—Ebola:  Is again short.  No new cases this week.  They never did find the definitive source of the last mini-outbreak, which is mildly concerning, but only mildly.  Fruit bat for all we know.  Of course, after “Plandemic”, we cannot rule out a subtle ripple of the fabric of space and time either.  Anyways, there were renewed demonstrations around Beni, the location of the mini-outbreak, which disrupted Ebola monitoring activities a bit.  The WHO remains worried about vaccine access given the focus on COVID.

—Coronavirus:  Lot to cover here too, and I will try to stick to high points.

—First, the remdesivir study which led to its rapid approval for use against SARS-CoV-2 was published.  The publication dropped late Friday before the Memorial Day weekend here in the US, so you may have missed it.  Again, because it is a competitor’s newly approved product, I will not be saying much.  However, this commentary is in the public domain for those interested in perusing it:
https://www.zerohedge.com/technology/remdesivir-study-finally-out-drug-only-helped-those-oxygen-finds-mortality-too-high

—Second, there was a large observational study of 96,000 or so patients of hydroxychloroquine or chloroquine with or without a macrolide antibiotic published in the Lancet.  This was not a randomized controlled study, which is the gold standard, but is a very large observational study, and they did some statistical measures to control for the most important variables like age, sex, comorbidities, and COVID-19 severity.  There was no clear evidence of benefit for any of the drugs.  There was evidence that mortality may be increased with the treatments.  Shortly thereafter, the WHO paused ongoing randomized studies with hydroxychloroquine.  Some of that was a planned pause, though, and is mostly to look for a “safety signal”—that bounce in mortality that this observational study in the Lancet found.

While not the gold standard, and again just my opinion, I would have a hard time getting excited to go forward with hydroxychloroquine.  We’ll see what happens with the paused studies.

—As far as I am aware, Donald Trump has NOT caught malaria in the past week though.

—Third, the CDC issued a couple of updates.  The CDC is estimating likely overall mortality of SARS-CoV-2 in the US to be much closer to the flu.  Not a surprise.  However, all cause mortality is being reported as up in most if not all COVID affected countries, including the US, during the period when the virus was running most rampant.  Without knowing more details, I think that is less likely to be an undercounting issue (as some articles would have it), and more a “bed’s taken” issue, especially in regions of those countries where the local healthcare system was nearly or definitely overrun.  Again, I cannot stress enough that it is the -local- hospitalization rate that is the biggest threat of SARS-CoV-2.  The next available ICU bed, -locally-, is the most important bed.

For example, let’s say the US has some arbitrary number of ICU beds.  68,000 adult ICU beds, give or take.  Let’s say the virus hammertimes New York City.  We need 66,000 ICU beds, right now.  “Great!”, you think, “We have 2,000 more beds than we need!”  Hold up, home slice.  By the time you are spreading 66,000 ICU patients across the 68,000 beds nationwide, you are starting to fill ICUs located in Honolulu and Anchorage.  I haven’t pulled up Google maps, but those sound far for an ambulance transporting an ICU patient from New York, no matter how fast the ambulance is going.  So the -nation- may have 68,000 ICU beds.  You, in NYC, -don’t- have the ability to transport, as quickly as you need to, 66,000 critically ill patients.  If you have 66,000 ICU requiring patients in New York, you realistically have ONLY the ICU beds available within the city, PLUS -maybe- a day’s ambulance or helicopter ride.  You’re going to have a hard time keeping the patient stable in an ambulance all day though—ambulances are not designed for that.  The entire nation may have 68,000 ICU beds total, but the total of what is realistically available to you is far less than the 66,000 beds you need right now.  That’s how New York/New Jersey ran into trouble in all of this.  That’s why Lombardy, Italy ran into problems.  Hell, that’s what happened in Wuhan to kick all of this off.  

