June 11, 2020 - Some (Revealing, but Probably Deeply Skewed) Numbers about COVID by Race or Ethnicity

Willamette Week has this great chart today, showing the incidence of COVID-19 in Multnomah County by race or ethnicity.   The article has an arresting headline and subhead:  

Multnomah County Discloses Massive Ethnic Disparity in COVID-19 Testing 

as It Moves Toward Reopening June 12

Hispanics get COVID-19 at a far higher rate than whites and other groups

 but get tested far less often.*

There are two charts in the article, but one is easier to grasp than the other. That's the chart of cases, which shows two percentages:  one bar represents the percentage of cases by racial or ethnic group, and the other represents the percentage of that group in the population at large.

It's easy to see the two outstanding disparities, for white and Hispanic groups. Hispanics get COVID at a high rate given how big a proportion of the population they make up. And whites are the opposite. 

It gets more complicated, when you consider testing:

In Multnomah County, Hispanics are 10% of the population, but account for 27% of the cases. ... White residents of Multnomah County have received about 10 times the number of tests Hispanics have received, yet whites account for only about 1.5 times the number of cases.

I had to squint at this over and over. In Oregon, we don't test many folks. (Oregon! We're number 49! We're number 49!) But I think we must test a lot of our white population when they go into the hospital (because they have health insurance and can afford to go), or when they ask for a test and have the time to go and get one, and we must test Hispanic folks mainly when, for instance, there is an outbreak in a food processing place that has a heavily Hispanic workforce, and everyone in that whole workplace gets tested. That's my best guess. 

In a (simple) world where COVID was infecting and being passed to everyone equally, the percentage of the population represented by each racial or ethnic group would equal the percentage of cases in each group. (I know you know this. I had to recite it to myself when I was looking at the chart.) If Hispanics made up 10% of the population, they'd have 10% of the cases. And if white people made up 70%, they'd have 70% of the cases. That's not what our numbers reveal, though again, if you test members of one group only when they have had some possible contact with an infected person, and members of another group when they feel like getting tested, your numbers will get pretty sketchy.  

*In the article, there's also a chart showing the incidence of testing, but it doesn't do a good "compared to what" job, as the "cases" chart does.