On Tuesday, I wrote We Never Needed to Flatten the Curve. That post has gotten a lot of attention: it has been read over 175,000 times on this site, is all over Facebook, and was discussed by Rush Limbaugh on his radio show yesterday. The point of the post was that shutdowns, social distancing, etc., are intended to prolong the time during which the COVID-19 virus is active. The idea is to “flatten the curve” so that hospital resources are not overwhelmed by everyone getting sick at once. The assumption is that over time, a similar number of people get the virus, but fewer die because there are adequate medical facilities. You can see that in the classic chart (this is from the Centers for Disease Control and the Economist) that illustrates curve flattening:
In that post, I pointed out that hospitalizations due to COVID-19 are far below what modelers projected, while claiming that they had taken the states’ shutdown orders into account. In fact, COVID-19 hospitalizations have been far fewer than the normal variation in flu hospitalizations from year to year. Thus, the whole curve flattening exercise has been overkill, at best, which calls into question whether the shutdowns have any justification.
In this post I want to take up the utility of the shutdowns from another perspective: what will happen when they end?
I think many people probably assume that the COVID-19 virus will somehow die off while we are hunkered down, and we will emerge when it is safe to come out. But I don’t believe that is true. The virus isn’t going away. Whether the shutdowns end tomorrow or next year, the virus will still be there, and people will still catch it.
The spread of an epidemic depends in part on how many people the average person will infect during the time he has the disease. Epidemiologists represent this as R0. The idea behind social distancing in all its forms is to reduce R0 so that the virus spreads more slowly, thereby prolonging the epidemic and flattening its apex. The current shutdowns likely have been effective, to some degree, in doing that. If so, it stands to reason that R0 will increase when stores and businesses reopen and people are allowed to gather together again.
So: what do the models tell us about the likely impact of ending the shutdowns? Not much.
The University of Washington’s IHME model is the one most people refer to, even though, if you look carefully, you realize that at times it is irrational. IHME says that all of its predictions “assume[e] full social distancing through May 2020,” and suggests that the country can be reopened in June. According to IHME, at that point the epidemic is basically over. It predicts just 619 deaths between June 1 and August 4, when the model ends:
What is the evidence for the claim that it is safe to stop the shutdowns on May 31, but not at any earlier date? IHME has explained, sort of:
Our model assumes social distancing stays in place until the pandemic, in its current phase, reaches the point when deaths are less than 0.3 per million people. Based on our latest projections, we expect social distancing measures to be in place through the end of May.
In a country of 330 million people, 0.3 per million means around 100 deaths per day. And that is what IHME’s model projects for the end of May. But why is 100 deaths per day the magic number? Why not 50? Why not 200? Or 500? That would move the reopen date back to around the middle of May. Is there some scientific reason to regard 100 deaths as the trigger for reopening America? If so, what is it? IHME is silent, as far as I can tell.
I take it that we are meant to assume that ending the shutdowns on, say, May 1 could lead to a resurgence in the disease that is not otherwise predicted. But again, what calculations underlie this assumption, and what is the scientific basis for those calculations?
It turns out, further, that IHME is hedging its bets. For one thing, the rosy scenario painted for the Summer is by no means guaranteed:
Our forecasts of zero deaths in July and August assume that appropriate measures are put in place to guard against the reintroduction of COVID-19 from another state…
…or country. These measures may include mass screening, contact tracing, testing of all individuals entering the country, and quarantine of people who test positive. Details on what these strategies need to be will be analyzed in future editions of the forecasts.
IHME further hedges its predictions by referring to the current epidemic as the “first wave” of COVID-19, something you don’t see if you look at the charts and graphs on the IHME site. Thus, IHME’s description of its April 13 revision includes the phrase “first wave” no fewer than eight times. So, does staying shut down until May 31 avoid a second wave, or is some kind of resurgence inevitable?
IHME has referred to the May 31 date only in national terms. But death rates vary widely from state to state. Should a given state reverse its shutdown order (assuming it has one) when its pro rata share of 100 deaths per day is reached?
I don’t know whether there is a scientifically coherent reason for regarding 100 deaths per day as a magic number that allows our economy to restart. (I think that by the end of May, the shutdown itself will be causing far more than 100 deaths per day.) Nor do I know whether it is plausible to expect that restarting the economy, at any time, will not lead to faster spread of COVID-19 and, at least to some extent, a resurgence of the disease.
Given the fact that hospital shortages have not materialized, it seems that advocates for drastic limitations on our freedoms have shifted their rationale away from the original curve-flattening concept. Now they imply, at least, that considerably fewer Americans will eventually catch COVID-19 as a result of shutdown orders. Whether that is true, what the net health benefits of shutdown orders will ultimately be, and when it makes sense to terminate those orders, are questions that to my knowledge have not been answered in any coherent or transparent way by any modeler.