I have always wondered about the mechanisms through which 'pundits' (like Matt) affect, or not, the conversation. It all feels fuzzy and vague to me, and causality is so difficult to attribute.
Question for discussion: how much credit should Slow Boring take - was this 12/18 article amongst the first, was it published when the idea was already in the marketplace, or was it a laggard and thus just added to the weight of the evidence?
Is it too controversial to suggest that boomers have ruined everything and the longer they stay in power the worse the world becomes so we shouldn’t prioritize them in vaccine distribution? The Quintessential Boomer President botched the pandemic response to cause this in the first place. On a pure political power and help the world perspective, the more old people die the better? That seems to be most utilitarian path to take
I love how Matthew says "I know there's lots of ways to look at it but can we please just simplify and move forward" then tons of comments are"yeah but I like my way of looking at it".
This is why policy problems that appear super simple on the surface always get bogged down and then no one is happy. Sometimes we just need to accept some people will be left behind.
I understand the need for simplicity and if we distributed vaccines on this basis, I wouldn’t be mad. However, if we’re talking about who can’t telework (nurse, doctors, grocery store workers, Amazon warehouse workers etc.) we are valorizing them in our rhetoric and asking them to take one for the team. The right thing to do is to deliver a tangible benefit like prioritizing them in vaccinations. Otherwise, we haven’t done anything more than say nice words.
Mostly makes sense, but I think that ppl with no immune system should get it about the same time as the elderly. I'm thinking of relatives and family members actively going through cancer treatment.
I truly don't know the most ethical way to distribute the vaccine. My issue is that I'm a primary care doctor and I am absolutely dreading the arguments/fights/threats and pleading that I foresee coming my way. Of note, at least right now, the group after healthcare workers/nursing home residents and front line workers is "adults with high risk medical conditions and adults 65 and older." Oh boy, that's a large group and what constitutes a "high risk medical condition?" Matt alludes to this in his article: do smokers qualify? slight asthmatics? I already had these kind of fights with people requesting letters for work so they could work from home. I don't want to be party to people using me as an excuse to cut in line, but I project that we primary care providers will be forced into the role of gatekeeper. I hope that when it comes down to group 1c , there is firm clarification from CDC as to what constitutes "high risk" but I doubt it. And then, despite everyone's best intentions to equitably roll out the vaccine, you can bet people will show up with to the pharmacy with a note from their doctor that they qualify. (Think Trump and the bone spurs).
Matt alluded to this, but isn't the biggest problem with the more nuanced, complicated solutions implementation? We're already having serious implementation issues with categories that are simple and the recipients are easy to locate. Is it realistic to think we're going to slice the population into several dozen pieces, then develop a plan to target each priority that won't slow things down? Are we really going to be able to "smart bomb" the distribution problem, or are we better off relying on old-fashioned dumb bombs. Complexity is going to be the enemy of success.
How do we know it saves the most lives? Perhaps we should have each state do a different variation, and gather some good data. Then, when the next pandemic comes, we will really know which method saves the most lives. That probably saves more lives in the long run.
The strongest argument in favor of this simplistic approach is the public's acceptance of the initial allocation to states. Distributing based on population is less optimal than distributing based on disease burden or some index of risk (e.g. FL gets more than UT given higher average age). But the public accepted the simplicity of just distributing by population, even if some experts carped about it. I personally prefer complex, optimized solutions. But public acceptance is so critical to the long-term goal of vaccinating everybody that prioritizing acceptance also prioritizes health impact.
The logistics of prioritizing essential workers are hard to understand. If it's by occupation, do grocery clerks to to a centralized location and just claim that this is their job? Do employers need to send lists of employees? If it's by business, then a bunch of non-essential employees will get vaccinated. My company is an essential business, manufacturing medical products in NY. Vaccinating our production staff sounds great, but how do you limit vaccination to them versus somebody like me who hasn't been to the office in my 5 months at this firm?
I work for a healthcare company. It's up to us to identify the "Phase 1C" high-risk population. Our analytics people are working on this now. One of them e-mailed the other day and basically said, "Of all risk factors, age is - by far - the highest one". I'm almost sure this will be how it goes.
