On hydroxychloroquine, expertise, and the power of persuasion.

On hydroxychloroquine, expertise, and the power of persuasion.

Recently, a friend who works in the ER wrote to ask me about hydroxychloroquine.

Yes, I know. I was shocked, too. But my friend was sincere. Although most reputable news outlets have publicized that hydroxychloroquine doesn’t work against Covid-19, my friend read an article from Harvey Risch in Newsweek that seemed really compelling.

Risch has impeccable credentials – he’s an M.D. Ph.D. and a professor of epidemiology at Yale’s School of Public Health. And a lot of what he wrote for his July 23rd article is quite sensible:

Why has hydroxychloroquine been disregarded?

First, as all know, the medication has become highly politicized. For many, it is viewed as a marker of political identity, on both sides of the political spectrum. Nobody needs me to remind them that this is not how medicine should proceed.

#

Medical data isn’t perfect, and confirmation bias is very real. So there’s a chance that medical doctors really could hoodwink themselves into discounting a helpful medication, the same way that so many medical doctors get suckered into overprescribing drugs after pharmaceutical companies bribe them with gifts. Yup, medical doctors are human, too.

I know that I’m so dismayed by our current president that I’m inclined to distrust hydroxychloroquine just because he says the drug is great.

So it was a shock for me to read Risch’s article. He wrote that there was data showing that hydroxychloroquine, when used in a combination therapy early during a high-risk person’s Covid-19 infection, could dramatically reduce the risk of serious complications. If more people took hydroxychloroquine, he wrote, fewer would die.

Risch acknowledges that hydroxychloroquine is dangerous – it might kill 1 out of each 10,000 people who take it – but Covid-19 is obviously dangerous, too – it kills 3 out of each 1,000 people who contract it:

In the future, I believe this misbegotten episode regarding hydroxychloroquine will be studied by sociologists of medicine as a classic example of how extra-scientific factors overrode clear-cut medical evidence.

But for now, reality demands a clear, scientific eye of the evidence and where it points. For the sake of high-risk patients, for the same of our parents and grandparents, for the sake of the unemployed, for our economy and for our polity, especially those disproportionately affected, we must start treating immediately.

Those are strong words. And, really, the Newsweek article felt persuasive to me. And so I looked up Risch’s research in the American Journal of Epidemiology, hoping to see the actual data in support of his claims.

I’m lucky, that way. I’m a scientist, so I don’t have to trust the words of a supposed expert. I’m an expert. I get to look at the data.

The data are much less compelling than Risch’s words.

Risch discusses the results of an uncontrolled study by Vladimir Zelenko, a medical doctor in Monroe, New York: “For example, among Connecticut cases 60 years of age or older, at present the mortality is 20%. Thus it would be ballpark to estimate that some 20% of the 1466 treated high-risk patients in the Zelenko cohort would have died without outpatient hydroxychloroquine plus antibiotic.

This is an egregiously inaccurate statement. The high death rate cited – 20 – is for older patients who test positive for Covid-19 and have such severe symptoms that they need to be hospitalized.

As described in the short statement released by Zelenko, he treated 405 people who visited his office complaining of mild cough, fever, headache, sore throat, or diarrhea. His patients were not given a Covid-19 test. Presumably, many were never infected with Covid-19.

It is not a surprise to see that a 60-year-old patient who takes hydroxychloroquine after developing a sore throat from seasonal allergies is less likely to die than a 60-year-old patient who is diagnosed with Covid-19 in the hospital.

Of Zelenko’s 405 patients, at least two 2 died. This is lower than the expected 1% mortality rate of high-risk patients who contract Covid-19. But this set of 405 patients included low-risk patients experiencing shortness of breath and high-risk patients experiencing mild headache, many of whom never had Covid-19.

Zelenko’s report is two pages long and written in extremely lucid prose. Risch either totally misread it, which is galling, or intentionally mis-described it, which is worse.

#

So, why was Zelenko giving people hydroxychloroquine in the first place?

Well, I’d heard that an in vitro study – which means “inside a test tube or petri dish, not a person” – showed that hydroxychloroquine reduced Covid-19 viral replication. But I hadn’t read the original paper. So I looked it up.

