Yesterday, I published this graph:
I can’t make sense of their report/line chart. Y Axis label is “Cases Per Million Population”. Does that mean “Positive test results Per Million people who are alive”? It occurs to me that a chart indicating the ratio of “Positive test results of tests conducted” whereby the resulting number is properly dated and charted against the date whereby the positive test subject was tested. With a ratio measurement, one could view whether the number of positive results is increasing or decreasing relative to the number of tests detected.
I will do my best with this, but I’m going to have to turn over some rocks to answer this question properly. First off, with respect to the nation-wide green line, there is no universal standard for defining a “case”. My US case counts come from Worldometer, which aggregates state counts. I can speak to the California Department of Public Health methodology, which is used for the other two lines. The CDPH considers a case to be an individual who tests positive for Covid-19. The number of times that person is tested, as long as it is at least one, doesn’t affect the case count.* The CDPH reports cumulative cases every day, and I subtract the day’s count from the previous day’s count to get the number of new cases. I don’t think whether the person testing positive is alive or dead enters into the calculation, they are both a new case, although I think there are a vanishingly small number of people in California who contract the disease and die before they are tested. There may be some significant number who contract the disease and die before their test results are registered.
You bring up the issue of where a new case is entered on the x axis. I am using the date the test result is reported for both California and Monterey County. Monterey County also provides information about cases categorized by date of symptom onset. I don’t use this information, for several reasons. The first is that I don’t know what they do with symptomless people who test positive. The second is that I’ve seen no definition of “symptom onset”. The third is that they are continually making revisions to those numbers, and it’s hard to keep current.
You ask for a chart that you call “Positive test results of tests conducted”. I don’t know how to produce the chart that you ask for, since more than two dimensions seem to be necessary. You say “With a ratio measurement, one could view whether the number of positive results is increasing or decreasing relative to the number of tests detected.” I do have this chart which plots the positivity ratio versus the number of tests performed:
I produced this a while ago to shed some light on the counterintuitive claim that more testing raises the positivity ratio, and it updates itself with new data, so it is current.
*California changed its methodology for counting tests performed last spring. Before that time, if a person was tested more than once, that still counted as one test. Ever since then, each test is counted separately. This means that the positivity ratios aren’t quite what they seem, since the numerator is per individual and the denominator is per test.
Leave a Reply