Researchers trying to model how many Americans can be hospitalized or killed by COVID-19 have been criticized by several quarters, with accusations that they unnecessarily intimidated people or gave leaders false assurances that it was OK to reopen businesses.
In Colorado, the latest forecasts that the state would run out of hospital beds as early as September seem likely, as new coronavirus infections subsided after a rapid increase in mid-July.
Other forecasts far underestimated the toll of the outbreak and predicted that about 300 would die from the virus in early August. As of Friday, 1,736 Coloradans had been killed by the virus and another 1
Experts point to at least three factors to explain why projections have not hit the mark: the inherent limitations of models, holes in our knowledge of the new virus, and people who change their behavior faster than expected.
Jimi Adams, an associate professor of health and behavioral sciences at the University of Colorado Denver, said there is often a misunderstanding of what models do. Some, like new two-week predictions from Google, are trying to predict what will happen. Others are trying to present what could happen under different possible scenarios, including if no policy changes, which is not likely during a pandemic, he said.
Think of the difference as Ebenezer Scrooge’s question to Ghost of Christmas Future: “Are these the shadows of the things that will be, or are they the shadows of the things that can only be?” Basically, Scrooge asks if he has any power to avoid a future where he dies alone and Tiny Tim dies young. Even if the ghost does not respond, Scrooge changes his behavior to try to avoid the worst care scenario, and the future to come is less bleak.
We saw something like this happen because states and cities required people to stay at home or wear masks, and when people responded to a resuscitation in cases by avoiding high-risk settings, said Beth Carlton, associate professor of environmental and occupational health at the Colorado School of Public Health. .
It is unusual for people to change their behavior as quickly and dramatically as they have in response to COVID-19, she said. It is good for public health, but makes it harder to project what can happen.
“People have the power to change the direction of these forecasts,” she said. “Part of power is individual and part is politics.”
Location data for mobile phones used to estimate social distancing and survey responses on masking show that people become strict about distance when the media reports cases and relaxes when the image begins to improve, which was not what the researchers originally expected, said Ali Mokdad, a professor at University of Washington Institute for Health Metrics and Evaluation. Although it is good that people take precautions when they feel in the risk zone, it is relaxing when falls fall right up to the next peak, he said.
“Unfortunately, we will see these ups and downs unless people adapt to the virus,” he said.
The IHME’s latest model was near deaths in Colorado in early August, projecting 1,869 deaths by Wednesday – 18 more than the total number of people who died from the virus. Going forward, it estimates between 2,149 and 5,006 people could die on December 1 if almost all of Colorado’s wear masks in public, and tolls can range from 2,744 to 18,244 if they do not.
Initially, models had to rely on data from China and Europe, which did not always apply to the US situation, Mokdad said. The picture was also limited during the first months of the virus’ spread in the United States because the lack of tests meant that no one really knew how far the virus had spread, he said.
“Our model evolved when we have more data,” he said.
Katie Colborn, research director for surgical results and applied research at CU’s Anschutz Medical Campus, had previously worked to predict malaria outbreaks in countries where the disease is common. With malaria, there is about a decade’s worth of high-quality weather and case information, so it is possible to predict that an unusually rainy season is likely to result in more cases, at least in the short term, she said. For COVID-19, the early warning signs are not nearly as clear.
Researchers are still trying to find out how the disease is transmitted, including how long it can stay in the air after an infected person speaks, Colborn said. Other questions, such as why it is widespread in nursing homes, but apparently not in health centers, also need answers, she said.
PolicyLab at the Children’s Hospital in Philadelphia is also refining its strategy as knowledge develops. In May, they predicted that there was little risk of another outbreak in Colorado Springs this summer, but that Denver could be in danger. Both cities have seen an upward trend in coronavirus infections, but growth is steeper in El Paso County, where residents appear to be practicing less social distance.
In recent forecasts, they have shifted to focus on the next to four weeks, said Brian Fisher, senior lecturer in pediatrics and epidemiology at the University of Pennsylvania. The further out you look, the more likely you are to be off, because behavior or policies may have changed, he said.
They have also learned that the effects of temperature and humidity are more complicated than originally thought, Fisher said. This spring, they thought hot days would work against the virus, but that has not happened, perhaps because people began to gather in air-conditioned spaces, he said.
The PolicyLab model uses cell phone data on visits to non-essential businesses, such as shops and restaurants, to estimate social distancing, Fisher said. It does not stand for other precautions that people may have taken, such as wearing masks or getting a roof instead of sitting in a restaurant, so it is possible that some counties will not see the expected recovery in cases when people go out more, he said.
“They may have reduced social distance, but they do it in an intelligent way,” he said.
Models will vary based on the quality of their data, as well as their manufacturers’ mathematical choices and assumptions about human behavior, says Aditya Khanna, director of network modeling at the University of Chicago. Although specific figures may not reflect what is actually happening, most have been correct in predicting that if people started living as they did before the pandemic, it would be a nail in the coffin, he said.
“There are some things that are very clear,” he said. “In the big picture, there is a lot we have already learned.”
Subscribe to the newsletter every week to have health news sent directly to your inbox.