—So in short, no one has a real good idea yet what the butcher’s bill for SARS-CoV-2 is going to be.  Total mortality for just SARS-CoV-2 will be flu+ as we have discussed before.  All cause, especially in places that did run out of the next marginal bed, is some part of or above that total too.  However, what I am seeing in the news is either “see?  It was just the flu!  All of these restrictions were unnecessary!”, which misses the next marginal bed problem, or “they’re undercounting deaths—it’s much worse than we know!!!”, which is almost certainly wrong too because the direct mortality is flu+.  In fact, if you tell me which article slant you have been seeing, I can pretty much guess where you get most of your news.  Just realize they are both playing an angle.

—In other CDC news this week was a slight change in the recommendations for being considered recovered and no longer infectious for SARS-CoV-2 after you catch it.  Long story short, if you had symptoms, stay away from others until you have gone 3 days without fever (and other symptoms improved) AND until it’s 10 days from when you first noticed symptoms.  If you are a known exposure, someone in close contact with a known positive, try to isolate for 14 days still.

—And lastly for our CDC round up, they issued a warning on antibody tests, claiming that a test which says you -have- antibodies to SARS-CoV-2 may be right only about 50% of the time.  “But how can that be?  After all, this crazy pathologist whose e-mails I read said some antibody tests now available are 99% sensitive and specific.  That seems pretty accurate!  Did the companies lie?”

No, they did not.  The CDC is making this warning largely because we still have -no- good idea how prevalent the virus is, since we know a lot of people are NOT symptomatic when they do get exposed (and never get tested in a window where PCR has a fighting chance to detect it), but we don’t know how many of them actually make enough antibody to be detected after the fact.  The -prevalence-, how common a disease is in the population, and the size of the population tested determine how many false positives you need to worry about.  This is the part where we will have to do a little math—sorry.  I will keep it simplified.

So let’s assume we have 100,000 people that we are going to test.  The CDC seems to be assuming that the PCR testing on asymptomatic patients (about 1.5% or so positive, at least in Indiana) is the prevalence.  So 1500 people actually have SARS-CoV-2.  If our antibody test is 99% sensitive, it will detect 99% of the people who actually have SARS-CoV-2 and, presumably, antibodies (1,485 of 1500).  If it is 99% specific, it will correctly tell 99% of the people who do NOT actually have SARS-CoV-2.  In our 100,000 people, that is 98,500 people who do NOT have SARS-CoV-2.  Our 99% specific test will tell 97,515 of them, correctly, that they do not have SARS-CoV-2 antibodies.

BUT, that means 985 people who do NOT actually have SARS-CoV-2 get told that they have antibodies to SARS-CoV-2.  Those 985 are the false positives in this example.  Obviously, if that becomes a condition for travel or work (as some of the more extreme ideas out there would have it), that’s a problem.

More importantly, when you get tested or your doctor tests you, it’s not known for sure if you have it or not.  So what we call the “positive predictive value” addresses this.  The positive predictive value is simply a calculation for how likely you are to actually have the antibodies IF you test positive (strictly speaking, it’s the ratio of true positives to total positives [true positives plus false positives]).  The positive predictive value is 60.1% in this example.  So, if you are one of these 100,000 people tested, and only 1.5% of them actually have the virus, and you test positive, the chance that you actually have the antibodies is only 60.1%.
Even though the test is 99% sensitive, and 99% specific.

BUT, we can change the assumptions and change the positive predictive value.   The positive predictive value depends A LOT on how prevalent the disease is.  If a lot more people have the disease, a positive result is MUCH more likely to have been in an actually positive person.  

So let’s run the numbers again.  We’ll test 100,000 people once more.  1.5% of them are PCR positive, but our rule of thumb has always been that some multiple of the people testing positive have the disease, but just don’t know it.  We’ll assume that 10 times as many people who are PCR positive actually have SARS-CoV-2, but don’t know it.  So of our 100,000 people, 16,500 of them actually have SARS-CoV-2, and have antibodies.  Our test is 99% sensitive, so we identify 16,335 of the people who actually have SARS-CoV-2 antibodies out of those 16,500.  And it’s 99% specific.  So of the 83,500 people without SARS-CoV-2 antibodies, our test screws up and calls only 835 of them false positive. 

