This story was originally published by Mother Jones and appears here as part of the Climate Desk collaboration
Ask any parent in the Mother Jones office and it’s clear they are eager to get their kids back to school. But as more schools move to reopen this fall—and as President Trump increasingly pressures and even threatens schools to reopen—experts warn there’s still quite a lot we don’t know about kids and COVID-19. What role do children play, for instance, in spreading this virus? Are they just as likely to get infected as adults? Are they just as infectious? And why have a small number of children developed a potentially deadly inflammatory illness after testing positive for COVID-19?
At this point, “following the science” to safely reopen schools simply doesn’t mean much.
Reopening schools may very well be a risk worth taking—after all, so much of our economic recovery depends on it—but it’s unlikely that we’ll have clear answers to many of these questions in the near future and know with any certainty just how safe schools would be for kids, teachers, and communities. While it feels like we’ve been living with the coronavirus forever, we’re still in the early phase of understanding it; the research on children and COVID-19 is relatively limited, and the studies that do exist have shown conflicting results. At this point, “following the science” to safely reopen schools simply doesn’t mean much—and it’s near impossible to try to rely on data to help you decide whether you should send your kids back to school.
As David Abramson, a professor of social and behavioral sciences at NYU’s School of Global Public Health, explains, getting kids back in classrooms means that students “will serve as potential vectors back into the community.” But at the same time, “it’s almost impossible to imagine not opening schools given all that is at stake. And so it’s like a devil’s bargain.”
It’s probably no surprise then that there’s no clear consensus among health experts about schools reopening. In late June, the American Academy of Pediatrics issued a recommendation that school administrators aim to have children “physically present in school” for the next school year. Meanwhile, the Centers for Disease Control and Prevention maintains that virtual-only classes are the lowest risk option for schooling. And while some countries, including Denmark, Austria, and Germany have had success in reopening schools, other countries, including Israel and Japan, and South Korea, have opened and re-closed schools after seeing a surge in new cases.
So, what gives? Why don’t we have a clearer picture of how kids transmit and are impacted by the coronavirus? Part of the reason is that conducting research on a new virus at lightning speeds is difficult enough in adults—for kids, it is significantly harder. Here’s a breakdown of a few key reasons why studying kids and COVID is so difficult—and why reopening schools now would mean doing so with a lot of uncertainty.
What people are reading
Many children with COVID-19 appear to be asymptomatic. That makes it hard to understand the full scope of the problem.
When it comes to kids and the coronavirus, one of the few things that is clear is that children can become infected with SARS-CoV-2, the virus which causes COVID-19, at all ages. “There are a number of papers that show that children can become infected,” explains Dr. Steven Zeichner, a professor in the Department of Pediatrics at the University of Virginia School of Medicine. “Some of them are symptomatic, there’ve been a few deaths”—but, he adds, “a large number are asymptomatic.” And for children who do show symptoms, he says, research shows their “disease course” tends to be milder than in adults.
A study of 2,135 pediatric COVID patients in China, initially published in mid-March in Pediatrics, for instance, found that more than 90 percent had asymptomatic, mild, or moderate cases. And in the US, a recent report from the CDC shows that about 5 percent of lab-confirmed COVID-19 cases between January 22 and May 30 were in people 19 years old or younger, or nearly 70,000 individuals; less than 50 people in that group died from COVID, according to the CDC’s analysis.
But because so many children are likely to show mild symptoms or be asymptomatic, the true number of cases is likely to be much higher than we know.
But because so many children are likely to show mild symptoms or be asymptomatic, the true number of cases is likely to be much higher than we know. “At this point, primarily, we’re testing people who are symptomatic, except in the case of health workers and others where we need to know if there’s been a lot of exposure,” says Dr. Cynthia Haq, a clinical professor and chair of the Department of Family Medicine at the University of California, Irvine, “and because children are less likely to be symptomatic, you’re less likely to be tested.”
All this begs the question that’s crucial for schools in particular: If children show milder symptoms, does that mean they aren’t spreading the virus as much? That’s not totally understood, Zeichner says. “The likelihood of transmission depends on the amount of virus that somebody is producing, and the interaction of the person with the virus with the person who isn’t infected,” he says. “It is likely that children and adults with fewer symptoms may be producing less virus, which probably makes them less likely to transmit the infection.” But researchers are still investigating whether that’s the case.
To study children, researchers have to jump through lots of hoops.
“When we conduct research on children, there are special levels of protection because children are more vulnerable,” Haq says. “We don’t want them to be exploited for research purposes.” As a protected group, children can’t formally consent to testing. In general, she adds, “we don’t like to conduct studies on children unless there’s clearly no evidence of harm from the study or definite evidence of benefit.”
On top of that, “you have to be really mindful of what you’re saying to a child,” says Abramson, who also directs a health disaster research center. “You can’t be saying things that are going to upset them. You can’t ask them questions in ways that will be potentially harmful to them. If you’re going to talk directly to children, that’s going to be a very difficult and challenging event.”
“We don’t like to conduct studies on children unless there’s clearly no evidence of harm from the study or definite evidence of benefit.”
For these reasons, random testing of children on a large scale is not always feasible. And if you’re going to rely on parents to report on their children’s health or behavior, research shows that’s not always reliable. For example, in a 2017 study, Abramson and colleagues asked a group of children and their mothers about the child’s mental health following the BP oil spill; they had a difference of opinion about a third of the time.