It should have taken me less than a minute to find this paper. Unfortunately, people have been pretty sloppy with their references. I get it. Covid-19 is scary, and it’s urgent, so people are publishing faster than usual.

I assumed that I could pull up almost any paper on hydroxychloroquine and Covid-19 and quickly find the citation for the original study. Indeed, most purport to be citing it. But in this, the citation that ought to have pointed to that study instead sent me to a paper on the differentiation of lung stem cells, and in this, the relevant citation incorrectly points to a paper on the drug lopinavir.

Ugh. I mean, these bungled citations aren’t that big a deal for me, personally – just means I had to give up on piggybacking and instead search Pubmed. But it undermines trust when you can’t get the little things right.

Anyway, the earliest reference that I found was from Liu et al., their study “Hydroxychloroquine, a less toxic derivative of chloroquine, is effective at inhibiting SARS-CoV-2 infection in vitro.” And, yes, I’ll admit – I thought about putting in the wrong link just to mess with you. But, if I did that, would you still trust me about the rest of this?

Liu et al. used Vero cells – a cell line derived from a kidney cancer in African green monkeys – and for Figure 1, they measured both how much hydroxychloroquine it takes to kill cells (about 200 micromolar is a cytotoxic dose) and how much hydroxychloroquine it takes to inhibit viral infection (about a 10 micromolar dose).

Okay. To me, that’s already sounding a little spooky. The bigger the difference between an effective dose and a lethal dose, the safer you are.

That’s why a bunch of hippies died after The Teachings of Don Juan was published. That book touted jimsomweed as a psychedelic. Indeed, the plant contains a high concentration of scopolamine, which can give people nightmarish visions of flying. It’s a powerful hallucinogen. But the effective dose is quite close to the lethal dose – when curious kids try to get high off it, they’re flirting with death.

Everyone’s body is a little different from everyone else’s. Maybe a dose that’s safe for you would kill me. The odds of disaster are worse when the effective dose and lethal dose are similar.

So, Liu et al. saw cytotoxicity kick in at around 100 micromolar hydroxychloroquine, getting pretty high by 200 micromolar. And for their visual assay of viral infection, they bathed their Vero cells in 50 micromolar hydroxychloroquine.

To block viral entry, they were coming pretty close to just killing these cells with the drug.

And the problem is even worse inside a human body. You take a drug and it gets into your bloodstream. It’ll reach some concentration there. This is the concentration that matters most for toxicity.

But the drug will only be effective against Covid-19 when it reaches your lungs. When Marzolini et al. used mass spectrometry to measure how much of hydroxychloroquine was actually getting from a patient’s blood to their lungs, they found that it wasn’t at a high enough concentration to reproduce any effects seen in vitro.

#

Indeed, a randomized clinical study showed that hydroxychloroquine fails as a post-exposure prophylaxis. The drug was given to people who were worried about exposure because they’d spent time with someone who tested positive for Covid-19. The drug didn’t help – these people contracted the infection at the same rate as people who were given a placebo.

A randomized clinical study also showed that hydroxychloroquine fails as a cure. People who visited a hospital and tested positive for Covid-19 but had mild symptoms were given the drug. Their disease was just as likely to progress as people who received a placebo.

Hydroxychloroquine doesn’t work, and it’s toxic.

#

I was left wondering: why would Risch write these things? Why would he write that article for Newsweek? He’s clearly intelligent, and, from the tone of his writing, I feel confident that he wants to help people.

He might even believe wholeheartedly in the conclusion he’s presenting.

That’s generally true among scientists. Confirmation bias is insidious.

That paper from the team at Harvard? They did some modeling and argued that, if Covid-19 is seasonal, we will save most lives by periodically shutting down. But their model left out the waning immunity that would cause Covid-19 to be seasonal! Whoops. That’s why they reached the wrong conclusion.

Or the recent New York Times editorial from Iwasaki and Medzhitov, both professors of immunobiology at Yale, reassuring readers that they won’t get Covid-19 twice. Well, that’s not correct.

Some antigens confer immunity that lasts about as long as our lives. Most don’t. Influenza immunity lasts months, not years. The paper that Iwasaki and Medzhitov cited in their article, a study in which people were intentionally infected with a less dangerous coronavirus, found that immunity to that virus lasted months, not years.