I'm not sure I agree that "saving the most lives" should be the goal, as opposed to saving the most QALYs. If your 95 year old grandmother who likely had 1-2 years left dies of/with Covid, that's sad. If a 35 year old grocery store worker gets Covid and is one of the long-haul patients who suffers permanent lung scarring and brain damage from oxygen deprivation -- not killing her, but making her life much worse, and making her less able to provide for her two young kids for their entire childhoods -- that is a _catastrophe_, even though she isn't dead.
I'm glad you brought this up, I think people are getting a bit heated about _deaths_ when that's not the only criteria that the CDC is using, the ACIP slides from the 19/20th meeting take into account the potentially lifelong health effects of catching Covid which Matt seems to not mention. I realize this thread was opened on the 18th but we've about these health problems for months at this point. The CDC very well could make the wrong decision about how much value is gained by focusing on reducing infection rather then reducing deaths, but it's a valid decision to make and they're right to consider it. (afaik they won't actually make the decision until the meeting on the 20th)
Death rate increases more than linearly with age. In fact increases *exponentially* -- just eyeballing the graph, each increase of 20 years increases the death rate by a factor of 10.
LY's remaining decreases linearly with age.
Therefore we do know, by math, that if we treat the oldest first we will save exponentially more years of life than if we do something else.
It isn't close at all -- the increase LYs saved will be exponentially more if we go oldest to youngest than anything else, by more or less the distance from the strategy implemented to an "oldest to youngest" strategy to whatever exponential one calculates.
No "QA" -- in this calculation. I'm treating a LY for a 70 year old same as a LY for a 25 year old.
But that's just the problem. The quality adjustment _matters_. If a 35 year old was going to have fifty years of health, and instead they will lose ten years, but also spend the other forty years significantly impaired, unable to maintain their career or care for their kids, that is _almost as bad_ as if they just fully lost the fifty years. And don't really have a solid estimate of what % of younger patients live, but have serious damage. Like, I've seen estimates anywhere from 2% to 30%. We probably won't really know for sure until we've had a decade of longitudinal followups.
I've come around to the view that we simply don't have good enough data or good enough administrative resources to solve this problem in a way that wouldn't instantly get gamed, and so age is the best we can do in the real world. But I think if you pretend it isn't a problem, you're not really serious about the ethics of healthcare.
I thought QALYs were being used to discount the remaining years of life of old people, not to add a cost factor on the "young" side of the balance sheet.
Thinking this through...
Even so, due to the exponential increase in death rate, the cost term from impacted LY's on the "young" side of the ledger would have to be huge to make up for the exponential increase in death rate -- the severity and frequency would have to be high.
I think that if young people were basically getting crippled by this disease at a rate comparable to the old person IFR, we'd be hearing about it. I think this is unlikely especially as it's apparent that the disease generally is more severe in old people than young -- frequency of negative long term impacts is likely to follow same trends as the IFR.
Suppose that a 75 year old would've lived to 85, gets a severe case, and drops dead.
Suppose a 35 year old would've lived to 85, gets a severe case, and then lives a significantly-worse life and dies at 75.
The cost at the tail end of life is identical, but then you _also_ have to account for the way everyone connected to that 35 year old has a worse life for the interening forty years. The 35 year old doesn't have to be "crippled" for long-term lung, heart, and mental problems to impose a _huge_ cost on society. Look at the enormous gains we see when we do stuff like just reduce the impact of air pollution on kids ( https://www.vox.com/2020/1/8/21051869/indoor-air-pollution-student-achievement ). Having one of your parents go from healthy and active, to suffering chronic asthma and brain fogs, will have significant knock-on effects.
And then add in the fact that the the young person is more likely to also transmit the disease -- old people outside the nursing home context generally have an easier time reducing social contacts.
If in fact it is true that the distribution of "severe cases" is identical among the old and young, but "severe cases" for the young are much less fatal in the short term, then the total social cost of infections among the young is _obviously_ much higher.
