with Paulina Firozi

A prominent model highlighted by the Trump administration early in the coronavirus crisis projected in March that deaths in the United States would total about 82,000 by now. 

But the U.S. death count, now at 153,000, is nearly double that. 

If anything has become clear over the past five months, it’s that accurately predicting how quickly the novel coronavirus will spread and how many people it will kill several months into the future is nearly impossible. 

Policymakers are under intense pressure to respond to the public health crisis – and citizens also understandably want accurate numeric predictions on how many coronavirus infections and deaths to expect in the coming months. Cases and hospitalizations now appear to be dropping, but the seven-day average of new U.S. covid-19 deaths topped 1,000 for the ninth day in a row yesterday.

Future deaths are even harder to predict than the course of a hurricane, such as the one that worked its way up the Eastern Seaboard this week. 

And one of the biggest mistakes is focusing on just one model — like the one from the University of Washington’s Institute for Health Metrics and Evaluation (IHME) above — instead of taking many different projections into account, epidemiologists stress.

“You make better forecasts when you make multiple models,” said Nick Reich, an epidemiologist at the University of Massachusetts School of Public Health. “If you rely on just one model, you’re going to be disappointed.” 

That’s why the Centers for Disease Control and Prevention is relying on a mash-up of 32 models.

The CDC is funding a University of Massachusetts team run by Reich, who has developed a way to compare and merge more than two dozen different models from universities and analytics groups around the country. The models include the IHME model, along with modeling from the University of California, the University of Michigan, Oliver Wyman and other prominent centers.

The group’s website, dubbed the Covid-19 ForecastHub, updates the individual models on a weekly basis and also depicts an “ensemble” curve that combines them. That merged curve predicts 172,446 deaths in the United States by the end of August, although the numbers will be updated today. 

Before the pandemic hit, Reich had spent the last four years working on seasonal influenza forecasting for the CDC. He stressed that drawing from a portfolio of models is crucial for accuracy — and he applauded the agency for pursuing that approach, even though it has been heavily criticized overall for its pandemic response.

“I know CDC has been sort of dumped on throughout this, but for the last eight years they have been a leader in creating this vision for collaboration,” Reich told me. “This is far more than any one agency could do or any one company could do.”

When I asked the CDC to detail which projections it is relying on, a spokesman pointed me to a Web page that lists the models used by ForecastHub and links to the website. 

ForecastHub only makes predictions four weeks out.

Modelers stress that the further out a model tries to project infections, hospitalizations or deaths, the less accurate it will be. Reich feels any projection beyond six weeks into the future is virtually useless.

That’s because modelers have to plug in a number of estimates, including how infectious the virus is, how many sickened people will die it and the extent to which people practice social distancing. (Here’s an interactive explainer on how modelers arrive at their verdict.)

If any of those estimates are off, they’ll result in inaccurate projections. The longer the time span, the further off a projection can be.

“We have limited ourselves to four weeks because we feel beyond that, the models are heavily dependent on untested assumptions about what will happen in the future,” Reich said.

Reich recently tweeted more of his thinking about modeling:

It’s especially challenging to predict what kinds of social distancing measures will be in place, because states are revising them all the time based on local developments. According to the CDC website, 25 models assume existing distancing measures will remain in place during the prediction period, while the others make different assumptions about how social distancing will change in the future.

Public officials are still under pressure to make predictions far into the future.

Speaking to CNN over the weekend, Deborah Birx didn’t rule out the possibility that some 300,000 Americans will have died by the end of the year. The White House coronavirus response coordinator had been asked about that estimate from Scott Gottlieb, former head of the Food and Drug Administration.

She said the death toll will depend on the mitigation efforts by states and how compliant people are with social distancing regulations.

"Public health is called public health because it has a public component, and we need all of the public to help us get in control of this virus," she said, urging people to wear masks in public, avoid crowds and practice social distancing.

But Birx noted that human behavior can dramatically alter outcomes. Such an outcome would be far less likely if people were vigilant about social distancing and avoiding large gatherings.

Michael Osterholm, chairman of the Center for Infectious Disease Research and Policy at the University of Minnesota, said the media’s treatment of modeling has presented “a real challenge” in helping Americans understand its limits.

“I’ve not necessarily been a big fan of models; I think they’re all wrong,” Osterholm told me. “In this pandemic, it’s been a real challenge because the media tend to focus on a new model estimate when it comes out, and for 36 or 48 hours there’s all kinds of fanfare.”

Ahh, oof and ouch

AHH: A group of seven states have formed a pact to pressure companies to ramp up rapid-detection test production. 

The states — Virginia, Louisiana, Massachusetts, Michigan, North Carolina, Ohio and Maryland — banded together to come up with a plan in the absence of a national testing strategy. They’re led by three Republicans and four Democrats and say other states and cities may join as well, Erin Cox reports. 

The governors say “talks have already begun with one of the two companies approved by the FDA to sell point-of-care antigen tests that can detect the virus in less than 30 minutes,” Erin adds. “Maryland Gov. Larry Hogan (R) negotiated the deal during the final days of his tenure as chair of the National Governors Association.” 

The plan is for each state to request 500,000 rapid tests to help alleviate their reliance on the traditional testing infrastructure bogged down by long delays. 

“With severe shortages and delays in testing and the federal administration attempting to cut funding for testing, the states are banding together to acquire millions of faster tests to help save lives and slow the spread of COVID-19,” Hogan said in a statement. “We will be working to bring additional states, cities, and local governments on board as this initiative moves forward.”

OOF: More than a dozen nursing home companies that allegedly misused federal funds received hundreds of millions in coronavirus relief. 

