COVID-19 in the classroom: fortunate and unfortunate events
Paul Tupper, Caroline Colijn
School has been back in session for two months in BC and the pandemic is nowhere near ending. As case numbers rise in BC there are two important questions: How safe are schools for children and teachers? — — and how much will classroom contact contribute to the spread of the virus in the community? School has been open long enough now that we begin to have some answers to these questions.
There have been many exposures: situations where an infectious student has attended class. But the vast majority of these incidents have not led to transmission. As of October 23rd there were 213 exposure events in BC schools, 6 of them considered clusters, since there were more than one infected student detected. In most of these situations it is unclear whether there was more than one detected case because of transmission in the classroom, or because of multiple students contracting the virus outside and then bringing it into the school. But on October 23rd an outbreak was reported at a Kelowna school with 11 detected cases. The term outbreak implies that transmission likely happened within the school.
How often do we expect clusters of cases and outbreaks to occur? There are two components to the question: Firstly, how often is there an exposure? Secondly, how many other students and teachers are infected by the single exposure event?
As we calculated earlier, the number of exposures we see per week will depend on the prevalence of COVID-19 in the population. We can do some rough calculations to get an idea of how the number of exposure events depend on daily reported cases. The data reported above was for the first 6 weeks of classes, so we are seeing on average 35 exposures per week.
During this period the number of reported cases per day grew rapidly, but the average over the period is about 70 reported cases per day. Assuming this ratio remains roughly constant, we estimate the number of new exposures per week is about half the number of reported cases per day. The exact ratio is not crucial, but the point is that if the prevalence is high, there will be many exposures. If we can keep prevalence low, there will be few exposures. Currently the prevalence of COVID-19 is growing rapidly so we can expect the frequency of exposures, clusters, and outbreaks to increase.
How bad will outbreaks be? The good news is that most school exposures appear to lead to no transmissions at all. This is what is observed in other countries and regions that opened schools earlier, and also looks like what is happening in BC. Our good fortune is partly thanks to the preventative measure we have taken in BC schools. Elementary and high schools were restructured so that students spent most of their time with the same group of classmates. Telling students with symptoms to stay home and get tested — — and only returning after a negative test result is received — — likely contributes a great deal to reducing transmission. Use of face masks and handwashing have also played a part. But even under our current system, outbreaks can occur, as we have seen in Kelowna. What can we do to mitigate this? What do various policies cost in terms of the disruption of the lives of students and parents?
To investigate these questions, we created a computer simulation of what happens in a classroom when a single infectious student attends. Using what we know about the coronavirus and its transmission, and the structure of a student’s day in the BC school system, we predict whether other students in the classroom will be infected, and if so, how many. We consider who may be directly infected by the first student (known as the index case) as well as the students who may be infected by these students, and so forth. (See our preprint: COVID-19’s unfortunate events in schools: mitigating classroom clusters in the context of variable transmission for full details.)
There are three important aspects of COVID-19 transmission that influenced our results:
Presymptomatic Transmission — People are familiar with the idea that if you catch a cold, there will be some time before you develop symptoms and are able to give it to others. One of the features of COVID-19 that has allowed it to be a worldwide pandemic is that you can transmit to others for a few days before you develop symptoms yourself. Even if everyone was perfectly careful about isolating when they develop symptoms, there are still plenty of opportunities to transmit first.
Asymptomatic Transmission- Even worse, some infected people, maybe as many as 40% of children and youth, will never develop symptoms at all. Besides meaning even more opportunities for the virus to be transmitted to others, this means that we will often not know the true size of clusters unless we test widely. And as we see in our simulations, two cases in the same school may appear unrelated, but actually be linked by a chain of asymptomatic infections.
Aerosol Transmission- The now famous 2m (or 6 foot) rule is based on the idea that the virus is transmitted primarily through expelled droplets that fall to the ground fairly quickly before dispersing around a room. This would mean that people would have to be quite close or have direct physical contact in order for transmission to occur. In the early days of the pandemic it was hoped that this was the only way it was transmitted. There is ample evidence now that transmission over longer distances is possible. The 2m rule still makes sense and helps reduce transmission, but the balance of evidence shows that transmission over longer distances is possible too, especially indoors.
