Monday, May 04, 2020

Models Shift to Predict Dramatically More U.S. Covid-19 Deaths

Virginia's governor Ralph Northam - the only US governor who is a physician - just extended Virginia's business shutdown for another week as reported by the Virginian Pilot. Whether the May 15, 2020, opening date gets further delayed will depend on whether Virginia sees 14 days of improvement.  Meanwhile, a number of GOP lead states are opening up businesses and playing a game of Russian roulette with their citizens' lives.  New reports on models predicting the Covid- 19 death rate suggest that  Northam may be taking the wiser choice given the new predictions that 134,000 Americans will die of Covid-19 between now and August (243,000 if one includes the upper end of the margin of error).  A Johns Hpkins report predicted 3,000 deaths per day by June (that would be a daily death rate not too much less than the 9-11 fatality number).   It goes without saying that Trump is cheering on governors reopening their states focusing as always only on himself and what a weak economy could mean for his re-election hopes. The Washington Post looks at the disturbing model predictions.  Here are excerpts:
A key model of the coronavirus pandemic favored by the White House nearly doubled its prediction Monday for how many people will die from the virus in the U.S. by August – primarily because states are reopening too soon.
The Institute for Health Metrics and Evaluation at the University of Washington’s School of Medicine is now projecting 134,000 coronavirus-related fatalities, up from a previous prediction of 72,000. Factoring in the scientists’ margin of error, the new prediction ranges from 95,000 to 243,000.
Dr. Christopher Murray, the director of IHME, told reporters on a call Monday the primary reason for the increase is many states’ “premature relaxation of social distancing.”
For the first time, Murray explained, the model is factoring in data from four different cell phone providers showing a major uptick in Americans’ going out in public. This rise of mobility in the last week or 10 days is likely leading to an increase in transmission, he said.
Even with its latest forecast, the University of Washington model is still far more optimistic than a model developed by Johns Hopkins for CDC predicting as many as 3,000 deaths per day by June.
In a statement on Monday evening, the Johns Hopkins Bloomberg School of Public Health called the leaked models “preliminary analyses,“ saying that they had been provided to the Federal Emergency Management Agency to help in scenario planning, were not a final version and were not meant to be used or presented as forecasts.
Still, the statement added, “the information illustrates that there are some scenarios, including the premature relaxation of social distancing, that are likely to cause significant increases in the number of COVID-19 cases and deaths in the United States.“
In a statement to reporters, White House deputy press secretary Judd Deere said that the internal report obtained by The New York Times was “not a White House document” and hadn’t been presented to the coronavirus task force.
The CDC document projects north of 175,000 new cases of Covid-19, the respiratory illness caused by the novel coronavirus, each day. That’s up from about 25,000 new cases per day last week and more than four times the peak of about 37,000 new cases per day.
The alarming modeling comes as some states are already beginning to put parts of the White House’s phased reopening plan into motion despite concerns that the administration’s guidelines for doing so have not yet been met. It also underscores fears that moving too fast to relax strict social-distancing restrictions could fuel a dangerous second wave of infections.
The CDC document found some reason for optimism, noting that nationwide, the trajectory of new illnesses in "multiple counties, including hard hit areas in Louisiana and in the New York City region" has continued to decrease, and that incidence rates have recently plateaued around Chicago.
Still, it found that there "remains a large number of counties whose burden [of illness] continues to grow or are in an elevated incidence plateau, including in the Great Lakes region, parts of the Southeast, Northeast, and around southern California." The document includes a color-coded map of the country with darker spots peppered throughout, and it states that the goal "is to have all communities be represented in the lighter colors, demonstrating little to no disease burden and no increase in trajectory."
Murray also noted that the updated University of Washington model also now takes into account the ramping up of testing in most states, as well as warming temperatures heading into late spring and early summer. But he cautioned that the impact of temperature on coronavirus transmission is not yet fully understood, and likely will not be for several more months.
For now, IHME is assuming that every degree Celsius the temperature goes up will lead to a 2 percent decline in transmission. “Are we sure about that? No,” he said.

No comments: