Forecasting a Healthy Workplace: Simulation as a COVID-19 Mitigation Tool


The endgame for the COVID-19 pandemic is clear: we will find ways to live with the virus, which will allow us to return to business as usual. Earlier last year, the highly effective mRNA vaccines gave hope. 

But the rise of the Delta and Omicron variants and other factors have slowed progress; many employers are struggling to find ways to co-exist with the coronavirus. Mitigation protocols have upset some employees, who may not accept “new normal” working conditions, like wearing masks, physical distancing, regular testing or vaccine requirements. Even businesses reluctant to implement these efforts may now have to comply with new directives from the federal government mandating vaccination or testing for companies with over 100 employees.

Employers will have to develop new strategies that work for their employees, maintain compliance with these legal directives and minimize business interruptions. The good news is that many companies are already using tools and approaches in other facets of their work that can set them up for success in the new post-pandemic era.

From retail and transport to food services and e-commerce, businesses have adopted simulation models to forecast, manage and predict supply, demand and resource needs. Expected in today’s business environment, infectious disease monitoring is even built into some of these models. 

Applying this same thinking to employee health — using forecasting models to simulate COVID-19’s spread and the impact of mitigation approaches — is key to pinpointing the plan that will work best for ensuring workplace productivity, safety and morale. 

Building the Viral Model

Early in the pandemic, Professor Paul Romer, an economist at New York University and Nobel laureate, shared a visual simulation of COVID-19’s infectious potential.Opens a new window In a community depicted as 200 small, blue triangles moving at random, five became infected with the coronavirus and turned into red dots. When a red dot passed near a blue triangle, the infection spread, turning that blue triangle into another red dot, which, in turn, infected others until it recovered and became noninfectious and immune. 

Romer ran the simulation fifty times, and, typically, 80-90% of the blue triangles became infected and infectious red dots. That meant that with no mitigation efforts in place, the COVID-19 virus stood to infect nearly everyone on earth, overwhelming global health systems. Like supply chain software that depends on weather forecasts and transportation updates, the more accurate the assumptions and data used in these epidemiologic models, the better the predictions. Thanks to worldwide research efforts, we now have a much stronger understanding of how this virus spreads: 

  • Most viral spread results from respiratory droplets — talking face-to-face, sneezing, and coughing. 
  • Infection risk is highest where droplets are most concentrated — within a few feet of an infected person. The finest droplets and aerosol particles can remain airborne for minutes to hours. Masking reduces the probability of transmission by about one-third. 
  • The virus is prone to mutations that can make it more infectious. For instance, the original COVID-19 virus had an R0 of about 2.5, meaning that each infected person was likely to pass it to 2.5 other people, on average. The Delta variant that now makes up most U.S. cases has been estimated to have an R0 of between 5 and 8. And the new Omicron variant is estimated to have about twice the infectivity of the Delta variant.
  • The COVID-19 virus has other pesky traits. Many people who get infected never become symptomatic, even though they can still infect others. And many who develop symptoms can be contagious before those symptoms arise. For the initial COVID-19 virus, people could be asymptomatic for 7-14 days, while with the Delta variant, that figure is about 2-4 days. One studyOpens a new window estimated that about half of all COVID-19 spread is caused by asymptomatic and presymptomatic people.

These are a few of the more predictable facets of the coronavirus that inform better modeling. What’s less predictable is the people. That’s where simulations come into focus.

Putting Simulations to Work

In his original work, Romer introduced one more feature: a mitigation strategy, namely COVID-19 testing. He suggested regularly testing a random sampling of people, symptomatic and asymptomatic. If any of those tested were infectious, they would isolate until they were no longer contagious. In this simulation, even with imperfect testing, the situation was much more manageable. Instead of 80-90% of the population facing infection, he found about 20% in total were infectious and never more than 10% at a time.

Romer’s model proved correct. For those who could afford it last year, daily testing was a highly successful strategy. Everyone else asked, “What amount of testing will be just right — balancing safety and affordability?” 

The forecasting modelsOpens a new window that our company and others built have helped address this and other questions. Working with each employer for each site, the answers depend on key factors, such as the prevalence of COVID-19 in the surrounding communities where employees and their families live, local worksite conditions (factory floor versus office, for example), viral variant types, rates of immunity and risk tolerance of the employer. 

In general, we have shown it is less risky if everyone coming back is screened first — tested for the coronavirus before or upon arrival. That helps identify who needs to be isolated and, if necessary, treated. Baseline testing also establishes the magnitude of the COVID-19 problem, which can inform appropriate policies. Together with masking, physical distancing, handwashing, and other precautions, testing programs like these kept the likelihood of a serious outbreak low last year for many colleges and businesses.

The Challenge of Back to Work

Like the pandemic itself, models and simulations evolve over time. More real-world data allows us to hone our models and make them more accurate. We factor in county-level variations in COVID-19 prevalence to account for geographic variations, include new variables like viral variants and vaccinations that confer varying immunity levels, and build into the model information about different vaccination rates across worksites.

While we cannot ever precisely predict the behavior of a mutating virus in a chaotic system, these tools help guide local, near-term decision-making. Leaders can simulate potential results of certain decisions, like implementing mask mandates or different testing regimens. And they can evaluate each decision’s impact on worksite health vis-a-vis employee sentiment. 

This knowledge is key for finding ways to put the pandemic behind us. As we gaze into our cloudy crystal ball, one thing is certain: businesses that invest in forecasting and managing their employees’ health successfully will be at the forefront of our return to a more normal and brighter future. 

How are you using computer models and simulations for making staffing decisions? Let us know on FacebookOpens a new window , TwitterOpens a new window , and LinkedInOpens a new window .