Small groups of people and a man walking a dog keep their distance from each other on the Eastern Prom in Portland on Monday. Credit: Troy R. Bennett / BDN

The BDN is making the most crucial coverage of the coronavirus pandemic and its economic impact in Maine free for all readers. Click here for all coronavirus stories. You can join others committed to safeguarding this vital public service by purchasing a subscription or donating directly to the newsroom.

AUGUSTA, Maine — The Maine Center for Disease Control and Prevention has spent a month modeling how the new coronavirus might spread. The scenarios are wide-ranging.

Under one extreme, Maine could have thousands of new cases each day by June, with deaths rising rapidly. In another, cases are close to peaking now and will decline in the coming weeks. In one model, the difference between a 60 percent and a 70 percent decline in social contact amounts to hospitals being pushed past capacity or numbers being manageable.

[Our COVID-19 tracker contains the most recent information on Maine cases by county]

Not all outcomes are equally likely. Many of the scenarios modeled indicate that the peak Maine will come sometime in the next few weeks. Rather than estimating the number of Mainers who might become infected with or die from the disease, the Maine CDC has used the models to focus efforts on areas where the state is likely to run into issues, including hospital capacity.

“The reason we do this is to plan, not to predict,” said Nirav Shah, director of the Maine CDC.

As of Tuesday, there were 519 confirmed cases of the virus in Maine, with 12 deaths. The future spread depends on individual behavior that health officials are beginning to understand while they prepare for worst-case scenarios including a shortage of critical care beds.

Individual behavior affects the spread of the virus. It’s not the only unknown.

The Maine CDC looks at predictive models developed by the University of Washington, Johns Hopkins University and Imperial College London. The results of those models vary based on levels of social distancing — which can only be roughly estimated — while also relying on other imprecise assumptions about the nature of the virus and how much it has already spread.

The models clearly show that lowering the rate of physical contact reduces the number of people each person who contracts the virus goes on to infect. That key number, known as the basic reproduction number, is estimated to be around 2.2 on average for the coronavirus prior to public health interventions, though it could be anywhere between 1.4 and 3.9.

For the spread of the disease to slow, that must drop below 1, meaning everyone with the disease must transmit it to less than one other person on average.

Made with Visme

In one iteration of the Johns Hopkins model, an increase in social distancing from 60 percent and 70 percent swung the basic reproduction number below 1 within the next few weeks. According to the model, that was the difference between cases continuing to increase steadily — eventually requiring a surge in hospital capacity — or cases beginning to decline.

“Which one are we going to be?” Shah said. “We don’t know.”

That’s because it is hard for the Maine CDC to say whether the state has reduced social contact by 60 percent, 70 percent or some other amount. School closures, Shah said, might account for a 30 percent reduction in social contact. Transportation and cellphone data indicate that Mainers are traveling less, but it is hard to say exactly how that correlates to physical interactions.

In Wuhan, the Chinese city where the virus emerged, researchers estimate that residents reduced social contact by between 80 and 90 percent, which effectively stopped the spread of the virus. The effects of any public health intervention are unlikely to show up in case counts for two to three weeks in Maine, Shah said, in part because the period between exposure and symptoms can be up to 14 days.

“The course of the epidemic fundamentally in Maine will depend on how much public health interventions reduce the contact rate,” Shah said. “The virus is not going to change.”

Critical care beds may be Maine’s biggest constraint.

The Maine CDC has also spent weeks surveying hospitals on their supply levels, including the numbers of intensive care beds and ventilators, and comparing those numbers to the scenarios predicted by the models.

Shah said that helps the CDC decide whether to focus additional resources on issues such as surge capacity, supplies of equipment or staffing, among other issues.

The models generally suggest that the number of critical care beds is the most likely constraint on the state’s capacity, followed by the number of respiratory therapists. That has led the Maine CDC to prepare for temporary coronavirus treatment sites across the state.

Under some of the worst case scenarios, temporary sites, which have been set up in some of the places hit hardest by the virus, including Wuhan and New York City, could mitigate the overcrowding of hospitals. In other scenarios, such sites would not be needed.

Models are ever-changing and subject to wide ranges.

The models that the state is using change daily because new data is always coming in, so they may look different depending on the day one sees them. For example, the model developed by researchers at the University of Washington was updated on Monday to reflect additional data based on outcomes in Germany, Spain and Italy.

That update had the effect of reducing the projected cases and deaths sharply in the United States and in Maine, with the model now predicting 277 Mainers requiring hospitalization for coronavirus at the epidemic’s peak. It projects 115 total deaths by the summer, but that projection is deeply uncertain, though researchers expect that the number will lie somewhere within a relatively wide range of 58 deaths and 216 deaths by August.

Shah compares modeling the disease to forecasting the weather, saying that predicting future case counts was in some ways similar to predicting “what the barometric pressure in Presque Isle will be on July 17 at 2 p.m.”

“Models are inherently uncertain,” he said. “No weatherman, almost never, will say there’s a 100 percent chance of something happening tomorrow or a zero percent chance.”