The illusion of control

 I've avoided posting about the pandemic for awhile now. Partly this is because I made a few predictions that hadn't panned out, as well as others that sometimes looked to be wrong but hadn't panned out yet. I'll start with a mea culpa on the two big predictions that didn't pan out, although I have devoted some space to these before:

Mea Culpa

The Nordic Experiment: I already wrote about how I got this one wrong, but I want to add some recent data to really put it to rest. This was always an experiment with a small n and a large number of confounding factors (such as, will Swedes act the way their government tells them or will they still practice extra precautions, will other Nordic countries act similar to the Swedes regardless of national policy, will other policies - such as poor cocooning of nursing homes - swamp the policy effects?). Still, the claim was that Sweden's strategy would drive them toward herd immunity earlier, so they would have a large spike in cases, but then be immune from subsequent outbreaks. This has not happened, and since the expected November resurgence of COVID-19 was the most significant event to prepare for (and therefore the reason to pursue this policy in the first place) I think we can call this strategy ineffective at producing herd immunity.


I placed deaths before cases, as that's the more important and consistent figure (as-in, rate of testing hasn't been consistent, so comparing number of tests across time is less interesting than comparing number of positive tests across countries). What's clear from both cases and deaths is that Sweden's curve was neither shorter than the other Nordic countries, nor did their prolonged curve prevent additional outbreaks. This prediction was wrong.

Early Signal of Decline as measured by cases per test: I already made an update to this on the post itself. At first, I thought this had been entirely disproven by subsequent events, but then the recent case spike demonstrated that a lot of what we were observing over the summer was smaller local outbreaks in various locations. This should be distinguished from a true nation-wide outbreak like what we saw beginning this November. The prediction was still not useful at the time, but that's largely because the data available weren't consistent enough to provide a useful signal. It was not data that should have been used to make a prediction to begin with, so sorry about that.

Vaccine Prospects: I believe I made this claim in the comments section of a previous post. I, like many others, claimed it was highly unlikely there would be a viable vaccine within two years of discovering the virus. Huge amounts of resources were poured into this project, which I figured would happen and had planned on. I based my estimate on my personal experience in clinical development, which is what I currently do for a living. No matter what you think of specific political parties, I think the people who worked hard to get the vaccines approved deserve respect for accomplishing something truly unprecedented.

Mea Victoria

Seasonal Coronavirus: My first post on the pandemic was a simple observation that the generic term "coronavirus", used before the pandemic for the type of virus SARS-CoV-2 is, trends to increase in Google search results around November/December of each year. Therefore, I predicted that - if the virus is seasonal like its cousins - we would see a significant outbreak in November. Please note that I called this long before other people did. I'm not saying they got the idea from me, as the information was freely available, just that I didn't get it from anywhere else. Thus, when we started to see a dramatic increase in case counts in November I was not particularly concerned.

In contrast, I noticed a lot of people casting about for an explanation as to why we were suddenly seeing a surge of cases. What had we done wrong? Who was to blame? There was a lot of needless public shaming, name-calling, and general nastiness. More on this later as we get into the reason for this post's title.

Social good will: In an early post about the pandemic, I wrote about how it was obvious most societies would engage in lockdowns and other measures intended to control the virus. These measures were taken under the assumption that the virus carried significant morbidity and mortality risks (an unknown with a wide margin of probability at the time) and that at the very least we should do what we could to make the experiment work - if for no other reason than that at least we could learn something to pass on to future generations faced with a similar situation. When I wrote that article, we knew a lot less about the virus than in early May when I wrote a different article about my concerns in spending social good will. Over the intervening month, I felt we had learned enough about the virus to answer certain unknowns about whether the virus would impact large enough numbers of sub-populations that it might cause a major public health crisis if it got out of control. From what I saw, we didn't identify new populations that should be cocooned (besides elderly with preexisting conditions), and this new information should be used to adjust course for a public that had largely complied with unprecedented intrusions.

