2.2 Million American Deaths from Covid-19?

The strange career of a fanciful projection

You have probably seen the headlines. The Intercept, for example, blared: 2.2 MILLION PEOPLE IN THE U.S. COULD DIE IF CORONAVIRUS GOES UNCHECKED. A Nicholas Kristof column was headlined The Best-Case Outcome for the Coronavirus, and the Worst Will we endure 2.2 million deaths? Or will we manage to turn things around? Kristof reported, "Dr. Neil M. Ferguson, a British epidemiologist who is regarded as one of the best disease modelers in the world, produced a sophisticated model with a worst case of 2.2 million deaths in the United States." News reports suggest that President Trump's Covid-19 advisers told him about the 2.2 million death worst-case scenario, and that helped spur him to extend federal social-distancing policies through the end of April. He has also suggested, falsely, that if we fall far short of 2.2 million deaths, it would mean that his policies have been successful.

Given all the attention the 2.2 million "worst-case scenario" figure has received, it's worth exploring where it came from. On March 16, a group of public health specialists in the UK published on March 16th what has become known as the Imperial College study. I'll leave it to the professionals to debate whether their math regarding virus spread and so forth is right, what I want to focus on here is something that requires no expertise to discuss, which is the underlying assumption that drove the 2.2 million figure. Here is the relevant paragraph from the study, with the key assumption in bold:

In the (unlikely) absence of any control measures or spontaneous changes in individual behaviour, we would expect a peak in mortality (daily deaths) to occur after approximately 3 months (Figure 1A). In such scenarios, given an estimated R0 of 2.4, we predict 81% of the GB and US populations would be infected over the course of the epidemic. Epidemic timings are approximate given the limitations of surveillance data in both countries: The epidemic is predicted to be broader in the US than in GB and to peak slightly later. This is due to the larger geographic scale of the US, resulting in more distinct localised epidemics across states (Figure 1B) than seen across GB. The higher peak in mortality in GB 16 March 2020 Imperial College COVID-19 Response Team is due to the smaller size of the country and its older population compared with the US. In total, in an unmitigated epidemic, we would predict approximately 510,000 deaths in GB and 2.2 million in the US, not accounting for the potential negative effects of health systems being overwhelmed on mortality.

You got that? The 2.2 million figure was a projection based on a scenario where not only do the government and private companies not engage in any "control measures," but individuals don't on their own change their behavior to avoid contracting or spreading the virus. The study refers to this possibility as "unlikely," but let's be blunt: it's entirely fanciful. The notion that no one is going to do anything different even as the death toll from the virus mounts into the five and then six figures is not "unlikely," it's entirely contrary to common sense and common human experience, not to mention data about how people said in early February they would react if the virus spread. (I, for one, started carrying around and using hand sanitizer and trying to avoid crowds in late February.)

Some will argue that the statistic was worth putting out there anyway, just to give us an idea of what toll a completely uncontrolled virus spread would look like. Perhaps. Unfortunately, the media generally failed to make clear that this was not a real-world projection, and were abetted in that malfeasance by the lead author of the study, Neil Ferguson. For example, Dr. Ferguson told the New York Times on March 16th that the potential health impacts were comparable to the devastating 1918 influenza outbreak. That outbreak killed approximately .6% of the U.S. population, which today would amount to around two million people, or very close to the fanciful 2.2 million projection. Nor does Ferguson seem to have made any effort to correct Kristof et al. when they wrongly claimed that 2.2 million was a realistic worst-case scenario.

And the media continues to misreport what the study said. For example, here is Wired yesterday: "The report, which also predicted 2.2 million American deaths if the government [what about private parties?] did nothing…" (One of the few journalists to get it right was Jacob Sullum of our own Reason.com: "Although those horrifying numbers got a lot of attention, they were never plausible, as the paper itself said, because they were based on the clearly unrealistic premise that 'nothing' is done to contain, suppress, or mitigate the epidemic.")

When I pointed out on social media that the 2.2 million figure was fanciful, some accused me of being in league with virus deniers and/or Trumpism. Now that Trump has embraced the figure, perhaps we can lay that one to rest. Others have argued that to the extent the figure spread and scared people, that was a good thing, because it spurred governments and individuals into necessary action. Perhaps in the short-term that's true. In the long-term, providing false or exaggerated information to the public that supposedly reflects the judgment of "experts" will erode confidence in both those who are reporting those judgments and in the experts themselves, a prospect which may have devastating public health consequences in the future.

Moreover, while the spread of exaggerated predictions may compensate to some extent for "virus deniers" and other forms of underprediction, there are also some immediate costs. If the government overreacts, there is the toll on the economy from unnecessary precaution. Beyond that, overprediction feeds anxiety disorders, and also leads to asymptomatic people or people with minor symptoms demanding testing, going to the emergency room, etc., which not only helps overwhelm the medical system, but may itself increase the spread of the virus when these people leave their homes to seek medical attention. At the very least, we should recognize that exaggerated projections reported without caveat have significant potential costs.