Stopping transmission will give a bigger bang for the buck. The overall deaths may actually be less by focusing on transmission first. Paying healthy, active people to take the vaccine may also be advantageous, and overcome the reluctance of the people most likely to be spreaders.
I don't think so, numerically, in the short term. If we could all the transmitters in the first month, that would be an interesting calculation. But we will have, approximately, 50 million by the end of January (if everything goes right). That's not enough to stop transmission (and transmission effects are still being studied and are unclear; it will likely stop some but not all transmission) but it is enough to vaccinate everyone over 65.
The current vaccines were approved on the basis of blocking disease symptoms, not blocking transmission. We don't know if they block transmission and so therefore it's not prudent to distribute them based primarily on transmission-blocking benefits
How do you identify the key vertices? I saw a recent article about using the Friend's Paradox (on average, your friends have more friends that you), but all of this seems impractical and fraught with (as Zeynep says) fraught with lobbying, etc. I like the approach of just doing it (after the first round) by age. Maybe not optimal, but it can be done faster/quicker. But I do recognize there is a lot of power in your approach.
Under an assumption (reasonable, but to my knowledge no studies asked this) that the vaccine is effective at making one not contagious in addition to being effective at preventing infection, the greatest effect in reducing spread would be among those who have the most contact with the most other people. If half (time/closeness weighted) of one's interpersonal contacts are vaccinated, that would reasonably be expected to reduce one's chance of being infected by half, i.e. one'd get half the effect of the vaccine without being vaccinated.
Combine with the longer-term effects of not dying from CV19 being uncertain (there definitely seems to be some risk of long-lasting consequences), and I think the argument for going for R-reduction (which has the side effect of reducing death) is stronger than trying to minimize deaths.
Granted, the process of identifying who are the key vertices in contact graphs (and thus those for whom vaccination would have the greatest effect on spread) is difficult to impossible to determine exactly, and in only some cases matches the lobbying (I would think that flight attendants and restaurant/hospitality workers would be high up: I'm not sure that firefighters would qualify). There's also perhaps a defensible argument that, say, the 5 million "most key" vertices are also the most likely to have had it already, though the evidence against any herd immunity having been reached argues against that.
I agree with Thaddeus and Levi here; if transmission is significantly reduced (thinking here mainly about in-house transmission but also about lack of adequate protection/conditions for some essential workers) then the question of who should get in it a nursing home (the worker or the resident?) becomes more complicated.
Except that not great evidence that the vaccine stops spread, so those transmitters would still be transmitting (haters gotta hate). Moderna has better data for stopping trans than Phizer.
Yeah, there are unknown there and officially speaking, we don't have evidence that it stops transmission (in that it cannot be an official claim). Unofficially, I'd be surprised that if both vaccines do not dampen transmission to a great degree, when we get around to more precisely measuring it. (Which goes a bit against my argument for prioritizing the elderly, but I still think the severity of the disease is so exponentially dependent on age that we should do that. Plus, the elderly definitely seem to transmit more--probably because they shed more due to higher viral loads).
When the UK began to vaccinate according to risk I asked myself a similar question. Is there more power in using the vaccine as a tool to break chains of transmission between the young and socially mobile, or should it be wielded as a preventer of death and suffering in the elderly and at risk? In my mind, as the son of a firefighter, it seems at first sensible to throw water on the raging flames so to speak; and vaccinate those with high degrees of social contact. However, like you and countless other folks mention, the rate of death in long term care and hospitals in the elderly age groups in stunning. Care homes are dangerous places to be. I’m glad that my country is taking the risk based approach, and I can only hope that the US manages to take enough of the same path that the behaviour of a strategy like that becomes emergent. Do you think that a mix of the two in differing municipalities depending on varying population age factors and trends in viral behaviour might show some interesting results? Or will it all be terrible as we’re becoming accustomed to?
Indeed. The shortage is real, and targeting transmitter population is a great idea I think once we get the mortality/hospitalization under control and that clearly points at age as the key factor to take into consideration.
