to the disease—has experienced higher rates of hospitalization and death (the average age of COVID-19 deaths statewide is 82). Further, Massachusetts nursing homes—whose congregate living situations add further vulnerability—are 88 percent Whitei, which is more than 10 percent higher than the White share of the total population. Age adjusting COVID-19 death rates allows for a more accurate comparison between groups with different age distributions.
Age-adjusting COVID-19 death rates essentially normalizes the age distribution across the different racial and ethnic groups in the state. Because the Massachusetts Department of Public Health (DPH) does not release deaths by age and race jointly (a necessary metric for precise age-adjustments), an estimate was calculated using data by race and age separately, supplemented and weighted with Census data (see methodological note in the figure below). DPH stopped reporting deaths by age on August 12th, and deaths by race on November 2nd. Therefore, rates are calculating using data from August 11th, the latest release including deaths by both age and race. The resulting estimates show what the death rates for each racial and ethnic group would be if the age distributions were the same across each group. Toggle the buttons below to observe the difference between age-adjusted and crude mortality in Massachusetts. When age-adjusted, the rates for Asian, Black, and Hispanic/Latinx deaths increase dramatically, while the White death rate decreases. The Hispanic/Latinx community in particular, largely concentrated in COVID-19 hotspots such as Chelsea, Lawrence, Revere and Lynn, sees a whopping 238 percent increase in deaths per 100,000 people when adjusted for age.
In short, if all racial and ethnic groups had the same age distribution across communities in Massachusetts, death rates for the Black, Hispanic/Latinx and Asian populations would be significantly higher than those for the White population. Ultimately, age-adjusted rates highlight a more accurate picture of COVID-19’s impact on the state’s communities of color.
The different waves of COVID-19, place and equity
Communities hardest hit during the “first wave” of the pandemic were also communities with the highest concentrations of populations of color, people living in overcrowded housing, low income households, and frontline workers.ii Since March, the Commonwealth has made a concerted effort to increase access to free, asymptomatic testing in known COVID-19 “hot spots” such as Chelsea, Brockton, Everett and other Gateway Cities in an attempt to “Stop the Spread.”iii Despite these efforts, municipalities with the highest per capita rates in Wave 1 (April 22–June 10) remain strikingly similar to those in Wave 2 (September 30–November 18).iv
Overall, cases per capita in Wave 2 are higher than they were in Wave 1. This increase can be partially attributed to increased access to testing. While significantly higher than in the summer and early fall, the number of daily confirmed hospitalizations has not reached even half of what it was in April and May, partially due to more younger populations contracting the virus (less likely to experience severe or fatal symptoms) than in the spring.
Cities ranking high in the first wave still rank high in the second wave, though most have seen a significant drop in cases per capita. In Lawrence, where 80 percent of the population is Hispanic or Latinx and 17 percent live in poverty, there has been a dramatic increase in cases per capita, with only a modest increase in testing compared to the state average. The persistent inequalities that underscored the first wave are just as present in the second, as the state’s most vulnerable populations continue to bear the brunt of COVID-19’s most harmful outcomes.
What are the most important community factors in understanding the COVID-19 outbreak?
When we first examined equity and the COVID-19 outbreak in Massachusetts, we focused primarily on the social determinants of health, particularly in our Gateway Cities, and the incidence of COVID-19. Our analysis showed a clear and strong correlation between the outbreak of COVID-19 in a given community and that community’s population share that is people of color, lives in overcrowded housing, works in frontline jobs, and earns lower incomes. Additionally, average household size and population density were correlated with COVID-19 spread in a community, albeit a bit less strongly. Of course, a number of these concepts are intertwined in the social world. When the relationship between two variables is examined in isolation, any association observed may be in fact due to one or more additional variables that are correlated with both variables in question. For instance, frontline workers often earn low incomes and may also live in crowded housing in densely populated areas. As a result, it is important to determine which of these factors are most strongly associated with the likelihood of being infected with COVID-19 and subsequently the most likely to drive COVID-19’s spread. To do this, we developed a multiple regression model to examine how these factors impact municipal COVID-19 rates. Simultaneously including multiple variables enables us to disentangle the results of multiple factors. We included the following variables:
- Percent of “overcrowded” housingvi
- Population density (persons per square mile)
- Percent of workers who are ”frontline”vii
- Percent of population who are persons of color
- Percentage of the population who are over age 65
- Average household size
- Poverty rate
- Per capita income
We found that the strongest predictors of COVID-19 rates were overcrowded housing, average household size, per capita income and population density. Specifically:
- For each additional percentage point of households in a community that are “overcrowded,” the number of COVID-19 cases increases by 35.0 per 10,000 members of the population.
- For each additional thousand dollars of per capita income, the number of COVID-19 cases decreases by 3.6 per 10,000 members of the population.
- For each additional 1,000 persons per square mile, the number of COVID-19 cases increases by 11.8 per 10,000 members of the population.
- As the household sizeviii increases by 0.1, the number of COVID-19 cases increases by 24 per 10,000 members of the population.