Notice as more people actually have SARS-CoV-2, the number of negative patients goes down, and we get fewer false positives as a result.  In this scenario, the positive predictive value, or the chance that you actually have SARS-CoV-2 antibodies if you test positive is 95.1%.  
That’s a big swing from 60.1% chance of really having antibodies to 95.1% chance of really having antibodies with the same test based solely on how much SARS-CoV-2 there is floating around out there (prevalence).   This applies to more than just SARS-CoV-2 testing—for example, an inadequate test for a rare disease can a make a drug to treat that disease look less effective than it really is.  We may revisit that some time post-COVID, because it’s important, and not nearly as widely appreciated as it should be.

That said, NO ONE, not even the CDC, has a real good idea how much SARS-CoV-2 is floating around out there—again, because a lot of people get exposed to SARS-CoV-2, but never get sick and never get tested.  I think the CDC put this caution out -not- to criticize the tests that are available (99% sensitivity and specificity is REALLY good for a clinical test), but to brush some of the crazy back off the plate.  Specifically, relying on antibody tests as a “passport” or proof that you can be safely out in society as a privileged “immune” survivor.  That sounds like a good idea, but it’s not a real good idea.

—That said, we may get a better sense of how much SARS-CoV-2 is floating around in the US though, and how much herd immunity is really already established thanks to all of our participants in the Great American Memorial Day Weekend SARS-CoV-2 Prevalence Study.  You saw pictures of these courageous volunteers all weekend, crowding beaches and pools with no regard for social distancing or antiviral precautions whatsoever.  When I urge you to “be prudent”, to be clear, -that- is NOT prudent. 

But, going back to the Black Death example, this is expected behavior.  Especially when you have people cooped up in quarantine for so very long.  They’ll take nice weather, re-opening states, and falling new positive cases and take the YOLO option from our “choose your own adventure.”  Avoid the temptation to be judgy.  We are ALL already exposed, no matter how religiously you follow every guideline.  The 14th century was mostly rural (so high social isolation already), and even densely populated urban centers were not nearly as densely populated as we have now.  Travel was sooooo much slower.  And Yersinia pestis still marched inevitably to pandemic.  Once it’s out, it’s out, and remember, no matter your choice in our “Choose Your Own Adventure”, they all got the plague at about the same rate.  The argument for social distancing and precautions is I want to reduce the risk that I am one of the unlucky ones whose immune system goes a little too nuts. I think vitamin D [up to 4,000 IU per day], exercise, good diet, good sleep are as or more important than a mask for that though.  But I’m not a hermit either.  We’re all already exposed.   The other compelling argument for social distancing and mask use is that I want to actively reduce the chance I might expose someone who is in a high risk category.  Not that I see too many of those people, but I mask up and keep your distance, because the mortality stakes are higher for them–and you can’t always identify them by sight.  They gotta’ go to the grocery store at some point too.

Anyways, courtesy of our Memorial Day YOLOs Experiment, if we DON’T see a big uptick in new cases through the weekend and into next Wednesday, the US, at least, will be well on the way to coming through this wave of SARS-CoV-2.

—Again, your local results may vary, and “second waves” are more likely to be local/regional—city, county, state level.

—Speaking of second waves, got this paper, a model of potential “second wave” dynamics for SARS-CoV-2, from one of our new readers:  https://science.sciencemag.org/content/368/6493/860.full

Two things to remember in medical research: 
1) “Every model is wrong; some are useful.”2)  https://xkcd.com/1217/  (hover your mouse over the picture–there is a second caption)

So, let’s look at this model and see if it’s useful.  They have a good idea here—they look at how other, less severe coronaviruses have tended to spread, and how long people are immune to them, and ask “what if SARS-CoV-2 does the same?”   They lay out some of the major assumptions behind that idea in the very first sentence, which ends with “we still do not know enough about postrecovery immune protection and environmental and seasonal influences on transmission to predict transmission dynamics accurately.”  Neither I, nor they, have any idea how likely, how probable, those assumptions about SARS-CoV-2 acting exactly like certain other coronaviruses are.  The key to how useful this will be as a model is how close those assumptions are to the actual biological behavior of SARS-CoV-2.  We have no idea right now.

