COVID-19 deaths and infections appear to have slowed down for now, but as we begin to emerge from our homes in an attempt to return to normal, we need to brace for the storm to surge back.

The nation has watched Georgia, my home state, just now end its shelter-in-place order. But depending on how strictly the public practices physical distancing measures, the total number of deaths in the state, which reached 1,300 in the first week of May, could range from 6,000 to 18,000 by August, and at the pandemic's peak the number of new infections per day could range from 40,000 to 80,000. The peak demand for hospital beds could range from 11,000 to 25,000; for ICU beds from 1,800 to 4,000; and for ventilators from 900 to 1,900. These demands are likely to far exceed the capacity that may be available for COVID-19 patients in the state.

Our research suggests shortages across 14 coordinating hospital regions in Georgia, especially under low or medium compliance with physical distancing after the shelter-in-place order lifts. In some scenarios, even if all available hospital resources were used for COVID-19 patients, at the peak, it would still not be sufficient for the majority of the regions.

In some regions, the deadly shortfalls could extend for several weeks. This further emphasizes the importance of strict compliance with voluntary shelter-in-place and physical distancing behavior after the end of state-ordered shelter-in-place.


These projections are based on an agent-based model we developed to predict the spread of COVID-19 geographically (at the census-tract level) and over time (the simulation model in this paper follows a structure similar to the model in two of our earlier papers—here and here—on pandemic influenza). Unlike curve-fitting models, the agent-based simulation enables us to model disease progression (how the sickness runs its course in an individual, depending on age) and behavior (e.g., compliance with physical distancing recommendations) at the individual level. It captures the disease spread by modeling person-to-person interactions in households, workplaces, schools and communities, utilizing detailed data such as household size, workflow and population demographics.

In the simulation model, we tested various scenarios of shelter-in-place and voluntary quarantine along with school closures with different durations and time-varying compliance levels. The model results suggest that during the first 180 days of the disease (starting around mid-February), if there had been no intervention, the percentage of the population infected could have exceeded 64 percent, and the peak would have occurred around mid-April. School closures reduce this percentage to around 55 percent and delay the peak by about a week. In other words, despite their significant impact on the society, school closures alone are not sufficient to make a big dent on COVID-19 spread.


Though hopeful pharmaceutical candidates are appearing on the horizon, none is available so far to slow or stop COVID-19—no antivirals, no vaccine. Our primary weapon against this highly infectious disease for the foreseeable future is physical distancing (including school closures), shelter-in-place, isolation of people confirmed with the disease, and voluntary quarantine.

The majority of schools are closed in the United States for the remainder of the academic year, and most states issued shelter-in-place orders in March or April of four weeks or longer. These interventions have depressed COVID-19’s advance. In our models, we found that shelter-in-place significantly helped slow the disease spread and delay the peak.

That was good, but also temporary, and now we are at a critical juncture.

As states start lifting shelter-in-place orders, if very strict compliance with physical distancing guidelines slips, COVID-19 will spread fast and far. It may help to look at each social distancing measure for its effectiveness, and our model found particular potential in voluntary quarantine.


Under voluntary quarantine, the entire household must stay home if there is a person with cold- or flu-like symptoms—even in the absence of testing or confirmation of COVID-19—until the entire household is symptom-free. Our research suggests that compliance levels with voluntary quarantine have a tremendous impact on the percentage of a population that gets infected and the number of new infections per day at COVID-19’s peak. After the lifting of a four-week mandated shelter-in-place in Georgia, the percentage of the infected population could range from 28 percent to 46 percent by August, and the peak could occur anywhere between June and August, for high and low compliance with voluntary quarantine, respectively.

High compliance with voluntary quarantine delays the peak and significantly reduces the total number of infections and deaths. That would reduce the stress on health care workers, hospital beds, ICU beds, ventilators and other resources.

However, we would like to caution that this intervention alone is not sufficient. Some households may have individuals infected with COVID-19 even if all household members are asymptomatic; voluntary quarantine would not impact such households and they could infect others.


COVID-19 threatens almost all aspects of human life as we know it, from public health to supply chains and the economy to relationships. The health and well-being of the population is of utmost importance, but there is also a growing urge to relax physical distancing, to go back to normal, as economic and social pressures mount.

While it is highly effective in reducing infection spread, shelter-in-place can be socially and economically disruptive if it remains in place for a long time. As an alternative, voluntary quarantine only targets households with symptomatic individuals, hence, the number of people staying home at any given time is significantly less under voluntary quarantine—even under high compliance levels—compared to shelter-in-place.

There is no easy way back to normal yet. But as we move into a new reality, we must consistently follow physical distancing and also never forget that asymptomatic people can still infect others. States must strongly advocate voluntary shelter-in-place, voluntary quarantine and other physical distancing measures. Our collective patience and commitment to continuing physical distancing will save lives and help us return to a healthier and stronger society, which is a foundation for a strong economy.

Read more about the coronavirus outbreak from Scientific American here, and read coverage from our international network of magazines here.