A University of Washington model of coronavirus deaths was updated—and increased 10%— Wednesday to include a presumed increase in nursing home deaths (according to CNN), ultimately predicting the country will see 66,000 deaths from the disease by August.
The model, created by the University’s Institute for Health Metrics and Evaluation, includes mortality data from Johns Hopkins University, along with data from the World Health Organization and governments across the globe.
The increased estimate comes from states beginning to count nursing home deaths as COVID-19 cases, the institute’s director Chris Murray told CNN Tuesday.
In the update, the model included a state-by-state estimate of when social distancing guidelines could be relaxed, stressing that the timetable is dependent on containment measures: testing, contact tracing, isolation and limiting gathering size.
According to the model, states that could consider relaxing social distancing between May 4 to May 10 are: Alaska, Hawaii, Montana, West Virginia, North Carolina and Vermont.
Meanwhile, the model says states that should wait until June 8 or later include Georgia and Florida (both have been criticized for relaxing restrictions too quickly).
“If people start to go back to normal social interaction or even progressively go back, the risk of transmission will go up,” Murray told CNN.
“That the [University of Washington] model keeps changing is evidence of its lack of reliability as a predictive tool,” epidemiologist Ruth Etzioni, who once worked with the institute, told Stat News after the model’s April 17 update. “That it is being used for policy decisions and its results interpreted wrongly is a travesty unfolding before our eyes.”
“Model outputs *should* change in response to new data and when you’re dealing with nonlinear systems, small changes can have a reasonably large effect,” tweeted Nate Silver of FiveThirtyEight in response to Etzioni’s comments, above. He said the model’s revised death toll was “reasonable,” but its projected three- to five-fold decrease in hospitalizations was “not great” because the revision was large.
“Forecasts of deaths will help inform public health decision-making by projecting the likely impact in coming weeks,” according to the Centers for Disease Control website. The CDC says it considers data from University of Washington model and three other organizations to inform its efforts. But the White House in a March 31 press briefing appeared to rely on the Washington model, along with a model created by Imperial College in London. As many as 2.2 million Americans could die without mitigation efforts like social distancing, according to Imperial College’s forecast. At the time, the University of Washington model predicted 38,000 to 162,000 U.S. fatalities would occur through summer. Since then, the model has been revised at least six times, according to a Forbes review. Critics and unnamed White House advisors expressed doubts about the Trump administration’s estimated COVID-19 death toll, the Washington Post reported April 2. The University of Washington institute has warned officials not to rely solely on its predictions. “If I were a governor of a state, I would certainly not make a decision based just on our model,” Murray said in a virtual briefing last week.
COVID-19 estimation updates (IHME)
COVID-19 Forecasts (CDC)