Beside the obvious ones, one other “outcome I care about” is whether, with a vaccine, you might still get a case of COVID-19 that is mild, but then followed later on by a rather disabling case of so-called “long COVID”. (There seems to be examples of just that.)
As someone who's already living with one chronic condition of the “unexplained” variety, likely some kind of weird autoimmune problem, I don't particularly fancy adding more chronic symptoms to my situation. Of course, this existing chronic condition doesn't qualify me for early vaccination, so I have to wait my turn and will have to remain very careful until I can get a shot. (It'll be a while, here in Nova Scotia.)
So one question that is not addressed anywhere (and not your column either) is: how good are the vaccines at preventing “long COVID”? I am tempted to believe that the ones with higher efficacy might be better, but of course I have no evidence of that. And it'll likely be a long time before we have any data on this (if ever).
I guess the bottom line is that I'll still have to take the first vaccine I can get, whichever it is. But the whole “long COVID” issue will remain a big question mark for a long time, I am afraid.
Since fairly early in the pandemic, I've been wondering about the four coronaviruses that are usually considered "common colds." (Human coronavirus 229E (HCoV-229E), Human coronavirus OC43 (HCoV-OC43), Human coronavirus HKU1 (HCoV-HKU1) and Human coronavirus NL63 (HCoV-NL63)). I've read that some of these can lead to pneumonia and occasionally death in older or immunocompromised people; and I've also read about the similarity between long COVID and Chronic Fatigue Syndrome / myalgic encephalomyelitis, and the fact that some researchers are now considering the possibility that ME/CFS might have something to do with one of the "common cold" coronaviruses. To my layperson's ears this sure sounds as if the following might be true: that all the "common cold" coronaviruses are just as bad as SARS-CoV-2, with the gigantic difference that they are endemic, and the vast majority of people contract mild cases when they are very young, providing immunity from the more serious effects for the rest of our lives. The majority, but not all. Some catch one of the "common cold" coronaviruses at a less advantageous age and are plagued by symptoms for years afterwards. Anyway, that's a layperson's guess. I'm wondering if any immunologists are following similar lines of reasoning, (or why they might dismiss this idea out-of-hand), and more to the point, I'm wondering if there is any research into a "broad-spectrum" coronavirus vaccine, one that might provide or boost resistance to all of the coronaviruses that affect humans.
The large messaging challenge here is that we are fighting a ubiquitous mental model - that of percent based comparison - that no matter how many discussions we have of more nuanced things like end points and statistical power, I'm not sure can be overcome. I wonder if it would be better to redesign the service model? We acknowledge a difference and "prescribe" the mRNA vaccines for higher risk people and J&J for lower risk (age, co-morbidities). Let people select into their higher/lower risk clinics. If they select into higher, they come to high risk clinic day, where you know you have mRNA vaccines. It would be one more operational variable but would give people agency and possibly negate some of the need for this debate. And possibly, it would remove some of the inevitable operational mess that will come with people getting to a clinic and rejecting what they perceive as a less effective vaccine, as they have done with AZ in Israel. You also could run the clinics very differently, as the J&J is so much easier to handle (cold chain, no second dose scheduling). So, the pros may outweigh the con of an extra selection variable.
It's probably no accident that the most problematic mental models always seem to be the ones that are flattening, reductionist, binary, static or excessively linear in structure and function. This is especially true when the thing being modeled is a complex organism like a human (including its immune system, or behavior, or anything else). Even the use of "immunity" as an abstract noun evokes the image of a monolithic, fixed thing (like a wall).
Having said that, I wonder if there may be a greater diversity of mental models already informing people's response to the vaccine than Zeynep gives credit. Not only does this vary according to personality, values, training, what their social network is telling them, whether they read Zeynep's column, etc.; it's quite likely one individual doesn't rely on a single model in appraising the vaccine but may employ several complementary ones reflecting different layers and aspects, which may even interact with one another. For instance, there is the binary model jlb proposed below as a crude heuristic for deciding to take the vaccine. There is Meaghan's percent-based comparison model for those now wondering *which* vaccine they should take and obsessing over tiny differences. There is the Long COVID model where all breakthrough symptoms are equally dangerous since there's no rhyme or reason (so it's like holes right in the middle of the wall that no one else seems to notice or acknowledge because they're only focused on height; or the wall toppling over on us, or getting hit by friendly fire, or getting ambushed from behind and then held captive and tortured). And there is the "having stuff rammed down our throat" model where authority and agency is the big preoccupation.