I was - and still am - concerned that a refusal to change course, open up, and accept a bad flu season (while still taking care of the at-risk population) would likely generate significant strife and sour people on lockdowns. I still see a lot of people talking about the importance of social distancing and mask wearing at reducing the spread of the virus. But even though we're seeing higher daily case counts (and about to surpass our previous zenith of daily death counts) I'm hearing much less support for lockdowns. I'm going to double-down on my prediction that the COVID-19 pandemic will not prove to be deadly by historical standards. This is still concerning to me, as a future pandemic that's as deadly as prior moderate examples could easily kill hundreds of millions of people. My only consolation is that people tend to have short memories about this kind of thing. Perhaps if the next pandemic takes at least twenty years to get here we'll be able to rebuild social good will enough to make the next lockdown politically feasible again. If subsequent minor pandemics like this one get a similar treatment, my confidence in this rebuilding of social good will diminishes. It certainly feels like a poor decision to spend all our social good will now in the hope that we don't get a truly deadly pandemic in the next two decades.

Mea Educatus

I think we've learned a few things over the past few months with the virus. As an immunologist I've learned it's not a good thing for your field to get a lot of attention by laypeople. By this I mean that large numbers of people have fooled themselves into believing a few internet searches make them informed about complex ideas they continue to misunderstand. The only difference after the internet search is they now feel they're 'knowledgeable' about the subject and feel it's okay for them to authoritatively share that illusion of knowledge with everyone. I've had a few people who don't know my background try to share things with me that are obviously false. Too often, they cite some story they read, but just as often those stories become so much 'common knowledge' that nobody remembers where it came from. It becomes difficult to track down the source of the rumor, yet everybody believes the false fact.

The worst offenders here are probably journalists, not because they're less informed than anybody else, but because they have a larger readership and yet don't take the time to become responsibly more informed than your uncle Ted passing on misinformation over social media. Too often I'll see an article that cites some new study. In a bid for clicks, they'll cite those studies and make claims that the study demonstrates something new, when in fact the new thing we learned is either not what the journalist claims, or directly in line with what we already knew or at least suspected (often both).

I'm not giving a pass to social media, and I'm explicitly not giving a pass to social media companies. They, too, have been dangerously irresponsible with whatever aura of 'legitimacy' they continue to enjoy but have not earned. For all their intent to prevent the spread of 'dangerous fake news' they've entirely missed fake information that's actually dangerous, sometimes even promoting it because whoever they have doing the 'fact checking' is as uninformed as the journalists they rely on. Here are just a few examples of what I'm talking about:

Three-month immunity: I kept hearing people quoting this 3-month figure for antibodies. That your antibodies only last for three months, and then you could get infected again. This flies in the face of decades of immunology, and all the theory. Now, I'm not going to claim a pathogen is incapable of overcoming something like memory B-cells, because they do that. But the most common way they do it is by successive slight mutation over serial infection of new individuals so they're no longer recognized by the antibodies that were created during the last infection by the time they get back around to you. An analogy would be someone dying their hair slightly imperceptible shades over time, such that their close friends might not notice any changes, but if they met someone they hadn't seen in a couple of years that person would be shocked to learn their friend's hair went from blond to black. For coronavirus, we can test this claim by tracking sub-strains. From what I've seen, the virus mutates rather slowly, so this isn't as much of a concern (though still a concern) and unlikely to be driving an major phenomena of mass reinfections so far.

But that's not what we're talking about when people cite this 3-month figure for antibodies. What they're claiming, instead, is that your body can create antibodies against SARS-CoV-2, but that those antibodies go away after 3 months. This may seem like a simple claim, but it's far and away not simple. To my knowledge, we've never seen a virus capable of actively disrupting memory B-cell persistence after those cells were initially established. We've seen viruses that evade the immune response, hide from the immune response, or mutate so quickly they can make memory B-cells ineffective. But those things are fundamentally different from actively eliminating memory B-cells after they've formed. And indeed, finding a virus capable of doing so would be a huge discovery.

(As in, Nobel-worthy. More than that, it would change and save countless lives each year - far more than the virus has taken. It would be so momentous that future generations would look back at our time and cite SARS-CoV-2 as the quintessential example of a blessing in disguise. God's gift to mankind. How could they possibly see such a bleak situation this way, just because a virus is able to do such a simple-seeming thing? Imagine if we could turn that discovery to our advantage - and indeed watching Nature do this kind of thing is very often exactly how we make such discoveries. With this power we could selectively eliminate problematic memory B-cell populations at will. We could eliminate all auto-immune diseases, like Crohn's, IBD, asthma, food and seasonal allergies, etc. We could improve transplants, so patients wouldn't have to go on immunosuppressant drugs for the rest of their lives. We would probably even reduce our dependence on getting exact HLA matches for transplants, so that donor matching would become a thing of the past. And that's just the short list; I could write a whole post just about the potential in this technology alone. It would be revolutionary to millions of people!)

