A: Let’s be sure we know what these numbers are telling us, and what they are not. The Florida Sun Sentinel recently reported that “Nearly one-in-three children tested for the new coronavirus in Florida has been positive.” Specifically, that’s 17,000 total positive tests among children. What’s more, 4 Florida children have died of COVID-19.
What this means: AMONG THE CHILDREN WHO WERE TESTED in Florida, a pretty large number were positive (aka percent positive or positivity rate).
It does not mean that children are at higher risk than adults, higher risk than we thought, or that 1 in 3 children in Florida are infected.
Several of the contributors here at Dear Pandemic have demography training, and there is almost nothing that demographers love more than a denominator problem. This is a great example of when we need to be asking ourselves this key question: who is in the denominator?
The denominator here is kids-who-were-tested. When we divide the number of positive tests by the number of tests performed (for one age group only, in this case), we get the percent positive, aka positivity rate. In recent days in Florida, we have 30% positive tests for kids. That is a high percent positive. (See also our previous post on positivity rates).
Instead, we might want to know just how common a pediatric COVID-19 case is in Florida, or how many kids have gotten sick so far in the pandemic relative to the number of kids overall (the incidence proportion). To get that, we can take those 17,000 positive tests among children and divide them instead by the total child population of Florida, which is about 4.2 million. This comes out to an incidence proportion of ~4 cases per 1,000 children so far this year. That is also pretty high.
Or, we might also want to know whether kids are at higher or lower risk than adults. To do that, we need to compare the incidence proportion of kids to adults. So far this year, in Florida, the incidence proportion for adults is about 16 cases per 1,000 adults. That’s also a lot–and it’s much higher than the incidence proportion for children. There’s a lot of COVID-19 circulating in Florida (in all age groups).
Four pediatric deaths is unquestionably four more than we would like to be reporting. If we want to compare risk of death for kids vs. adults, we need to look at the case fatality rates. Four out of 17,000 confirmed cases so far in Florida works out to a case fatality rate of 0.02% for kids. Compare to the observed case fatality rate of 1.9% (~5,000 deaths/271,000 confirmed cases) for those over the age of 18. (Here’s our previous post on case fatality rates too).
One final note is that the kids-who-are-tested number itself is not a random sample of all kids. The kids who are being tested are mostly also kids who are symptomatic. In fact, kids may be even less likely to get tested when they’re not symptomatic than adults, because let’s be honest… no one really wants to take their perfectly well child to a testing site to have an uncomfortable double-nose swab performed.
This is supported by the numbers: the percent positive for kids (30%) is MUCH higher than the percent positive for the population as a whole (11%). So keep in mind that we don’t get much information about asymptomatic kids from this, and that we might not be comparing apples to apples when we look at confirmed cases for kids and adults.
What have we learned?
– There’s an awful lot of disease circulating in Florida.
– These numbers, though scary, support what we already knew: compared to the percentage of adults with infection, children are less likely to have symptomatic COVID-19 and have lower risk of death.
– Kids are probably being tested mostly when they have symptoms, which means we still know very little about how many asymptomatic kids there are.
– Always ask yourself who is in the denominator.
Emily Oster had her own helpful take on the Florida report, and other new kid-related COVID info, in her special weekend update on Saturday.
Original story in the Florida Sun Sentinel.
