# Screen the Kids!

## Those little humans who are not (yet) eligible for a coronavirus vaccine may prolong the pandemic.

As we face a new struggle to get covid-19 vaccination rates up in this country, we need to remember that there is a group of people with virtually zero vaccine uptake. This group often congregates together in indoor gatherings, coming into close physical contact for extended periods. Fully 24% of Americans are part of this group.

We call them children.

And, as I am putting this together, there is currently no FDA authorized vaccine for kids.

Might that population of children form the reservoir for subsequent COVID outbreaks? While data is pretty clear that safe school re-openings don’t drive community COVID-19 positivity rates, there is still concern that kids — who are more likely to be asymptomatic — may become vectors of infection anyway. Are they our Achilles heel, and if so, short of authorizing some vaccines for them, is there anything we can do about it?

A new paper in JAMA Network Open suggests we need to be screening kids for asymptomatic infection. We need to be doing it now, and we need to be doing it fast.

Now, this is a simulation study — a complex set of mathematical models that operate under varying assumptions. As an empiricist, I tend to favor real-world evidence, but simulations can be useful — particularly when they reveal relatively large effects that are robust to a bunch of assumptions — and that’s what we have here.

Ok here’s the setup. Assume you have a population of adults and kids, age distribution exactly like that of the US. COVID-19 is spreading in that population at a certain rate (we can play with that rate, called the effective reproduction number — in the base case they use 1.2 meaning that on average each infected person infects 1.2 additional people). This is the sort of transmissibility we see now given good social distancing and masking — but new variants and loosening of restrictions could push it higher.

Adults can be vaccinated, and we can play with how fast and what percent of adults get the vaccine (they range coverage from 40% to 60% of adults, which I hope is a bit too conservative).

We can estimate how effective the vaccines are at preventing *disease *(95%, based on the mRNA trials), and how effective they are at preventing infection (which includes asymptomatic cases — they model that they get around 90% efficacy here).

Finally, we can ask how quickly we can identify, and isolate a new — potentially asymptomatic infection.

This final term — the speed with which we can find a newly infected person — is the basis of this analysis. Put that all together and you can ask a big question — what percent of the population will be infected after a year under these various conditions — the population attack rate.

For context, we’ve had coronavirus for about a year in the US — and our best estimates are that somewhere between 20 and 30% of us got infected in that period of time. The authors target a 5% infection rate over the next year as “success”, allowing them to figure out exactly how fast we have to detect new silent cases (which include asymptomatic and presymptomatic cases) , to achieve that level of control.

To keep things simple, they first modeled the outcomes in the absence of any vaccine and without any effort to detect silent infections. In that situation, the model predicts that around 10% of the population would be infected by the end of the year. That number isn’t 20 or 30% thanks to the fact that a bunch of us have already been infected, which reduces the effective reproduction number.

But what if we could identify just 10% of silent infections in the population using screening tests and contact tracing? Well, if we could do that within 2 days of the individual becoming infectious, we’d drop the overall attack rate to 3%.

But the model is clear — speed is of the essence. If there is a 5-day lag between infection and detection, we’d only drop the attack rate to 9%.

In fact, speed is so important, that detecting 40% of silent infections within 5 days has the same effect as detecting just 5% within 2 days.

Of course, we have vaccines. At least, for adults.

So the authors asked — what if we focused our efforts entirely on detecting silent infections in kids?

I’ve collated the results in this graph. It looks pretty similar to what I’ve shown you already — the base case without detection of silent infection gives us a population attack rate of around 9% in the next year.

But detecting 10% of silent cases in children within 2 days of infection, cuts that rate to 5%. Children are just one quarter of the population in the US. We’re talking catching 2.5% of silent infections to essentially cut the attack rate in half. That starts to get interesting.

Yes, this is all simulation, but we can learn a lot from the big picture. The math says it’s not really how many silent infections we detect, it’s how quickly we detect them.

So — how do we detect silent infections quickly?

Well, screening for one. New rapid tests, some of which are saliva based, could be deployed where kids are — schools that is — to detect new infections before it’s too late.

Also contact tracing — remember that? If we can get the infection rate down to a reasonable number, professional tracers can seek out and test those who may have been exposed to a newly infected COVID patient quickly.

The important thing is that we don’t have to detect 100% of silent infections to make a dent. Just capturing 10% of infections in kids alone may have dramatic impacts — as long as we do it quickly.

What if kids become eligible for vaccines? Does that make this issue go away? Surprisingly, the model says no.

Without any effort to identify silent infection in kids, you’d need to vaccinate 81% of them to keep the overall population attack rate below 5%. In other words, more aggressive screening of kids may provide some pretty big bang for the buck.

Variation of all those assumptions — including how susceptible kids are to infection and how transmissible kids are when they are asymptomatic didn’t change the math too much. Changing the infectivity of the virus — the effective reproductive number — did a bit — with more dramatic effects of rapid screening seen when infectivity was high, and, as you’d probably expect, less exciting results if that infectivity rate was below 1.

It’s interesting — recently so much of our effort has been on the vaccination program, that we may have forgotten just how important testing and contact tracing are. They really do make a difference — and, until we have a lot of them vaccinated, those little humans who, thankfully, don’t get too sick when infected with COVID-19, may keep the pandemic going longer than it needs to.

A version of this commentary first appeared in medscape.com.