So I was obviously interested to find out where this claim was coming from, but had a hard time finding the scientific literature it was based on. If true it signaled something momentous. Then someone linked to what I think is the original source of the rumor. It was a peer-reviewed paper on antibody isotype switching. In other words, it was nothing new. Literally exactly what we would expect to see in a normal immune response to any virus. The authors didn't design a study capable of demonstrating whether memory B-cell populations disappear over three months, and they didn't claim that. In fact, they specifically noted the limitations of their study, which would have precluded that claim. (They didn't specifically mention the memory B-cell idea, as I assume they didn't imagine laypeople would come up with such a dramatically wrong interpretation of their data when they went to publish it. Instead, they noted that their control group's population was younger than their COVID-19 group by well over a decade. Any claim of missing memory B-cells would be confounded by the fact that a population whose mean age was 53 would be a poor comparison to a group whose mean age was in the high 30's.) Where did the three month claim come from? That's how long the study went for. The authors had demonstrated their hypothesis of antibody isotype switching, so they stopped it so they could publish.

In order to believe this one, I'm going to need significantly better evidence. Remember that this is an extraordinary claim. If you do find a credible peer-reviewed study that substantiates the 3-month claim, please send it my way. Otherwise, based on what I've seen so far this claim is bunk.

(An aside for those interested in what antibody isotype switching is and why the authors were writing a paper about it: I've written before about how biology often has many more layers of complexity than you'd expect. This is also the case for antibodies, and understanding isotype switching is one more layer down. There is more than one type of antibody, and indeed many types of antibody have sub-types to them. Each type and sub-type of antibody has certain specialties - things they do better than other antibody types. There's a lot of cross-over, so don't imagine distinct lines of capability for each antibody isotype. When B-cells first start making antibodies, they usually start with an antibody isotype we call IgM. These M-type antibodies are big, bulky, and capable of soaking up a lot of stuff. They're also not good at a lot of specific tasks we'd want them to be able to do. Over time, through a process called affinity maturation (which is a post I'm excited to write in the offing - partly to explain why we do multiple doses of vaccines, but mostly because it's simply amazing) antibodies get better at detecting specific antigens. As this is happening, they often change to a different sub-type of antibody, in the case of  SARS-CoV-2 they unsurprisingly bias toward IgG. This allows them to be better at attacking the kind of thing they're built to defend against. IgA helps defend mucous membranes and gets antibodies across the placenta/breast milk, IgE helps defend against parasites and harmless house dust, IgG does a little of everything, given it has a bunch of sub-types. So why do this study? To better understand how the body reacts to the virus to build its immune response. What did the authors find? It reacts exactly as we would expect.)

Comorbidities: There was another report that widely circulated, citing data from the CDC's website, that stated most people who die of COVID-19 have multiple comorbidities. The people who passed this along usually claimed that reports of COVID-19 deaths were exaggerated because most of these people died of something else (the comorbidities). Now, there's a difference between claiming that COVID-19 accelerated the deaths of people who were already in fragile health, and thus might likely have died of something else within three months anyway, and claiming that they didn't actually die of COVID-19. The first claim is difficult to figure out, but usually we can get a feel for the population-level impacts by looking at something like all-cause mortality. If more people are dying of COVID-19, but fewer people are dying of cancer, we might suspect the cancer patients on death's door are dying of the virus a little earlier than they would if they'd been taken by cancer. If all-cause mortality doesn't go up at all, despite the virus, we would suspect the virus is selecting its victims largely from the ranks of those who are already terminal. If all-cause mortality goes up, we assume there's an external cause, which we might attribute to the virus.