Isn't a big problem, beyond how we picture the wall, that "severe" isn't a quality of the virus itself (the wave) but a quality of the outcome? And that the vaccines and the immune system are conceptually and functionally distinct? So in Zeynep's model there could be three parts: the vaccine as the initial barrier that deflects and breaks up the wave before reaching the wall; our immune defenses as an adjustable wall that can alter its height and thickness or even shape-shift as needed, with guerilla soldiers waiting on the other side to pick off invaders; and severity of illness as whatever's left after making it past all this stuff. But for many the mental model may be as simple as: two barriers one after the other, which may or may not be enough to protect.
The statistical power of studies of drugs as well as vaccines is misunderstood. We have over 50 independent studies of Ivermectin which in toto are far larger than the "Phase 3" the FDA feels secure with. But they are from independent researchers all around the world. At least 20 of the studies are well controlled double blind placebo controlled. And none of them are biased by the profit motive, because no one makes money on a non-profit generic drug that's a huge commodity in the vetinary world as well as being regularly taken by 1/24th of Earth's human population.
For someone that understands stats, independent unbiased, small but excellent studies are WAY MORE POWERFUl than one phase 3 trial paid for by a drug vendor.
But the FDA sits on it's hands and tries to talk people out of taking one of our safest drugs.
And the NIH Lies by cherry picking the two worst studies which are totally out of date now. One that says mouse liver cells would need a lot of drug, which is irrelevant because we know human lung cells take less. And the pick the worst possible study, another early study from Peru that proves that Sicker patients die more, and some of them are given Ivermectin. (Really, that's what the study proves! Not even close to an attempt at an unbiased sample. )
Some mental models cannot be changed, because - well - they're generally a good guide to making decisions. 95% is always better than 67% and you cannot wave this away by saying that one applies to apples and the other to oranges or by saying a 1 in 20 chance is no different than a 1 in 3 chance.
So rather then try to change that model, use a binary model, e.g. if you get a shot of ANY vaccine then you won't die or end up in the hospital.
Parenthetically, a message, to be effective, must be targeted to the mental model of the recipient.
And thank you Zeynep for creating a space to think hard and carefully.
Fantastic illustration of the immune system. Love it. Also like the explanation of statistical power and sample sizes. I've been struggling with that challenge as well. In a slightly different context, I like this comparison: Imagine you have to find out if a die is 10% off, that is it is 10% more likely to roll a 1. Even throwing the die a 100 times, is not going to make it clear, that the die is skewed, since you are only going to get a couple more 1s than you would expect. Sample size needs to be quite big. But for the vaccine challenge, the MVP scenario is certainly better.
Can hospitals start a reporting system (to the CDC?) whereby new inpatients being treated for COVID are asked if they have been vaccinated? We presently have almost 50,000 hospitalized patients. Even if only half of the hospitals contributed this information along with patient ages, we would quickly acquire valuable data with good statistical power.
I've been working with a mental model where the immune system of the body works like the public health system of a city or nation. The fever and other immune systems are like the economic shutdowns, and the immune response to a vaccine is like developing a quick and easy virus test. This analogy helps me understand why the size of the initial dose makes a difference to the severity of the illness - a city with 10,000 cases at the time they discovered is going to need to go through a longer and harder lockdown than a city with 1 case that they discovered, in order to clear it. It also helps me understand why a vaccine (or prior infection) would be able to prevent severe illness more effectively than it can prevent any symptomatic illness - even with good tests, Australia and New Zealand occasionally need a brief lockdown, but they never need something like the UK ones.
Obviously there's a lot more disanalogies, but thinking about it this way has seemed helpful to me.
Beside the obvious ones, one other “outcome I care about” is whether, with a vaccine, you might still get a case of COVID-19 that is mild, but then followed later on by a rather disabling case of so-called “long COVID”. (There seems to be examples of just that.)
