Health and policy experts dissect what makes some states more vulnerable to COVID-19 overload

On Friday, a three-person panel of experts discussed via webinar the various models being used to make predictions about the spread of the new coronavirus and how local and national leaders should be using that information to make decisions. 

The webinar was organized by The Commonwealth Fund, a private foundation based in New York that advocates for a higher quality of health care. The conversation focused on three big questions: Will hospitals have enough beds, ventilators and staff to care for people with severe symptoms from the new coronavirus? How soon can the community lockdowns end? And how soon can people return to work and school and their normal lives?

Last week the group released a detailed analysis of several federal data sources to better understand how COVID-19 might impact states differently. They considered the health risk factors of residents, the number of hospital beds and ventilators, the ratio of skilled providers to population, access to health insurance coverage, and state spending on public health systems, among other statistics.

The Commonwealth Fund analysis concluded that as many as 44% of adult Americans could be at risk for the worst COVID-19 symptoms because of age or chronic health conditions. The report also noted that the U.S. currently has one ventilator for every 4,878 adults and one ICU bed for every 2,941 people. By counting doctors who are not currently providing direct care — such as those in managerial positions — the fund estimated that the country might be able to add another 6.6 people to the frontline of COVID-19 care.

The figures varied significantly state to state and county to county. 

For instance, the percentage of residents the report deemed to be at elevated risk ranged from 52% in West Virginia to 37% in Utah. Both Kentucky and West Virginia data showed more of their elderly residents also had chronic health conditions than people older than 60 in the country as a whole. The number of ventilators per 100,000 residents ranged from 13.3 in Oregon to 77.6 in the District of Columbia. 

“It’s not necessarily a situation that we think one state is more prepared or less prepared,” said David Radley, a public health scientist and researcher with The Commonwealth Fund. “All states are going to face big challenges if and when COVID cases surge in their states. We hope the resources we’re putting together gives context to the people trying to manage resources.”

In reality, leaders of governments, hospitals and businesses are inundated by the volume of information available, which grows every day. 

Numerous scientists, economists and technology company analysts have published a variety of similar data models using traditional information sources, like testing and deaths data, demographic estimates, and hospital staffing and equipment reports. Others have developed novel approaches, like aggregating data from digital thermometers to identify communities where COVID-19 fevers might not match testing information and GPS tracking data from cell phones to measure whether people are obeying shelter-in-place orders.

The toughest part can be understanding which predictions to use when making critical decisions about health resources or shelter-in-place orders, said Andy Slavitt, who served under President Barack Obama as acting administrator for the Centers for Medicare and Medicaid Services. 

“Several governors have been told their peak date may be in May or in June, but other people are telling them their peak date is right around the corner,” Slavitt said. 

The cost of making a wrong choice can be high.

“If they underestimate, they end up with people dying in hallways in their hospitals, which is hard for anybody to stomach,” Slavitt said. If they’re wrong the other way, states that buy too many ventilators, for instance, might deprive those machines from other states that ultimately need them.

Dr. Nirav Shah, co-founder of Covid Act Now and a scholar at the Stanford University Clinical Excellence Research Center, urged leaders to ask five questions when evaluating which data to consider and which models should guide their decisions. 

Does the model match reality at a high level — for instance, is it matching what’s happened in other countries where outbreaks started sooner? Does the model make assumptions that are risky and what would happen to this model if it’s 10% wrong or 50% wrong? Is it clear how the model was made? Is it clear what the limitations are for the model and what could cause it to be wrong? Is it vetted and endorsed by professional scientists?

The answers leaders will find, of course, is complicated because the quality of data used in many models is poor, Shah said. 

Because the U.S. has limited testing, it’s impossible to have an accurate, real-time picture of where the virus is and who is infected. The information that is available is a dated snapshot because it can take days for people to develop symptoms and days more for them to receive test results and even more time for public health officials to trace all of their community contacts then report it. People with the worst symptoms might spend two weeks in the hospital before recovering or dying, again creating a delay in information available to leaders.

Slavitt called for states and the federal government to share even more information with the public than they are now and to do so as quickly as possible. He urged them to go beyond simple case counts to publish clear tallies of health care resources, such as ventilators, beds and workers. 

“There is no success other than saving as many lives as possible and I don’t know how we can save as many lives as possible when we don’t know what’s available,” he said.

In the end, the experts on the panel largely agreed that leaders don’t need predictive models to do the right thing. Decades of infectious disease research clearly says what is most effective at saving lives: early, decisive and sustained action. 

“Ventilators are saving half the people who are put on them. Ventilators are not the answer,” Shah said. “It’s the upstream — social distancing and lockdowns — that are going to save lives.” 

Slavitt admits it will be tough for leaders to make the right choice even if it is obvious to them.

“There’s going to be enormous pressure on governors to start to relax things,” he said. “The scientists and security experts I talk to tell me, ‘We know exactly what we need to do and we know for sure we won’t be able to do it.’”