What the research says about reopening schools and why values, not data, are winning the day

Who knew being a school board member would involve so much philosophy?

Let me explain. Since March, school boards around Utah (and really, the world) have been put in the uncomfortable position of deciding when and how to reopen schools. Very few decisions in a pandemic are easy, but ones involving our children are particularly emotional because of that biological impulse to see our children succeed at nearly any cost. So when, for example, the Salt Lake City School District board met Tuesday to decide whether and when to hold in-person classes, it was always going to be a heated discussion.

But for most of the pandemic, we’ve been working to get data to help us understand these problems. Informed decisions are always better decisions, after all — frequently, we even delay making decisions until we get good data.

The good news is that in the past month, we’re getting some really quality research on how school opening decisions have impacted the pandemic, with multiple important studies with big sample sizes being published. The bad news? It’s not so clear what conclusions to draw from these studies.

And oh my, is it ever a rabbit hole of research to dig into. Let me show you what I mean, beginning with the most basic data: the number of cases in American kids.

On Jan. 1, the Centers for Disease Control and Prevention released accumulated data on cases among children in all 50 states. Smartly, they split it up into five age groups: 0-4, 5-10, 11-13, 14-17, and 18-24.

And they found huge differences. Overall, there were 2.8 million positive COVID-19 cases among these age groups between March and Dec. 12. A majority were in the 18-24 age group: 1.65 million cases, or 57%. On a per-capita basis, it’s the age group with the single largest prevalence of the coronavirus, children or adult.

The other youth groups saw way fewer positive tests. The high-schoolers recorded 467,000 cases, while 5-10 year olds had about a third less, with 313,000 cases in a slightly larger age group. The 11-13 year olds had 227,000 infections, while those kidlets 0-4 saw 212,000 infections.

Looking at the incident rates, you can make a case for both opening and closing schools. You may have heard people say “Cases for everyone started to rise greatly when schools went back in session.” That is true. But it is also true that “cases for kids in age groups who attend elementary, junior, and high school are significantly lower than the those for the rest of society.”

As we know, these case counts are dependent on the number of tests performed. And there are so many states and districts with different rules for returning to school that it’s hard to get a sense of what’s going on from such a basic picture.

So let’s take a look at a more direct study. The National Center for Research on Education and Choice compared school opening rules with coronavirus hospitalizations in essentially every U.S. county. They ended up with a lot of data to try to find a correlation between more open schools and fuller Intensive Care Units.

What they found is that there’s no sign of hospitalization increases in counties that opened schools — so long as the weekly rate of hospitalizations begins below 40 per 100,000 people. Salt Lake County’s hospitalization rate has always been well below that. Data collected by Utah company Qualtrics tried to look at that question using nationwide school data, and infection rates were found to be low.

But wait! There are plenty of reasons why places that prioritized virtual learning might have higher increases in hospitalization that have nothing to do with school. For one, school opening has been highly politicized — indeed, the decision to go virtual is less correlated with COVID prevalence than it is with Democratic vote prevalence. Democrats also tend to be elected in places with high minority populations and in places with denser households, both factors that correlate with sickness and hospitalizations. It is possible these studies are looking a bit too broadly.

So what if you looked at specific cases and whether they are tied to specific schools?

Governmental contact tracing operations would help, but our country’s efforts haven’t been good enough to do that. Americans don’t answer the phone calls from health departments, so it takes a huge effort to be able to do studies on much smaller groups.

Utah’s CDC study is an example, where 12 kids contracted the coronavirus in child care settings, then passed it on to 12 adults. But North Carolina researchers found 773 cases of coronavirus among 11 school districts, then found only 32 secondary cases from those students — with the huge caveat of we really have no idea how successful their contact tracing methods were.

Since other countries have been more successful with contact tracing, you might go overseas to find an answer. One study from Germany indicates that school openings did not increase cases, and another German case study only found a few cases from schools. Another found low transmission in Norway and Denmark. While Austrian studies that included regular testing of every student indicate similar levels in COVID prevalence and transmission in kids in school and adults.

Phew: this is the rabbit hole I was talking about.

I’ve steadily transitioned from describing one study with four paragraphs... to describing four studies with one paragraph. I’ll save you the trouble of the eight-study paragraph but I promise you, it’s possible. If you’re so inclined, the Education Policy Innovation Collaborative has a terrific bibliography of dozens of other studies.

You get an idea of the back-and-forth that this quantity of research presents. Still, we can tease out some trends. If I had to summarize these studies, I’d say we have:

• Really good and consistent evidence that open colleges lead to significantly enhanced community spread of the coronavirus.

• Some evidence that open high schools contribute to community cases.

• A majority of evidence that shows elementary schools contribute insignificantly or not at all to community cases.

Great. And yet whether or not school openings lead to community spread is just one of the factors our leaders are considering.

We also have to think about the long-term health of the kids who get COVID-19 in schools (even those without symptoms may face long-term concerns). We have to consider the short-term and long-term health of the kids who don’t go to school. We have to consider the economic costs of reopening with even a modicum of safety, and if those funds could be better used elsewhere. Oh, and the lives of teachers and staff are important too — one study estimated 42% to 51% of them may be at elevated risk in one way or another.

In each of these topics, you can research down to a huge degree of specificity. The school opening problem is one that touches nearly every aspect of our society.

It all reminds me of one of the most fascinating properties of the online encyclopedia Wikipedia. Starting on any random page of Wikipedia, if you click the first link of that page, then repeat the process, there is a 97% chance you will end up on the encyclopedia’s page for philosophy. Heck, try it out.

In the end, there’s so much information out there about school reopenings that any school board member’s decision is likely to come down to their philosophy about which research to emphasize. Even that might be charitable: much of the decision might just be from the core values they find most important.

That explains the disagreements among Salt Lake City’s school board, which is moving to a hybrid in-person/online model for older students in February. It explains why other Utah districts moved to allow students in class earlier and why some Utah lawmakers are frustrated at Salt Lake City’s hesitance. All of them can point to really quality studies to back up their opinion.

This sort of thing is very frustrating to scientists — they want there to be one right answer, and it’s their job to find it! But the truth is in hyper-complex areas of decision-making, there may not be that definite right answer. In circumstances such as these, even people with all of the data at their disposal are going to go with their prior convictions. And of course, those decisions are going to impact the data we see in the future.

It pains this data nerd to admit it, but sometimes, the data just doesn’t have much of an impact. Sometimes, philosophy rules the day.

Andy Larsen is a data columnist. He is also one of The Salt Lake Tribune’s Utah Jazz beat writers. You can reach him at alarsen@sltrib.com.