My biggest technical quibble with this model in the paper is that it’s deterministic, which is to say it’s a plug-and-chug once they set those assumptions.  My preference, given the unknowns around their assumptions, would have been for a stochastic, or probabilistic model, with a range of possible outcomes for the same input of assumptions, and then report the -range- of outcomes as they ran the model over and over again with the variation in their assumptions.  That’s a -really- technical quibble though.

The long story short on this model is that -IF- SARS-CoV-2 has immune response and spread characteristics similar to coronaviruses that cause -LESS- severe symptoms (and may be less rapidly transmitted), you may see bouncing “second waves” all the way out to 2025.  

In the famous reply of the Spartans to Philip II of Macedon though, “If.”

Arguing against is that SARS-CoV-1 (the cause of SARS) produced longer immunity than the viruses they will wind up comparing SARS-CoV-2 to in their model, and SARS-CoV-1 (and MERS) are more similar to SARS-CoV-2 than their comparison viruses.  Further, we know from last week’s update that the coronaviruses are similar enough to each other that some immune cells reacting to these more common coronaviruses can cross react and respond to SARS-CoV-2 in the lab.  So their assumptions about immune response, just for starters, may not be accurate.  Ultimately, they may be comparing to the wrong set of similar coronaviruses.  On the other hand, neither SARS nor MERS were as infectious as SARS-CoV-2 appears to be, and the viruses they do compare to in this model are causes of the common cold.  So assumptions about infection rate may be well modeled by what they use in this paper.  

I don’t know how close to truth their assumptions are.  Neither do they.  Only time will tell.  So I would view their model as useful as kind of a “worst case” for second waves.  If SARS-CoV-2 is the Worst Cold Ever, this -specific- model may be pretty accurate.  Is this outcome possible?  Sure.  Is it probable, or even THE most probable?  Tough to say.  

And what they don’t really touch on, but is the most important unknown for a second wave, is does SARS-CoV-2 have the same hospitalization rate (and high risk group mortality) when it comes back in a second wave?  The threat is the next marginal bed (and sheer mortality for high risk groups).  If that threat is less, as more herd immunity is present, or the virus on re-introduction does NOT produce the “Ah-nold and Commandos Level the Forest” immune reaction that appears to do the damage, second waves are going to be much less of a deal.  

Regardless, I would expect the huge amount of research and brainpower being devoted to SARS-CoV-2 right now to have either found a way to vaccinate or a treatment that keeps it from threatening that next marginal available bed, well before 2025.  Either one defangs SARS-CoV-2.

—Also got sent this machine learning based model from a UK based reader:  https://covid19-projections.com/

Notice it’s a different set of assumptions and inputs from the paper.  Again, all models are wrong, some are useful.  This one is interesting–you can play around and find your state or country.  This is the first I have seen predicting a late July/August “second wave” for the US in general based largely on reopening plans (I think, if I understand their model’s weights correctly).  Looking at the model’s Indiana numbers, Indiana is in is more of a downslope in new cases than they have in their model at the moment (which looks pretty flat on their graph).  I can also tell you the chances of reaching their assumed 44,000 tests per day capability in Indiana by June 1st is wildly optimistic.

So also a maybe for that July/August peak?  This machine learning model will get a significant test from all the YOLOs who were out and about Memorial Day Weekend here.  If there is no big upsurge in new cases through the weekend into mid-next week, their prevalence estimate is probably too low, and that late July/August peak is less likely.  Check back later.

–Following up a previous conspiracy theory, the computational biologist “on the cusp of a great coronavirus breakthrough” who was killed in Pittsburgh?  Murder-suicide in a love triangle.  No apparent conspiracy.

—Lastly, in the realm of social ramifications.  Rumors are things like Nintendo Switches are tough to find due to a component shortage in China.  Again, expect bizarre supply chain disruptions all year long.  It will be a very unusual year.

