A new model of the potential spread of COVID-19 model was released recently, and I encourage anyone who’s back to college this fall to copy it to their own Google Drive and plug some numbers in and read the accompanying paper. This model was built with the college environment in mind, and President Elizabeth Bradley led the invited commentary on the work.
Under the model’s base assumptions, the authors recommend testing every two days in order to maintain a manageable number of COVID-19 cases. Vassar, as far as I am aware, only plans to test students once every two weeks. President Bradley, in her invited commentary, stressed that it would be possible to do this and still maintain a controllable number of infections if the original model’s original assumptions of virus spreadability and introduction were dampened by the safety measures put into place at Vassar.
The publicly available model has four main variables to experiment with regarding Vassar: the number of initial infections, the number of new infections introduced from outside the environment, the reproduction number—the rate at which the sick infect the non-sick—and the frequency of tests. The model also includes a variable for test sensitivity and test specificity (the level of false negatives and false positives, respectively), but President Bradley has confirmed test sensitivity of 80 percent and specificity of 99 percent, so that number is relatively solid.
President Bradley and her co-authors’ Vassar-specific inputs into the model were five initial infections—although it is unclear if that means undetected among the populace, or if it means successfully quarantined—on campus, one new infection from outside per week, a reproduction number of 1.25, and tests every two weeks. Model authors Paltiel, Zheng and Walensky’s best case scenario was 10 initial infections (although they said varying this number did not significantly change the results), five new infections from outside per week, a reproduction number of 1.5, and tests every two days. In other words, Bradley’s model calls for a better than best-case scenario. This is possible if everyone follows health and safety guidelines and Vassar’s contact tracing proves effective, but even under this scenario 78 students are infected with COVID-19.
If we harshen some of the assumptions just a bit, we can see how narrow the path for success is: Assume 10 initial infections instead of five, two new exogenous infections per week instead of one, and a reproduction number of 1.5 (what would still be considered a successful country or state) and we instead have a scenario in which 351 students are expected to be infected. Vassar stated that if more than five percent of on-campus students have COVID-19, the college will likely shut down for the semester. That equates to about 125 students, a number that could be reached if the only change from Bradley’s estimates was one more outside infection being introduced per week. Even if we keep the number of exogenous infections to just one per week, if COVID-19 transmission is just a little higher than expected, 1.4 rather than 1.25, then we still get a total of 125 students infected.
These are all just best case scenarios too. David Paltiel, one of the authors of the original study, was clear about this in a statement to Inside Higher Ed: “My problem with [Vassar’s plan] is it should be sufficient if the planets align and if everything that could go wrong goes right. I’m a little bit distressed by how confident Vassar is to weed out infection by re-entry and to keep the students inside that walled garden and to regulate that behavior.” Even with a community care pledge, contract tracing and the experience of handling COVID in the spring, our level of safety is on a razor’s edge. Anything that could even slightly increase the reproduction number of the virus could be what pushes the college over the edge. If you don’t believe me, by all means try the model for yourself. I suggest trying it with a reproduction number of 0.75 (contained), 2 (approximately a cruise ship, with shared dining and living spaces similar to a college), and 4 (pre-pandemic New York, the absolute worst-case scenario) to get a sense of scale, and then try various combinations of exogenous infections and a reproduction number between one and two.
The bare facts are that each student has to determine what level of safety they are comfortable with and take appropriate action, be it not taking in-person classes, committing to a voluntary self-isolation or not returning at all. This model is probably our best guess regarding the safety of a college campus, and its conclusion is that testing every two days is what is required for a safe return to campus for the students, employees and vulnerable members of the community. I hope President Bradley’s assumptions prove correct. After all, it’s our lives she’s betting.