How This U of L Researcher Is Forecasting The Spread of Coronavirus In Kentucky
A team of researchers including a mathematician from the University of Louisville are poring over the numbers trying to estimate how far the coronavirus will spread in states like Kentucky.
The number of confirmed cases has accelerated since January with global cases surpassing 400,000 cases on Thursday, according to Johns Hopkins University. The U.S. Surgeon General said this week “it’s going to get bad” as Kentucky Gov. Andy Beshear warns it will get worse before it gets better.
Researchers are tracking the virus like a hunter stalks its prey — picking up clues along the path. It’s not easy with only weeks of data at their fingertips, but their estimations can help governments, hospitals and communities make more informed decisions.
“If you know what is the worst case, what is the best case, then you know how to prepare for it,” said Dan Han, University of Louisville associate professor with a doctorate in applied mathematics.
Han is using her Ph.D. to develop a disease outbreak model for the coronavirus in conjunction with researchers from Johns Hopkins, University of North Carolina at Charlotte and University of Michigan.
“I created the model based on those statistics, and I think it looks scary,” Han told WFPL News on Friday. And that was before the team revised the numbers… upwards.
There are inherit risks in making predictions because of the limits of available data. A tweak of a figure may dramatically change the outcome. Han is working with about three weeks of data for the state of Kentucky and building an outbreak model based on the available information she can glean from scientific studies on the coronavirus.
“We are only in the early stages of this,” she said.
It’s especially difficult to make estimates for cities like Louisville, where there is even less available detailed information.
Han admittedly began with a more optimistic outlook with her figures. For example, researchers do not yet truly know the average number of secondary cases generated by a first case. That's called the basic reproduction rate.
Han initially assumed that disease would likely spread more slowly in this country then in China because the U.S. is less densely populated. As a result, she decided on a basic reproduction rate of 1.9.
How The Model Works
But Han’s outlook has changed. Based on the rapid increase in transmission seen over the last week in the U.S., she revised the reproduction rate to 2.24 so now she’s assuming that for every one person who catches COVID-19, 2.24 new cases will be generated. Her new reproduction rate is more closely aligned to the figures in a study the Imperial College of London released last week, which used a value of 2.4.
Another largely unknown factor is the number of asymptomatic, undiagnosed patients. Using data from the World Health Organization, Han estimates the virus has an incubation period (period in which symptoms develop) of 5.5 days and an infectious period of 15 days.
Han’s forecast also does not take into consideration the impact of social distancing, which could cause the forecast to overestimate the number of confirmed cases depending on the effectiveness of government policies. The forecast is also based on the number of confirmed cases, but the U.S. has limited tests so it is also likely undercounting the actual number of cases in the country.
“I think all of the estimations right now are not that accurate, but this is the best we can do,” Han said.
What She Found
As of Wednesday, 124 people have tested positive for COVID-19 in Kentucky with 32 of those cases coming from Jefferson County. Here are Han’s estimations for the next month:
- 7 days: 68
- 15 days: 107
- 30 days: 232
- 7 days: 394
- 15 days: 635
- 30 days: 1,354
The researchers’ model assumed that the number of confirmed cases in Kentucky will peak after 150 days: on Thursday, August 20.
In that worst case scenario, without taking into consideration the impacts of social distancing, Han predicts by that time, there will be nearly 87,000 confirmed cases in Louisville and around 635,000 in Kentucky.
Looking At Fatalities
The New York Times reports the CDC estimates between 200,000 and 1.7 million Americans could die over the course of the pandemic — if there is little effort to contain the virus. The Imperial College of London estimates between 1.1 and 2.2 million people will die in the U.S.
Han’s predictions skew toward even lower estimates. Han initially assumed a fatality rate of 3.4% based on information from the World Health Organization. But on Wednesday, the CDC came out with a new study looking at the COVID-19 outbreak on cruise ships. In that case, researchers found the fatality rate to be around 1%, Han said.
That report has provided crucial additional information because it includes results for both people who did, and did not have symptoms — giving a more complete picture of where the virus was spreading.
So based on a 1% fatality rate with the number of confirmed cases peaking at 150 days, the model finds 869 deaths from COVID-19 in Jefferson County, 6,353 deaths in Kentucky and 465,434 deaths across the country, Han said.
Impact On Hospitals
Han’s model highlights more than anything, the need to suppress the spread of the disease. The faster the virus moves through a community, the more quickly it can overwhelm local hospitals.
"Now we know how to prepare for it, like how many masks do you need, and how many medical facilities do you need to provide, and how many nurses should get ready," Han said.
There are a total of 13,301 licensed short-term care hospital beds in Kentucky with an average occupancy of 47%, according to a 2018 Kentucky Cabinet for Health and Family Services report.
The average occupancy is already down because Gov. Beshear canceled elective procedures at the same time hospitals are preparing to expand intensive care units to brace for an influx of cases.
"We have plans in place to convert hotels into hospitals with additional beds, if needed,” Beshear said.
Kentucky’s hospitals are also reporting the number of open beds and their overall capacity to the state, but it’s not clear if the state plans release that data to the public, said Ginger Dreyer, a spokesperson for the Kentucky Hospital Association.
When, where and if hospitals could be overwhelmed: those are the kind of questions Han thinks she could answer, if she access to that data.