These for-profit nursing home providers have settled civil lawsuits in recent years with the Justice Department over accusations of fraudulent Medicare billing, substandard care and other abuses, according to a Post analysis of federal spending, Debbie Cenziper, Joel Jacobs and Shawn Mulcahy report. 

“The companies repaid the government a total of more than $260 million and nearly all are under active corporate integrity agreements with the inspector general of the U.S. Department of Health and Human Services — the same department that distributed the coronavirus relief payments. The five-year agreements require independent audits, employee training and other enhanced reporting protocols,” they write. “… All told, nursing home companies sued for Medicare fraud in recent years received more than $300 million in relief payments.”

The “no strings attached” funding, which was distributed via the $175 billion Provider Relief Fund, is meant to pay for coronavirus-related expenses or lost revenue. 

“Watchdog groups and some independent experts say the government should have assessed the financial needs of the nursing home industry, scrutinized the track records of providers and attached some spending restrictions before distributing the payments,” Debbie, Joel and Shawn write. “They say they are most concerned about for-profit companies, some owned by private equity and other investment firms, that in the past have slashed costs and cut staff to boost profit.”

In addition to the nursing home providers facing accusations of misusing funds, millions more in funding “went to nursing homes with widely publicized breakdowns during the pandemic.”

OUCH: Scientists aren’t yet sure if plasma from covid-19 survivors works as a treatment.

The Trump administration has poured $48 million into a program at Mayo Clinic to allow tens of thousands of patients to get infusions, and Trump tweeted this week calling on survivors of the disease caused by the coronavirus to donate their plasma. 

But researchers haven’t been able to finish studies, because unexpected demand for plasma has inadvertantly undercut the studies that could prove it works, the New York Times’s Katie Thomas and Noah Weiland report

“The only way to get convincing evidence is with a clinical trial that compares outcomes for patients who are randomly assigned to get the treatment with those who are given a placebo," they write. "Many patients and their doctors — knowing they could get the treatment under the government program — have been unwilling to join clinical trials that might provide them with a placebo instead of the plasma.” 

The easing of the virus outbreak in many places across the country has also made it harder to recruit infected people for research. 

Katie and Noah add: “In a statement, a spokeswoman for the Food and Drug Administration said that the expanded access program was meant to bridge the gap until trials could get underway and ‘was never intended to substitute for randomized clinical trials, which are critically important for the demonstration of efficacy.’ ”

Reopening schools

U.N. Secretary General António Guterres warned of a “generational catastrophe” as schools remain closed and students stay home. 

In a video message, he urged countries to suppress the virus sufficiently to allow schools to reopen, calling the coronavirus pandemic “the largest disruption of education ever.” A policy brief published alongside Guterres’s message emphasized that suppressing transmission of the virus is “the single most significant step” leaders can take toward reopening schools, Anne Gearan reports

The Trump administration is doubling down on its push for schools to reopen.

“White House press secretary Kayleigh McEnany was asked Tuesday about the experience of Israel, where some schools that opened have closed because of a surge in coronavirus cases,” Anne writes. “She replied by quoting Robert Redfield, director of the Centers for Disease Control and Prevention, who said last week that children need to return to school, but omitting his cautions about doing so safely.” 

Schools are making a patchwork of decisions as students begin classes. Many public schools, under heavy pressure from teacher's unions, are starting with virtual-only learning, while smaller private schools are more likely to start in-person.

“Trump’s demands that schools reopen while infection rates are increasing in most states have politicized reopening decisions being made at the local and state levels. Many district leaders, including in Republican-led states, have said they are starting the academic year virtually because it is too dangerous to reopen school buildings and risk the spread of the coronavirus,” Anne reports. 

Mississippi’s conservative governor, Tate Reeves, for one, reversed course and announced a delay the return to school for middle- and high-schoolers as he also expanded mask-wearing requirements. 

Agency efforts

The CDC is warning about a seasonal spike in a rare polio-like condition that largely affects children. 

Prevalence of acute flaccid myelitis has surged in even-numbered years since observation of the disease began in 2014, Lenny Bernstein reports. 

The disease may be caused by any of several viruses. It can progress over hours or days and lead to permanent paralysis or respiratory failure, according to the CDC. 

“Among 238 cases in 2018 reviewed by the CDC, 98 percent of patients were hospitalized, 54 percent required intensive care, and 23 percent were placed on ventilators to help them breathe,” he adds. “… It is unclear whether mask-wearing, social distancing and other measures that have been taken against the coronavirus will limit the outbreak of acute flaccid myelitis expected this year.”

Coronavirus latest

Here are a few more stories to catch up on this morning: 

Congress on the coronavirus: 
  • The White House and Democratic leaders agreed to try to finalize a deal on unemployment benefits and eviction restrictions by the end of the week, as negotiations continue on the latest coronavirus relief bill, Seung Min Kim, Erica Werner, Carol D. Leonnig and Jeff Stein report.
The hardest hit: 
  • The cost of groceries is spiking for basic staples, as supply chains are disrupted by the pandemic. The price increases have hit struggling households hardest — particularly unemployed Americans who saw their unemployment benefit expire last week, Rachel Siegel reports.
More on schools: 
  • The support staff members who keep colleges and universities running are worried about the return to campus. They fear inconsistent safety protocols and students who may not wear masks and could put their health at risk, Danielle Douglas-Gabriel reports. “They are the people who clean the classrooms, tend the grounds, sort the mail and keep students fed, but every interaction raises their risks. And sitting the semester out is not an option if they want to earn a living,” she adds.

Sugar rush

For 20 years, reality TV dating shows have provided escapism and entertainment, but at times have perpetuated damaging stereotypes. (The Washington Post)