A new forecast says the United States can expect a total of over 410,000 deaths by Jan. 1, according to a model from the University of Washington. That’s an additional 225,000 deaths from now until the end of the year.
The estimates are based on real-time data from local and national governments, hospitals, the World Health Organization and other sources, rather than assumptions about how the disease will spread.
It even predicted that if a herd immunity approach is pursued, which President Donald Trump has suggested as a way to push through the pandemic, the death toll could reach 620,000 by Jan. 1. Meanwhile, if more people wear face masks as advised, about 122,000 lives could be saved from now until year’s end.
The White House has been using the Institute for Health Metrics and Evaluation forecast to inform public policy, so the model is often coined as the “key” one to pay attention to during the pandemic.
But what about all the others that exist? Why do different models forecast different numbers and which one should the public follow?
For perspective, the Centers for Disease Control and Prevention uses 35 models from dozens of research centers across the country to forecast what the U.S. coronavirus death toll might look like.
The agency calls it their “national ensemble forecast,” which has predicted that total COVID-19 deaths is likely to reach anywhere between 200,000 to 211,000 by Sept. 26. The CDC’s data does not extend beyond that date, making it hard to compare it to other models that release further predictions.
Some experts think it’s “irresponsible” to publish such a deep forecast that extends to the end of the year like the University of Washington model.
“There’s just so much uncertainty. … There are too many variables going on and no one really can know for sure what’s going to happen,” Youyang Gu, a data scientist who runs a separate model called Covid-19 Projections, told CNBC. “There is just really no data to work off of for a winter season.”
The variables in question include state’s constantly changing mandates, government decisions, hospitalization rates, community spread and more. Some experts assume more people will wear masks whereas others predict lockdowns will ease earlier than reality; some go for the statistical modeling approach while others focus on disease transmission instead.
Models are intended to help policymakers plan for the months ahead, which means they need to change along with the data. But just because they change doesn’t mean they’re wrong, experts say.
Estimates are to be taken with a grain of salt.
In June, the University of Washington model predicted 200,000 Americans will die from COVID-19 by October, which is less than a month away, McClatchy News previously reported. As of Sept. 4, more than 187,000 people have died in the U.S., according to a Johns Hopkins tracker.
And in August, the same model released their most recent predictions on death toll until the end of Dec. 1, which stated that nearly 317,000 Americans will die from COVID-19.
There’s no telling which model is most accurate or dependable, but all modelers agree that continued devotion to mask wearing and social distancing can prove many forecasts wrong for the better.
“Looking at the staggering COVID-19 estimates, it’s easy to get lost in the enormity of the numbers,” Dr. Christopher Murray, director of the Institute for Health Metrics and Evaluation model, said in a news release. “The number of deaths exceeds the capacity of the world’s 50 largest stadiums, a sobering image of the people who have lost their lives and livelihoods.”
“But the science is clear and the evidence irrefutable: mask-wearing, social distancing, and limits to social gatherings are vital to helping prevent transmission of the virus.”