Policy and Punditry Need to Adapt to New Virus Data
When we began our foray into quarantine seven weeks ago, there was a unifying and eminently sensible rationale behind it: “Bend the curve.” The idea was this: If allowed to go unchecked, COVID-19 would overwhelm hospitals, leaving patients without beds. Short on ventilators, patients would be left to suffocate. In short, by slowing the spread of the virus we would prolong the amount of time it spread through the country, but would reduce the total number of deaths. Moreover, we would buy time for the nation’s testing apparatus to ramp up, to produce more ventilators, and to expand hospital capacity.
This concept went viral. Vox produced one of the most memorable images of the epidemic with the chart below showing cases (without protective measures) reaching a sharp peak, blowing through the capacity of the health care system, while imposing protective measures results in a longer, flatter curve that fails to overwhelm the health care system. The image spread rapidly on social media -- former President Barack Obama even retweeted it – and “Flatten the curve” became the number one hashtag on Twitter for quite some time. (Click to enlarge the image.)
Notice what the hashtag storyline was at that time: We need to give the country time to get its testing program going, to expand hospital capacity, and to get enough ventilators in place. The virus will spread longer, but in the end fewer people will die. Pop culture “explainers” spread like wildfire on sites like Medium, with articles such as “The Hammer and the Dance” receiving over 40 million views and being translated into dozens of languages. The thesis of that article was explicit: Bring the hammer down on coronavirus – which, under the assumptions of the article, would kill 2% of people infected – and then begin to dance, tamping down outbreaks as they pop up here and there. “Exponential growth” entered the lexicon of people who had never taken a statistics course, as charts of coronavirus cases accelerating (and often never decelerating) made their way around the web.
If this really was the goal, then “job well done,” as they say, or at least largely so. Consider the following chart. It shows the daily number of cases for each state, indexed to the largest number of daily cases in that state. In other words, the red square for New York doesn’t mean the same number of cases as the red square in Wyoming. Instead, it means that was the date that New York had the most cases for New York in the time series, while the square for Wyoming means that was the date that Wyoming had the most cases it ever had in its time series.
There are many interesting stories within these data, but the main takeaway should be relatively clear: No states are on anything resembling an exponential growth trajectory, almost all states are past a peak, and most states are substantially so. This would suggest that in many states, the question really should be how to reopen while keeping hospitals from being overwhelmed again.
This is especially true given that the situation on the ground has changed dramatically since early March. Most states have substantially expanded hospital capacity, both by securing emergency locations to be used in case of overflow and by suspending elective surgeries, to the point where many hospitals are facing financial crises. Moreover, the arrival of the first COVID-19 therapeutic, Remdesivir, will help, since hospital stays are shortened when the drug is used. Personal protective equipment and ventilator availability has expanded, we’ve developed techniques for sanitizing PPEs, and ventilators may not be as useful as once thought. As of this writing, we’re testing over 200,000 people a day, which eclipses the rate South Korea achieved when containing its viral outbreak.
Perhaps most promisingly, the death rate looks lower than initially expected. It isn’t clear how much lower – studies disagree – but most of the serological studies find an implied fatality rate lower than the 1% used to arrive at the conclusion that 2 million people would die if the virus were allowed to run its course. Likewise – this is much less commented upon although it might be more important – the hospitalization rate looks substantially lower than initially anticipated. That isn’t to say we should just let the virus run wild – even if the base fatality rate (which could vary due to quality of care, multiple exposures, etc.) is a quarter of a percent, that’s still a half million people dead – but it does suggest that the risks are different than we initially expected, and which have in many ways been mitigated from where we stood last March.
But in the meantime, there seems to have been a subtle shift in the discourse. Some of this has been a refusal to update prior assumptions – some people seem to believe not much has changed since early March – but other analysts have subtly moved from “bend the curve” to what we might call “crush the curve.” Under the latter approach, rather than looking to keep hospitals from becoming overwhelmed, which raises the fatality rate, we should look to avoid all fatalities. If you look at Obama’s timeline, you can see this play out in real time: On March 4 he urged people to wash their hands and stay home when sick. On March 12, he defended canceling large gatherings. Three days later he suggested everyone stay home to the fullest extent possible. This was all consistent with the growing conventional wisdom that we needed to close down in order to flatten the curve as the viral spread became more acute in the United States; on March 18 he seemed to endorse the “hammer and dance” strategy, warning of potential multiple rounds of social distancing.
But by April 8, he acknowledged that the curve had been bent, but also seemed to suggest that relaxing current policies demanded additional measures to ensure there wouldn’t be another breakout. A few weeks later, he retweeted an article that suggested we would need as many as 35 million tests per day before people could return to work.
We see this shift expressed in public policy away from reopening as well. In California, for example, the governor closed the beaches everywhere last weekend as warm weather brought thousands of people flocking to the ocean. . This is despite the fact that the evidence that hot, sandy beaches are good places to transmit the virus is not that great and, more importantly, that California appears to have peaked quite some time ago. This isn’t to argue that everything in California is going swimmingly, but it is striking that restrictions in the state are moving in the opposite direction of facts on the ground there, especially given the standard of a month ago. It might make more sense to retain restrictions in a place like New York City, but Mayor de Blasio’s move to arrest people gathering in large groups seems discordant when cases in New York are receding. Note too the shift in rhetoric: “This is about stopping this disease and saving lives. Period.”
But the shift has probably been the most pronounced among pundits. Perhaps the strongest statement of the “crush the curve” point of view comes from an article published in The Atlantic, with the (frankly unhinged) headline “Georgia’s Experiment in Human Sacrifice,” with the subtitle “The state is about to find out how many people need to lose their lives to shore up the economy.” Infections in Georgia appear to be trending downward, and it is beginning to reopen its economy, including gyms and hair salons. The upshot of the title and the article (which avoids the hyperbolic language from the headline) is that people will die as a result of the decision to open early.
It seems likely that this is the case, but the idea behind bending the curve wasn’t that we would bend the curve until there were no more cases. Indeed, it was expressed that we might end up with a similar number of cases, but that by spreading them out we would lower the number of fatalities. This, then, is something different: The idea that we should use the shutdowns to eradicate the virus as best we can, and that weighing lives against the economy reflects a choice tantamount to sacrificing some portion of the population.
Reasonable minds can disagree over whether Georgia has it right (although Colorado, which is reopening amid expanding caseloads, has avoided similar criticism), but this absolutist stance is nonsense. We engage in cost-benefit tradeoffs all the time. As I’ve noted before, we engage in a similar cost-benefit exercise every year with the seasonal flu. The stakes appear higher here, which weighs heavily on the benefit side of shutting down compared to seasonal flu. The flu also has a vaccine that allows people to protect themselves against some strains, but as parents quickly learn, the efficacy of that vaccine varies wildly from year to year. But, regardless, we ought not pretend that we don’t weigh a substantial number of lives against the economy every year, and occasionally make decisions that will undoubtedly “sacrifice” people.
This seems to reflect a wider phenomenon of people being driven into “teams” regarding the shutdown. We’ve become polarized on the issue, and indeed this polarization is beginning to reflect our underlying politics. This is an unwelcome development. One of the dynamics about team sports is an inability to see the other side’s point of view; indeed, that is in many ways the point of teams. As this virus develops, flexibility will be crucial in determining how well we come out of it, and a willingness to listen to the viewpoints of those we don’t generally agree with is probably the most important trait we can have. But, as with so many other things, that seems to be one more fatality resulting from this virus.