The findings follow an assertion by Dr. Scott W. Atlas, the president’s science adviser, that masks are ineffective, in a tweet later taken down by Twitter for spreading misinformation. On Wednesday, the Centers for Disease Control and Prevention released new guidance recommending mask use in public settings, including public transportation.
A surge of infections, driven in part by neglect of safety precautions, has begun to overwhelm hospitals in much of the nation. More than 75,000 new cases were reported in the United States on Thursday, the second-highest daily total nationwide since the pandemic began. Eight states set single-day case records.
These numbers are likely to continue through the fall and winter, with a steady rise in cases and deaths until January and staying high after that point, said Christopher Murray, director of the Institute for Health Metrics and Evaluation at the University of Washington and lead author of the report.
“We strongly believe we are heading into a pretty grim winter season,” Dr. Murray said.
The new study, published in the journal Nature Medicine, also offered a rough estimate of the pandemic’s toll in the United States: perhaps 500,000 deaths by March 2021, even with social distancing mandates reinstated in most states.
Other experts cautioned that, as with any model, the new estimates are based on many assumptions and should not be seen as predictions.
“It’s not a prediction or forecast, because we can will this number out of existence,” said Shweta Bansal, an infectious disease modeler at Georgetown University who was not involved in the new work.
Instead, she said, the model should be seen as a “sophisticated thought experiment” whose conclusions can significantly change if people alter their behavior.
“I’d like for people to see this study as a call to action, sort of a wake-up call, especially for those individuals who are unconvinced by the devastation that this pandemic is causing,” she said.
Epidemiological models that try to predict trends far into the future, as the new one does, are particularly prone to flaws “given how dynamic the situation is, and how quickly things can change,” added Ashleigh Tuite, an infectious disease modeler at the University of Toronto.
Still, she and others said, the numbers seem reasonable as a rough estimate of the toll by March 2021 if current trends continue.
Dr. Murray and his colleagues analyzed the number of cases, testing rates, mask use and cellphone data to estimate people’s movements from the first recorded case in each state through Sept. 21. They then estimated the death toll until March 2021 for each state, with or without mandates for social distancing and mask use.
If many states continue to roll back the mandates in place, the team found, the number of deaths by Feb. 28 could top one million, with one-third occurring in California, Florida and Pennsylvania.
More plausibly, states might reinstate distancing mandates when daily deaths reach a threshold of eight deaths per million. That would result in 511,373 deaths by the end of February 2021, according to the model.
Other models don’t look as far into the future or haven’t taken seasonality into account, and have underestimated the number of deaths that could result, Dr. Murray said.
Such models “feed the not very science-based views that are circulating out there that the epidemic is over, or the worst is behind us,” he said. “And that’s a pretty risky strategy.”
But Dr. Tuite said she was unsure whether even accounting for seasonality, deaths would peak in the spring, as the model estimates. Dr. Murray’s model does not take into account the treatments available now for people who are hospitalized, she added.
For example, deaths among hospitalized patients have dropped to 7.6 percent from 25.6 percent in the spring, according to one study.
The new research rests on other flawed assumptions, Dr. Bansal said. The model offers estimates for individual states but does not account for age- or location-based variations within states, and the figures are based on limited testing and death data from the early part of the pandemic.
Because of these and other assumptions, the estimated number of deaths is at best an approximation. Still, the figure underscores the need for individual and population-wide precautions.
Dr. Murray and his colleagues showed that mask use, in particular, has a considerable impact, cutting down the risk of infection at both an individual and population level by about half.
As of Sept. 20, just under half of Americans reported that they always wear a mask. But regular mask use by 95 percent of the population would save 129,574 lives, according to the new analysis. Regular mask use by just 85 percent of Americans could prevent 95,814 deaths by March 2021, possibly forestalling restrictive lockdowns, Dr. Murray said.
“Increasing mask use is one of the best strategies that we have right now to delay the imposition of social distancing mandates and all the economic effects of that, and save lives,” he said.
Mask mandates and penalties for not wearing a mask can raise the numbers of people wearing the face coverings, he suggested.
The mask estimates are also likely to be rough approximations, but even so, Dr. Tuite said, “the qualitative finding is really important, which is that it has an impact, and an impact in a way that’s far less disruptive than lockdowns or other more restrictive types of interventions.”
Masks are an effective and inexpensive tool to stem the spread of the virus and yet have unfortunately become politicized, like much else in the pandemic, said Dr. Carlos del Rio, an infectious disease expert at Emory University in Atlanta.
“If you wear a mask, you’re a Democrat,” he said. “If you don’t wear a mask, you are a Republican. And I think that’s what’s totally wrong.”
“The fact that we continue making masks such a political issue is really upsetting,” he added, “because quite frankly, I don’t want to see people die.”