Covid-19 immunity will not last forever. The relevant question isn’t whether you can be infected again, it’s how soon you can be re-infected. With the data we have so far, it’s reasonable to expect that the answer will be measured in months, not years.

There’s some good news – the second time you contract Covid-19, it’ll probably be less severe than the first. In addition to antibodies, your immune system has “T cell memory” to help you fight off subsequent infections. But, as is also described in the paper cited by Iwasaki and Medzhitov, even people who felt fine were shedding virus again the second time they were infected.

During the second infection, the research subjects were shedding viral particles for a shorter period of time. But, especially with Covid-19 – a virus that can be transmitted simply by talking – a person who sheds virus for a short time while feeling fine is probably more likely to transmit the disease than somebody who sheds virus for a whole week while feeling like garbage.

The person who feels like garbage will stay home. The person who feels fine won’t.

Still, though, I was left wondering – what underlying beliefs would sway Risch enough that he’d make these blunders?

Eventually, I decided to lump his motivation in with mine. Maybe that’s fair, maybe it’s not. Really, I have no idea what he was thinking, so this is just my best guess.

But I imagine that many of these people – Risch, Iwasaki, Medzhitov, John Ioannidis, David Katz, all of whom are very smart, and all of whom mean well – understand that the strategies we’re using against Covid-19 are both ineffectual and are causing harm.

No shutdown will eliminate Covid-19 – the best we can do is to delay it. And we can delay it only as long as we maintain the shutdown. Maybe that seems fine if you’re an older, wealthy person brimming with optimism about vaccine development, like Anthony Fauci who thinks we’ll have a working vaccine early next year, but it’s unconscionable if you think a working vaccine might be five or more years away.

I don’t think we should try to pause children’s development for five years.

Still, there’s no mathematical or logical way to prove what we should do. School closures definitely slow the spread of Covid-19. How do you balance the good of delaying an elderly person’s infection by three months (which is equivalent to a drug that extends a patient’s life by three months) with the harms we’re causing?

I know what I’d do, but other people have different priorities than me. And that’s okay!

I’d like to think, though, that I’m not trying to hoodwink anybody about the science in order to deceptively get them to do the thing I think is right.

Like, yes, I think schools should be open. I think we owe it to children. Right now, children are suffering, but this is our fault, the fault of grown-ups.

We have known for over a decade that we ought to make coronavirus vaccines – we didn’t devote enough resources to it, and now we don’t have one. We’ve known for decades that eating animals – both those sold in meat markets like in Wuhan and the ones raised in “concentrated animal feeding operations” throughout the U.S. – will create more zoogenic diseases, and we kept doing it. We know that a guaranteed basic income would’ve given people the resources they needed to self-isolate during an epidemic – we don’t have one. We know that guaranteed access to health care would keep our death rate down.

Climate change will make pandemics more frequent, in addition to making our world unliveable for future generations. And we haven’t taken action to stop it.

None of these failings are children’s fault. We, older people, have failed. We fucked up. And now we’re asking children to make sacrifices to dampen the impact of our mistake (although, again, it won’t work – it’ll just delay the eventual repercussions).

I think today’s children deserve a fair shot at a good life, and I think that school is an essential part of that.

But don’t let anybody try to convince you that it’s safe to re-open schools because hydroxychloroquine will stop Covid-19.

On Ann Leckie’s ‘The Raven Tower.’

On Ann Leckie’s ‘The Raven Tower.’

At the beginning of Genesis, God said, Let there be light: and there was light.

“Creation” by Suus Wansink on Flickr.

In her magisterial new novel The Raven Tower, Ann Leckie continues with this simple premise: a god is an entity whose words are true.

A god might say, “The sky is green.”  Well, personally I remember it being blue, but I am not a god.  Within the world of The Raven Tower, after the god announces that the sky is green, the sky will become green.  If the god is sufficiently powerful, that is.  If the god is too weak, then the sky will stay blue, which means the statement is not true, which means that the thing who said “The sky is green” is not a god.  It was a god, sure, but now it’s dead.

Poof!

And so the deities learn to be very cautious with their language, enumerating cases and provisions with the precision of a contemporary lawyer drafting contractual agreements (like the many “individual arbitration” agreements that you’ve no doubt assented to, which allow corporations to strip away your legal rights as a citizen of this country.  But, hey, I’m not trying to judge – I have signed those lousy documents, too.  It’s difficult to navigate the modern world without stumbling across them).

