How can USA schools open safely w/ community virus levels 20-500x the other countries that opened?

Updated: 2 days ago

This post is following up on my previous post on how difficult the school opening decision is. A lot of people have been mentioning to me how successful other countries' school openings have been, and expecting we'll be fine here in the USA when we open as well. A COVID-19 literature report team at Washington State department of health published a report that detailed school opening details from countries for which data was available, which included several Asian countries (Japan, Taiwan, South Korea, Vietnam), Scandinavian Countries (Norway, Sweden, and Denmark), other European countries (Germany, France, Belgium, Greece) and other countries (Israel, New Zealand).

Here are some points from these summaries:

  • Grades: Some countries opened school only for elementary schools, not secondary (Belgium, Denmark, Norway, Greece, France) and Sweden never closed elementary schools, but closed secondary schools from March to June. Initially Germany only opened for graduating seniors and gradually allowing other younger children to come back.

  • Schedule: Most schools did some sort of schedule adjustment to increase physical distancing, with some splitting between morning and afternoon shifts (e.g. Germany, South Korea) and others alternating days (Belgium Switzerland), and many also staggered start/end/class change times

  • Masks: Most had masks either mandatory, or mandatory inside (Belgium, France, Germany, Norway, Greece, Korea, Switzerland), with some exceptions at times for younger students.

  • Class Size: Most had greatly reduced class size (most 10-15 students, Belgium, Denmark, France, Germany, Norway, Greece, South Korea, Switzerland).

  • Dividers: Some countries used dividers between students (Israel, Sweden, Taiwan, Vietnam)

  • Screening and Quarantining: Asian countries (Japan, South Korea, Taiwan, Vietnamese) required fever checks at school entry and isolation at treatment clinics if positive. Germany has done routine screening, and many are set up for contact tracing (Israel, South Korea, Germany and New Zealand). Germany has self-tests with <1 day turnaround.

  • Closures: Many schools have guidelines to quarantine students who test positive and close schools with outbreaks, with numerous closures in Israel and South Korea.

Many of these openings went quite well, without too much community transmission, except Israel who opened all at once and did not keep proper precautions in place so this opening helped ignite a huge second surge, Germany had evidence of child-to-child transmission but not much child-to-adult transmission. This is all somewhat encouraging, especially if we can incorporate many of these guidelines that successful countries used.

However, there is one HUGE difference between the USA and those countries, and it is the level of viral prevalence in the country. To assess this, I used the COVID-track app that my PhD student Emma Zohner developed to look at the viral levels at the time of opening. To assess this, we computed the daily number of new cases per day per million residents after computing a 7-day moving average to smooth out spikes. Following are the approximates new daily cases per million residents in the countries in the Washington study at the time they opened.:

  • <1: South Korea, Vietnam, Japan, Taiwan, New Zealand, Greece,Israel

  • 5-10: Switzerland, France (and only opened some regions with lower counts), Belgium, Norway (only opened elementary), Germany

  • 30: Denmark (only opened for kids <11, and only in small groups of 12 all day with no interactions with other groups)

  • 50: Sweden (never closed K-9 -- closed secondary March-June)

Most countries had very low viral levels in society when they opened. All Asian countries had virtually no virus in the community, and Greece, New Zealand and Israel also had exceptionally low levels. Other European countries were also very low, and even with that low viral load France only opened regions with very low counts. Denmark was higher, but also only opened schools for smaller children and kept them in groups of 12 that were together all day with no interactions with other groups, which would greatly reduce the threat of spread, and Sweden had the highest levels in June when secondary schools were reopened at 50 new daily cases per million.

When we look at the USA, the daily new case count is 200 per million, which is 20-500 times higher than any of the above opening countries except Denmark and Sweden, for which it was 4-7 times higher. USA has tested at higher rate than many of these countries (except Israel, Greece, and Denmark), but the ratio of cases is much higher than ratio of tests done, so this is not an artifact of testing. Said another way, the USA testing positivity rate of 8% is much higher than most of these other countries.

Here is a plot of the data to put it into better perspective:

This is the factor that gives me by far the most concern about opening schools. This is a first-time experiment, and unprecedented. NO country during this pandemic has tried to open schools with viral levels anywhere near what is experienced in the USA right now. Not even close. And even some countries with minuscule viral levels still had problems, with Israel having an enormous second surge and South Korea having to close numerous schools.

Of course, different regions in the country have very different community levels of virus. Here is the approximate new daily cases per million residents by state,

  • >300: FL, MS, AZ, TN, AL, NV, GA, SC, LA

  • 200-300: ID, TX, MO, OK, CA, AR

  • 150-200: AK,UT, NC, NM, MD, WI, NE, ND, IA

  • 100-150: KY, KS, VA, MN, MT, IN, WA, DC, RI, CO, DE, OH, IL

  • 50-100: WY, OR, MI, SD, WV, PA, MA, NJ, HI

  • Others: CT 40, NY 33, NH 23, ME 15, VT 7

Florida leads the country at 450, followed closely by Mississippi at 430 (Mississippi is quietly having a dramatic surge this past month, with hospitals overwhelmed, overlooked perhaps because it does not have high population levels). Other surging states are >300, and another set >200. Another large group of states have levels between 100 and 200, still much higher than any country that has opened. 8 states have levels between 50-100, still higher than any of the opening countries in Europe, but at least relatively close, and only 5 states have numbers less than Sweden, all in New England.

Here is a plot of these numbers across the 50 states plus DC (thanks Emma Zohner!)

Caveat: just like country-level summaries do not precisely reflect the situation in each state, so state-level summaries do not precisely reflect the situation in local communities, e.g. at the county level since the states are not uniform in their community viral levels. Thus, it is advisable to look at any summaries of viral spread at the local level as well.

