Modeling gives data-driven picture of COVID-19 risks

Cornell researchers found no evidence of students transmitting COVID-19 to employees or the broader community during the fall semester, but students who traveled were eight times more likely to test positive within two weeks of their return, according to a new report that is helping the university shape its ongoing pandemic response.

The Cornell epidemiological modeling team, led by Peter Frazier, associate professor in the School of Operations Research and Information Engineering in the College of Engineering, has updated their mathematical model to reflect data from the fall semester about student and employee COVID-19 cases, as well as the rising prevalence of the coronavirus in the Southern Tier and across the country, and the evolution of faster-spreading virus variants.

“The success of our COVID-19 plan very much depends on our ability to make informed decisions based on sound science,” said President Martha E. Pollack. “We’re extremely fortunate to have the expertise of world-leading faculty whose input is allowing us to make decisions that are not just data-driven, but based on our own data – including modeling based on all of our own specific parameters.”

One of the most important outcomes of the new modeling is the university’s decision to increase the testing frequency for undergraduate varsity athletes, students living in group housing and those participating in social Greek-life organizations, from twice a week to three times a week.

By increasing the frequency of testing for these specific groups, modeling shows that the university could potentially reduce the number of infections in the student population by more than 50%.

More than 10% of the 86 student infections detected in the fall were tied to travel.  Therefore, the university is now scheduling at least one surveillance test for all undergraduate, graduate and professional students on Friday, Saturday or Sunday to discourage nonessential travel.

In addition to not finding evidence of students transmitting COVID-19 to employees or the broader community during fall semester, they found little evidence of transmission from the community to students.

However, the modeling tells a different story for university employees. The team found a strong correlation between infection rates in the counties surrounding Ithaca and infection rates in faculty and staff, with 75% of employee infections stemming from exposure to family members and at social gatherings, and from travel beyond Ithaca.

“We’re extremely fortunate to have the expertise of world-leading faculty whose input is allowing us to make decisions that are not just data-driven, but based on our own data.”

Martha E. Pollack

“We saw a lot of employee cases in December and January, and that was a period where we had really high prevalence in the broader community, mostly from holiday gatherings,” Frazier said. “And so you saw a lot of staff and faculty getting infected at home or during travel, and then we would find out about it in the surveillance program.”

To help address this, Cornell now offers the same comprehensive assessment of campus exposures and supplemental testing programs (such as adaptive testing) to employees that have been available to students.

“That is a really useful thing to have because many of our employees don’t live in Tompkins County,” Frazier said. “What that means is that when you do contact tracing for employees, one of whom lives in Ithaca, and another one lives in Elmira, and another one lives in Binghamton, it gets very complicated if that entails coordinating across three public health departments.”

The new modeling doesn’t reflect the cases reported at the start of the spring semester, but the report states that its conclusions are consistent with what has been seen so far.

The researchers have simulated several scenarios to anticipate how the transmission rate may change in the coming months. Reflecting a mix of pandemic fatigue, an increase in indoor gatherings during the winter and the impact of faster-spreading variants, their nominal assessment points to just over 200 student cases during the spring semester, a roughly 15% rise over the fall’s numbers. In more pessimistic scenarios, projections rise to more than 300 student cases.

Because vaccine availability for most students will not come in time to have a major impact on the spring semester, and employees already have very low on-campus transmission rates, the modeling doesn’t factor in the impact of vaccinations on infections. They will, however, reduce prevalence in the broader community and protect the most vulnerable.

Frazier said he remains cautiously optimistic about the coming fall semester.

“We have a system in place and it works,” he said. “One thing we’re always doing is, we review every case and we ask ourselves, ‘what can we do better?’ So it is very possible that either conditions on the ground will change, or we’ll start to see behavioral changes or new risk factors, and then we may end up tweaking policies in order to address those.”

Media Contact

Abby Butler