But it could be a worse unusual year.  For an example, check out 1816—the “Year Without A Summer.”  And then realize that kind of global crop failure, from that specific cause, averages about once a millennium as near as we can tell in the historical record.  At least food is still being produced, even if there are distribution hiccups, and an unusual year could be a whole lot worse.

I think a big part of the psychology problem around this is we all assumed that we were post-historical, and post-nature, and that with our modern science, modern healthcare, globalized world, modern politics, watch dogs like the CDC, WHO etc., that unusual years could not happen.  After all, we have machine learning and AI now.  They can win Jeopardy.  They can predict anything.  Anything.  Especially with Big Data.  Unusual years can’t happen, because they are predicted and prevented and handled before they could ever be so vulgar as to inconvenience  us. 

They always happen though.  We may have been through with history and nature.  History and nature are very much not through with us.  And you should always protect the downside from a highly unlikely, but highly-significant-and-world-changing-if-it-happens event.  Taleb has an entire book called “Anti-Fragility” on this if you are interested and out of Netflix.
The age of coronavirus should have taught us this much–humility about what we can actually predict and prevent, and better knowledge of things that can and cannot happen.  Let us find the courage to continue the thought from there.

—So you have a population with 40 million unemployed in like 10 weeks, massive disruption in everyone’s lives, higher incidence of anxiety, depression and suicide from all of it, with clear and present failure of nearly every major institution, where everyone can see that the people “in charge” do not deserve to be, are in it for themselves, and it’s not clear who can replace them to fix this. 

And then George Floyd dies in Minnesota.
This is the nation that threw tea in the harbor, and made “tar and feather” an actual, historic thing.  And I think that nation is starting to remember that.

Gonna be a long, hot American summer.

Oh, and the Onion nailed it today:  https://www.theonion.com/protestors-criticized-for-looting-businesses-without-fo-1843735351

—The Norwegian head of their version of the CDC came out this week and said that the pandemic probably could have been handled without a lockdown in their country, had they been more aggressive with testing and implementing social distancing.  Boris Johnson, in the UK, flat out admitted the UK was unprepared for the pandemic.  Must be nice to have honest politicians.  For what it’s worth, and my opinion only, where I think the US lost the plot was when flights to/from China were cancelled, but not to/from South Korea and Europe and especially Italy, when it was clear that the virus had jumped in healthcare threatening mass from China to those locations at least.  You couple that with insufficient testing availability for reasons already discussed, and you suddenly had a lot of cases in the US, but no way to know how many, and to avoid the healthcare system from getting overwhelmed everywhere, draconian measures became more attractive.  I get the good Director General from Norway’s point, but I am not sure, given the disinfectant/mask/glove shortage that happened at the same time, enough people would be prepared to make the social distancing changes quickly enough to avoid a lock down to keep as many marginal beds open as you could.  You can easily argue at least some mass quarantine was necessary to let the PPE and testing catch up.  But how long was necessary–tougher question.  Same for businesses, who would have had bingo time to get social distancing changes in place.  The lockdown did give some time to think how to accomplish all of the necessary distancing required to keep the next local marginal bed still available going forward.  

I am glad I was not in position to have to make that kind of “lockdown / no lockdown and how long” call in the face of such tremendous uncertainty—even if some of that uncertainty was the result of significant self-inflicted wounds.

—Lastly, there will come a day when I am not recommending a read on Epsilon Theory.  Today is not that day.  Rusty absolutely nails it—and as a bonus, you’ll learn where tuxedos and rules of etiquette actually came from:  https://www.epsilontheory.com/a-new-gilded-age/

—Alright, this is getting epic length even for me, and there is even a “social ramification” section to go that I am just going to save for next week.

Continue being prudent.

Your chances of catching Ebola this week are equivalent to Donald Trump agreeing by saying the US was unprepared for a pandemic.

Your chances of catching coronavirus this week are equivalent to Donald Trump agreeing with Boris Johnson that the United Kingdom was unprepared for a pandemic.

<Paladin>