Simulations show how dangers of reopening universities amid COVID-19 could be avoided

March 17, 2021
Simulations show reopening universities isn't without risk. (AP Photo/Ben Margot)

Simulations show reopening universities isn't without risk. (AP Photo/Ben Margot)

Summer is approaching, COVID-19 appears to have an end in sight and colleges are heading toward normalcy for the 2021 academic year, but after running simulations of the virus' transmission rate at U.S. universities, researchers are urging that reopening only be considered if the pandemic's familiar safety precautions and bulk rapid testing policies are both strictly enforced.

Their comprehensive analysis, published Wednesday in Nature, includes data collected from over 80 universities. Results indicated that because a profuse amount of transmission is attributed to asymptomatic carriers, present methods of contact tracing and symptomatic testing are not nearly enough of a precaution to reopen educational institutions.

Ujjal Mukherjee, the lead author of the study and a professor at the University of Illinois at Urbana-Champaign, explained to The Academic Times that rather than looking at when to reopen schools, his team decided to ask the question, "What will it take to reopen?"

Their motivation follows the race toward reopening among college administrators. For instance, the provost of New York University, in the former epicenter of the pandemic, announced in February that NYU will resume fully in-person instruction starting this fall, noting growing vaccine availability.

Decisions to return to in-person classes come despite the fact that last fall, universities were shown to be potential superspreaders. Mukherjee hopes his study's results will help large educational institutions understand what precautions need to be taken to safely control the virus from being transmitted, and avoid that risk. His team also plans to expand on the work by taking vaccine rollouts into account.

"We have started a project to look at the effect of combining testing with vaccination at a community level," he said, mentioning that even with the rollout of vaccines, they believe that, "Decisions [of relaxing policies] need to be carefully vetted out through regular surveillance testing and estimation of infection loads in large educational institutions."

One technique considered safe and effective by the Centers for Disease Control and Prevention is contact tracing. Scientists use this term in reference to locating which individuals have a confirmed case of COVID-19, tracking who's in their social circle, then asking all of these people to voluntarily reside in an isolated location of their choice for a standard quarantine time

But Mukherjee showed that, on its own, contact tracing is not sufficient to ensure the safety of students and staff at educational institutions. 

While the simulations indicated that contact tracing is important at the initial stages of reopening when infection rates are naturally low and chains of transmission can be detected more easily, as time went by, bulk, randomized, rapid testing was more effective. This method involves testing many individuals for COVID-19, regardless of whether they have active symptoms or admit an exposure, in a cost-effective and quick manner.

"At later stages with wider spread of the infections these chains multiply and break, then bulk testing performs a better job," he said. The numbers revealed that, ideally, one-quarter of an institution's entire population should be tested per day.

"Our simulations show that for large-sized institutions like large public universities, testing at the rate of 20% of the population per day [would be] safe," he said, adding that beyond this fraction, incremental gain is actually marginal.

However, Mukherjee stresses that this testing rate ultimately depends on compliance with safety factors such as social distancing, mask-wearing and restricting mass gatherings that often take place at universities.

The findings also indicated that, within reason, the speed with which test-takers receive their results is more important than test sensitivity. This is because until test results come back, asymptomatic carriers tested at random would likely go about their day without extra precautions.

"If the results delay after testing then the testing essentially becomes ineffective," Mukherjee said. "A one-day delay is equivalent to a 50% reduction in effectiveness of testing."

He added, "The surrounding society's infection rate should play a big role, since no institution is an island," conveying that the simulations addressed how students and staff would probably interact with people outside the institution by including the surrounding community's COVID-19 transmission behaviors.

The simulations' design had several components for accuracy. First, all transmission data was acquired from various universities around the country. The pool included private as well as public institutions, in both urban and suburban areas. Then, the team decided on parameters that would describe the correct characteristics of a university-like institution, such as size and group events. 

The model, called an analytic epidemic model, tested for asymptomatic transmission of COVID-19, the effect of contact tracing, bulk rapid testing and the rate at which people would come into contact with others within and outside the institution. 

"We tried to model the infection dynamics of a small population in an analytical setup that does not consider the normal simplifying assumptions of a large population model — which are generally used for community population-level transmission," Mukherjee explained. 

After designing the model, the team performed computational simulations that predict outcomes of complex systems. These models are called agent-based simulations, and the complex system they tested for included the criteria set aside during the parameter selection phase. These simulations also tested specific, small-scale populations to prevent oversimplifications.

"Using real data from different universities, we are able to estimate these parameters by using a curve fitting technique," Mukherjee said. "We show that we are able to trace back and predict the infection dynamics at institutions quite well."

The team controlled for factors such as high-risk category individuals and efficiency of quarantine, and Mukherjee does not expect that typical model-based limitations of the study would adjust the results significantly. 

Going forward, Mukherjee is growing his team and so far has included a doctor specializing in infectious diseases. Because he and his current team are policy-focused and speaking from a mathematical perspective, they hope this addition can help provide a biological perspective of test effectivity and how vaccine rollout might adjust the findings. 

The paper, "Evaluation of reopening strategies for educational institutions during COVID‑19 through agent based simulation," was published March 17 in Nature. It was authored by Ujjal K. Mukherjee, Subhonmesh Bose, Anton Ivanov, Sebastian Souyris, Sridhar Seshadri, Ronald Watkins and Yuquian Xu, University of Illinois at Urbana-Champaign; and Padmavati Sridhar, University of California, Berkeley.

We use cookies to improve your experience on our site and to show you relevant advertising.