“Flatten the curve” won’t work. We need to “Reverse the curve”
Many people, including major media and many if not most health care professionals, are sharing or endorsing the article or associated graphic of the now “viral” crowdsourced article “Flatten the Curve”,as a way for people to make sense of the actions we we need to be taking to deal with the Covid-19 pandemic. The graph feels right, the phrase is quite catchy and makes a good hashtag (have to have a good hashtag!), the associated article recommends strong measures which seem to match what we are doing now, and to be fair I believed it all myself until I reviewed the data that proves with little doubt why it does not represent the right approach, and indeed may create a complete catastrophe if it is allowed to guide our actions.
This approach in public health terms, is also known generically as “mitigation”, and it is not a novel idea. The basic premise is that you aim to slow the growth of new cases enough so that health care capacity can handle the demand for hospitalization and treatment, and therefore avoid being overloaded, with all the bad consequences that flow from that. This is what the now famous graphic represents visually.
What you will notice about the graph is that there are no numbers under the axes.
The numbers are the problem.
Lets put some numbers in.
The Covid-19 virus is said to be somewhat more contagious than the seasonal flu, under a scenario of uncontrolled spread (meaning no interventions,we just let it go). With every new case infecting roughly 2 more (~R=2 to R=2.5), and a doubling rate in number of cases at somewhere between 3–5 days. That’s exponential growth — where you get big numbers really fast.
The seasonal flu clocked in at about 50 million cases in the US over a short period of months this flu season, and is said to be less contagious.
So, with no interventions any reasonable estimate of total number of Covid-19 cases is at least 50 million. In fact probably quite a bit more, like 100 million. Some estimates have it considerably higher. But lets say 50 million.
The hospitalization rate for Covid-19 is roughly 20% of cases (source: China data). That’s 10 million new hospitalizations over say a short window of months. About 1 million will require intensive hospitalization, meaning ICU.
Lets say, per the graphic, you “flatten” the curve,substantially slow the growth of cases, and reduce the total cases over the next 3 months by 2/3 (pretty successful for a mitigation approach). That’s still 3 million new hospitalizations, and about 300 thousand intensive hospitalizations, and with each hospitalization requiring 2–6 weeks, many of those hospital stays will be overlapping.
How many ICU beds does the USA have currently?
About 45,000. And many of these are full — its not as if the ICU system has 100% vacancy waiting for the Covid-19 pandemic. There is a bad flu season as well.
Here is an interesting data set I found online:
Hospital beds per 1,000 people (from OECD):
Germany:8.0
South Korea: 12.3
Germany: 8.0
France: 6.0
China: 4.3
Italy: 3.2
United States: 2.8
Consider that Italy’s pretty state of the art health care system is completely overloaded right now, and severely rationing care.
Under even the more optimistic outcomes of flattening the curve, we still overload the health care system. Some estimates Ive seen that claim to be conservative, talk about 20 patients for every bed.
What is the outcome of the health system overloading? Its very important to understand this next point, if there are not enough ICU beds, or even regular beds, people will not be treated. Many of them will therefore die. The case fatality rate will jump from 1–3% with best available treatment, to perhaps 4%, 6%, or even 10%.
This represents an enormous amount of additional death just due to the overloading, never mind the fatality rate of Covid-19 by itself. Possibly hundreds of thousands additional deaths, as well as the fact many hospitals wont be able to do anything but treat Covid patients, so there will be increases in death from all the normal things that people die from if they don’t get treatment, like heart attacks. It is difficult to estimate this, but it will not be trivial. What frame of reference do Americans have to understand this level of additional death over a short period? 9/11? That was 3000 people. HIV/Aids, high numbers in the US, but that was over a larger period of time. Only wartime provides a reference point, and the last war we participated in to take lives at that level was WWII.
Flattening the curve isn’t enough.
So what is?
There is another, more proactive approach, with a different name — in the academic paper this article is based on, and presumably the public health world generally, its called “suppression”.
Id like to call it “reversing the curve”, a lame bid to coin my own vaguely catchy hashtag and free-ride off the popularity of the very phrase I am critiquing. I don’t expect it to catch on.
Suppression, or “reversing the curve”, means really one thing — getting the reproduction rate of the virus, known as R, to go below the threshold where new cases continue to rise (R<1). This isnt a slowing — per flatten the curve — this is a complete reversing. New cases reach a peak (fairly quickly, in mere weeks, not over an extended period), level out, then start to go down, and continue to go down, until there are almost no new cases. Taken far enough, it can bring new cases to zero (tho they can spike back up at any time, a separate issue not covered in this article).
China and several other Asian countries have successfully accomplished the goal of “suppression”, using a variety of overlaid interventions, many very strict and disruptive, as has been reported. So far no other counties have accomplished this same feat — they are trying but too early yet to tell — however, this is now the model that most other countries have settled on for its response to the virus.
The actions taken for “flatten the curve” (mitigation) and “reverse the curve” (suppression) are quite similar even if the goal is very different (slowing vs reversing).