As someone who's already living with one chronic condition of the “unexplained” variety, likely some kind of weird autoimmune problem, I don't particularly fancy adding more chronic symptoms to my situation. Of course, this existing chronic condition doesn't qualify me for early vaccination, so I have to wait my turn and will have to remain very careful until I can get a shot. (It'll be a while, here in Nova Scotia.)
So one question that is not addressed anywhere (and not your column either) is: how good are the vaccines at preventing “long COVID”? I am tempted to believe that the ones with higher efficacy might be better, but of course I have no evidence of that. And it'll likely be a long time before we have any data on this (if ever).
I guess the bottom line is that I'll still have to take the first vaccine I can get, whichever it is. But the whole “long COVID” issue will remain a big question mark for a long time, I am afraid.
Since fairly early in the pandemic, I've been wondering about the four coronaviruses that are usually considered "common colds." (Human coronavirus 229E (HCoV-229E), Human coronavirus OC43 (HCoV-OC43), Human coronavirus HKU1 (HCoV-HKU1) and Human coronavirus NL63 (HCoV-NL63)). I've read that some of these can lead to pneumonia and occasionally death in older or immunocompromised people; and I've also read about the similarity between long COVID and Chronic Fatigue Syndrome / myalgic encephalomyelitis, and the fact that some researchers are now considering the possibility that ME/CFS might have something to do with one of the "common cold" coronaviruses. To my layperson's ears this sure sounds as if the following might be true: that all the "common cold" coronaviruses are just as bad as SARS-CoV-2, with the gigantic difference that they are endemic, and the vast majority of people contract mild cases when they are very young, providing immunity from the more serious effects for the rest of our lives. The majority, but not all. Some catch one of the "common cold" coronaviruses at a less advantageous age and are plagued by symptoms for years afterwards. Anyway, that's a layperson's guess. I'm wondering if any immunologists are following similar lines of reasoning, (or why they might dismiss this idea out-of-hand), and more to the point, I'm wondering if there is any research into a "broad-spectrum" coronavirus vaccine, one that might provide or boost resistance to all of the coronaviruses that affect humans.
Thanks to Zeynep and all the thoughtful posters!
The large messaging challenge here is that we are fighting a ubiquitous mental model - that of percent based comparison - that no matter how many discussions we have of more nuanced things like end points and statistical power, I'm not sure can be overcome. I wonder if it would be better to redesign the service model? We acknowledge a difference and "prescribe" the mRNA vaccines for higher risk people and J&J for lower risk (age, co-morbidities). Let people select into their higher/lower risk clinics. If they select into higher, they come to high risk clinic day, where you know you have mRNA vaccines. It would be one more operational variable but would give people agency and possibly negate some of the need for this debate. And possibly, it would remove some of the inevitable operational mess that will come with people getting to a clinic and rejecting what they perceive as a less effective vaccine, as they have done with AZ in Israel. You also could run the clinics very differently, as the J&J is so much easier to handle (cold chain, no second dose scheduling). So, the pros may outweigh the con of an extra selection variable.
It's probably no accident that the most problematic mental models always seem to be the ones that are flattening, reductionist, binary, static or excessively linear in structure and function. This is especially true when the thing being modeled is a complex organism like a human (including its immune system, or behavior, or anything else). Even the use of "immunity" as an abstract noun evokes the image of a monolithic, fixed thing (like a wall).
Having said that, I wonder if there may be a greater diversity of mental models already informing people's response to the vaccine than Zeynep gives credit. Not only does this vary according to personality, values, training, what their social network is telling them, whether they read Zeynep's column, etc.; it's quite likely one individual doesn't rely on a single model in appraising the vaccine but may employ several complementary ones reflecting different layers and aspects, which may even interact with one another. For instance, there is the binary model jlb proposed below as a crude heuristic for deciding to take the vaccine. There is Meaghan's percent-based comparison model for those now wondering *which* vaccine they should take and obsessing over tiny differences. There is the Long COVID model where all breakthrough symptoms are equally dangerous since there's no rhyme or reason (so it's like holes right in the middle of the wall that no one else seems to notice or acknowledge because they're only focused on height; or the wall toppling over on us, or getting hit by friendly fire, or getting ambushed from behind and then held captive and tortured). And there is the "having stuff rammed down our throat" model where authority and agency is the big preoccupation.
