Hospitals using AI to predict the decline of Covid-19 patients — before knowing it works

Hundreds of hospitals are using an AI system to help predict which Covid-19 patients will likely become seriously ill, often without knowing whether the tool is actually effective, STAT's Casey Ross reports. The rapid deployment of the platform — created by electronic health record vendor Epic — shows how the pandemic is not affording the luxury of taking months or even years to properly test the tool before using it to inform care.

Here's more of the latest on the pandemic:

In new data released by accident, Gilead Sciences' antiviral drug remdesivir failed to speed the improvement of patients with Covid-19 or prevent them from dying in a long-awaited clinical trial conducted in China. The study was stopped early due to a low number of patients, but a spokesperson for the company said that "trends in the data suggest a potential benefit."
STAT's Nicholas Florko is out with another story on Rick Bright, the recently-ousted head of BARDA. The big question now is why Bright, who claims to have been demoted for pushing back against authorizing the use of hydroxycholoroquine for Covid-19, was involved in the decision to approve the drug at all, since there's no precedent for someone in his position to have been involved in drug approval decisions.
In a long essay posted to his blog, Bill Gates cautioned that innovation in the form of a highly effective treatment and a vaccine would need to be in place before the world could fully reopen.
In a STAT First Opinion, psychiatrist Morgan Medlock writes that although the quick steps that medicine and society have taken to adapt to Covid-19 are good, a larger, longstanding problem — racism — ought also to have been given equal attention and the benefit of swift action.
In another First Opinion for STAT, emergency medicine resident Caroline Schulman writes about the large volume of calls she's made in the past two weeks to patients who had tested positive for Covid-19, many of whom have questions that she is unable to answer. Read more here.

Treating Covid-19 is much more expensive than other common infections

The ongoing Covid-19 pandemic is expected to make hospital and government costs soar, and a new analysis finds that treating the disease is more than four times as expensive as treatment for other infections. A team of researchers at the City University of New York and other institutions simulated how different sections of the U.S. population could get Covid-19, and calculated associated costs for medical personnel, equipment, and medications. Covid-19 treatment for a single case costs around $3,000, the scientists estimated, which is 5.5 times more expensive than treating pertussis, and about four times as expensive as treating a case of the flu. The researchers also found that if 20% of the U.S. population is infected with Covid-19, then treating the infection, including follow-up care, would cost more than $214 billion. The findings highlight the need to aggressively reduce the number of Covid-19 infections, the authors conclude.

1 in 5 cancer survivors report not being able to change jobs over fear of losing health insurance

A new survey of cancer survivors reveals that around 20% of them report not being able to move jobs freely because they are worried about losing health insurance. This phenomenon, known as job lock, can adversely affect people's chances of progressing in careers and can also have a negative impact on their quality of life. Looking at responses from more than 1,300 cancer survivors who responded to a national survey about medical expenses in 2011, 2016, and 2017, scientists found that nearly 1 in 5 say they couldn't move jobs out of fear of losing health insurance. Around 11% of partners of cancer survivors who were surveyed reported job lock, while overall, around 33% of survivors said they or their partners experienced job lock. Among survivors, females were more likely to report job lock, while among partners, males were more likely to say so. Those under the age of 74 were also more likely to experience job lock.

Inside STAT: Coronavirus 'serosurvey' results are coming. Here's how to kick their tires

A WOMAN'S BLOOD IS COLLECTED FOR TESTING OF CORONAVIRUS ANTIBODIES AT A DRIVE-THROUGH TESTING SITE IN HEMPSTEAD, N.Y. (SETH WENIG/AP)

The results from the first studies to use blood-based tests to determine how widespread the Covid-19 outbreak in the U.S. is are trickling in. These tests, known as serological surveys, can serve as crucial tools in determining the extent of the pandemic, largely because the tests can help identify those who had Covid-19 even if they were asymptomatic or hadn't received an official diagnosis. But in the rush to get people tested, a flurry of unvalidated and otherwise unreliable blood-based tests flooded the market, and these initial studies may have used some of these tests to make their observations. As such, relying on the findings from these studies to paint an accurate picture of the crisis would be unwise, experts warn. STAT's Helen Branswell and Andrew Joseph have more here.

Nearly 800,000 people could die of malaria due to Covid-19 disruptions to prevention campaigns

A new WHO analysis finds that nearly 800,000 people in sub-Saharan Africa could die from malaria this year due to interruptions to anti-malaria campaigns brought on by Covid-19. The main campaigns that could be affected provide insecticide-treated nets, malaria diagnoses, and antimalarial medicines. In the worst-case scenario, nearly 769,000 people — or twice the number of people who died from malaria in sub-Saharan Africa in 2018 — could die this year. Even the best-case scenario, in which only some anti-malaria campaigns are disrupted, could lead to more than 411,000 deaths. The new analysis "reinforces WHO’s call for maintaining essential, life-saving services to prevent, detect and treat malaria during the COVID-19 pandemic," WHO Director-General Tedros Adhanom Ghebreyesus said in a statement.

More people are sleeping for six or fewer hours per night now than a decade ago

As the weekend approaches, here may be one more reason to stop doomscrolling and catch up on sleep: New data from the CDC find that more people are now reporting getting six or fewer hours of sleep per night. Previous research has linked fewer than six hours of nightly sleep to increased risk of an early death. In the new analysis, researchers compared responses from employed U.S. adults who completed national health surveys in 2008-2009 and 2017-2018: Around 33% of people in the later survey reported getting six or fewer hours of sleep per night, compared to around 29% of people who said the same a decade prior. Those employed in the private sector were most likely to say they got fewer than six hours of sleep, while those who were self-employed were least likely to say so.

What to read around the web today

A scientific experiment in going back to work amid coronavirus pandemic. The Boston Globe
Scientists criticise UK government's 'following the science' claim. The Guardian
Born into a pandemic: Virus complicates births for moms and babies. Kaiser Health News
A gulf emerges in health tech: While some companies surge, others lay off their workers. STAT Plus
Coronavirus shakes the conceit of ‘American exceptionalism’. Associated Press