Both recommend some degree of closures, social distancing, improvements to hygiene and so on.
The difference is that suppression is by definition more aggressive and disruptive, and must be dialed up until one achieves the the desired outcome. It is indeed very harsh, and looked very harsh to us when China instituted it from our comfortable position of not yet being at risk. We look at it very differently today.
Each Asian country that employed this approach created their own mix of actions from a grab bag of possible measures, making it very difficult to unwind the effect of each measure by looking at all of them together. Nevertheless, the linked paper by Imperial College in the UK, attempts to use sophisticated modeling techniques to take a shot at doing just this.
Their conclusion is that social distancing, work at home, closures of schools and universities, combined with testing, quarantine, and isolation of cases, have in tandem a good chance of reducing the basic reproduction rate of Covid-19 to under 1, “reversing the curve”, and averting the severe catastrophe of overloading the health system.
Every Asian country that employed this sort of approach used slightly different measures, and timed them a bit differently relative to their own outbreaks, yet all were still ultimately successful at reversing the curve. This flexibility suggests the suppression approach can be successful here despite the many differences between these counties and the US, both in tools available, and culture.
Is it worth all the severe disruptions to normal life and potentially economy killing effects of all these draconian measures if they even have no better than half a chance of preventing the the worst case outcome?
One way to answer this is to say economies can bounce back, but dead people cannot un-dead (leaving aside zombies) .
In the absence of clear indications of exactly what mix of measures will accomplish the desired goal of suppression,as cases start to rise decision makers will likely end up in a race to institute the most restrictive measures, decreeing more and more draconian restrictions, so as to not seem “soft” on the virus response relative to other jurisdictions and taking the political hit if things go south. This is a political consequences dynamic, and while not the only dynamic in play it may be the predominant one right now. There will likely be some overreaction, bound to be, but perhaps only in hindsight. It is not clear yet what constitutes “overreaction” without more data and time. The very best science, is still quite fuzzy on the exact measures to take,how much of each ingredient to throw in the pot — only the broad outlines. We cannot model any closer than that without much more data (and time).
In any case, I expect further drastic measures to be instituted due to this political dynamic — more severe lockdowns,more closings, more shelter in place orders, movement restrictions, and so on. How this will play out as secondary effects, no one really knows yet. It will without a doubt be very challenging on almost every level.
Nevertheless, it is reasonably clear there is SOME LEVEL OF INTERVENTION THAT WILL REVERSE THE CURVE, even if we only know the broad outlines of what to do. The virus is not so contagious that its rate of spread does not respond to large reductions in contacts between people, as long as enough people are influenced or affected by the interventions chosen (it wont be everyone — and doesn’t have to be — to succeed).
If we do reverse the curve and do not overload the health system (or at least only in a few early hotspots like New York and hopefully not for very long), we will have won the first phase of this, at least from the standpoint avoiding a very bad outcome. We will have dodged an extremely large bullet for the time being. This will be what winning looks like and it is hugely consequential. Because to keep cases at a manageable level for the health system, by definition this will pretty much be a level we can manage as a society, given how slowly health care capacity can ramp up under even the most extreme circumstances. It very likely does not get any better than this outcome, at least until a vaccine, effective treatment, or testing for everyone is available.
What else can we do besides social distance?
Most of the media stories right now are focused on the details of “social distancing”, a phrase that we will not stop hearing about for a very long time. I get that, people need to get the message. However, there are two other key spokes of the suppression approach that are just as important, because each aspect of suppression amplifies the effects of the others. Any one of them is not nearly as effective as all of them layered on top of each other. More critically, it may take all of them to achieve the overriding goal of reversing the curve. The problem is because of various fails mostly at the governmental level, we are not yet up to doing these other things at needed scale. It would be excellent if the media could focus as much on pushing the government to act quickly on these as they do on other things. Their importance cannot be overemphasized and there is little evidence we are moving as fast as we could.
Item 1 — Implement widespread testing ASAP
If we had enough tests we could test vastly more people and locate people who are not yet symptomatic so they could self-quarantine earlier and avoid the most contagious period (when symptoms first start to appear per current understanding). They could then warn close contacts earlier (the most likely infected) to quarantine or test. And we could give peace of mind to people with mild symptoms or those who have had recent contact with someone sick (this will eventually be a very large number of people) but are actually negative themselves. So many benefits to widespread testing. Yet we are not doing this yet and we are still weeks if not months(!) from being where we need to be.
New testing options are now coming online — the FDA is finally moving quickly to approve them, but it will take weeks more at best just to get them to hospitals. Home tests (samples collected at home picked up and brought to a lab) would be great too — there is a risk of exposure when people go out to get tested. Drive thru testing, stochastic testing (recommended by Larry Brilliant, the expert who helped eradicate smallpox), quick turnaround testing for hospitals (speed is everything), targeted mass testing based on AI driven predictions of where outbreaks are emerging, all of this and anything else that makes sense. We have the potential to do more and better testing than even the Asian countries given our full spectrum of capabilities and resources but we have to act so much faster to get the needed testing infrastructure up and running, and deploy it where it can do the most good.
