Does closing schools slow the spread of coronavirus? Past outbreaks provide clues
As the societal disruptions from COVID-19 spread and intensify, a question for many in the United States is, what about schools? Schools in Japan, Italy, parts of China, and elsewhere have shuttered. A small but increasing number in the United States are following suit, whether for a day, a week, or longer.
But does shutting a school help a broader community, especially when the role played by children in spreading COVID-19 remains uncertain? Nicholas Christakis, a social scientist and physician at Yale University, thinks it does, but he recognizes that questions around closing schools are difficult. Christakis studies social networks and is developing software and statistical methods to forecast an epidemic’s spread before it happens.
This interview has been edited for clarity and length.
Q: There seems to be a lot of variability in how schools are managing this. What are the different approaches schools can take—and have taken in past epidemics—and how might they help?
A: I’d like to emphasize the difference between reactive and proactive school closures. Reactive is when a school decides to close when a student or parent or staff member is sick. Most people don’t argue with that. If the pandemic is at the school, you would want the school closed.
There have been a number of studies that have looked at reactive closures. These analyses, including a paper in Nature in 2006 using math models [of an influenza pandemic], typically find that such reactive school closures for a moderately transmissible pathogen reduces the cumulative infection rate by about 25% and delay the peak of the epidemic [in that region] by about 2 weeks. When you postpone the peak, you also typically flatten the epidemic and space out your cases. This has value. It means that the incidence on any given day is lower, so we don’t overburden our health care system.
Q: So a reactive closure would occur after a case of COVID-19 in a student, parent, or staff member. Should one case like this close a school? Does it depend on circumstances at all?
A: If, for example, a person flew to your town from Italy and brought the condition with them, that means something different than a community-acquired case, where we don’t know how they got the disease. A community-acquired case is like a canary in the coal mine. When you detect one case there are probably dozens or hundreds of others.
Q: A community-acquired case should close the school?
A: Yes. By that time, they’ve probably transmitted it to other people. It’s the tip of the iceberg. One paper I looked at [on an influenza pandemic] looked at closing a class or a grade. It doesn’t gain you much.
Q: What about the parent returning from a trip in Italy? Should that school close?
A: Maybe. They could isolate people close to the infected individual. I would probably close it, but I could certainly understand not doing so.
Q: How about proactive school closures, before there are any infections associated with a school? Are they helpful?
A: Proactive school closures—closing schools before there’s a case there—have been shown to be one of the most powerful nonpharmaceutical interventions that we can deploy. Proactive school closures work like reactive school closures not just because they get the children, the little vectors, removed from circulation. It’s not just about keeping the kids safe. It’s keeping the whole community safe. When you close the schools, you reduce the mixing of the adults—parents dropping off at the school, the teachers being present. When you close the schools, you effectively require the parents to stay home.
There was a wonderful paper published that analyzed data regarding the Spanish flu in 1918, examining proactive versus reactive school closures. When did [regional] authorities close the schools relative to when the epidemic was spiking? What they found was that proactive school closing saved substantial numbers of lives. St. Louis closed the schools about a day in advance of the epidemic spiking, for 143 days. Pittsburgh closed 7 days after the peak and only for 53 days. And the death rate for the epidemic in St. Louis was roughly one-third as high as in Pittsburgh. These things work.
Q: How should jurisdictions decide when to pursue a proactive closing?
A: How many cases are there in the region? And what is the epidemiologically relevant region? If you’re in a mid-sized town you might say, as soon as there’s a community-acquired case in my town, whether it’s in my school or not, I’m closing my school.
Q: Let’s consider the community-acquired case of a rector at a church in Washington, D.C., whose COVID-19 diagnosis was announced this past weekend. Should schools close in the region because of that one case?
A: If the rector was in the epidemiologically relevant area (however you define it), and if you believe that you will close your school in a reactive fashion when there’s a case at your school, and a community-acquired case appears—it’s going to appear [in the school]. So why not close now ahead of time, and gain all the extra time for your staff and students to minimize spread?
Q: This sounds hugely disruptive.
A: Unsurprisingly, there are substantial costs—health costs and economic costs. Many children get school lunches; their health might suffer from closing the school. Health care workers could be taken out of commission to care for their own children precisely when we need them at hospitals. Parents could lose job opportunities. This is why in Japan, they’re providing basic income to parents during the closure. It should be rightly seen as a state expense.
Q: Are there social distancing efforts short of closing schools, especially if there are no cases associated with a particular school? For example, canceling big events that bring together lots of families?
A: Yes, I’m so glad you mentioned that. We don’t have to have an all-or-nothing policy. We can have intermediate steps. For example, why not allow families who want to keep their kids home keep them home? Why not cancel all activities, like sporting events and musical performances that have large groups present?
When we engage in social distancing, it’s not so much that you don’t get infected yourself. The real advantage is that by removing yourself from circulation, you stop all the paths of this virus through you. You are doing a social service, you are helping the community. Employees who want to work from home [and are able to] can work from home.
Q: Many schools have closed for 1 day to sanitize. Is that a useful strategy?
A: I don’t know the answer to that. It depends on the circumstances.
Q: Another big question is what the endgame is. If schools close, when should they reopen?
A: Honestly, I don’t know how much research has been done. It’s a number of weeks when you close. The Chinese have had their schools closed for 6 weeks. The Japanese have had their schools closed for four. What is the reopening rule? I don’t know the answer.
Q: School closures are pretty controversial right now. Some articles, like this one and this one, suggest they risk accomplishing little. And because this outbreak is still so new, we are drawing on previous outbreaks of other diseases to guess what shutting schools can do for this one. What would you say to people who argue that closures hold little value, especially when a community has relatively few cases of the virus?
A: Let’s do a thought experiment. If there is an outbreak in your school, would you advocate for closing the school? If the epidemic is occurring around your school, you know that it is going to strike the school. And so if you’re prepared to close the school when it arrives at your school, it makes much more sense to close when it is near the school.
We know from past epidemics of multiple types of viruses that school closure works. We know that it interrupts adult transmission even if the kids are not vectors. Here, it’s likely that kids are vectors, and preliminary evidence from China suggests that they can be. I absolutely recognize it’s a difficult calculus. But what we’re talking about here is a pandemic.