How Risky Is It to Send My Kids to School in a Pandemic?
When elementary schools opened this fall, my family opted in. We felt that the small risk of COVID-19 was worth the huge benefits.
Yet, as we kissed our three monkeys goodbye, and returned home to a quiet house, our jubilation was tinged with trepidation.
We knew that our best efforts to estimate the back-to-school risks were crude at best. A month later, we have a much clearer picture of the risks we are all assuming — kids, families, and communities.
As a parent, my big concern is that my kids will get infected, and possibly spread it to others. This story shares what we have learned about exposure risk, transmission risk, and community spread, all of which can help you understand your own risk.
How likely is it that my child will be exposed to COVID-19 at school?
The first piece of the risk puzzle is your COVID-19 exposure risk: the risk that your child is exposed to somebody who is carrying COVID-19 and may spread to others.
What is an exposure? An “exposure” is when someone shows up at school with COVID-19 during their infectious period (presumably unaware). Importantly, an “exposure” does not tell you whether or not COVID-19 was acquired at school, or brought in from the outside; it also doesn’t tell you whether or not this person spread it to anybody else.
News tip: When reading COVID-19 news, pay close attention to whether or not the exposures resulted in any spread. For example, when Berlin re-opened schools in August, story after story talked about 41 “exposures” (out of 825 schools). Yet, none of these exposures led to outbreaks. An exposure without any transmission is, in many ways, good news! See example here from Global BC.
As expected, the risk of COVID-19 at a given school is tightly linked to rates of COVID-19 in the surrounding community. The higher the COVID-19 rates in the community, the more exposures — people bringing COVID-19 to school — we see.
In my area, the Vancouver School Board, we saw about 23 exposures in the first month of school (23 school days) in elementary schools — about 1 exposure per day across 20,000 students (per Vancouver Coastal Health website).
Using these numbers, I estimated our annual exposure risk — the likelihood that we will receive one of the dreaded “your child has been exposed” letters at some point this year — to be about 1 in 10 per kid (see Appendix for math).
In the neighbouring suburb of Surrey, where community COVID-19 rates are about double the Vancouver rates, the school exposure rates were roughly twice as high. In the last two weeks, Surrey had about 1 exposure per day per 10,000 elementary students (28 exposures in 10 school days across about 30,000 elementary students, per Fraser Health Website).
It’s a similar story in the United States (US). A national school dashboard led by Professor Emily Oster, reported average daily exposure rates in the range of 1 per 10,000 students in elementary secondary schools, with higher rates in regions with higher COVID-19 rates, as expected.
The exposure data not only confirmed our use of community rates as a key risk factor, but also told us that many of the predictions we made this summer, such as those published in the New York Times (and my own back-to-school story), were overly pessimistic. The observed exposure rates are far lower than those predicted based on raw community COVID-19 cases.
In both Vancouver Surrey, the exposure rates are nearly 10 fold lower than predicted by simple models using raw community COVID-19 rates (see Appendix for calculations). The US figures above are also far lower than predicted by simple models.
Why are observed exposures lower than predicted?
Many predictions made the simplistic assumption that every active case of COVID-19 is out there lurking in the community. Yet, this assumption is flawed. First, anyone with even mild symptoms should not be at school. Simply heeding symptoms will take many (but not all) people out of the circulating pool after the first few days. Second, when contact tracing is successful, most people will be contacted within a few days of exposure, and be quarantining as a precaution. Third, younger children (under age 10) typically have lower rates of infection than adults — though the exact rates in children are still unclear due to incomplete testing.
Key messages: School exposure rates are linked to community COVID-19 rates — and are often lower than raw case counts.
How likely is it that COVID-19 will be spread at school?
We all know that some environments, and interactions are riskier than others — that indoors, crowded, poor ventilation, and lack of protective layers, means higher risk. Yet, concrete estimates for absolute transmission risks are only now coming into focus.
What is transmission risk? Transmission risk is the likelihood that an infected person passes it on to someone else.We can estimate this risk by comparing the number of exposures (potential transmission events) to the actual number of transmission events. If one out of every 10 contacts, on average, becomes infected, then the transmission risk is 10%.
Here in British Columbia (BC), the transmission risk in schools has been reassuringly low. According to our provincial health officer, Dr. Bonnie Henry, our first month yielded 100 exposures across the province, yet no “outbreaks” (using a fairly liberal definition of outbreak as uncontrolled, widespread transmission), per CBC news.
We did, however, have at least one cluster of cases in elementary schools. In Caulfield Elementary in West Vancouver, parents reported that nine kids in the same cohort tested positive, which led to nine additional cases among household members (frustratingly, no official numbers have been shared). Regrettably, it’s unclear whether or not there were any other case clusters, as this this information has not been made publicly available. Thus, parents are left relying on general updates from our public health officer — and the news.
Tip: Proceed with caution when interpreting small COVID-19 “clusters”.
When we see many COVID-19 cases in the same school, this likely due to in-school transmission. It’s much less clear what to make of a small “cluster” of cases. Clusters are often defined as more than one exposure in two weeks at the same school.
A cluster does not necessarily represent in-school spread. In fact, many “clusters” are actually be the same person showing up to school more than one day, or two completely unrelated exposures. When we assuming that any two cases must represent an in-school transmission, we are essentially painting a worst-case scenario, as depicted by this citizen-generated map.
BC’s experience, that that most exposures do not result in further spread, and that most clusters are small and contained, is consistent with published studies and with back-to-school experiences elsewhere.
Encouraging stories are piling up around the world, including accounts from Germany and Norway. In addition, a new study in the United States found that childcare center staff did not endure a greater COVID-19 rate early in the pandemic.
