The (absolutely huge) cost of delay

This post was put up on my other blog on Friday, March 20.

As of now, on the Worldometers site, there are 14,400 coronavirus cases in the US, and there have been 218 deaths. Yesterday, there were 10,000 cases, but only about 120 deaths. I’d say the cases figure is on track to reach 30,000 by Sunday, as I suggested yesterday – although it may actually be pointing to a higher figure. My calculus skills have rusted since college, but if someone wants to look at reported cases for the last two weeks (based on Worldometer, since the CDC site is worthless) and calculate the curves of first and second derivatives, I’d be very interested in seeing that.

Of course, the 83% growth rate in deaths between today and yesterday won’t continue (in which case the entire US population will be dead before long), but this is going to be more and more visible as we go on, especially when we start having over 1,000 deaths a day, which will I’m sure be sometime in April. As I said yesterday, hell on earth – except May will be worse (and maybe June. Hopefully not longer than that!).

It seems very clear to me from everything I’ve read that the only thing that will stop the virus cold (and even then, new infections will continue to grow for 14 days) is a total shutdown of all business and government, except for essentials like grocery stores and gas stations – as well as medical and emergency services, of course. California just imposed this yesterday, although New York City didn’t, as had been hinted two days earlier.

Of course, this isn’t going to happen on a national level today, although I certainly hope it happens next week and I’m just about certain it will happen the following week. What is the cost in total cases and deaths (throughout the pandemic) of each week of delay in doing this? As a very rough estimate, and assuming there is no acceleration in the rate of growth (which of course is probably unlikely), I’m going to assume the current seven-day growth rate continues into the future.

As of last Friday, there were 1500 cases; today there are roughly 14,500 – that’s a growth rate of 866% per week! I’m going to assume the order to lock down comes next Friday and it’s totally successful, meaning that two weeks after that, there are no new cases reported (as is the case in China now). This means that, if we look at the total cases and deaths (based on the level of cases on that date – not on deaths reported up to that date) say two months from today – and we compare the two scenarios, we’ll have an idea of how much it cost to delay the decision by a week (although that will still be well below the total cost across the pandemic, since we can’t calculate that until the pandemic is over and we know all the numbers).

If the order had been given today, the total cases four weeks from now would be the same as it will be two weeks from now (since by assumption new cases fall to zero two weeks after the order), which is 1.1 million. If it’s given next Friday (amazing if it happened, although probably not realistic at all), the total cases four weeks from now are the same as the number three weeks from now, which is 9.4 million. So the number of cases saved by making the order today are 8.3 million.

Now let’s look at deaths, and use the two percent mortality figure I’ve been using (which is probably pretty optimistic). In the case where the decision is made today, there will be 22,000 deaths over the course of the pandemic. If the decision is made a week from today, there will be 188,000 deaths, so the difference is 166,000. In other words, a one week delay in the decision – which is just about guaranteed – will cost about twice as many lives as all US combat losses since World War II.

Let’s look at the cost of delaying two weeks, meaning the order is given two weeks from today. The total cases four weeks from now goes up to 79.3 million, so the difference from giving the order today is 78.2 million cases. What about deaths? If the order goes out two weeks from now (and – again – is 100% effective), deaths will be a mere 1.6 million, and the difference between that number and the number if the order is given today is statistically insignificant – i.e. waiting two weeks to make the decision will cost 1.6 million lives.

Does that seem like a big number? In a must-read op-ed by Nicholas Kirstof in the NY Times this morning, Dr. Neil M. Ferguson, one of the leading epidemiologists in the world, says he expects total deaths in the US to be between 1.2 and 2.2 million. So I’d say my 1.6 million cost for the delay is actually very realistic (I’m sure the difference between Dr. Ferguson’s two figures has to do with what measures are and aren’t taken to contain the virus).

Let's say total pandemic deaths are 1.2 million, Dr. Ferguson's low estimate. What does that compare to? According to Wikipedia, total US military deaths (from all causes) during all wars since 1775, are 1.35 million. Total deaths in the 1918 flu pandemic in the US were 675,000. 

Of course, as the Wall Street Journal said in an editorial today, there will be a big economic cost to shutting down the economy, and they actually recommended we loosen some social distancing that has been put in place. They point out that ultimately the health of US citizens will be hurt if the economy literally collapses (and Treasury Secretary Mnuchin talked about possible 20% unemployment yesterday – that’s very bad, I’ll agree).

And I must say they have a point. Maybe we should just all get infected as soon as possible, and let the pandemic run its course. You can be sure the economic pie will ultimately come back to what it was a couple weeks ago, and keep growing nicely after that. Best of all? There will be 2.2 million fewer people demanding a share of that pie. More for the rest of us – unless you or I happen to be one of the 2.2 million, of course. In case you don't know me, I tend to get very sarcastic when I'm mad, and the idea that the WSJ would even suggest this is maddening. They clearly haven't thought the numbers through.

P.S. The same math works in reverse. If this order had gone out two weeks ago (March 6) and been 100% effective, meaning new cases had fallen to zero today, the total cases for the pandemic would be 14,400, and total US deaths from the epidemic would be 288.

You might say that making the order two weeks ago would have required an extraordinary bit of prescience on the part of our leaders. Well guess what, this is pretty much what happened in South Korea, and China outside of Hubei province. Their leaders didn’t have a crystal ball, but they could see what worked in Wuhan, and also the speed with which infections had grown before that city (and province) was locked down. South Korea’s total infections and deaths so far? 8,652 infections, and their new infection rate is close to zero. Their total deaths so far are 94. And how about China’s? 80,967 cases and 3,248 deaths as of today – and those are unlikely to grow much at all, since the only new cases are from outsiders coming into China, all of whom are quarantined for 14 days.

P.P.S. Someone just forwarded me a new article by Thomas Pueyo, whose article last week sparked these posts. It looks very good, and he has a good summary up front: https://medium.com/@tomaspueyo/coronavirus-the-hammer-and-the-dance-be9337092b56

Highly recommended!

P.P.P.S. (4 PM ET) The governor of Illinois is set to announce a shelter-in-place order today, and the NY governor did that earlier today. So we've started to do a total lockdown already. But what's needed is a nationwide lockdown for at least 14 days, with shutting down travel except for emergency needs. If it isn't a total lockdown, the virus will reappear as soon as the local lockdowns are lifted.

P.P.P.S. I just checked the Worldometer site. The total cases in the US now (4PM) are 18,100! In other words, they've gone up 3,600 in about 4 or 5 hours. This, folks, is what exponential growth is. Remember, the hourly growth number itself is growing. We had an increase of 4,500 cases between yesterday around noon and today around noon, and in another hour or so we'll already have logged that for the day, meaning the number tomorrow at noon will very possibly be 25,000 - or 10,000 more than today. Since the growth since yesterday was 5,000, this means the number of new cases is now doubling every day, but in a day or two it will be more than doubling, then tripling...every single day. Does the multimillion deaths figure really seem that unrealistic?