Click here to download a PDF version of this post
The Virginia Department of Health (VDH) is changing the way it presents race and ethnicity data. Since COVID-19 has impacted some communities more than others, It is important for the public to understand these health disparities so they can respond appropriately. In some cases, reporting data in large groups, for large geographic areas, can mask disparities. However, reporting data at too fine detail can put individual privacy at risk. This is especially true for groups with relatively small numbers in Virginia such as Native Americans or Pacific Islanders.
Previously, we published data for ethnicity and race separately. With the exception of the “White” and “Black or African American” categories, all other races were combined into a single “Other Race” category. The new reporting categories are:
- Asian or Pacific Islander – Non-Hispanics who identify as “Asian” or “Native Hawaiian or Pacific Islander”
- Black – Non-Hispanics who identify as “Black or African American”
- Latino – Individuals of any race who identify as “Hispanic or Latino”
- Native American – Non-Hispanics who identify as “American Indian or Alaska Native”
- White – Non-Hispanics who identify as “White” alone
- Other Race – Non-Hispanics who select “Other Race” alone
- Two or More Races – Non-Hispanics who select more than one of the above race categories
The new reporting categories allow us to provide more detail on racial and ethnic health disparities while preserving individual privacy. Additionally, combining the ethnicity and race questions into a single reporting category reduces the amount of missing data from about 33% for both race and ethnicity questions, to about 26%. This is because we only need a response in one or the other categories for reporting purposes.
While these categories allow us to report data in most categories for the majority of local health districts, in a few cases, local health districts were combined for specific groups:
- Asian or Pacific Islander
- Lenowisco + Cumberland Plateau
- Eastern Shore + Three Rivers
- American Indian or Alaska Native
- Lenowisco + Cumberland Plateau + Mount Rogers
- Pittsylvania-Danville + Southside
- Eastern Shore + Three Rivers
- Alleghany + Roanoke
The Results
With these changes, some results stand out. For instance, It is clear that COVID-19 is disproportionately affecting Virginia’s Latino population. Although Latinos make up just 10% of Virginia’s population, current data suggest that they account for 45% of cases, 35% of hospitalizations, and 11% of deaths. It is important to note that there is no evidence that migration or travel from Central or South America played a role in bringing COVID-19 into the United States. While it is difficult to make direct comparisons due to some anomalies in the disease reporting system, it is clear that this population is being impacted significantly.
What about the missing data?
Although this new reporting method is an improvement over previous methods, it is hard to ignore the elephant in the room: the large amount of missing data. This missing information makes it difficult to draw any firm conclusions about the cause(s) of these disparities. For example, it is possible that factors such as age and geography may play a more important role in the observed health, and related social and economic disparities, than race and ethnicity alone.
VDH is pursuing multiple strategies to address the problems caused by missing data. First, we are encouraging individuals, health providers, and laboratories to be more diligent in reporting race and ethnicity data. Second, we are trying to match records from multiple data systems to fill in some gaps. Finally, we are exploring statistical imputation methods that use existing information (e.g., census tract data, geolocation, surname) to impute missing data (Note: VDH will always report and label imputed data separately from official data).
What does this mean?
Race and ethnicity are complex issues. A government organization like VDH reporting data using race and ethnicity groups can never fully reflect how individuals identify. While individuals select race and ethnicity categories themselves, we use r categories based on Federal standards. This allows us to match with federal data and better measure disparities in health at the population level.
Health disparities between populations are not caused by a group’s race or ethnicity. They reflect societal factors like geography, access to healthcare, poverty, and racism that may disproportionately affect people of color. Statisticians with the Office of Health Equity will use the data collected from the current pandemic to identify those health inequities and inform tracking and reporting guidelines so that Virginia will be better prepared to address similar crises in the future.
