Reexamining the Data on COVID-19 Case Rates for in-Person Teachers and School Staff
As a scientist, the spouse of a former K–12 educator, and a parent of two young children, I have closely followed the debate over whether children in America should attend in-person schooling during the COVID-19 pandemic. My interest in this topic piqued when a neighbor shared with me the Rockefeller Foundation’s recent plan for school reopening. This plan claimed that in-person schooling could resume as early as February 1st with no additional risk of COVID-19 spread both for students as well as school teachers and staff, if schools implement mitigation procedures, including testing and mask wearing. However, my analyses of updated data show that reopening schools now will put teachers and school staff at risk.
I understand the arguments claiming that in-person schooling is safe for children. But I remain concerned for America’s more than eight million school teachers and staff, who are likely to be more susceptible to COVID-19 than children.
The argument that in-person schooling is safe for teachers and school staff is based on the idea that these groups do not show greater COVID-19 case rates when schools are open. To support this claim, the Rockefeller proposal highlighted data showing similar COVID-19 case rates for teachers and school staff as compared to the baseline rates in their surrounding community. These data came from the COVID-19 School Response Dashboard, a project created by a team led by Dr. Emily Oster. The Dashboard consists of a public website that aggregates various datasets on COVID-19 infections across schools, including both voluntarily uploaded data from individual schools as well as government data. The Dashboard also provides tools for analyzing these data.
It seemed bold to suggest that infection rates for teachers and school staff would not be elevated during this pandemic, given recent observations of increased COVID-19 infections for people of all ages as well as new scientific research suggesting a link between in-person schooling and increased community transmission. Therefore, to test the robustness of these claims, I examined the data from the COVID-19 Dashboard myself using its own built-in analytic tools.
COVID-19 spread among school staff over time
I specifically conducted these analyses to understand the claim that in-person schooling did not add COVID-19 infection risk for teachers and school staff. Dr. Oster made this claim in an article on November 20, and it was repeated more recently by the Rockefeller Foundation plan, as well as by politicians, academics, foundations, and others.
This claim was based on a comparision of COVID-19 case rates for in-person teachers and school staff versus matched community members. Oster’s analysis measured these numbers for New York state data from Oct. 12 through Nov 6. She found similar COVID-19 case rates for teachers and school staff compared with matched community members, which she interpreted as indicating that in-person schooling was safe for educators.
I wanted to confirm this result myself and to explore whether it was reliable over time, especially given the recent surge. I opened the Dashboard website and selected it to display data from schools in New York state that implemented mask wearing. The site looked like this:
On this page, the lines in the right plot show COVID-19 infection rates for different time periods; the black popup box highlights the New York case rates for the period of 10/26–11/8, which overlapped with the date range from Oster’s report. For these dates, the mean COVID-19 daily case rates (per 100,000 people) were 8, 12, & 12 for students, teachers and school staff, and the community, respectively. Because the case rates in this period were fairly similar for teachers and school staff compared with the community, it seemed consistent with her interpretation that in-person schooling was not associated with large additional risk for those groups.
I then focused on the rest of the plot, which showed that COVID-19 case rates began to rise on Nov. 9. To visualize these numbers, I entered the Dashboard’s numbers into a spreadsheet and plotted them over time:
In this plot, the colored lines show case rates among school teachers and staff for different school levels and the black line shows the community. Overall, the lines have a positive slope, matching the nationwide COVID-19 case surge during this period. But looking more closely, one sees that the colored lines show steeper increases than the black line. Thus, during mid November through December, COVID-19 case rates increased more sharply for in-person teachers and school staff compared to the rest of their community.
To directly quantify the additional COVID-19 case risk that teachers and staff face from in-person teaching, I computed their percent change in case rates relative to their community. As shown in Figure 3, for Nov. 30–Dec. 13 in-person teaching was associated with increased case rates for teachers and school staff at all levels. This increase was especially notable at the elementary-school level, where in-person teachers and school staff showed a 53% higher COVID-19 case rate.