That's not what people were claiming with the CDC reference, though. They were claiming the second idea - that people are actually still dying of cancer, or diabetes, or whatever; but that death is being counted in the wrong statistic. However, it's not possible to make that second claim based on the information on the CDC's website. You can actually go to that website and find the listed comorbidities, as well as their frequencies. (At least you could a month or two ago when this was circulating a lot and I went to check it out.) What were the comorbidities? Things like pneumonia, lower respiratory distress, shortness of breath, collapsed lungs, etc. In other words, they were descriptions of the symptoms you'd expect from someone who was dying of COVID-19. Nobody walks into the doctor and tells them, "Doc, I have dangerously elevated AST and ALT!" Those are laboratory diagnoses. People go in for what we call clinical complaints. Usually these are objectively observable (though some like pain are subjective) by the patient and the physician. Doctors often list these clinical diagnoses, as a laboratory diagnosis is an assumption, whereas the clinical diagnosis is an observation of what they personally witnessed. The list on the CDC website is long, and the majority of reports are from COVID-related symptoms. If anything, the CDC's reported comorbidities biased me to believing that the hypothesis of overreported COVID-19 deaths, if true, must be a much smaller factor than some have. Exactly the opposite of what has been claimed. I'm sure you're seeing a pattern here.

Missing B-cell immunity: I want to emphasize that this kind of misinformation spreading has happened among otherwise-intelligent people who are just operating outside their field of expertise. I saw a study being circulated, which the people citing it reported that among people who had recovered from COVID-19 about a third did not have any detectable antibodies. The intelligent people who circulated this paper claimed this was concerning as it meant a lot of people weren't creating antibodies and therefore would never be immune after COVID-19 infection. They could be reinfected, and we may never reach herd immunity - even with a vaccine! I went to the study, and discovered that - once again - any reasonable immunologist would come to the opposite conclusion. It's true the authors found there was no B-cell immunity in half the population tested. Instead, they found only T-cell immunity. In other words, half the population was still developing adaptive immunity (arguably it's the T-cells that are pulling most of the weight in protecting you from re-infection, especially for a virus) but in a way that's invisible to a test that only looks for circulating antibodies!

Regeneron: When the president got sick and they announced they were giving him a course of experimental treatment, which Trump called 'Regeneron' even though that's the name of a company not the name of a treatment, I was interested in the nature of the treatment. Trump claimed this was a new 'miracle cure' that would be rolled out to all hospitals around the country and could solve the crisis. Friends on the political Left derided the whole thing as advertising, assuming without evidence that the president was dying of COVID-19 in a hospital room somewhere but that he was lying saying everything was fine. Obviously, the claim the president's health status was being hoaxed the whole time is a claim that never panned out. However, I think many people on the Right and Left could have used a little more information on exactly what the president was getting, and why it was neither a miraculous invention created specifically in response to the crisis by amazing policy, nor was it the panacea that would solve everything if it could get out there to the masses.

The treatment was something we call polyclonal antibodies. Without going too deep into the immunology in a post that has already gone too deep as it is, I'll summarize by saying that we've been able to crate polyclonal antibodies to practically anything we want for many years. These aren't new, which if my friends on the Left had realized they'd understand that it's both dumb to question such a basic advance as 'apply creation of polyclonal antibodies to COVID-19' and also to question whether such an intervention might be beneficial. After all, it's essentially the idea behind plasma donation from patients who recover from COVID-19, and that works just fine.

Meanwhile, a polyclonal antibody is a little less expensive to create than a monoclonal antibody. That's cold comfort, given that monoclonal antibodies are among our most expensive drugs today. As a grad student, I often joked that instead of being worth my weight in gold (at this writing that's over $5 million) I'd rather be worth my weight in monoclonal antibodies. That is tough to estimate, but would be closer to $150 billion (at a price near $1,750g, which I grabbed from a random antibody I used to use), easily making me the richest person on the planet. Even though polyclonals are less expensive than monoclonals, they're not that much less expensive. They also take a lot longer to make. They're not as easy to manufacture mass quantities of to distribute around the country. It's great that we have this as an option for severe patients, but the treatment regiment the president got - getting shots before displaying symptoms - was never going to be an option that would become generally available to the public. (I don't know, maybe it will at some point, but not in the near term.)

What does control look like?

