A closely watched coronavirus model is now forecasting nearly 170,000 coronavirus deaths in the United States by October 1. The projections paint a grim picture of what could come when summer turns to fall, with a steep rise in daily deaths forecast in September.
The model, from the Institute for Health Metrics and Evaluation at the University of Washington, previously only offered projections until August. It was cited often by the White House earlier in the pandemic, and it’s one of many models currently featured on the US Centers for Disease Control and Prevention website.
It now projects that 169,890 people will die from Covid-19 in the US by October 1, with a possible range of about 133,000 to 290,000 deaths.
Daily deaths are expected to decrease through June and July then remain relatively stable through August. But the model forecasts a sharp rise in deaths come September.
In the model, projected daily deaths double from 410 on September 1 to 1,018 on October 1. Notably, though, the model’s uncertainty increases as time goes on. At the beginning of October, for example, the model offers a range of 96 to 4,382 possible daily deaths.
“If the US is unable to check the growth in September, we could be facing worsening trends in October, November, and the following months if the pandemic, as we expect, follows pneumonia seasonality,” IHME Director Dr. Christopher Murray said in a statement.
In a press release, IHME said its model is based on data through June 6. “Large gatherings in some states due to lifting of social distancing restrictions, gatherings on national holidays, and public protests are reflected in the general trend toward increased mobility,” the institute said.
The model looks at cell phone data, and increased mobility means people are moving around more. That could lead to more contact and opportunities for coronavirus transmission, but how exactly mobility corresponds with infections remains unclear. Behavioral changes like physical distancing and mask wearing could decrease the risk.
IHME says it includes other factors in its model as well, including data on testing, pneumonia trends, mask use, population density, air pollution, altitude, smoking and “self-reported contacts.”
According to the institute’s analysis, mask use results in a 50% reduction in Covid-19 transmission. But IHME’s data on masks is also self-reported, and it may not be representative of the population at large.
The IHME model has been closely watched, but it’s been criticized before for its assumptions and performance. It used to project that US coronavirus deaths would stop this summer, for example, which some experts said was unrealistic. IHME has since made a number of revisions to its methodology.
Still, the model is just one of many – and it’s not the only forecast offering longer-term projections. Another model, from independent data scientist Youyang Gu, projects that 201,550 people will die from Covid-19 in the US by October 1, with a range of about 147,000 to 284,000 deaths.
That model, which is also highlighted on the CDC’s website, projects an increase in daily deaths in July before a decrease in August and September.
Last Thursday, an ensemble forecast from the CDC projected more than 127,000 coronavirus deaths in the United States by June 27. The short term forecast relied on 20 individual forecasts from outside institutions and researchers. The CDC regularly releases new projections, so another ensemble forecast could be released later today.