April will be hell on Earth. May will be worse

I won’t publish a cybersecurity/NERC CIP post today. I want to make sure this post gets picked up by the email feed tonight. The stakes are currently MUCH higher in the public health arena.

On Monday, there were 3,100 confirmed cases of coronavirus infection in the US. In the post I wrote that day, I pointed out that this number had been doubling every three days during the previous six-day period, and I would be watching to see if that trend continued – since it undoubtedly meant a huge number of total (i.e. including ones not currently counted because there aren’t enough tests available) cases three weeks from Monday, which I estimated at the time to be 1.9 million.

Well, that trend didn’t continue – it got worse. Johns Hopkins released a figure of 9,300 confirmed cases as of 10 PM ET last night – meaning, of course, that the confirmed cases tripled in three days. The Worldometer site lists 9,477 as of 8 AM ET, so that is comparable. And what does the CDC site say? It says the total as of noon yesterday was 4,484! However, that’s just the sum of the confirmations that CDC itself has done, and of all “public health labs”.

Up until a few days ago, those numbers would have been the same, because CDC was only allowing its own lab in Atlanta to do tests! So the fact that there are already more confirmations from outside the CDC’s labs means that – finally! - more testing is getting done outside of CDC than in it. That’s good news, but it goes in the “Thank God for small favors” department. The US as a whole has conducted fewer than 40,000 tests in total, whereas South Korea – with 15% of the US population – is now doing 20,000 tests a day.

And there are still people in the US government who try to say that “everyone who needs one” is getting a test! I assume they’re saying that just to show they have a sense of humor, but the time for humor about this debacle is over. For a great story on the totally bungled “rollout” (“trickle-out” would be a better term) of Covid-19 tests in the US, see the article in today’s WSJ (and if you can’t see it because it’s behind a paywall, you should take a trial subscription and seriously consider subscribing when the trial is up. The Journal has great reporting, although I take issue with a number – but certainly not all – of their editorials. But I can’t understand why they’re not making their coronavirus stories available for free, as the NY Times is).

So think about it. There are now more tests being done outside of the CDC and “public health agencies” than inside it (and the difference is certain to grow rapidly), yet the CDC probably isn’t going to change their reporting. So they’re doing a great job of faithfully reporting a totally useless number. Who gives a s___ how many cases have been identified just by the CDC, since it’s not anymore the total identified so far in all of the US? 

Of course, both the Johns Hopkins and CDC numbers are huge underestimates of the actual number of cases. In fact, the British report – which also said there could be 2.2 million deaths in the US unless drastic social distancing is put in place – that was used by the White House on Monday to justify the 10-person limit on gatherings said that the true number of cases is between five and ten times the identified numbers. Meaning that the true number of cases at this hour (and going forward you’ll pretty much have to just rely on hourly estimates if you can get them. Later you’ll want minute-by-minute estimates) is between 47,500 and 95,000.

Let’s do some math now (I’m dreading this, and you should too): Let’s go back to say confirmed cases are doubling every two days, although they’ve clearly tripled in the last three days. The reason I’m doing that is that a lot of the increase in confirmations is due simply to finding cases from people that were never tested before because of the super-restrictive testing requirements. And, as testing finally ramps up (although I’m sure 20,000 tests a day like South Korea’s number is still distant for the US), the delta between actual and confirmed cases will narrow, meaning the actual rate may “only” (!) be doubling.

If there are 47,500 actual cases now (5 times reported – the low British estimate), there will be 190,000 cases on March 25; 760,000 on March 31; and 3,040,000 on April 6 (i.e. 1 million more than my estimate on Monday). And let’s say that growth in new cases stops cold on that day, because next week the Federal government locks down the whole country and halts all but essential travel by ground and air (this is absolutely essential, and I’m actually close to positive it will happen, or at least by the following week. Of course, it’s almost certain it will be widely violated, at least at first, and it will likely be phased in – just because the Feds always seem to want to push the total US pandemic deaths a lot higher. They’re doing a wonderful job of that, too). There are now two questions to ask about this 3 million number.

The first is whether the hospital system will be overwhelmed or not. I certainly don’t know enough to answer that question, but I would think it’s possible. The second question is how many deaths will ultimately result from that level of infection. Last Saturday I repeated the mortality rate estimate in the great article that prompted me to start writing these posts (and that article has had 40 million hits in the last week): between .5 percent and 4.0 percent. The rate depends very heavily on whether and how much the hospital system is overwhelmed (as is happening now in northern Italy and as happened at first in Wuhan).

Given the huge gap between what hospitals need and what they have (and the lack of freely-available testing greatly increases that gap. For example, if a doctor or nurse has common cold symptoms but can’t be tested, they have to stay home. You really need to test every health care provider every day), I’d guess it’s a lead-pipe cinch that the mortality rate will be at least two percent. And since I’m hoping the Federal government can redeem itself by actually doing something about this problem (rather than telling governors, as president Trump did last week, that they need to find ventilators on their own. Earth to Trump: There are no ventilators at all anywhere in the world, and only governments that are commandeering private factories to produce them will have new supplies in say the next month –and they’ll all be reserved for domestic consumption. Meanwhile, the US hasn’t even started to do that yet. There are far fewer than 100,000 ventilators in the US, and a great number of those are already in use treating people who are sick from other diseases, like simple pneumonia), I’ll use 3 percent as my high estimate of mortality rate.