Another important feature of COVID-19 is that the rate of transmission can vary a lot from one person to the next, and from one situation to another. How many people get infected in a given setting is unpredictable. For example, suppose an infectious person attends a choir practice. How many of the attendees will be infected? In one case 87% of the attendees were infected; in another there was no transmission. We considered a range of infectiousness for both the index case, and for the classroom and activity. In order to match the features of transmission in schools observed in BC and other places, we needed to consider transmissions that lead to no transmission, to small clusters, and to outbreaks that encompass over half the class. Then we asked how different interventions change the resulting cluster size. We describe four different interventions of increasing strictness.
Our baseline assumption was that children and adults who developed symptoms would go home and isolate, at least not infecting other children in the classroom. This simple measure alone has a big effect. But for the reasons we described above, this is not sufficient to stop transmission in all cases. When either the index case was especially infectious, or the environment was bad for transmission, there could still be clusters that encompassed half the class.
The next level of intervention we call the contact model, and it is close to what we are doing in BC currently. Symptomatic students go home and get tested. If the result is negative, the student can return to school when symptoms subside, but if the test is positive, then the student’s close contacts in the class are identified and they are asked to isolate and perhaps get tested too (especially if they develop symptoms). We modeled this in our simulation by imagining that each class of 25 is split into 5 groups of 5 students. Within each group, the rate of transmission is higher than between students in different groups, because they have more intense, direct contact. This protocol is moderately effective relative to the baseline. Under conditions where the median cluster size is 12 (nearly half the class), the contact model reduces the cluster size to a median of 8, saving 4 students from being infected. The problem is that even with reduced transmission between contact groups, inter-group transmission still occurs. If it didn’t, we wouldn’t see clusters like we did in Kelowna.
A way to be even more stringent is to treat the whole classroom as one big contact group, intervening with all students once there is a single positive test result. This is definitely more disruptive, but we hoped that it would do even better with the contact model in reducing the size of clusters. Surprisingly, it didn’t do that much better. Looking at median cluster size, a further two students were spared infection, so that on average, cluster size was reduced by half from the baseline.
The difficulty with all these interventions is that they act after considerable transmission has already occurred. If the index case is asymptomatic, they can transmit the virus to others in the class for a long time, because we don’t find anything until someone else gets infected, passes through the incubation period, develops symptoms of their own (if they ever do), gets tested, and receives a positive test result.
What would be truly effective in preventing large clusters is for testing to occur before symptoms appear. We don’t have the resources to give every student their own test every day, but there is an alternative: pooled testing. In this procedure, samples are collected from all students in the class on a regular schedule, pooled together and then tested. A positive result indicates that at least one student in the class has the virus, and then the students can be told to isolate while more careful testing identifies the individual cases. In the most stringent version of this that we modeled, clusters of size 12 were reduced to just two students: there was just a single transmission in the classroom that wasn’t prevented. However, this would require a massive increase in testing infrastructure, and it would be a significant departure from Canada’s approach to the pandemic thus far.
Our work on rapid, regular testing in particular was picked up in the media (Study suggests frequent rapid testing of students can curb outbreaks before they begin CTV October 23, 2020; Universal testing may be needed in schools to prevent the transmission… Alkhaleej Today — October 25, 2020; SFU study suggests frequent rapid testing of students could prevent outbreaks ; CFAX 1070 — October 26, 2020; Universal testing in schools may be needed to prevent SARS-CoV-2 transmission The Medical News — October 25, 2020).
And indeed some countries, sectors and industries are using mass testing to identify potential index cases before they develop symptoms and test positive, thus intervening before chains of transmission have built up. So would we say we are “calling for universal testing” or that it’s required in schools (in particular)? Not necessarily. If we did decide to make the shift to massive testing, it might be better used in adults to prevent workplace and social transmission, limiting exposures in schools. It would depend how quickly we could get test results and how well we could act on them.
So what can we do in the meantime to minimize transmission in schools? Much of it is staying the course: When students develop symptoms, they need to stay home and get tested if recommended by the CDC. And we need to make this easy for students and teachers, by providing a way for them to keep up with school work while they are away. The way schools have been restructured to minimize contacts has been very good. This has likely greatly reduced the number of clusters and their magnitudes. We have to maintain these new structures. We must continue to use masks, continue distancing, and continue having consistent bubbles of students interacting. Contact tracing and intervening with known contacts of students is very valuable, and poses a good tradeoff with the disruption of closing entire classes. This requires us to maintain strong contact tracing capacity. Most important of all: keep prevalence in the community low. If there is no COVID-19 in the community, students cannot bring it into the schools. There are no outbreaks without exposures.