A careless sentence could doom a god.

But if a god were sufficiently powerful, it could say anything, trusting that its words would reshape the fabric of the universe.  And so the gods yearn to become stronger — for their own safety in addition to all the other reasons that people seek power.

In The Raven Tower, the only way for gods to gain strength is through human faith.  When a human prays or conducts a ritual sacrifice, a deity grows stronger.  But human attention is finite (which is true in our own world, too, as demonstrated so painfully by our attention-sapping telephones and our attention-monopolizing president).

Image from svgsilh.com.

And so, like pre-monopoly corporations vying for market share, the gods battle.  By conquering vast kingdoms, a dominant god could receive the prayers of more people, allowing it to grow even stronger … and so be able to speak more freely, inured from the risk that it will not have enough power to make its statements true.

If you haven’t yet read The Raven Tower, you should.  The theological underpinnings are brilliant, the characters compelling, and the plot so craftily constructed that both my spouse and I stayed awake much, much too late while reading it.

#

In The Raven Tower, only human faith feeds gods.  The rest of the natural world is both treated with reverence – after all, that bird, or rock, or snake might be a god – and yet also objectified.  There is little difference between a bird and a rock, either of which might provide a fitting receptacle for a god but neither of which can consciously pray to empower a god.

Image by Stephencdickson on Wikimedia Commons.

Although our own world hosts several species that communicate in ways that resemble human language, in The Raven Tower the boundary between human and non-human is absolute.  Within The Raven Tower, this distinction feels totally sensible – after all, that entire world was conjured through Ann Leckie’s assiduous use of human language.

But many people mistakenly believe that they are living in that fantasy world.

In the recent philosophical treatise Thinking and Being, for example, Irad Kimhi attempts to describe what is special about thought, particularly thoughts expressed in a metaphorical language like English, German, or Greek.  (Kimhi neglects mathematical languages, which is at times unfortunate.  I’ve written previously about how hard it is to translate certain concepts from mathematics into metaphorical languages like we speak with, and Kimhi fills many pages attempting to precisely the concept of “compliments” from set theory, which you could probably understand within moments by glancing at a Wikipedia page.)

Kimhi does use English assiduously, but I’m dubious that a metaphorical language was the optimal tool for the task he set himself.  And his approach was further undermined by flawed assumptions.  Kimhi begins with a “Law of Contradiction,” in which he asserts, following Aristotle, that it is impossible for a thing simultaneously to be and not to be, and that no one can simultaneously believe a thing to be and not to be.

Maybe these assumptions seemed reasonable during the time of Aristotle, but we now know that they are false.

Many research findings in quantum mechanics have shown that it is possible for a thing simultaneously to be and not to be.  An electron can have both up spin and down spin at the same moment, even though these two spin states are mutually exclusive (the states are “absolute compliments” in the terminology of set theory).  This seemingly contradictory state of both being and not being is what allows quantum computing to solve certain types of problems much faster than standard computers.

And, as a rebuttal for the psychological formulation, we have the case of free will.  Our brains, which generate consciousness, are composed of ordinary matter.  Ordinary matter evolves through time according to a set of known, predictable rules.  If the matter composing your brain was non-destructively scanned at sufficient resolution, your future behavior could be predicted.  Accurate prediction would demonstrate that you do not have free will.

And yet it feels impossible not to believe in the existence of free will.  After all, we make decisions.  I perceive myself to be choosing the words that I type.

I sincerely, simultaneously believe that humans both do and do not have free will.  And I assume that most other scientists who have pondered this question hold the same pair of seemingly contradictory beliefs.

The “Law of Contradiction” is not a great assumption to begin with.  Kimhi also objectifies nearly all conscious life upon our planet:

The consciousness of one’s thinking must involve the identification of its syncategorematic difference, and hence is essentially tied up with the use of language.

A human thinker is also a determinable being.  This book presents us with the task of trying to understand our being, the being of human beings, as that of determinable thinkers.

The Raven Tower is a fantasy novel.  Within that world, it was reasonable that there would be a sharp border separating humans from all other animals.  There are also warring gods, magical spells, and sacred objects like a spear that never misses or an amulet that makes people invisible.