I mean, you're right of course. But the age effects are SO strong that they mostly overwhelm this kind of thing. It'd be one thing if the IFR for a 35 year old were 0.1% and the IFR for a 60 year old were 0.2%. But it's not. The IFR for a 35 year old is 0.05% and the 60 year old is 0.5% -- and the 95 year old grandma is who knows, quite possibly 20%. Even adjusting for QALYS, you'll *mostly* approximate the same "the older you are, the higher priority you should be," and the differences are small enough that adding the complexity and potential for unfairness aren't worthwhile.
I'm not sure we really _know_ what the comparative QALY loss is. Infection Fatality Rate is not useful for assessing that. Yes, only 0.05% of the healthy 35 year olds _die_. What % of the healthy 35 year olds experience serious long-term problems, though? Do we have that well-established even, yet? If one 35 year old who otherwise would've lived 60 healthy years gets, instead, _zero_ more _healthy_ years -- and is instead unhealthy and dealing with expensive problems for the rest of her life, and dies a decade or two earlier than she would've -- that is arguably _worse_ than one 85 year old dying who o/w would've lived in middling health to 90 or 95.
I think it is fair to argue, though, that getting this right is too hard a problem to solve in the next week or two, and anything we could cobble together would be subject to gaming by people with connections. So probably the age rule is in-practice better than anything we could actually come up with.
I agree that we don't know for certain, but there's very little evidence that suggests that "severe cases that might lead to highly reduced quality of life" are distributed any differently from "severe cases that might lead to death," and lots of good prima facia reasons to believe that they aren't.
I disagree strongly with your last clause. My impression is that the evidence suggests that the difference between the old and the young is more or less that what's rising exponentially with age is the likelihood that a severe case kills you, instead of just damaging you a lot but leaving you alive. The percentage of cases that are severe, and leave lasting evidence of damage to the heart, lungs, and cranial vasculature, is high even in the young -- possibly as high as 30%.
As another epicycle, I'd also add that outside the nursing home setting, the 35 year old grocery store worker is _much_ more likely than an 85 year old who has secure private housing to pass the virus on -- to their kids, and to other workers -- so in theory you should be pricing that into your model as well. But again, this means more uncertainty and complication.
I feel quite certain that it would be better / more-just to vaccinate at least some essential workers ahead of elderly who have secure housing and can afford to have food and whatnot delivered to them. I just am not sure we have any way of clarifying _which_ essential workers, and who among the elderly counts as secure.
"Should clergy be considered eligible so they can minister to their flocks in person?" Nothing says ethical like gathering people to spread viruses while you yourself are immunized!
Color me dubious. I'm retired, approaching 67, and am happy to continue hunkering down for months to let other people go first. But to go to the more abstract, I'm not sure I accept the premise that protecting life, especially of the aged, should necessarily be the highest priority, nor is vaccinating them the best way to pursue it. I might target nursing home workers first, as those most likely to bring COVID into those homes. And I would much prefer we vaccinate teachers early so all excuses for keeping schools closed (especially for poorer students) are eliminated. How much more lifelong damage do we want these kids to suffer?
And I also wonder if we're overthinking this prioritization thing. We could get *lots* of people vaccinated in the next few months; we may see significant decreases in outbreak earlier than we might anticipate. By the time we've worked through all the intricacies of how to best prioritize, we may be well on the road to wide population coverage anyway.
And lastly, I'd prefer the rules states are using to remain fairly vague. I would hate to see those decisions become polarizing; big fights are the last thing we need, especially as the herd immunity goal is maybe just a few months away.
I have always wondered about the mechanisms through which 'pundits' (like Matt) affect, or not, the conversation. It all feels fuzzy and vague to me, and causality is so difficult to attribute.
With that caveat, reading this: https://www.axios.com/trump-coronavirus-vaccines-seniors-doses-faster-ca22e892-71d4-4e04-9219-b41fec799524.html and remembering the post above from Dec 18 made me sit up and go "hey, maybe people making decisions DO read thoughtful 'pundits' sometimes and it does sometimes makes a difference!"