Targeting transmission at some point in the future would be great, but first would't we need a vaccine that stops transmission? The current vaccine phase 3 studies looked at their ability to stop disease symptoms, not stop transmission . . .
If one were developing an algorithm for prioritizing different groups, two things would be required: clearly defining the outcome measure one wishes to maximize, roughly "potential spreader" vs. "medical vulnerability of the individual", and then the factors predicitive of same.
If focusing on medical vulnerability, some comments implicitly argue for, approximately: maximizing "years of quality of life" vs. "lives saved". (Setting aside for the moment the fraught ethical issue of whether anyone can ever assess someone else's quality of life... )
Counterintuitively, maximizing years of quality of life may argue *for* continued weighting of age.
As people get older, on average, there are more and more things for which *in person* medical appointments are necessary on a fairly regular basis. For some conditions, there may be irreversible cost to additional postponement of those appointments another half year or so, e.g. incipient glaucoma, heart disease.
In this case, age functions as a very imperfect, population-level proxy for the time-sensitive nature of such care. Imperfect, objective, population-level proxies are of high value when there's a need to quickly prioritize on the "wholesale vs retail" level, which is the nature of public health.
However, there are several other large groups for whom continued postponement of in-person appointments has a high cost:
+People with chronic and/or progressive illnesses which do not fall into the narrow definitions of the few "additional conditions" which decrease the survival rate if Covid were contracted, e.g., auto-immune disorders.
+People with a strong family history of certain cancers or progressive neurological illness
+ People in the middle of cancer treatments.
Age isn't a great proxy for these. Taking them into account would either require an inevitably incomplete list of such conditions, or else input from individual clinicians. The latter might be seen as too subject to favoritism (and is biased against those w/o a relationship w/ a particular clinician/clinic) ; the former might lead to protracted disputes about inclusion/exclusion of specific conditions. But it would still be worth trying to consider these groups in assigning priority.
Perhaps the biggest demographic variable that ought to bump people forward in the line is being a person of color, and/or, poor. These highly correlate with all three major outcome measures: having a public-facing job which increases the chance of being a spreader; and at the individual-harm level, both the likelihood of getting Covid in the first place, as well as decreased survival rates once illness is contracted.
It's difficult to imagine having a sensible public discussion of race as a priority factor at this point in time. But it's past time.
They have gone to a lot of effort in their APIC documents to create analyses output slides which purport to show that prioritizing the 65+ population before essential workers doesn't make a difference (though their analysis does not actually model this prioritization and therefore does not prove this), so you'll be fighting an uphill battle with CDC officials who only have a cursory knowledge of the work that was done and will just say "our analysis showed there isn't a benefit from prioritizing people who are 65+" -- and those officials may really think there's a high quality analysis that shows that, even though there isn't.
The question of whether to initially prioritize those who are categorically at higher risk of death vs. those who are more likely to spread illness invites thoughtful inquiry.
But the whole, careful structure of how to achieve the greatest reduction in morbidity and mortality depends on widespread trust that patiently waiting your turn won't amount to self-harm, if your turn continues to recede while the estimates of vaccine availability decrease. Only if the rollout is perceived as fair and predictable, is it rational to be patient.
All this disquiet about who should go first-- either because of a different opinions re weighting of variables or because of fear for oneself and one's cohort-- was made much worse when 45 & co inexplicably cut the initially available supply by half: "as recently as October, federal officials had turned down [Pfizer's] entreaties to lock in another 100 million doses", only to beg for those doses after they were committed elsewhere. It was the easiest way possible to achieve more widespread vaccination more quickly, but they didn't do it. If you have twice as many doses, the questions about priority order are immediately less fraught.
Today, we read that states had the number of doses promised in the current wave cut by as much as half, and "States report confusion as government reduces vaccine shipments, while Pfizer says it has ‘millions’ of unclaimed doses". It's hard to know which is more disturbing: attributing this to malice or incompetence on the administration's part.
