The St. Louis city skyline with Gateway Arch photographed at sunset. (Photo: Stephen Emlund/iStock)

Even as vaccines begin to roll out in large numbers, new coronavirus variants present tough challenges for leaders in setting health policy. With the TRACE (Testing Responses through Agent-based Computational Epidemiology) simulation model, the City of St. Louis now has help.

Since we first reported on TRACE last July, SFI External Professor Ross Hammond, a senior fellow in Economic Studies at the Brookings Institution, along with a team at Washington University, have been using the model to help policymakers at different levels of scale manage the pandemic uncertainty — addressing both knowns and unknowns to produce multiple potential outcomes for any policy option. And early results of the model make the benefits of agent-based modeling clear.

A petri dish for policy

Hammond and his colleagues pitch TRACE as a petri dish for policy, allowing for all kinds of policy experiments you can’t do in the real world. For St. Louis, the team simulated approximately 10,000 policy combinations across 16 different epidemiological scenarios and looked at millions of discreet scenario simulations.

While most other models do meta-analysis based on averages, TRACE goes the other direction, looking at all the possible values and covarying them. This makes for robust results which are critical for policymakers dealing with uncertainty.

“If you want to know how many ER beds you’ll need next month, conventional forecasting models are great,” Hammond explains. “But if you want to understand all the intervention options across a wide range of scenarios and in a heterogeneous population, these models don’t have the completeness of the policy choice set or the uncertainty set. Only by using the complex systems approach,” he emphasizes, “can we get these kinds of insights.”

Findings point to the value of masks

St. Louis is a big, diverse city that crosses multiple state lines. They face several health challenges — including disparities — and are now grappling with the risk of a fourth wave of COVID-19.  For local policymakers, perhaps the biggest takeaway was the importance masks will continue to play in fighting the virus.

The model shows that if you increase the numbers of people wearing masks and wearing them correctly, you can counteract any variants effectively without doing much else. This vital information is now being shared on their health department website.

“Even this far into the pandemic, the scientific literature is unclear about what masks are achieving in the real world,” says Hammond. "We were able to show how good you have to assume them to be in order to control the variants.”

Lessons learned

For Hammond, the value of working with policymakers has never been more obvious. “Working collaboratively from early in the process to build a model that actually represents their setting and suits their needs is key,” Hammond says. “They then understand and have confidence in the model, and more willingness to act on it.”

Once the immediate COVID-19 crisis has passed, Hammond and team will look at lessons learned. “There are real risks of future pandemic events that we have to prepare for,” he explains, “but there are also broader lessons about the role of modeling in public policy and its ability to tackle hard problems quickly, and about the way this experience has highlighted disparities for our nation. These models, and the role of a complex systems approach, can be a part of how we address those disparities.”