Question for discussion: how much credit should Slow Boring take - was this 12/18 article amongst the first, was it published when the idea was already in the marketplace, or was it a laggard and thus just added to the weight of the evidence?
Is it too controversial to suggest that boomers have ruined everything and the longer they stay in power the worse the world becomes so we shouldn’t prioritize them in vaccine distribution? The Quintessential Boomer President botched the pandemic response to cause this in the first place. On a pure political power and help the world perspective, the more old people die the better? That seems to be most utilitarian path to take
Are we going to get a lot of regression discontinuity design job market papers based on differences in state vaccination plans?
I love how Matthew says "I know there's lots of ways to look at it but can we please just simplify and move forward" then tons of comments are"yeah but I like my way of looking at it".
This is why policy problems that appear super simple on the surface always get bogged down and then no one is happy. Sometimes we just need to accept some people will be left behind.
I understand the need for simplicity and if we distributed vaccines on this basis, I wouldn’t be mad. However, if we’re talking about who can’t telework (nurse, doctors, grocery store workers, Amazon warehouse workers etc.) we are valorizing them in our rhetoric and asking them to take one for the team. The right thing to do is to deliver a tangible benefit like prioritizing them in vaccinations. Otherwise, we haven’t done anything more than say nice words.
Mostly makes sense, but I think that ppl with no immune system should get it about the same time as the elderly. I'm thinking of relatives and family members actively going through cancer treatment.
The barrage of articles for months now about the disproportionate harm to minorities are now bearing fruit.
https://www.nytimes.com/2020/12/05/health/covid-vaccine-first.html
I truly don't know the most ethical way to distribute the vaccine. My issue is that I'm a primary care doctor and I am absolutely dreading the arguments/fights/threats and pleading that I foresee coming my way. Of note, at least right now, the group after healthcare workers/nursing home residents and front line workers is "adults with high risk medical conditions and adults 65 and older." Oh boy, that's a large group and what constitutes a "high risk medical condition?" Matt alludes to this in his article: do smokers qualify? slight asthmatics? I already had these kind of fights with people requesting letters for work so they could work from home. I don't want to be party to people using me as an excuse to cut in line, but I project that we primary care providers will be forced into the role of gatekeeper. I hope that when it comes down to group 1c , there is firm clarification from CDC as to what constitutes "high risk" but I doubt it. And then, despite everyone's best intentions to equitably roll out the vaccine, you can bet people will show up with to the pharmacy with a note from their doctor that they qualify. (Think Trump and the bone spurs).
Matt alluded to this, but isn't the biggest problem with the more nuanced, complicated solutions implementation? We're already having serious implementation issues with categories that are simple and the recipients are easy to locate. Is it realistic to think we're going to slice the population into several dozen pieces, then develop a plan to target each priority that won't slow things down? Are we really going to be able to "smart bomb" the distribution problem, or are we better off relying on old-fashioned dumb bombs. Complexity is going to be the enemy of success.
My goodness, Matt. You are such a clear, sober thinker.
How do we know it saves the most lives? Perhaps we should have each state do a different variation, and gather some good data. Then, when the next pandemic comes, we will really know which method saves the most lives. That probably saves more lives in the long run.
The strongest argument in favor of this simplistic approach is the public's acceptance of the initial allocation to states. Distributing based on population is less optimal than distributing based on disease burden or some index of risk (e.g. FL gets more than UT given higher average age). But the public accepted the simplicity of just distributing by population, even if some experts carped about it. I personally prefer complex, optimized solutions. But public acceptance is so critical to the long-term goal of vaccinating everybody that prioritizing acceptance also prioritizes health impact.
The logistics of prioritizing essential workers are hard to understand. If it's by occupation, do grocery clerks to to a centralized location and just claim that this is their job? Do employers need to send lists of employees? If it's by business, then a bunch of non-essential employees will get vaccinated. My company is an essential business, manufacturing medical products in NY. Vaccinating our production staff sounds great, but how do you limit vaccination to them versus somebody like me who hasn't been to the office in my 5 months at this firm?
Simplicity wins.
