A few weeks ago, the Massachusetts Department of Public Health began releasing municipal and zip code-level data on COVID vaccination, including variables for age group, race/ethnicity, and sex. As we work to promote equitable access to the vaccine, these data are so important to help us identify disparities, figure out what might be causing them, and take action to address those that are rooted in equity concerns. I wrote in my last reflection about needing better visibility into temporal and community-level trends, so I was eager to take a look – and in the process, to try using Tableau to create a visualization that would be helpful to community members, partners, and my colleagues.
The visualization currently lives here, and includes three weeks of data for the eight communities where my work is focused. There are two key measures shown throughout the dashboard:
- Category-specific percentage of fully vaccinated individuals: This measure tells us the percentage of people of a given racial/ethnic group, age group, or sex who are fully vaccinated. This can be a useful indicator of inequities in vaccination, as it accounts for differences in the number of individuals in each group, by town, showing which groups are being vaccinated at higher or lower rates compared to other groups.
- Ratio of group’s share of fully vaccinated individuals to share of population: This measure takes the proportion of vaccinated people who are members of a given racial/ethnic group, age group, or sex – and divides it by the proportion of the total population that same group accounts for. This can be another indicator of inequity, as it compares the proportion we might expect, if everyone had the same likelihood of being vaccinated, to the proportion we actually see. A ratio of 1.0 means that a group’s share of the vaccinated population is exactly equal to that group’s share of the total population; a ratio greater than 1.0 means that group is over-represented among vaccinated individuals; and a ratio less than 1.0 means that group is under-represented.
It has been so helpful to get feedback from a variety of people on this visualization, so I imagine it will evolve over time. Here are a few reflections on what it tells us so far, and what questions it raises.
Across towns, Hispanic individuals are less likely than people of other racial or ethnic groups to be fully vaccinated, and are under-represented among vaccinated individuals to a greater degree than any other racial or ethnic group. Black and Asian individuals are under-represented among vaccinated individuals more often than not, though not as starkly as Hispanic individuals. On the other hand, White and multi-racial individuals tend to be over-represented, though there is some town-by-town variation in magnitude.
Why might this be? Vaccine eligibility has been largely predicated on age, and we know that the median age of the White population in Massachusetts is older than other racial and ethnic groups. It could be the case that fewer Hispanic people are eligible for the vaccine, and these percentages will shift dramatically in the weeks ahead. Perhaps this is the case for Black and Asian communities as well.
However, if this were the full explanation, I might expect to see ratios farther below 1.0 among people in younger age groups, and a more consistent pattern across racial and ethnic groups based on population age structures. It really stands out that Hispanic individuals are so much less likely to be fully vaccinated. Furthermore, eligibility has encompassed factors other than age, too – health care workers, first responders, people with certain medical conditions, childcare workers and teachers, long-term care and congregate care residents and staff – groups in which Hispanic people (as well as Black and Asian) are certainly well represented.
The disparities seem to point more to true inequities in access – to lack of transportation to vaccination locations, to limited translation and interpretation services on appointment websites and phone lines, to occupational barriers like lack of paid leave or unpredictable schedules, to the climate of fear created by immigration policy that clouds interaction with government or institutional authorities, and to limited access to culturally and linguistically appropriate health care in general. Coalitions like Vaccine Equity Now! in Massachusetts have called attention to these barriers, and their roots in structural racism and other forms of oppression. Hesitancy and misinformation may play a role as well, and it is always essential to ensure people have answers to their questions, delivered by trustworthy sources. But, it’s equally important to understand what exactly people are “hesitant” about. One recent survey suggests that Black and Hispanic people often have very reasonable concerns about side effects, costs, safety, and effectiveness which, once clarified, increase their likelihood of vaccine acceptance – rather than the source of hesitancy being ideological opposition. As others have written, there is a risk in over-emphasizing individual hesitancy or over-simplifying mistrust, distracting from the current structural and systemic barriers that we need to dismantle.
How will we know if efforts to equitably expand access to vaccination are successful? One way may be to track how the ratios change over time – the last tab in the visualization. Town by town, and as eligibility criteria expand to all adults, we would hope to see the category-specific ratios trend toward 1.0 – in other words, any given group’s share of fully vaccinated people should end up being equal to that group’s share of the population. Given the disproportionate impact of COVID in communities of color, equity would mean aiming for an over-representation of BIPOC individuals among the vaccinated population, especially in the near term.
These are not perfect measures. Racial and ethnic categories are overly broad, and obscure meaningful within-group differences. Children are not yet eligible for vaccination, but they are included in the population denominators. These percentages are not age-adjusted, so we have to understand how age-graded eligibility criteria may explain certain disparities based on population age structure – some degree of disparity may be expected, like if there is a long-term care facility in a certain community, and the residents of that facility are predominantly White. That said, this raises an important conversation about the ethics of predicating eligibility on age, when structural racism and other drivers of inequities in COVID infection and mortality are the same drivers that lead to premature mortality and lower life expectancies in certain BIPOC communities. Asian, Black, and Hispanic populations all skew younger, overall, compared to White populations in Massachusetts. A seemingly race-neutral policy can still have racially disparate impacts.
My hope is that this visualization can be a starting point for conversation, for people in our communities to ask questions about why patterns might be occurring as observed, and for engaging in sense-making and problem-solving together. I am learning in real time here, and greatly appreciate feedback, comments, and questions.