Post #691: GIGO, CORRECTED

Source:  Unusable test count data from the Commonwealth of Virginia.

Edit:  Some time after I posted this originally, Governor Northam tweeted that the antibody-based (after-the-fact) tests account for just 9 percent of testing, and that he has directed the Virginia Department of Health to break out the two types of tests (viral DNA versus blood antibodies) separately in its data reporting.  Basically, you can ignore the rest of this now.

GIGI is a computer programming initialism:  Garbage in, garbage out.  It means that even if you have a program that does exactly what it’s supposed to do, if you feed false information into it, you’ll get false information out of it.

There are two types of tests for COVID-19.  One is a viral DNA-based test to tell whether you actively have the virus on your mucous membranes, often called a PCR (polymerase chain reaction)-based test.  It’s the test of whether you are actively infected.  It’s test used to determine how the health care system will address you, whether or not you need to quarantine, and whether you could as a newly-infected cases.

Separately, there’s a test for antibodies in your blood.  That’s an after-the-fact test, and tell you whether you were, at some time in the past, infected.  Typically, those antibodies only show up weeks after infection, at which point, you typically are no longer carrying or shedding the virus.

I got an email today, from a colleague pointing me to an article in yesterday’s Atlantic.  Turns out, the reason Virginia’s testing numbers started to go way up, without a commensurate rise in count of infected persons, is that they started combining the count of antibody tests with the count of DNA tests.  At least, that’s what was reported today in The Atlantic.  Thankfully, Virginia  has the sense not to include positive antibody tests among the count of infected cases, per their May 7 posting on how they calculate testing rates.

But the upshot is that this graph, from the Commonwealth, is no longer interpret-able as showing that testing to find infected people has gone up.  Fact of the matter is, unless the Commonwealth chooses to separate out the viral DNA (polymerase-chain-reaction or PCR test), you have no idea whether testing for active coronavirus infection has gone up or now.

All the more reason to understand that the increase in new infections isn’t an artifact of increased testing, as I discussed in an earlier post.  In fact, based on what Virginia reports, we can’t even be sure that there is “increased testing”.  At least, not of the sort that is used to find infected individuals.