This low transmission rate is also consistent with two massive contact tracing studies — one in South Korea, and one in India — that shed light on COVID-19 transmission rates in different contexts. Within households, we have seen transmission rates just north of 1 in 10 — only a one in 10 chance of infecting another family member. Transmission rates in “low risk” environments are in the range of 1%.
The science of contagiousness: the transmission rates reported in large studies are averages. I turns out that there is a lot of variation in how contagious people are. A handful of cases spread to many people, while most spread to none (70% of cases in the large Indian study). It’s best to assume that all cases are highly contagious.
Key message: Transmission rates in elementary schools are low. My best guess is in the low percents, maybe 2–3%.
How likely is it that my kid will get COVID-19 from school this year?
Above, I estimated an annual exposure risk of about 10%, and a transmission risk of about 2–3%. By multiplying these two these together, I get 1 in 400 (0.25%) as my educated guess for the chances one my kids will get infected with COVID-19 this year. Check out my Appendix to see the general framework and try it with your numbers.
This is a crude estimate, based on crude assumptions. Yet, for me, this is better than nothing — and more reliable than what we had before schools opened. For my family, this risk is still acceptable, though it may not be to those with different risk-benefit equations.
Key message: We are taking small — but real — risks, by putting our children in school for a year. Your family’s risk depends on the local COVID-19 rate, and the precautions in place at your school.
Our first month of school also sheds light on the potentially broader implications of schools being open.
Can communities keep COVID-19 rates in check with schools open?
Initially, there was some concern that opening schools would cause dramatic rises in community rates. Indeed, COVID-19 rates were on the rise in some places as schools opened. However, it seems that these increases were transient, and likely due to other factors, such as the return from summer holidays.
Here in Vancouver, rates rose for a while after schools opened (and summer ended) but have settled at levels not far from where they were before schools opened: We saw 307 new cases in the two weeks leading up to schools opening on Sept 10, 2020 (Aug 21-Sept 3, 2020) and 325 cases just over a month later (Oct 2–15, 2020) in the same population.
In addition, this German study vindicated schools as the culprit for post-summer rate changes by comparing regions that re-opened at different times (see School Re-Openings after Summer Breaks in Germany Did Not Increase SARS-CoV-2 Cases).
Key Messages: We can safely open schools without exploding COVID-19 rates in the community. Current testing practices are unlikely to be missing many “stealth” cases.
My Bottom Line
Our back-to-school strategy is working well overall. Hundreds of thousands of kids are happily back at school, and most families have not been harmed by COVID-19.
Yet, the risk is not zero. My heart goes out to the handful of people who have suffered as a result of school-related transmissions. When you are the one in thousands affected, the statistics telling you that your risk is “small” become irrelevant. This gulf between population level success and individual stories is an unfortunate reality of public health.
While our promising back-to-school story should provide some relief, and a better sense of the risk we are assuming, it’s not a reason to be complacent. Our success to date reflects our low community rates, our precautions at school, our diligence in staying home when mildly ill, and our efforts to manage out-of-school contacts. If we let our guards down, things can easily change. We all need to keep doing our part to minimize the spread of COVID-19.
In addition, we should strive to continually improve and learn from our failures. The Caulfield School story suggests that our government should reconsider its exposure policy, and consider keeping siblings and parents home when one child is exposed. In addition, it’s clear that lack of transparency is eroding trust and causing more anxiety among parents who feel poorly informed. A more detailed dashboard that shows numbers of cases per school, similar to those in Alberta and Ontario, would help a lot.
They say that “no news is good news”. I disagree. I want to hear the good news, celebrate it, and help my fellow parents to feel a bit less anxious.
See Appendix below for my “cocktail napkin” COVID-19 risk calculations.
Estimating Your Annual Infection Risk
This is the general formula. The actual numbers are very rough estimates. and
- Annual Infection Risk = Annual Exposure Risk x Transmission Risk
- Annual Exposure Risk = Daily Exposure Rate x Number of School days. Adjusted for non-independent exposures. Exposure risk varies based on local COVID-19 rates and pod size.
- Transmission Risk = Percent of cases that spread to others. Transmission risk depends on depending on protective measures.
Estimating Annual Exposure Risk
Here is my example:
We are seeing about 1 daily exposure per 20,000 kids. If my kid is exposed to 20 kids, this means a 1 in 1,000 chance of exposure per day. Each day we go to school, we roll the dice again. Multiply by 200 school days and you get 1 in 1,000 x 200 = 1 in 5. However, many exposure are not independent. I cut this number in half to reflect the fact that many of these exposures are to the same set of students.
Another way to look at is is that we expect roughly 4000 exposure events over the year (20 per month x 20 students per exposure x 10 months). Those 4,000 exposure letters will probably be sent to 2,000, not 4,000 families.
Predicted exposure risks using community COVID-19 rates
Vancouver: When schools opened, we were at about 1 active COVID-19 case per 2,300 people (based on 307 cases in an area of ~700,000 people over 14 days — see my back-to-school risk calculations). This rate predicts 9 exposures per day in an area serving 20,000 elementary school kids* (20,000/2,300).
Surrey: 1 case per 1,100 kids based on 741 cases in 14 days per 800,000 people (BCCDC, Oct 20, 2020). This predicts 28 exposures per day based on 30,000 students (30,000/1,100)*
*Student body: 30,000 students = 44,000 students x 70% in-person (best guess!).
Surrey vs Vancouver: Community rates
In the last 14 days, Vancouver saw 325 new COVID-19 cases in a population of roughly 650,000 people, or 1 in 2,000. Fraser South (where Surrey resides) saw more than twice as many cases (741 new cases) in a population of nearly 800,000, or one in 1,100 (Source:BCCDC, Oct 20, 2020)