BTW, for big picture, here it is broken out by region:

To illustrate the practical implications of the corresponding estimated infection levels in school settings, an app by researchers at Georgia Tech provides for each county in the USA the probability of a randomly selected group of 10, 25, or some other group size of individuals from that county having at least one infected individual, based on current case counts, population level, and estimate of uncounted:counted infections (5 or 10:1, which is strongly supported by literature). These are county level, so more locally relevant than the state-level summaries. For a group of 25 individuals, a typical sized classroom in the USA under normal circumstances, here are some example probabilities as of 8/1/20:

  • New York, NY: 13%

  • Suffolk, MA: 16%

  • Multnomah, OR: 28%

  • Philadelphia, PA: 30%

  • Baltimore, MD: 50%

  • Franklin County, OH: 50%

  • Polk County, IA: 54%

  • St. Louis, MO: 60%

  • Cullman, AL: 64%

  • Harris County, TX: 64%

  • Los Angeles, CA: 64%

  • Hillsborough County, FL: 74%

  • Orange County, FL: 79%

  • Clark County, NV: 79%

  • Maricopa County, AZ: 82%

  • Davidson County, TN: 87%

  • Miami-Dade, FL: 98%

Now, just because an infected person is in a room with 25 people doesn't mean they will automatically spread to the class. Infectiousness is thought to be maximized from several days before symptoms appear to several days after, plus hopefully vigilance to isolate individuals with symptoms will reduce probability of spread, and distancing and masks will help reduce the probability as well, as long as the school has a rigorous plan that is strictly followed. But these numbers provide some important perspective about how high the community viral levels seem to be, and also provide a useful tool to assess the relative severity of the current crisis in different counties around the country.

These facts make me extremely concerned about the wisdom of opening up in many places in the USA. I know how badly children and teens need to return to school for their educational, social, and psychological well-being, not to mention their parents' or their families' financial well-being. But getting large groups of children together with viral levels like this seems to be very risky, and as we saw above, unprecedented, and seems ripe to cause a surge.

Maybe with precautions, if they are followed, viral levels can be kept relatively under control and be safe. I have seen plans for many schools including hybrid plans to reduce number of students in school at a time, physically distanced rooms, and mask wearing, which are all good ideas. I don't see many with staggered start times or class switch times, so hallways could be crowded, and I've seen a mixed bag in terms of lunchtime, with some keeping lunch in rooms and others allowing lunchrooms. But in the USA we don't have the testing and contact tracing that many of the countries above have, and I haven't seen many school districts doing daily temperature screenings like European countries are doing. There also may be many schools with old buildings with potentially poor ventilation that could contribute to potential spread by accumulating aerosol particles.

Given all of these factors, it seems like most districts in the USA have less vigorous mitigation strategies than the countries that opened successfully, to go along with viral infection levels potentially orders of magnitude greater. Plus, even with well-laid plans for mask wearing and social distancing, how can teachers and staff get the high level of compliance necessary for these mitigation strategies to be effective?

This is a very difficult problem, and given that in many areas of the country their parents are doing a poor job of following these guidelines, can we expect the children to do that much better? Mask wearing, physical distancing, not touching your face all day at school?

I hate to be so gloom and doom, but I'm just calling it as I see it as I look at the data and put it into perspective of infection levels reopening would bring to our schools, and comparisons with other countries that have opened (mostly) successfully.

It seems like for much of the country, it might be necessary to consider online learning to start the year, with a vigorous push in the communities to get viral levels down to much lower levels so that children can return to school with a reasonable chance of avoiding sparking infectious surges. In the meantime, governments should request and use emergency funds to equip and prepare schools if they are not ready yet, and engage in a concerted effort to get fast testing capabilities as well as contact tracing capabilities to keep up with the infections when schools open and keep them under control. And this extra time will also help more districts share information from other districts to construct the most robust, detailed plan possible. There are so many factors to consider, combining information across districts can lead to more robust plans.

What do the numbers have to look like before schools open? I've been asked that question a number of times, and the answer is a really don't know, but for many places in the USA it has to get much lower than it is now. As I have discussed in other previous blog posts, raw cases counts are not sufficient for assessing emerging surges and epidemics, especially in the environment of increased testing.

  • To take population into account, the new daily counts per capital (e.g. per million like above) are a potentially useful measure, both the absolute level and the change over time.

  • Also, the testing positivity rate, number of positive tests over total tests, is another measure to look at, with values <5% considered "ready to open" by CDC but also whose slope over time is a measure of potentially improving or worsening situation in the local community. Note that because of variability in reporting of case counts and testing, 7-day averages are recommended to avoid artifacts and outliers on given days. Also, looking at the current level as well as the trend over time are useful to see whether the situation seems to be improving or worsening.

I have a son entering 9th grade, and we just moved into the district last year, so I desperately want him to get to go to school in person and get plugged in socially with the other teens, as well as the activities that can stimulate his development. He has had enough time with video games, YouTube videos, and Netflix, and being stuck at home all of the time makes it harder for him to be stimulated and grow. And knowing teachers in districts with many at risk and underprivileged children, I am also acutely aware of how may of these children need school even more desperately, not just for educational, social, and psychological reasons but in some cases also for nutrition, safety, and adult supervision and encouragement to spark their development.

I want schools to open so badly, but have trouble seeing how it can be done safely in most places I the USA. Maybe this can inspire us to finally come together as a society and take seriously the mitigation guidelines that have become abundantly clear need to be followed to limit the spread of this virus.


©2020 by Covid Data Science. Proudly created with