Researchers at the University of Washington now predict Virginia’s COVID-19 deaths will peak on April 22, a month earlier than their estimate just days before.
The updated prediction did not alter Gov. Ralph Northam’s plans for a surge sometime from late April through the month of May.
At a press briefing on Monday, Northam said his administration is looking at a number of projections.
“They change literally every day depending on the data that’s put into them,” Northam said. “So we haven’t made any adjustments.”
The model from the university’s Institute for Health Metrics and Evaluation has quickly become the most publicly-shared projection for the spread of COVID-19 despite disagreement on its methodology. It now predicts Virginia's hospital resource use will peak on April 20. It also substantially reduced its estimates for Virginia deaths, which it now expects to plateau at around 1,400 in early June.
In an April 5 post on its website, IHME researchers attributed the changes to better and more widespread data. The re-calculations take into account data from Spain and Italy on the effects of social distancing. Researchers said it also adds more state-by-state nuance on the types of social distancing rules in effect.
Some academics are skeptical.
Madhav Marathe, a professor in biocomplexity at the University of Virginia, said the model “has been a subject of significant debate in the scientific community,” in part because of concerns over the transparency of the methodology.
Marathe said he wasn’t sure how the model came to be so firmly embraced by policymakers, including the White House . He said all models, including one developed by his team, suffered from limitations caused by a lack of widespread testing.
Without that, scientists are left guessing what percentage of COVID-19 cases are picked up by testing, and how many patients are asymptomatic, Marathe said.
“That's the fundamental issue right now,” Marathe said.
Models may pull from different sample sizes, and data from one country may not be comparable in another, according to Pratha Sah, a postdoctoral associate at Yale University's Center for Infectious Disease Modeling and Analysis.
“It will take us many more weeks and many more months to completely understand this particular outbreak in the U.S.,” Sah said in an interview last month.
The limitations haven’t stopped Marathe and UVA’s Biocomplexity Institute from developing their own model, which was referenced in a speech last week by Northam. Their projections on Virginia were developed alongside the Virginia Department of Health, which provided the team with more granular data.
Marathe said they will release their own model to the state on Wednesday, though it remained unclear if VDH would release it publicly the same day.
Marathe declined to say when their model predicted deaths in Virginia would peak. He said the model would be updated weekly using data from VDH.
