When the surge of COVID-19 hit states varies wildly
By Jill Riepenhoff, Jamie Grey and Lee Zurik
Originally Published: April 10, 2020
(InvestigateTV) – When one coronavirus forecaster downgraded its COVID-19 projections in early April, it launched a range of reactions on Twitter.
Some wrote that the new model from the Institute for Health Metrics and Evaluation at the University of Washington proved strict social distancing and stay-at-home orders were working.
Others tweeted that the model proved state and federal officials overreacted and were crying wolf when the country was virtually locked down.
A model produced by COVID Act Now has been called both a liberal hack job and life-saving tool.
The IHME model shows that by mid-July no COVID-19 patients will be hospitalized.
The COVID Act Now model shows that even under the best of circumstances, that nearly 100,000 people still will be in the hospital fighting the disease in the summer.
Similar data points such as hospital data and confirmed COVID-19 cases can equal vastly different outcomes in these unprecedented times.
“In normal times, the models wouldn’t go directly from the scientists to the public. They would go through rigorous,” said Baruch Fischhoff, professor in the Department Engineering and Public
Policy and Institute for Politics and Strategy at Carnegie Mellon University. “It’s legitimately difficult for the public to figure them out.”
InvestigateTV analyzed the two models to illustrate how models interpret the data differently.
Both make predictions about how many hospital beds could be needed at different points in time and the dates of peak bed needs. But their results are vastly different.
In fact, there is an average 53-day difference between the peak hospital bed surge based on estimates both had as of April 7. For example:
• ICHE shows Vermont hit its hospital peak on March 23. Act Now predicts the surge will hit June 12.
• Nine states have already seen that bed surge come and go including New York, Michigan and Louisiana – three of the hardest hit COVID-19 states. Act Now predicts that on July 6, more than 28,000 total beds will be needed in those states to treat patients.
• By the end of May, ICHE shows that 20 states will need less than one hospital bed. Act Now shows that on July 6, every state will still need dozens, hundreds or thousands of beds.
Epidemiology models that are driving decisions on response to the coronavirus can be vastly different even as they examine similar hospitalization, death and infection-rate data. InvestigateTV compare two models: one from the Institute for Health Metrics and Evaluation at the University of Washington and another from COVID Act Now, a nonprofit organization created to make predictions about the deadly virus. InvestigateTV charted the dates in which each model predicted (in their respective best case scenarios) that each state would need the most hospital beds. These are the results.
Infographic: Cody Lillich, InvestigateTV
“These models give us different results because they use different numbers. They divide up the world differently,” Fischhoff said. “What’s missing is the synthesis… the overall picture of all the models.”
But while the models may disagree on when the virus will ebb, they have done a good job of helping Americans understand the social distancing and flattening the curve so hospitals like those in
New York City aren’t overwhelmed, he said.
But that’s little comfort to parents trying to work from home and educate school-aged children at the same time, to people trying plan a vacation or to a nursing-home resident who hasn’t been allowed a visitor in weeks.
“People are really tough as long as they feel they’re being treated fairly,” said Fischhoff, who also is a psychologist. “If somebody makes confident promises that can’t be fulfilled, then you really make life hard.”
Many states also are using different models tailored to their own state and COVID-19 circumstances. When matched against the national models, they, too, often disagree.
One created for Washington D.C. said the nation’s capital won’t peak until June 28 while ICHE said it hit that milestone April 3.
A University of Michigan model shows that by April 28, 1,400 of its states residents will be hospitalized. ICHE predicts on that day that between 136 and 741 beds will be needed.
In Ohio, researchers at The Ohio State University predict the Buckeye state will reach its peak April 19. ICHE shows the peak already happened. Act Now said the surge happens June 8.
Early models in Ohio also predicted that the state could see as many as 10,000 new COVID-19 patients a day. It prompted Gov. Mike DeWine in mid-March begin shuttering nonessential businesses and ordering people to stay home.
Because of those measures, the state likely will peak in mid-April with just 2,000 new cases a day, the Ohio State model shows.
“You have different models. We didn’t suppress that,” DeWine said at a press briefing. “Modelers didn’t think we’d do as good of a job as we did.”
The models show, he said, social distancing is working and that cases will grow if measures are relaxed too soon.
“We determine the way this story ends. It’s not the modelers,” DeWine said. “We write this story every single day.”