Isn't a big problem, beyond how we picture the wall, that "severe" isn't a quality of the virus itself (the wave) but a quality of the outcome? And that the vaccines and the immune system are conceptually and functionally distinct? So in Zeynep's model there could be three parts: the vaccine as the initial barrier that deflects and breaks up the wave before reaching the wall; our immune defenses as an adjustable wall that can alter its height and thickness or even shape-shift as needed, with guerilla soldiers waiting on the other side to pick off invaders; and severity of illness as whatever's left after making it past all this stuff. But for many the mental model may be as simple as: two barriers one after the other, which may or may not be enough to protect.
The statistical power of studies of drugs as well as vaccines is misunderstood. We have over 50 independent studies of Ivermectin which in toto are far larger than the "Phase 3" the FDA feels secure with. But they are from independent researchers all around the world. At least 20 of the studies are well controlled double blind placebo controlled. And none of them are biased by the profit motive, because no one makes money on a non-profit generic drug that's a huge commodity in the vetinary world as well as being regularly taken by 1/24th of Earth's human population.
For someone that understands stats, independent unbiased, small but excellent studies are WAY MORE POWERFUl than one phase 3 trial paid for by a drug vendor.
But the FDA sits on it's hands and tries to talk people out of taking one of our safest drugs.
And the NIH Lies by cherry picking the two worst studies which are totally out of date now. One that says mouse liver cells would need a lot of drug, which is irrelevant because we know human lung cells take less. And the pick the worst possible study, another early study from Peru that proves that Sicker patients die more, and some of them are given Ivermectin. (Really, that's what the study proves! Not even close to an attempt at an unbiased sample. )
Some mental models cannot be changed, because - well - they're generally a good guide to making decisions. 95% is always better than 67% and you cannot wave this away by saying that one applies to apples and the other to oranges or by saying a 1 in 20 chance is no different than a 1 in 3 chance.
So rather then try to change that model, use a binary model, e.g. if you get a shot of ANY vaccine then you won't die or end up in the hospital.
Parenthetically, a message, to be effective, must be targeted to the mental model of the recipient.
And thank you Zeynep for creating a space to think hard and carefully.
Off topic to where you seem to be going with this piece, yet on topic to your use of mental models, is a report in The Guardian today. It concerns a new study by the Election Integrity Partnership that uses the mental model of “superspreaders” to track the sources and spread of misinformation. https://www.theguardian.com/us-news/2021/mar/05/election-misinformation-trump-rightwing-super-spreader-study?CMP=Share_iOSApp_Other
Fantastic illustration of the immune system. Love it. Also like the explanation of statistical power and sample sizes. I've been struggling with that challenge as well. In a slightly different context, I like this comparison: Imagine you have to find out if a die is 10% off, that is it is 10% more likely to roll a 1. Even throwing the die a 100 times, is not going to make it clear, that the die is skewed, since you are only going to get a couple more 1s than you would expect. Sample size needs to be quite big. But for the vaccine challenge, the MVP scenario is certainly better.
Zeynep and Yong are a fantastic knowledge team.
Can hospitals start a reporting system (to the CDC?) whereby new inpatients being treated for COVID are asked if they have been vaccinated? We presently have almost 50,000 hospitalized patients. Even if only half of the hospitals contributed this information along with patient ages, we would quickly acquire valuable data with good statistical power.
I've been working with a mental model where the immune system of the body works like the public health system of a city or nation. The fever and other immune systems are like the economic shutdowns, and the immune response to a vaccine is like developing a quick and easy virus test. This analogy helps me understand why the size of the initial dose makes a difference to the severity of the illness - a city with 10,000 cases at the time they discovered is going to need to go through a longer and harder lockdown than a city with 1 case that they discovered, in order to clear it. It also helps me understand why a vaccine (or prior infection) would be able to prevent severe illness more effectively than it can prevent any symptomatic illness - even with good tests, Australia and New Zealand occasionally need a brief lockdown, but they never need something like the UK ones.
Obviously there's a lot more disanalogies, but thinking about it this way has seemed helpful to me.