LITTLE ROCK, Ark. — As cases increase in Arkansas three months into the COVID-19 pandemic, a data model made public by the University of Arkansas for Medical Sciences predicts people could be dying by the dozen over several days in October if the state fails to stick to social distancing guidelines.
The UAMS School of Public Health has been tracking the coronavirus for state officials since March, but its local advice has been overshadowed by high-publicized models by the Institute for Health Metrics and Evaluation out of the University of Washington.
The IHME model and others played a large role in convincing many public officials to issue lockdowns and shelter-in-place orders. They have since been criticized as their projections fell far short in places like Arkansas, which didn’t broadly close down.
Now that cases are increasing and other states that opened over the last month see new record numbers of cases, data scientists are warning everyone their dire predictions could still prove out if we don’t stick to mitigation measures.
“What the model is saying instead is, ‘all things considered, or if conditions at this time do not change, this is what could happen,‘” said Dr. Mark Williams, Ph.D., and dean of the School of Public Health.
He likened modeling to predicting hurricanes. They can project when conditions are right for tropical storms, but can’t tell you what day and where each one will strike.
Slides from his latest model presentation show the state well below capacity for intensive care beds, ventilators and hospital beds, but predicts a sharp rise beginning in July that trends beyond capacity at the end of September.
The IHME model sees COVID-19 deaths coming by twos and threes into August, but then a steady rise. By October, the line through the center of what’s called the “confidence interval” says 44 deaths in one day could be occurring.
“We don’t see those kinds of numbers in our model,” said Dr. Williams. “[At the start], what we saw here in Arkansas was there was no real steep increase in the number of cases day-to-day.”
But the outbreak centered in northwest Arkansas is a sign of a changing landscape.
“That steep [incline] is continuing to increase, and the rate at which it’s increasing is beginning to accelerate,” he said.
Whatever the reasons for that, Williams says these predictions can easily be altered. He nods to the challenge that comes when a model makes a dramatic prediction that scares a population into making changes and how that could be interpreted as making a bad call, especially as public health guidance has become politicized.
But without commenting on the decisions made by Governor Asa Hutchinson, Williams calls the messaging correct, even as guidelines relax. The insistence on wearing masks, practicing social distancing and avoiding big crowds for a while will make a liar out of his model.
“We’re asking for individual social responsibility and to do the right thing,” the governor said Monday after announcing 416 new cases.