We are back to the same crucial point: none of this works without trust in government. And that's been deliberately besmirched in this country.
I will continue to wait patiently, because I deeply believe in the common good. And I certainly never expected something of this magnitude and novelty to go completely smoothly. But I must admit that after the recent reports of 45&co actively making it worse, my "reptile brain"-- which I will overridge-- is clamoring for me to get out there and try to hustle a place in line.
The problem is that it is not just age that leads to increased risk, in some counties in the US a 60 year old Black or Latino man may have more risk of dying than a 75 year old white woman. And by targeting just those who are older you might end up disproportionately giving the vaccine to wealthier white people first. I do agree that those who live in LTCFs should be first on the list as we have seen they are at much greater risk than other groups.
I think LTCR are a clear priority, and everyone seems to easily agree on that given the horrific numbers. There are some gender differences, but the research I've seen says race differences are coming down not to biology but circumstances and comorbidities—which are also often tied up with circumstances. (Less like Tay–Sachs). People with co-morbidities are at higher risk, for sure but the age factor still overwhelms comorbidities. However, as we go down younger ages, comorbidities would/should certainly get priority. Just by age, someone who is above 75 has about 30 times the risk of dying compared the risk of someone in the 50-64 group (rough calculation will look up numbers). That's why I like the UK list.
It's less clear to me that death should be the only factor. I think we're crippling a sizable chunk of our population with "long COVID". If you try to take into account quality of life loss, then I think we just don't know enough to say what the best solution is. I think nursing homes should be a priority (COVID is very dangerous there, and logistically easy to do as well, and should include residents and workers), but past that it's less clear. So a police officer will get COVID in March so that an 80 year old person getting groceries delivered can get vaccinated first? I suspect we'll end up with "a little of everything", and split vaccines between essential workers and elderly, doing both at once, thereby slowing down completing each group. And each state will do it slightly differently.
I completely agree on studying long COVID and the aftermath of this virus, but those 80 year-olds have been severely isolated for almost a year now, and I think there's a quality-of-life argument there, too. There is no option here without trade-offs. Also, hospitals are under strain because of age (it's mostly 65+ people who get the severe disease), even though younger cases definitely exist (and somewhat understandably get a lot more attention).
I took a scan of the effects of prioritization paper and I did not see mention if they excluded care residents from their analysis. This seems relevant because most of the elder deaths have been in care homes (eg in BC 80% of the deaths over 70 were in care facilities). Since, as you note, care facility residents are being prioritized by everyone, it’s important to exclude their statistics from further analysis.
A peripheral point: I don't think it is quite accurate to say that the Marshmallow effects disappeared when controlled for various factors. It is true that a Psych Science study claimed this, but a follow up commentary was quick to point out that if one controls for everything, one can make an effect go away (cf. bad controls). See https://medicalxpress.com/news/2019-12-reanalysis-famous-marshmallow.html A snippet: "They tried to account for so many effects that it becomes impossible to interpret what these effects are telling us about the real relation between early self-control and later success." Falk, Kosse and Pinger have now performed a similar analysis. Crucially, however, they controlled only for confounding factors that could be clearly interpreted as such. Their re-examination of the data suggests that the replication study actually reveals a relatively strong correlation between readiness to delay gratification and subsequent scholastic success."
But yes, surely true that the effect is interconnected with society / the person who offers the Marshmallow.
Hi Zeynep, as always thanks for you excellent writing and analyses. Your intro on hope and your experiences in Turkey reminded me of a Jim Collin's passage in his "Good to Great" on what he called the Stockdale Paradox (https://innovationandcreativityinstitute.com/stockdale-paradox/):
“You must never confuse faith that you will prevail in the end —which you can never afford to lose —with the discipline to confront the most brutal facts of your current reality,
whatever they might be.” - James Stockdale
Stockdale said that the people who did not make it out of the prisons were the optimists, those who thought they were getting out soon. I have seen so many parallels today the past 9 months with the pandemic and people losing faith/getting angry/getting depressed.