But Kimhi purports to be writing about our world.

In Mama’s Last Hug, biologist Frans de Waal discusses many more instances of human thinkers brazenly touting their uniqueness.  If I jabbed a sharp piece of metal through your cheek, it would hurt.  But many humans claimed that this wouldn’t hurt a fish. 

The fish will bleed.  And writhe.  Its body will produce stress hormones.  But humans claimed that the fish was not actually in pain.

They were wrong.

Image by Catherine Matassa.

de Waal writes that:

The consensus view is now that fish do feel pain.

Readers may well ask why it has taken so long to reach this conclusion, but a parallel case is even more baffling.  For the longest time, science felt the same about human babies.  Infants were considered sub-human organisms that produced “random sounds,” smiles simply as a result of “gas,” and couldn’t feel pain. 

Serious scientists conducted torturous experiments on human infants with needle pricks, hot and cold water, and head restraints, to make the point that they feel nothing.  The babies’ reactions were considered emotion-free reflexes.  As a result, doctors routinely hurt infants (such as during circumcision or invasive surgery) without the benefit of pain-killing anesthesia.  They only gave them curare, a muscle relaxant, which conveniently kept the infants from resisting what was being done to them. 

Only in the 1980s did medical procedures change, when it was revealed that babies have a full-blown pain response with grimacing and crying.  Today we read about these experiments with disbelief.  One wonders if their pain response couldn’t have been noticed earlier!

Scientific skepticism about pain applies not just to animals, therefore, but to any organism that fails to talk.  It is as if science pays attention to feelings only if they come with an explicit verbal statement, such as “I felt a sharp pain when you did that!”  The importance we attach to language is just ridiculous.  It has given us more than a century of agnosticism with regard to wordless pain and consciousness.

As a parent, I found it extremely difficult to read the lecture de Waal cites, David Chamberlain’s “Babies Don’t Feel Pain: A Century of Denial in Medicine.”

From this lecture, I also learned that I was probably circumcised without anesthesia as a newborn.  Luckily, I don’t remember this procedure, but some people do.  Chamberlain describes several such patients, and, with my own kids, I too have been surprised by how commonly they’ve remembered and asked about things that happened before they had learned to talk.

Vaccination is painful, too, but there’s a difference – vaccination has a clear medical benefit, both for the individual and a community.  Our children have been fully vaccinated for their ages.  They cried for a moment, but we comforted them right away.

But we didn’t subject them to any elective surgical procedures, anesthesia or no.

In our world, even creatures that don’t speak with metaphorical language have feelings.

But Leckie does include a bridge between the world of The Raven Tower and our own.  Although language does not re-shape reality, words can create empathy.  We validate other lives as meaningful when we listen to their stories. 

The narrator of The Raven Tower chooses to speak in the second person to a character in the book, a man who was born with a body that did not match his mind.  Although human thinkers have not always recognized this truth, he too has a story worth sharing.

On artificial intelligence and solitary confinement.

On artificial intelligence and solitary confinement.

512px-Ludwig_WittgensteinIn Philosophical Investigations (translated by G. E. M. Anscombe), Ludwig Wittgenstein argues that something strange occurs when we learn a language.  As an example, he cites the problems that could arise when you point at something and describe what you see:

The definition of the number two, “That is called ‘two’ “ – pointing to two nuts – is perfectly exact.  But how can two be defined like that?  The person one gives the definition to doesn’t know what one wants to call “two”; he will suppose that “two” is the name given to this group of nuts!

I laughed aloud when I read this statement.  I borrowed Philosophical Investigations a few months after the birth of our second child, and I had spent most of his first day pointing at various objects in the hospital maternity ward and saying to him, “This is red.”  “This is red.”

“This is red.”

Of course, the little guy didn’t understand language yet, so he probably just thought, the warm carry-me object is babbling again.

IMG_5919
Red, you say?

Over time, though, this is how humans learn.  Wittgenstein’s mistake here is to compress the experience of learning a language into a single interaction (philosophers have a bad habit of forgetting about the passage of time – a similar fallacy explains Zeno’s paradox).  Instead of pointing only at two nuts, a parent will point to two blocks – “This is two!” and two pillows – “See the pillows?  There are two!” – and so on.