Confirming these results with other state data and national data
To confirm these results, I expanded these analyses to examine the Dashboard’s data from Texas and from across the entire nation. I created the plot shown in Figure 4, which indicates the relative increase in COVID-19 case rates for school staff compared to communities, for each region.
This plot shows that this same pattern of results—increased COVID-19 cases for in-person school teachers and staff—was consistent geographically and not limited to New York.
Summarizing the above analyses, in December there was a consistent trend for school teachers and staff to show elevated COVID-19 case rates compared to their communities, despite the use of masking in their schools. Thus, the pattern that the Rockefeller Foundation plan and Oster emphasized, where teachers and school staff showed similar case rates as the community, disappeared in mid-November when the pandemic surged.
Comparing case rates across regions with different overall case rates
Finally, I wanted to evaluate the claim from Oster and the Rockefeller Foundation report, where they showed data that in-person schooling was safe for teachers and school staff across regions with both high and low overall levels of COVID-19 spread. I felt this was an interesting analysis because it could show whether any added risk for in-person teaching was limited to certain regions, such as those where COVID-19 was rampant or rare overall.
Figure 5 below shows the results of this analysis for two distinct time periods. The left plot, which I duplicated from Oster’s article, shows data from Oct. 12–Nov. 6. The right plot shows my own analysis of data from Nov. 30–Dec 13. In these plots, each dot represents a group of schools, with the position along the x axis indicating the community case rate and the position along the y axis indicating the in-person case rate. In these plots, if a datapoint is above the diagonal 45° line, it indicates that there are greater case rates for in-person school teachers and staff than the community. Datapoints near the diagonal indicate that case rates are similar between the in-school and community groups.
Critically, focusing on the orange points (which represent teachers and school staff), the left and right plots show very different results. In the left plot (Oct. 12–Nov. 6), the orange points are fairly close to the diagonal line (although slightly above it). As Oster noted, this indicates that case rates were relatively similar between in-person teachers and school staff compared with their communities, and that this pattern is maintained across regions with both high and low overall case rates. In contrast, the right plot illustrates a different pattern of results for Nov. 30–Dec. 13. Here, the orange points are all substantially above the 45° line. These deviations are sizable—note the change in axis scale. Thus, these results show that during December in-person teachers and school staff consistently show an elevated risk of contracting COVID-19 relative to their community baseline.
Conclusions
Overall, the data from the COVID-19 School Response Dashboard show that in late November and December, with the pandemic surging, in-person school teachers and staff experienced greater COVID-19 case rates compared to others in their communities. The magnitude of this increase was quite substantial, and in-person teaching consistently posed added risk across communities with high and low overall levels of virus spread. Together, this pattern shows an abrupt shift from the earlier numbers from the Dashboard and Oster, cited by the Rockefeller plan, which had suggested that in-person school teachers and staff would show no excess COVID-19 risk.
The results I presented above come from a relatively straightforward set of calculations I performed by entering the data from the COVID-19 Dashboard into a spreadsheet. An analysis is only as good as its data. If the COVID Dashboard numbers have issues then perhaps all the analyses that rely on it are problematic, both the ones described here as well as those shown by Oster and the Rockefeller report. But if the Dashboard numbers are reliable, then my analyses indicate that the teacher safety reported by Oster and the Rockefeller report has vanished during the winter surge.
These results illustrate the dynamic nature of the pandemic, where the rate of virus spread may change dramatically over time. The COVID-19 dashboard is a very useful resource and we should expand and improve these types of data-collection efforts. However, while we are collecting this type of rich, dynamic data, we should simultaneously monitor the numbers and tailor our public health policies accordingly. It seems to me that many of our current policies and messaging related to the safety of in-person schooling for teachers and school staff are based on inaccurate and outdated data. In this fast-moving pandemic, merely collecting data is not enough; we need to continuously analyze the data and update our conclusions accordingly.
We owe it to our school teachers and school staff to be clear and transparent about the fact that the most recent data indicate that they may experience a relatively increased risk of COVID-19 infection if schools open as planned for in-person schooling. The newer numbers indicate that the situation has changed from October and that in-person schooling is now a significant source of COVID-19 case spread for teachers and school staff.