The above findings for overcrowded housing and per capita income are illustrated in the following figures, which hold constant all other variables except the one being modeled. The range of values on each X-axis corresponds to the entire range for the respective factor in the data set.
Benchmarking Massachusetts against other states on COVID-19 indicators
Massachusetts and other northeastern states were hit hardest during the pandemic’s first wave, experiencing high death rates and daily hospitalizations. Despite the recent spike in cases across the United States and including Massachusetts, the state has fared relatively well (emphasis on relatively). This is not to minimize the current crisis nor the concerns as we enter winter months. Overall Massachusetts boasts one of the highest rates of testing and lowest positivity rates nationwide. This pairing of information is a critical distinction because more expansive and readily available testing, particularly among people with mild or no symptoms, will naturally uncover more total cases in the population. With that, the positive testing rate is important in understanding the spread of the virus. If the positive testing rate is staying steady or going down, it would indicate some level of control over the spread of the virus (especially if the “steady” rate is relatively low). If the positive testing rate goes up, on the other hand, it would indicate increasing spread of the virus in the population. In the last seven days (as of Dec. 10) Massachusetts had the highest number of tests administered per capita in the nation, due to widespread access to testing for asymptomatic individuals. The share of positive tests has averaged around three percent, the fifth lowest rate nationwide. That said, as the case rate continues to rise locally and nationwide, public policy makers, public health officials, and the general public should continue to be vigilant in prevention precautions as we progress through the winter months and wait for broad circulation of the vaccine in 2021.
When considering death data over the full course of the pandemic, Massachusetts has the second highest overall deaths per capita, largely attributed to nursing home deaths concentrated in March, April and May, during the early moments in the pandemic. In Massachusetts, two-thirds of COVID-19 deaths have occurred in nursing homes (compared to 42 percent nationwide). This pattern is similar across New England, which tends to have older populations than other parts of the country. Narrowing in on deaths per capita over the last seven days, this rank falls 25 places, as young people now comprise a larger share of cases. Massachusetts also has one of the lowest shares of people with underlying medical conditionsix, which decreases the risk for severe COVID-19 illness or fatality.
The total number of COVID-19 related deaths is also likely higher than reported. The CDC, state departments of health and various news outlets have analyzed “excess deaths,” or deaths higher than the typical number during the same time period in previous years. Closer inspection of death data during the first wave of the pandemic demonstrates that COVID deaths were actually undercounted, rather than overcounted. We conducted an excess death analysis for the period from March 1 to May 31, and found that there were 6,846 reported COVID-19 deaths in Massachusetts, and an estimated 7,545 excess deathsx statewide, signifying an estimated 700 unreported deaths directly or indirectly related to COVID-19 (a roughly 10 percent undercount).xi
Conclusion
Massachusetts was put through the wringer during the pandemic’s first wave and, while the state has weathered the second onslaught somewhat better than most states, disparities in race, housing and income remain a paramount concern as we near 2021. Data limitations also persist, as statewide reporting on race, ethnicity, age and other factors remain inconsistent across time and place. As the Massachusetts Department of Public Health continues to release new detailed data on COVID-19, there is great potential for a more granular understanding of the virus as it affects the state’s communities of color and other vulnerable populations.
i Based on data from Brown University’s LTCFocus, the Centers for Medicare & Medicaid Services and the state of Massachusetts.
ii https://donahue.umass.edu/our-publications/donahue-data-dash-inequalities
iii https://www.mass.gov/info-details/stop-the-spread
iv Wave 1 is defined as April 22 to June 10; Wave 2 is defined as September 30 to November 18. Waves were based on a seven-day rolling average of COVID-19 cases and availability of analogous data. Cases were at an all-time low during summer months (mid-June – August). Since the writing of this brief, new data show that Wave 2 continued into December and cases are now dropping. These graphics therefore reflect only a portion of the second wave.
v July 22 is the earliest date for an analogous comparison of tests administered. Prior to this date, data were reported on number of persons tested.
vi 1.01 or more occupants per room as a share of total occupied units
vii SOC codes 21, 29, 31, 33, 35, 37, 41, 53. See bls.gov/soc/2018/major_groups.htm for more information.
viii All municipalities have an average household size between 2 and 3.
ix Includes chronic kidney disease, chronic obstructive pulmonary disease (COPD), heart disease, diagnosed diabetes and obesity. The data reflected are based on an analysis of the 2018 Behavioral Risk Factor Surveillance System (BRFSS) survey for which questions were available and U.S. Census population data among U.S. adults in 3,142 counties.
x The CDC defines excess deaths as the difference between the observed numbers of deaths in specific time periods and expected numbers of deaths in the same time periods. Expected numbers were calculated by the UMass Donahue Institute using five-year average monthly age-adjusted death rates in March, April and May from 2015 to 2019. Data courtesy of the MA Vital Statistics Registry via the Boston Globe.
xi It should be note there was a subsequent decrease in some other types of deaths, such as a flu, during the COVID-19 pandemic