shows a statistical model of coronavirus cases
A model created by the Institute for Health Metrics and Evaluation at the University of Washington of what North Carolina's surge could look like in the coming month. Prior models generated by the same organization showed a much more dire outlook. Economists, statisticians and epidemiologists have been examining data in various models to get a sense of what the coronavirus outbreak will look like as more people get sick. Image, data: IHME/ UofW

Are you a health care worker? We’d love to hear from you. Email editor at northcarolinahealthnews.org

By Anne Blythe and Rose Hoban

If North Carolina lifts its stay-at-home order at the end of April and goes back to business as usual before social-distancing measures were in place, the state’s health care system likely would be overwhelmed quickly with people infected with COVID-19 in need of urgent care, state-specific models show.

Got a question about COVID-19? You can ask us and we’ll try to find an answer! Click here.

Scientists affiliated with Duke University, RTI International and the University of North Carolina at Chapel Hill shared their findings with reporters on Monday as the number of COVID-19-related hospitalizations continues to climb.

They looked at two questions:

What would be the need for hospital beds if the state lifted the physical distancing restrictions included in the governor’s executive order set to expire on April 29?

What would happen if those policies remained in place longer?

“Maintaining some form of social distancing policy as those in place right now will give us the best chance to be sure our health system has enough capacity to manage the number of COVID-related infection,”  said Aaron McKethan, a senior policy fellow at the Duke University Margolis Center for Health Policy and part of the modeling collaborative. “We’ve shown that lifting all social distancing policies completely after April 29th, leads to a much higher, about a one in two — or 50 percent — probability, that our hospital acute and ICU capacity will be outstripped.

“We furthermore acknowledge in the brief that there are many challenges to maintaining social distancing.”

McKethan cited such issues as having a state with porous borders surrounded by states with different social distancing measures as well as the cost of keeping North Carolinians home and the impact on businesses that have had to close to the overall economy.

South Carolina is one of only nine states in the country that has not enacted strict social distancing measures, such as stay at home orders. As of Monday, South Carolina, with 48 deaths, has a death rate from COVID-19 that’s more than three times the rate of North Carolina’s, which has had 33 deaths in a population that’s more than twice that of the neighbor to the south.

‘Spaghetti plots’

The forecast from the independent collaboration, a dozen researchers with epidemiological, health care and actuarial expertise, draws from three models and functions much as weather modeling does during an approaching storm.

Statistics experts from UNC-Chapel Hill, RTI International and Duke likened predicting the surge of coronavirus cases to predicting the path of a hurricane. “It’s very hard to predict the precise day when a hurricane will hit our shores, especially when the weather is currently calm and the storm seems far away from our coastlines. But it helps to have some kind of informed probability, or weather forecast about this so that we can adjust our behavior and make plans,” said Aaron McKethan, a senior policy fellow at the Duke University Margolis Center for Health Policy. Photo of Hurricane Florence, Sept. 12, 2018, courtesy: NOAA/ National Hurricane Center

“We have been using a metaphor familiar to every North Carolinian who has dealt with hurricanes, or other extreme weather events,” said McKethan. “It’s very hard to predict the precise day when a hurricane will hit our shores, especially when the weather is currently calm and the storm seems far away from our coastlines but it helps to have some kind of informed probability or weather forecast.”

McKethan and others compared their estimates to the hurricane “spaghetti plots,” that predict possible hurricane tracks so familiar to Tar Heels.

What the modelers found was that without social distancing North Carolina could expect as many as 750,000 COVID-19 cases by Memorial Day compared to the 250,000 if similar measures to those in place now remained.

They issued their findings on the heels of adjusted model results from researchers at the University of Washington Institute for Health Metrics and Evaluation. Their newest model shows fewer deaths in this state than previously modeled as well as a peak next week, also much sooner than the institute’s previous forecasts have shown.

The local academics said their forecasts are not as rosy, in part because the information they’ve used in building their models is more specific to North Carolina.

An early model graph of bed demand generated by the COVIDACTNOW group based in Silicon Valley. Graph courtesy: CovidActNow

“For example, the RTI model is based on the North Carolina population, and has very specific attributes about the North Carolina population,” said Pia MacDonald from RTI and UNC Chapel Hill. “We can get to a granularized approach that is quite different from these other models that you read about in the newspaper.”

She also said that hot spots in the state can escalate very quickly because COVID-19 is so contagious, driving rapid growth in local areas, such as Charlotte.

“I worry that as some models have concluded that the peak is almost here, people fail to recognize that those models are often based on an assumption that social distancing continues for weeks to come,” Mark Holmes, director of the Cecil G. Sheps Center for Health Services Research and a professor at the UNC-CH Gillings School of Global Public Health, said after the news conference. “It can be easy to look at current trends and conclude that we can relax our social distancing.  But we need to recognize our history is a function of the policies that have been in place; most models recognize that relaxing social distancing would lead to a faster growth rate in infections.

An updated model from the advocacy group COVIDACTNOW shows that North Carolina’s peak hospitalization level has been reduced and the date of the peak has been pushed out several weeks into the future. Graph courtesy: CovidActNow

“That is, the fact that most models have continually revised downward their projections is a consequence of – not in spite of — the policies that have been enacted.”

The collaborative issued their forecast to help lawmakers and other policymakers figure out the best way to battle the pandemic during a time when there are no scientifically proven treatments for the ravaging pneumonia that accompanies some cases.

How many tests?

One of the big unknowns at present is how many people in North Carolina, or anywhere, have the coronavirus or have had it in the past.

Studies out of China and elsewhere indicate that many people, as many as 20 percent of the people who contract the disease, only have very mild symptoms or none at all. But at present, there’s no blood test to determine whether someone had the virus in the past, so there’s no way to gauge how many people might already have some immunity to the disease.

One idea that’s been floated by the press office of General Assembly Senate leader Phil Berger (R-Rockingham) is to randomly test North Carolinians to see what the actual prevalence of the disease in the population might be. In one release, he argues that this could be accomplished with as little as 1,000 tests, a number obtained from Stanford scientist John Ioannidis.

Ioannidis said in a follow-up email to NC Health news that while 1,000 samples would be useful, “Ideally, I would like a substantially larger sample, e.g. 10,000.”

The local academics who delivered their models Monday agree. They also said there are not enough tests in the state at present to test widely.

“We would need to repeat those tests over time,” said UNC-Chapel Hill epidemiology professor Kimberly Powers, who called testing an important goal.

Similar random testing regimens are proceeding in other countries such as Germany, where they’re testing about 350,000 each week. Even cities such as Miami have a testing rate that’s greater than simply a snapshot of 1,000 tests.

“Testing priorities right now in the middle of an outbreak really should be geared towards clinical care of people who are potentially infected and sick, and their close contacts,” MacDonald chimed in. “There are not unlimited testing capabilities, or testing supplies or ability to process tests right now. So the idea that we need to conserve testing for those that need it the most is very important.”

Republish our articles for free, online or in print, under a Creative Commons license.

Anne Blythe

Anne Blythe, a reporter in North Carolina for more than three decades, writes about oral health care, children's health and other topics for North Carolina Health News.