One of the things I've been saying since the beginning of the crisis - though I can't recall if I made the claim in this space - is that a lot of what we're seeing vis-a-vis international differences is probably driven more by population genetics than it is by public policy. To date, I think we're still figuring a lot of this out, but there was a recent study demonstrating that certain susceptibility markers are enriched in European populations (in other words, they'd be more susceptible to getting the virus) and are least likely to be found in African and East Asian populations. In other words, a lot of the stories we've told ourselves about why this or that nation has been successful at containing the virus, with comparable morality tales about those that haven't, are likely wrong. What you do probably matters less than the DNA you have no control over. Many rates of infection - or lack thereof - were baked into the genetic cake from the beginning. We've been running around thinking we're learning new things, and then applying them to the crisis (or complaining that others aren't applying them fervently enough) but it's likely we're hyper-focused on the wrong things, and have been all along.

This brings me back to the recent November surge. As I mentioned above, I saw a lot of people being nasty to friends and neighbors, blaming others for causing the recent surge in cases. This disregards the fact that we haven't done something new in the past two months that we could claim caused the surge. What we should have done is observe the results, assess whether the promised interventions were effective at preventing the feared outcomes (on masks the population-level data is not promising; areas with >95% compliance got the same outbreaks as areas with lower compliance; this isn't to say don't wear a mask, just that you shouldn't expect it to protect you - another claim I made early on). Instead, I see a lot of people leaning into those same old interventions, claiming we didn't hit them hard enough last time.

I have one more prediction to make. I'm not making this prediction based on evidence, so your confidence in this prediction should be low. If you go back to my first post about the crisis, I noted that the season would have more impact on COVID-19 cases than anything else. This is exactly what we saw, with the coming winter driving infectivity rates as already noted, but let's not stop there. If this trend holds, we should expect COVID-19 cases to once again subside in the springtime.  Right now, preparations are being made to ensure the most at-risk populations get it first, which is the right thing to do. That should help curb mortality, which is what we should care most about, but probably not do much to decrease daily numbers of cases, which is what has carried an outsized focus. We're told the vaccine won't become broadly available to the public until springtime next year. This is around the same time we should expect to see a natural decline in COVID-19 infections. When May rolls around, what do you think is more likely to get the credit for 'stopping the pandemic'? The vaccine (something we can control) or the season (something we can't)? I think the vaccine will get the credit, but the season will be the driving force.

People want control. Sometimes it's patients going to doctors, asking them for a cure to a disease we hardly understand. When the doctor tells them the best they can do is prescribe a series of drugs to treat the symptoms, I've seen people pursue dubious alternative medicines that promise a 'cure'. The alternatives aren't 'better', but at least they promise to solve the problem and when real solutions are lacking that seems to be good enough. (Usually the counter-claim is that anecdotal evidence is good enough, because no peer-reviewed studies have or ever will be done on various holistic/natural/traditional remedies. These claims are ignorant of actual studies about those claims that have been done. When 'alternative' natural medicines prove effective - as they sometimes have - they're incorporated into clinical practice and are no longer seen as 'alternative'. It's only when the studies fail that people claim no study was done to begin with.) I've seen the same thing with intractable political problems, economic challenges, personal struggles, and of course with the COVID-19 situation. My series on open questions is about cultivating the exact opposite of this impulse to declare a solution - and therefore reassert control - in the absence of any supporting evidence, or sometimes contrary to copious evidence to the contrary.

Let's not forget this when the pandemic is over. The current coronavirus isn't the only thing we still don't know a lot about. Many people, when confronted with something they don't know enough about - and more importantly don't have any control over - will grab onto anything they can to reassert the illusion of control. It's a common, possibly reflexive reaction. It isn't just driven by ignorance, but by a lack of alternatives. Nobody wants to be told there's nothing more we can do. Faced with the prospect of not being able to solve a problem, there is a large human drive to apply 'solutions' incapable of doing anything meaningful - and sometimes doing harm - simply because they mean doing anything at all. Despite the best evidence to the contrary, they get implemented. And after they demonstrate failure they're implemented more aggressively. Because sometimes an intervention isn't about actually solving a problem. It's about providing to the masses the illusion of control.

It's a trap. The only way to break free from the illusion is to accept there are some things we just don't know yet. There are some things we can't control.


Popular posts from this blog

A better addiction

Covid-19: Epidemiology is useful

Open Questions: The Origin of Life