Oops.
Texas health officials made a key change Thursday to how they report data about the coronavirus, distinguishing antibody tests from standard viral tests and prompting slight increases in the state’s oft-cited daily statistic known as the positivity rate.
The positivity rate is the ratio of the confirmed cases to total tests, presented by the state as a seven-day rolling average. The Texas Department State of Health Services disclosed for the first time Thursday that as of a day earlier, it had counted 49,313 antibody tests as as part of its “total tests” tally. That represents 6.4% of the 770,241 total tests that the state had reported through Wednesday.
Health experts have warned against conflating the tests because they are distinctly different. Antibody tests detect whether someone was previously infected, while standard viral tests determine whether someone currently has the virus.
Now that DSHS is reporting the number of antibody tests, it has recalculated its daily positivity rates starting Tuesday to exclude such tests. That led to a 0.41 percentage-point increase in Tuesday’s rate and a 0.55 point increase in Wednesday’s rate, according to DSHS calculations.
DSHS acknowledged last week that it was reporting an unknown quantity of antibody tests as part of the “total tests” figure. Despite that, Gov. Greg Abbott incorrectly claimed Monday that the state was not “commingling” the numbers while promising the state would soon break out the antibody test count.
[…]
When public health agencies combine antibody testing figures with viral testing figures, “I want to scream,” said Seema Yasmin, an epidemiologist and director of the Stanford Health Communications Initiative.
Viral tests, usually taken from nasal swabs, can detect an active coronavirus infection. If a person’s biological sample is found to have traces of the virus’s genetic material, public health workers can order them to self-isolate and track down any of their contacts who may have been exposed.
Antibody tests “are like looking in the rearview mirror,” Yasmin said, because they may show if a person has recovered from a coronavirus infection. That can be useful for public health surveillance, but it does not offer much insight about where the virus is currently spreading. Another issue is that many antibody tests have been shown to have high rates of inaccuracy, she said.
“As an epidemiologist, this level of messiness in the data makes your job so much more difficult, and it misleads the public about what’s really happening,” Yasmin said. “We’ve been talking about the capacity for testing increasing over the last few weeks, but now we might have to tell the public that might not be true.”
And dumping antibody testing data into the pool of viral testing data brings the overall positivity rate down, reflecting “a deceptive misuse of the data,” analysts for the COVID Tracking Project wrote last week. That’s because the numbers may make it seem like the state has grown its testing capacity even if a state’s viral testing capacity remains flat.
“This is crucial as we need increased capacity for viral testing before reopening to identify active infections even in the pre-symptomatic or asymptomatic stages,” the analysts wrote.
To be fair, Texas is not the only state to have done this. Florida and Georgia have been accused of manipulating their data in other ways as well. The bottom line here is that we’ll never get our arms around this pandemic if we don’t have good data. The data is messy enough as it is, we surely don’t need to be making it worse.
Pingback: How about some antibody tests? – Off the Kuff