As a child begins to speak, it becomes even easier to learn – the kid can ask “Is this two?”, which is an incredibly powerful tool for people sufficiently comfortable making mistakes that they can dodge confirmation bias.

y648(When we read the children’s story “In a Dark Dark Room,” I tried to add levity to the ending by making a silly blulululu sound to accompany the ghost, shown to the left of the door on this cover. Then our youngest began pointing to other ghost-like things and asking, “blulululu?”  Is that skeleton a ghost?  What about this possum?)

When people first programmed computers, they provided definitions for everything.  A ghost is an object with a rounded head that has a face and looks very pale.  This was a very arduous process – my definition of a ghost, for instance, is leaving out a lot of important features.  A rigorous definition might require pages of text. 

Now, programmers are letting computers learn the same way we do.  To teach a computer about ghosts, we provide it with many pictures and say, “Each of these pictures has a ghost.”  Just like a child, the computer decides for itself what features qualify something for ghost-hood.

In the beginning, this process was inscrutable.  A trained algorithm could say “This is a ghost!”, but it couldn’t explain why it thought so.

From Philosophical Investigations: 

Screen Shot 2018-03-22 at 8.40.41 AMAnd what does ‘pointing to the shape’, ‘pointing to the color’ consist in?  Point to a piece of paper.  – And now point to its shape – now to its color – now to its number (that sounds queer). – How did you do it?  – You will say that you ‘meant’ a different thing each time you pointed.  And if I ask how that is done, you will say you concentrated your attention on the color, the shape, etc.  But I ask again: how is that done?

After this passage, Wittgenstein speculates on what might be going through a person’s head when pointing at different features of an object.  A team at Google working on automated image analysis asked the same question of their algorithm, and made an output for the algorithm to show what it did when it “concentrated its attention.” 

Here’s a beautiful image from a recent New York Times article about the project, “Google Researchers Are Learning How Machines Learn.”  When the algorithm is specifically instructed to “point to its shape,” it generates a bizarre image of an upward-facing fish flanked by human eyes (shown bottom center, just below the purple rectangle).  That is what the algorithm is thinking of when it “concentrates its attention” on the vase’s shape.

new york times image.jpg

At this point, we humans could quibble.  We might disagree that the fish face really represents the platonic ideal of a vase.  But at least we know what the algorithm is basing its decision on.

Usually, that’s not the case.  After all, it took a lot of work for Google’s team to make their algorithm spit out images showing what it was thinking about.  With most self-trained neural networks, we know only its success rate – even the designers will have no idea why or how it works.

Which can lead to some stunningly bizarre failures.

artificial-intelligence-2228610_1280It’s possible to create images that most humans recognize as one thing, and that an image-analysis algorithm recognizes as something else.  This is a rather scary opportunity for terrorism in a world of self-driving cars; street signs could be defaced in such a way that most human onlookers would find the graffiti unremarkable, but an autonomous car would interpret in a totally new way.

In the world of criminal justice, inscrutable algorithms are already used to determine where police officers should patrol.  The initial hope was that this system would be less biased – except that the algorithm was trained on data that came from years of racially-motivated enforcement.  Minorities are still more likely to be apprehended for equivalent infractions.

And a new artificial intelligence algorithm could be used to determine whether a crime was “gang related.”  The consequences of error can be terrible, here: in California, prisoners could be shunted to solitary for decades if they were suspected of gang affiliation.  Ambiguous photographs on somebody’s social media site were enough to subject a person to decades of torture.

Solitary_Confinement_(4692414179)When an algorithm thinks that the shape of a vase is a fish flanked by human eyes, it’s funny.  But it’s a little less comedic when an algorithm’s mistake ruins somebody’s life – if an incident is designated as a “gang-related crime”, prison sentences can be egregiously long, or send someone to solitary for long enough to cause “anxiety, depression, and hallucinations until their personality is completely destroyed.

Here’s a poem I received in the mail recently:

LOCKDOWN

by Pouncho

For 30 days and 30 nights

I stare at four walls with hate written

         over them.

Falling to my knees from the body blows

         of words.

It damages the mind.

I haven’t had no sleep. 

How can you stop mental blows, torture,

         and names –

         They spread.

I just wanted to scream:

         Why?

For 30 days and 30 nights

My mind was in isolation.