New UW analysis lowers coronavirus death projections and suggests hospitalizations may have already peaked in Washington

Tribune Content Agency

SEATTLE — After a “massive infusion of new data,” modelers at the University of Washington are painting a much more optimistic picture of the novel coronavirus epidemic in the state, revising sharply downward their estimate of how many people are likely to die and suggesting Washington may have already passed the peak of hospitalizations.

The UW’s Institute for Health Metrics and Evaluation (IHME) says the state can expect roughly 632 deaths, compared to an estimate 10 days ago of more than 1,400 fatal cases of COVID-19, the disease caused by the coronavirus.

IHME Director Christopher Murray said the analysis shows that Washington — the state where the virus first took hold — is also among the first places in the country to begin bringing the epidemic under control. Gov. Jay Inslee cited the improved forecasts in his decision to return 400 ventilators to the national stockpile for use in harder-hit areas.

Nationwide, IHME is now projecting about 82,000 deaths through early August, down from an earlier estimate of 93,500. And while shortages of hospital and ICU beds will continue to be acute in hot spots like New York, New Jersey and Connecticut, many other parts of the country seem to have enough capacity to deal with a wave of severe infections expected to peak in most states by April 16, the analysis finds.

But the improved outlook is contingent on maintaining strict social-distancing measures, including closure of schools and nonessential businesses, at least through the end of May, Murray said Monday in a press briefing via Zoom.

“If you ease up prematurely … the potential for rebound is enormous.”

No model is perfect, and the numbers and trends are not guaranteed to be correct. But new information pouring in from Italy, Spain and the East Coast has greatly improved the researchers’ confidence in their results, Murray said. The UW group is currently the only source of state-by-state projections.

“We are acutely aware of the importance for us to get this right, because we don’t want to mislead decision makers,” he said. “At least for now, I think our models are the best that are available for … looking at the course of the epidemic across states.”

The possibility that Washington has already “flattened the curve” of the epidemic is borne out by observations at area hospitals. The number of patients hospitalized with COVID-19 infections at UW Medicine’s three Seattle hospitals has held steady at about 120 over the past five days, said Lisa Brandenburg, the facilities’ president.

The new UW forecasts also aligns more closely with patient counts at Swedish Health Services’ five hospitals in the Seattle area, said Chris Dale, chief quality officer. The peak number of hospitalized COVID-19 patients at Swedish so far was 102 on April 2. By Monday, it was down to 97.

Earlier in the spring, Swedish was bracing for up to 256 patients in intensive care units. But through Monday, there had been fewer than 60 at a time.

The UW hospitals haven’t had to add bed capacity or boost their staff above normal levels, thanks to the cancellation of most elective surgeries and other procedures. There also doesn’t appear to be a statewide shortage of ventilators, Brandenburg said.

“We are feeling increasingly confident we can handle the surge.”

The number of patients hospitalized with COVID-19-like symptoms statewide also suggests a downward trend, according to information released last week by the Washington Department of Health.

But DOH reported an additional 400 confirmed cases and 34 deaths from COVID-19 on Monday – which, if validated as one-day counts, would be unusually high.

Social distancing appears to have been particularly effective in reducing infections in King and Snohomish countries, where companies like Amazon and Microsoft began ordering employees to work at home long before Gov. Jay Inslee closed schools and nonessential businesses and issued a “stay-home” order, Murray said.

“That’s tremendously good news.”

Though testing levels are inconsistent, the number of new cases confirmed daily in King County has been generally declining since March 26. An apparent spike of 269 new infections reported April 5 included delayed results from previous days’ testing, health officials said.

“I think it’s too early to be confident, but I do feel we are moving in the right direction,” said Dr. Jeff Duchin, health officer for Public Health — Seattle & King County. “The progress is very tenuous because as soon as we start getting together again and having more contact, the number of serious infections and deaths will start to rise.”

The UW modeling has been in the national spotlight since President Donald Trump’s team cited it as one of the reasons for extending social distancing recommendations through April. Dr. Deborah Birx, White House coronavirus response coordinator, said the group’s projections were one of the sources used in the administration’s estimate that 100,000 to 240,000 Americans could die of the new virus.

Some critics have said the UW estimate is too rosy because it assumes uniform social distancing across the country, even though some cities and states still have not imposed restrictions. Murray said he thinks it’s likely those outliers will fall in line within the next few days.

The UW model relies on reported coronavirus deaths as its ground truth, and pulls in data from around the world on things like hospitalization rates, viral transmissibility and the impact of social distancing.

When the project started, the only evidence about peak hospitalization and lockdowns came from Wuhan, China, where the epidemic started. Since then, Italy and Spain have both imposed widespread restrictions, and have seen the number of cases peak, then begin to decline, Murray said. Now those experiences have been factored into to the U.S. projections.

The results’ lower estimate of hospital bed shortages is largely due to better data from New York and other states. Initial information from China suggested the number of people hospitalized was 11 times higher than the number who died, but in U.S. cities so far, the range is between five to seven times higher, Murray said.

The new data, along with improvements in forecast methodology, have also lowered some uncertainty levels. The March 26 projection of 1,400 deaths in Washington state fell within a range of statistical possibilities from 316 to 2,535. The new estimate of 632 deaths has a much narrower uncertainty range of 557 to 741.

The UW modeling effort began as a way to help UW Medicine plan for the epidemic and was expanded after other hospitals and states began clamoring for similar insights. Even though the estimates for Washington have steadily decreased, they were extremely helpful for planning, Brandenburg said.

Hospitals need to be ready for the worst-case scenario, she explained. While she’s delighted the new estimates veer more toward best-case territory, she still worries. “In the back of my mind, I’m still thinking we need to be prepared for those (higher estimates) because we don’t know for sure.”

Murray and his team will continue updating their model results daily, even as they begin to look ahead to what is likely to happen next. Society can’t remain locked down indefinitely.

But the very thing that has helped control the initial wave of the epidemic — a steep drop in transmission rates — also means that only a small fraction of people have been infected so far, Murray pointed out. That leaves a vast pool of susceptible human hosts for the virus.

In order to prevent another major wave, social-distancing measures will probably need to be eased gradually, accompanied by expanded testing and isolation of infected people.

“We cannot count on herd immunity, which simply means lots and lots of people got sick,” Murray said. “We’ll need other strategies.”


(Seattle Times staff writers Hal Bernton and Evan Bush contributed to this report.)


©2020 The Seattle Times

Visit The Seattle Times at

Distributed by Tribune Content Agency, LLC.