So if the total US population is locked down as of next week, there will be 3 million actual cases of Covid-19 as of April 6. At a 2% mortality rate, that translates into 60,000 total deaths, while at a 3% rate it’s 90,000. The former number is a little more than all US combat casualties in the Vietnam War. The latter number is about equal to total combat casualties since World War II.

Of course, the total deaths as of April 6 won’t be anywhere near even 60,000. They’re 150 today, and they’ll continue accelerating. Let’s say there are 1,000 total deaths as of April 6, and let’s use the 90,000 total deaths figure. And let’s further say that two thirds of those deaths occur in May.

So here’s where the title of this post comes in. In my scenario, there will be 30,000 deaths in April, and 60,000 in May. In other words, deaths from Covid-19 in the US will average 1,000 a day in April and 2,000 a day in May (or you could say 1,000 a day in April, May and June. No difference in the outcome). And if the British team’s high estimate of total cases now (ten times the reported number, rather than five in the low estimate) is correct, then all of these numbers should be doubled.

3/28: In the post I put up today, I found that the numbers in the above paragraph were wild underestimates. For example, on the last day of March, we'll probably have more than 3,000 deaths - i.e. more than on 9/11. That will continue - and of course grow exponentially - for a couple months. Can you say Apocalypse? 

And remember that I read an epidemiologist’s best estimate that total US deaths will be 400,000; and that the British team’s worst estimate is 2.2 million. Using these, we’re talking about 5-10,000 deaths a day.

Can you imagine what it will be like to live with the knowledge that a thousand people a day are dying in the US (when the total deaths for the entire pandemic in China will almost certainly be under 4,000), let alone 2, 3, 5 or 10 thousand a day? Remember, 3,000 people died on 9/11, and I know I brooded about them for months. The idea of having something like that hit you day in and day out for months is unfathomable to me.

April and May (and in all likelihood June) will of course be horrendous, even if you don’t personally know someone who dies of the virus (which will become increasingly unlikely as the deaths mount up). We’re all about to find out soon what it feels like.

Update at 2:30 CT: There's good news and bad news. First, Johns Hopkins has said the total infections is now at 10,000. So sticking with my old estimate of that number doubling every three days was wrong: They're more than tripling (again, this is partly because there's more testing available). So The 10,000 tested cases today will be 30,000 on Sunday, 90,000 next Tuesday, 180,000 next Friday, etc. 

But I had projected out estimates of total current cases (i.e. reported and unreported), based on the estimate of 5-10 times reported cases by the British team. Redoing that with the new figure of 10,000 reported cases today, I end up with 810,000 total as of March 31, which isn't too much higher than the 760,000 I said above. But here's a really scary demonstration of the power of exponential growth: I said above that there would be 3,040,000 cases on April 6. If reported cases are tripling every three days, that same number using the high British multiplier (10 vs. 5) is 7,290,000! Why the big difference? Simple: to get each six-day increment, assuming doubling every three days, I multiply by 4. But using the idea of tripling every three days (and it's actually grown faster than that in the last three days), the multiplier becomes 9. And for the following six-day period (i.e. April 12), that 7.2 million becomes - maybe you should turn off your computer at this point, if you have any sort of heart condition - 65.6 million! In other words, 20 percent of the US population. And keep in mind, April 12 is three and a half weeks from today.

There was one piece of good news in the Times: The "case mortality rate" in Wuhan - despite the serious problems they had with hospital capacity - was 1.4 percent, vs. 2 percent as the low end figure I used above in computing total deaths. I'll point out that, while a little belated, the Chinese government made an overwhelming response to that problem, including building some new hospitals in 10 days (which aren't needed now - maybe they could send them over here?). I don't think we'll see any ten-day hospitals here, even though we'll need a lot more than Wuhan did. So 2 percent is probably a much better estimate of the US mortality rate, although I suspect it will end up higher than that. 

Let's apply the 2% rate to the high end April 12 infection number of 65.6 million. That's 1.3 million people - still far below the 2.2 million worst case the British gave. But remember, 1.3 million deaths assumes we stop new infections cold as of April 12, which means we institute total social isolation (not just "distancing", whatever that means) as of March 30, or 11 days from now. So if that happens, we'll only have 1.2 million deaths! Cause for celebration, no?

What does this make my total deaths numbers? I estimated above that total deaths due just to the total level of infections on April 6 (i.e., not considering any new infections after that date) would be 60,000-90,000 (and these will be spread out over 2-3 months). But the total deaths due to the 7.29 million number I just estimated will be 102,000. So the lower mortality rate is overwhelmed by the much higher number of infections due to that number tripling every three days, not doubling.

Further, remember that we would have to implement near-total social distancing two weeks before either of those dates, for even those numbers to be valid for the entire pandemic in the US - i.e. we'd have to stop new infections in their tracks, which is pretty much what the Chinese did in Wuhan. Do you see what I meant yesterday, when I said we need to implement this within hours, not days? Of course, I'm sure it won't happen this week (I hope New York does it today as advertised. I wish Chicago would do it today also, although I'm already completely social distanced. I'm getting my groceries delivered and I'm not going outside except to go running by Lake Michigan). But it has to be done next week, and it has to include a grounding of non-essential domestic travel - by air, rail, car, mule cart, quantum teleportation, whatever.