I work for a healthcare company. It's up to us to identify the "Phase 1C" high-risk population. Our analytics people are working on this now. One of them e-mailed the other day and basically said, "Of all risk factors, age is - by far - the highest one". I'm almost sure this will be how it goes.
I'm not sure I agree that "saving the most lives" should be the goal, as opposed to saving the most QALYs. If your 95 year old grandmother who likely had 1-2 years left dies of/with Covid, that's sad. If a 35 year old grocery store worker gets Covid and is one of the long-haul patients who suffers permanent lung scarring and brain damage from oxygen deprivation -- not killing her, but making her life much worse, and making her less able to provide for her two young kids for their entire childhoods -- that is a _catastrophe_, even though she isn't dead.
I'm glad you brought this up, I think people are getting a bit heated about _deaths_ when that's not the only criteria that the CDC is using, the ACIP slides from the 19/20th meeting take into account the potentially lifelong health effects of catching Covid which Matt seems to not mention. I realize this thread was opened on the 18th but we've about these health problems for months at this point. The CDC very well could make the wrong decision about how much value is gained by focusing on reducing infection rather then reducing deaths, but it's a valid decision to make and they're right to consider it. (afaik they won't actually make the decision until the meeting on the 20th)
Death rate increases more than linearly with age. In fact increases *exponentially* -- just eyeballing the graph, each increase of 20 years increases the death rate by a factor of 10.
LY's remaining decreases linearly with age.
Therefore we do know, by math, that if we treat the oldest first we will save exponentially more years of life than if we do something else.
It isn't close at all -- the increase LYs saved will be exponentially more if we go oldest to youngest than anything else, by more or less the distance from the strategy implemented to an "oldest to youngest" strategy to whatever exponential one calculates.
No "QA" -- in this calculation. I'm treating a LY for a 70 year old same as a LY for a 25 year old.
But that's just the problem. The quality adjustment _matters_. If a 35 year old was going to have fifty years of health, and instead they will lose ten years, but also spend the other forty years significantly impaired, unable to maintain their career or care for their kids, that is _almost as bad_ as if they just fully lost the fifty years. And don't really have a solid estimate of what % of younger patients live, but have serious damage. Like, I've seen estimates anywhere from 2% to 30%. We probably won't really know for sure until we've had a decade of longitudinal followups.
I've come around to the view that we simply don't have good enough data or good enough administrative resources to solve this problem in a way that wouldn't instantly get gamed, and so age is the best we can do in the real world. But I think if you pretend it isn't a problem, you're not really serious about the ethics of healthcare.
I thought QALYs were being used to discount the remaining years of life of old people, not to add a cost factor on the "young" side of the balance sheet.
Thinking this through...
Even so, due to the exponential increase in death rate, the cost term from impacted LY's on the "young" side of the ledger would have to be huge to make up for the exponential increase in death rate -- the severity and frequency would have to be high.
I think that if young people were basically getting crippled by this disease at a rate comparable to the old person IFR, we'd be hearing about it. I think this is unlikely especially as it's apparent that the disease generally is more severe in old people than young -- frequency of negative long term impacts is likely to follow same trends as the IFR.
Suppose that a 75 year old would've lived to 85, gets a severe case, and drops dead.
Suppose a 35 year old would've lived to 85, gets a severe case, and then lives a significantly-worse life and dies at 75.
The cost at the tail end of life is identical, but then you _also_ have to account for the way everyone connected to that 35 year old has a worse life for the interening forty years. The 35 year old doesn't have to be "crippled" for long-term lung, heart, and mental problems to impose a _huge_ cost on society. Look at the enormous gains we see when we do stuff like just reduce the impact of air pollution on kids ( https://www.vox.com/2020/1/8/21051869/indoor-air-pollution-student-achievement ). Having one of your parents go from healthy and active, to suffering chronic asthma and brain fogs, will have significant knock-on effects.
And then add in the fact that the the young person is more likely to also transmit the disease -- old people outside the nursing home context generally have an easier time reducing social contacts.
