COVID-19 might just be four times less deadly than once projected, especially by the now discredited Imperial College Loneon models, a new estimate says.
“A review of antibody surveillance studies — which paint a much clearer picture of how many people have really been infected — suggests the coronavirus has a mortality rate of 0.25 per cent, meaning it kills one in every 400 people who get it,” The Daily Mail reports.
Most coronavirus modelling, including the grim Imperial College London projection that warned 500,000 Brits could die without action and convinced ministers to impose a lockdown, are based on a death rate of around 1 percent. For comparison, seasonal flu is estimated to kill around 0.1 per cent of patients.
The new estimate was based on figures from 23 different testing surveys carried out worldwide, which suggested the actual mortality rate ranged from as low as 0.02 to as high as 0.78 percent.
The Imperial College London model from March showed that as many as 2.2 million Americans could die from COVID-19. But the model was off — way off. And now experts say it was “totally unreliable.”
One computer data modeling expert said the Imperial model coding, done by professor Neil Ferguson, is a “buggy mess that looks more like a bowl of angel hair pasta than a finely tuned piece of programming,” The Daily Telegraph reported.
“In our commercial reality, we would fire anyone for developing code like this and any business that relied on it to produce software for sale would likely go bust,” David Richards, co-founder of British data technology company WANdisco, told the Telegraph.
The model has been a key part of recommendations from the White House Coronavirus Task Force and the Centers for Disease Control (CDC). Ferguson was also a scientific adviser to the British government, and he warned in mid-March that 500,000 people could die from the pandemic. U.K. Prime Minister Boris Johnson responded to the report by imposing a national lockdown.
Scientists from the University of Edinburgh say that the findings in Ferguson’s model were impossible to reproduce using the same data. The team got different results when they used different machines, and even different results from the same machines.
“There appears to be a bug in either the creation or re-use of the network file. If we attempt two completely identical runs, only varying in that the second should use the network file produced by the first, the results are quite different,” the Edinburgh researchers wrote.