If, say, Google and the other tech giants don’t have their best people on some aspect of this problem, in particular hotspot prediction, contact tracing, and targeted testing, as they employ an absolutely insane number of very talented people (Ive met quite a few) who can help — they should pay the price for this when it gets all sorted out. What bigger fish do they have to fry? Those who chipped in early will be remembered, as well as those who did not.
Item 2 — Get the quarantine monitoring house in order
We need to regularly check on infected people who are in self-quarantine (and their households) because people cheat less if you monitor them as well as many other good reasons. Cheating on quarantine has quite severe consequences as a person with an actual known infection is out exposing others willy nilly. A single quarantine cheater is almost certainly the same impact as a very large number of social distancing slackers, while I don’t have data to back this up it is pretty intuitive. Also, with regard to self-quarantine, many people are compromised in various ways and aren’t necessarily going to do the right thing even if they mean well. These people should probably be put into segregation or given more aggressive supervision. Monitoring is key.
Monitoring obviously takes more and more personnel as cases rise, and the needs will jump as rapidly as the virus. To my understanding we are only monitoring a tiny percentage of active cases at best, perhaps mainly as a result of lack of testing, but also because there simply isn’t sufficient staff and infrastructure to do so system wide at scale. Public health staffing for this sort of thing is low in normal times and absurdly low in these times. We need to hire the personnel ASAP and get the infrastructure working ASAP. This is both VITAL and doable, it just requires resources and action. I’ve barely seen this aspect mentioned anywhere.
We need to consider segregating infected people in special quarters — something China did by hospitalizing all cases, not just the ones that needed hospitalization. This takes infected people away from the people they are most likely to infect — their households — and also prevents cheating. It can potentially allow better care for people as well by streamlining the treatment process. It can also create concentration camps for the sick (think about the migrant detainment centers) that do more harm than good — we need to be careful how this is implemented if we go this route.
Finally, I would propose that one reason government officials are mostly touting “social distancing” as the move here is because on a certain level, it lets them off the hook — and puts the onus on the citizenry — by sending the message that that it up to the citizen to change their behavior, while allowing government to act more or less the same, only now with about 100x more control using their newly discovered emergency powers. In reality, its both us — and them, we are joined at this hip. Therefore all of us - citizens AND the authorities alike, will have to change how we think and act to most effectively address the situation we are in.
If they expect us to do our part to collectively “social distance” — they (the government) should also commit to doing their part — with appropriate levels of additional accountability if they don’t. Every classic gov-think box they are now thinking inside, and are used to thinking in — they must climb out of immediately to avoid the all too common organizational flaw of failure of imagination.
Every bit of improvisation and creativity that can be marshaled to address the crisis (and creativity is usually counter to the natural bureaucratic impulse) needs to be nourished to flourish on an immediate if not sooner basis. Either get it together or get pushed aside and let someone in who does have it together. Petty power struggles and “who will pay for it” arguments — save it for later.
I believe, perhaps optimistically, that we CAN beat this, based on the best modeling available and the real world experience of the Asian countries. Its not guaranteed, we have yet to see the effects of the interventions being tried in Europe for example, but its a solid chance. But we are going to have to continuously improve and tune our response and be absolutely ruthless in evaluating and re-evaluating both our biases and our actions. Lets be perfectly clear about what we are doing here — we are making things really hard for people now so they can be less hard later. We will create an absolute mess of our economy, but it will be easier to clean up without dealing with the psychic weight of vast numbers of additional dead, a raging pandemic, and a non functioning health system that won’t have the capacity to mend your broken leg after a bike accident.
To work, the “go big” response must happen before the coming spike up in cases that is clearly already baked in and unavoidable. Since the scope of the worst case scenario is essentially unbounded (because of compounding effects of things like health system overload) — we have no choice but go big right NOW on the response. Spare no expense that might expedite, spare no action that might help.
We shouldn’t aim for simply copying the Asian “suppression” measures — we should aim to do better however we can using that good ol’ american ingenuity, because in some ways we will do worse (we already have with regard to testing). In the meantime, lets retire #flattenthecurve and the flawed notion that the “mitigation” only approach will save us. If at all possible (and maybe its not), we must do more than flatten, we must reverse.
References:
This article (which I didnt see or know about prior to writing my own) comes to many of the same conclusions. The difference is it has millions of views. There are some problems with this article— in particular the mortality rate estimates are not well supported by the most current data, but it doesnt really effect the conclusions, which I by and large concur with.
https://medium.com/@tomaspueyo/coronavirus-the-hammer-and-the-dance-be9337092b56
Academic stuff, but lays out the detailed case for what I am largely paraphrasing with this article. It is also said (unverified) this is among the modeling that key decisions makers are looking at.
More evidence (if even needed) that widespread testing is the move. For those who believe its not “practical”, consider the alternative.
https://www.ft.com/content/0dba7ea8-6713-11ea-800d-da70cff6e4d3