If in fact it is true that the distribution of "severe cases" is identical among the old and young, but "severe cases" for the young are much less fatal in the short term, then the total social cost of infections among the young is _obviously_ much higher.
I mean, you're right of course. But the age effects are SO strong that they mostly overwhelm this kind of thing. It'd be one thing if the IFR for a 35 year old were 0.1% and the IFR for a 60 year old were 0.2%. But it's not. The IFR for a 35 year old is 0.05% and the 60 year old is 0.5% -- and the 95 year old grandma is who knows, quite possibly 20%. Even adjusting for QALYS, you'll *mostly* approximate the same "the older you are, the higher priority you should be," and the differences are small enough that adding the complexity and potential for unfairness aren't worthwhile.
***I mean, you're right of course. But the age effects are SO strong that they mostly overwhelm this kind of thing.***
Not only that, but "QALY" calculus gets overwhelmingly complicated, quickly, and we need a strategy now.
I'm not sure we really _know_ what the comparative QALY loss is. Infection Fatality Rate is not useful for assessing that. Yes, only 0.05% of the healthy 35 year olds _die_. What % of the healthy 35 year olds experience serious long-term problems, though? Do we have that well-established even, yet? If one 35 year old who otherwise would've lived 60 healthy years gets, instead, _zero_ more _healthy_ years -- and is instead unhealthy and dealing with expensive problems for the rest of her life, and dies a decade or two earlier than she would've -- that is arguably _worse_ than one 85 year old dying who o/w would've lived in middling health to 90 or 95.
I think it is fair to argue, though, that getting this right is too hard a problem to solve in the next week or two, and anything we could cobble together would be subject to gaming by people with connections. So probably the age rule is in-practice better than anything we could actually come up with.
I agree that we don't know for certain, but there's very little evidence that suggests that "severe cases that might lead to highly reduced quality of life" are distributed any differently from "severe cases that might lead to death," and lots of good prima facia reasons to believe that they aren't.
I disagree strongly with your last clause. My impression is that the evidence suggests that the difference between the old and the young is more or less that what's rising exponentially with age is the likelihood that a severe case kills you, instead of just damaging you a lot but leaving you alive. The percentage of cases that are severe, and leave lasting evidence of damage to the heart, lungs, and cranial vasculature, is high even in the young -- possibly as high as 30%.
As another epicycle, I'd also add that outside the nursing home setting, the 35 year old grocery store worker is _much_ more likely than an 85 year old who has secure private housing to pass the virus on -- to their kids, and to other workers -- so in theory you should be pricing that into your model as well. But again, this means more uncertainty and complication.
I feel quite certain that it would be better / more-just to vaccinate at least some essential workers ahead of elderly who have secure housing and can afford to have food and whatnot delivered to them. I just am not sure we have any way of clarifying _which_ essential workers, and who among the elderly counts as secure.
"Should clergy be considered eligible so they can minister to their flocks in person?" Nothing says ethical like gathering people to spread viruses while you yourself are immunized!
Color me dubious. I'm retired, approaching 67, and am happy to continue hunkering down for months to let other people go first. But to go to the more abstract, I'm not sure I accept the premise that protecting life, especially of the aged, should necessarily be the highest priority, nor is vaccinating them the best way to pursue it. I might target nursing home workers first, as those most likely to bring COVID into those homes. And I would much prefer we vaccinate teachers early so all excuses for keeping schools closed (especially for poorer students) are eliminated. How much more lifelong damage do we want these kids to suffer?
And I also wonder if we're overthinking this prioritization thing. We could get *lots* of people vaccinated in the next few months; we may see significant decreases in outbreak earlier than we might anticipate. By the time we've worked through all the intricacies of how to best prioritize, we may be well on the road to wide population coverage anyway.
And lastly, I'd prefer the rules states are using to remain fairly vague. I would hate to see those decisions become polarizing; big fights are the last thing we need, especially as the herd immunity goal is maybe just a few months away.
But boy do I agree that we need to keep rich assholes from jumping the line, as I read in today's LA Times: https://www.latimes.com/california/story/2020-12-18/wealthy-patients-scramble-covid-19-vaccine