A lot of studies coming out right now aren’t always the “gold standard” of research.
If you wanted to answer the question of how much children spread the coronavirus, there are a number of scientific avenues you could take—hypothetically speaking. Randomized controlled trials are considered the most rigorous way to conduct research, but in this case, that might not be possible: an experiment in which you expose a group of children to the coronavirus and see how many become infected is, for obvious reasons, unethical.
Aside from that, you could:
- Observe transmission dynamics in a single household or group of households.
- Analyze data from a large population.
- Build a model to estimate how an action (e.g., opening schools) will affect the public.
But each of the above methods of research can be biased. As Dr. Sheila Nolan, chief of pediatric infectious diseases at Boston Children’s Health Physicians, puts it, in general, “there’s lots of things being published and lots of stuff coming out, but most of it is either retrospective or somewhat anecdotal. And it’s not that gold standard of how to really do a research study, which is exceedingly difficult to do while you’re in the middle of a crisis.”
Take, for example, a recent survey in New York. Researchers went around the city testing thousands of people at grocery stores and other shopping locations for coronavirus antibodies. From the results, they estimated in April that nearly 14 percent of New Yorkers have likely had COVID-19. But if you only look at grocery store shoppers, Abramson says, you aren’t getting a representative sample of the population. You’d miss, for example, people in prison or nursing homes. In effect, the design biases the study.
“There’s lots of things being published and lots of stuff coming out, but most of it’s either retrospective or somewhat anecdotal. And it’s not that gold standard of how to really do a research study.”
The fact that so many childhood cases are asymptomatic can further complicate study design and introduce bias in research, says Justin Lessler, an associate professor of epidemiology at Johns Hopkins Bloomberg School of Public Health and a co-author of an early contact tracing study in China. To measure how this virus spread within households, for instance, researchers in early contact tracing investigations typically identified homes with at least one infected individual, tested everyone else in the household, and observed who picked up the virus. The problem is, “primary cases” tended to be adults with symptoms, so “if children are asymptomatic more frequently,” Lessler says, “it’d be harder to find them as a primary case because usually, the first person you find is sick.” As a result, you might only see how adults spread the coronavirus to kids—missing how kids spread it to adults—and could fail altogether to capture homes with infected children. (There are other times, Lessler explains, where you may see the opposite effect, like when school is in session during a seasonal flu epidemic.)
One researcher I spoke with, Dr. Yvonne Maldonado, a professor of pediatric infectious diseases and of health research and policy at the Stanford School of Medicine, told me her “ideal study” would be to follow children in school settings and their households. She’d track their infection rates and antibody development over time to see how COVID-19 spread among children. She and her colleagues had actually planned to do a study along these lines—until schools shut down. Now, she’s limited to only doing that study in households. “Because children aren’t actually interacting with other kids, etcetera, the whole interaction is gone now, it’s not there,” Maldonado says. “So we might have limited data here, but it’s still worthwhile to consider.”
Finally, researchers are trying to understand the virus in kids through modeling, which is “essentially generating data,” Abramson says. Models are built on a series of assumptions, so one model on its own might be an outlier. Therefore, Abramson says, it makes little sense to rely on just one model; when models are done well, he says, “they really give you a great understanding of how dynamics can shift an outcome.” But because models are not observations of real events, he says, “no model will accurately reflect reality.”
The research on kids and COVID that has been done hasn’t shown coherent results.
If anything, the coronavirus has reminded us that science is messy, it’s slow, and it doesn’t always make sense. For example, consider Lessler’s contact tracing study, which included 391 people in China with coronavirus infections and 1,286 of their close contacts. One of the key findings was that children under 10 years old were just as likely to be infected with COVID-19 as adults, but less likely to show symptoms. When it first published as a preprint in early March—that is, as a preliminary finding and before undergoing peer-review—it “really scared everybody,” Alasdair Munro, a pediatric infectious diseases researcher at University Hospital Southampton in the UK, told Nature in May.
If anything, the coronavirus has reminded us that science is messy, it’s slow, and it doesn’t always make sense.
But in the months since, other studies have shown that kids are less likely to get infected with the coronavirus. It’s still unclear if these differences in results were due to bias in the study designs, real differences in the population, or something else. “It’s possible that our study is an outlier,” Lessler told me early last month. “The numbers are small, so it’s perfectly possible that our result is not going to be borne out over time. And that’s the nature of science.”
Modeling studies have also produced head-scratching results. One study, which initially published in Science in late April, estimated that school closures could delay the pandemic and essentially flatten the curve by 40 to 60 percent. “My simulation shows that yes, if you reopen the schools, you’ll see a big increase in the reproduction number, which is exactly what you don’t want,” Marco Ajelli, a mathematical epidemiologist then at the Bruno Kessler Foundation in Trento, Italy, told the New York Times in May. But other modeling studies have since suggested that school closures don’t have much effect at all on slowing viral transmission.
Though frustrating, it’s not particularly surprising that we’re seeing conflicting results in research on a virus that nobody’s ever seen before, that’s being conducted at rapid speeds, and in the middle of a pandemic. And, of course, no single finding is definitive. As time goes on, we’ll almost certainly gain a better understanding of the transmission dynamics among children. Until then, administrators will be left to “safely reopen schools” without knowing what that really means. As Haq puts it, “We need to get children back to school to support their psychosocial development, but this will inevitably increase the risk of transmission. There are no easy answers. We’re in the land of tradeoffs.”