B.in April, President Trump chose a computer model of coronavirus distribution as his oracle of choice. Not surprisingly, this simulation initially had higher estimates than other algorithms, projecting far fewer Covid-19 deaths – and its unconventional calculations and fluctuating estimates drew sharp criticism from epidemiologists.
But the statisticians behind it have changed their methods, and their new figures, released on Friday, support what scientists have long said: Abolishing social distancing measures could lead to huge deaths and the widespread use of masks in society. it can save tens of thousands of lives.
“We think the key point here is a huge winter tide,”
In particular, the paper estimates that there could be about half a million Covid-19-related deaths in the United States by the end of February, and that about 130,000 of these tragedies could be prevented with the universal use of masks. But experts warn that the figures from any of the model’s hypothetical scenarios are less useful than comparing the various options. By placing these projections side by side, you can begin to see what effect something like wearing a mask can have at the population level, if you take the authors’ estimate that facial coatings can reduce an individual’s risk of infection by about 40%.
“The exact numbers are unpredictable,” said Ruth Ezioni, a biostatician at the Fred Hutchinson Cancer Research Center and the University of Washington, who is not involved in the new study. “What should drive politics here is the difference between scenarios with and without masks.”
The initial work by the IHME team last spring did not attempt to model the transmission rate and incubation periods of the new coronavirus, but rather attempted to adapt the United States to data from outbreaks elsewhere and show a bell-like curve, with cases up to largely increase and decrease symmetrically. Their predictions also faltered, and since Nicholas Reich, University of Massachusetts, Amherst, a biostatist whose team compares a number of different Covid-19 models, described them to STAT, “are inaccurate and do not reflect the consensus of the modeling community.” “
The IHME then moved to a more traditional way of modeling infectious diseases, which aims to calculate the mathematical cascade of transmission: the number of people susceptible to the disease, how many are exposed, how many are then infected and how many recover and therefore have at least temporary immunity.
After the institute made the change, Reich explained, “their more recent short-term forecasts, which were sent to the COVID-19 Forecast Center, performed quite well with an accuracy of one month in the future. They are not the best model, but they seem to make reasonably accurate short-term forecasts. “
However, this document exceeds this one-month limit, and the further you go into the future, the greater the uncertainty of the forecasts. The team uses past status data for the frequency of cases and any other variables – such as cell phone mobility data, seasonal fluctuations in pneumonia, mask usage levels – to design how changing these variables can change the results.
This type of analysis is not intended to tell us how effective wearing a mask is in reducing the spread of the disease. Rather, it is just one of the many bits of information that researchers have included in their model – and their assessment in this regard comes from a meta-analysis they have made from previous peer-reviewed studies and preprints that specifically address this issue.
Nor should we expect this model to provide any certainty about the future.
“This is not a prediction in itself, as the results depend on the model’s very specific assumptions about how effective the masks are and how big the absorption of wearing masks is,” Reich said.
Instead, as Zeynep Tufekci put it in the Atlantic, we must use this kind of research to “prune catastrophic branches of the tree of opportunity that lies ahead.”
Since all kinds of changes in policy and behavior take place at once, it is almost impossible to separate the influence of any part. What this kind of analysis can provide is a hint at the general direction we need to go – and the ones we need to avoid – to keep as many people safe as possible. Given that a scenario in which everyone wears a mask in public all the time results in far fewer deaths than one in which the use of masks continues at its current rate, an effective course of action seems quite clear. , say the authors.
As surprising as their conclusion may be, it can still be useful to say.
“We don’t need a model to tell us that we should all wear masks, we don’t need a model to tell us that if we continue on our way, we will see tens of thousands more deaths in the next few months,” Ezioni said. . “But sometimes when a person provides a model and you see these curves and you see these numbers, it helps to scare you in an appropriate way.”
She added that this could also help strengthen public health decisions: “A model like this can be very important for our governors, who are actually trying to do something to enforce the mandates. They are not popular, so politically it can make it difficult for a governor who is trying to do the right thing. Models like this can provide a kind of evidence to support this policy. “