If you look at slides 31-33 of the ACIP presentation, it is clear that the ACIP recommendations are mostly based on purely subjective determinations that they classify as “Ethics.” I see little scientific or technical justification for their proposed course of action, which seems mostly designed to appeal to a perceived audience.
Is it true that 10% of people w COVID end up as "long haulers"? The determination of who goes first is a difficult one and I'm not advocating for a different priority system. However, I do think that we should consider the impact, both personally and societally, of long haulers in our thinking. The amount of chronic disease and disability that will follow in the wake of this virus is distressing.
Isn't there a way around these ideas by creating a surplus of vaccine? As Stephen Beranyi suggested in the NYT yesterday (The Covid-19 Vaccine Doesn't Mean Big Pharma Is Your Savior https://nyti.ms/37o6zq1) can't we force the opening up of Covid vaccine creation to other companies and deliver a large enough quantity to vaccinate everyone - elderly and essential workers within less than 100 days? I'd love to see an elected leader like Biden act for the mass of people instead of weighing lobbyist-led insights for various populations.
I just co-wrote a piece for the NYT, based on that single-dose trial idea. https://www.nytimes.com/2020/12/18/opinion/coronavirus-vaccine-doses.html Yes, solving the vaccine shortage would obviously the best way forward, but we'd still face the question of what to do this month. We have 50 million doses now.
I might be wrong but it looks like from the CDC APIC website that they're basing their recommendation off of old (and now faulty) assumptions in an Oct. 30th analysis that assumed that the vaccine (a) would be only 35% effective in those 65+; and that (b) the vaccine would stop transmission [which as I understand it may or may not be the case, we don't know yet]. The 11/23 CDC APIC slide deck incorporates its quantitative analysis from a 10/30 epidemiological analysis that lists these as its assumptions on pages 7 and 8. On the CDC website at https://www.cdc.gov/vaccines/acip/meetings/downloads/slides-2020-10/COVID-Biggerstaff.pdf
Edit: There's a statement in the 11/23 CDC APIC slide deck saying they updated the vaccine efficacy assumption to 90% . . . though looking at the 10/30 document, I can't shake the feeling that something is very off about the analysis in it . . . it's the same feeling I get at work when I first realize someone's trying to use Powerpoint to hide something from me
I do too but it might not matter, because their 10/30 document scenario assumptions barely prioritize any group, and the outputs are used in later APIC docs to misleadingly argue that group prioritization doesn't have an impact on infections or deaths. Specifically, the models they run in the 10/30 doc do 20mm healthcare workers first, then either (a) 20 mm 65+; or (b) 20 mm essential workers; or (c) 20 mm high risk people. In each scenario they then give the next 160mm vaccinations broadly with nor prioritization between 80mm essential workers, 55 mm 65+, etc. Despite this scenario (a) leads to double the decline in deaths in their own analysis, but they use bar charts that obscure the magnitude of difference, and bury that page towards the back of the slide deck. They never show an analysis that totally prioritizes all 55m 65+, and if they did, even leaving the rest of their model numbers/assumptions the same, I suspect it would show a massive decrease in potential deaths.
Edit: They also (egregiously!) in their analysis emphasized scenarios where the vaccine is "infection-blocking" even though what we have are "disease-blocking" vaccines. In the "disease-blocking" scenario, age-prioritization saved more lives . . .
Maybe it is the risk of transmission that we should focus on, and not the risk of death.
Eventually, yes but why that first?
Stopping transmission will give a bigger bang for the buck. The overall deaths may actually be less by focusing on transmission first. Paying healthy, active people to take the vaccine may also be advantageous, and overcome the reluctance of the people most likely to be spreaders.
I don't think so, numerically, in the short term. If we could all the transmitters in the first month, that would be an interesting calculation. But we will have, approximately, 50 million by the end of January (if everything goes right). That's not enough to stop transmission (and transmission effects are still being studied and are unclear; it will likely stop some but not all transmission) but it is enough to vaccinate everyone over 65.
I always prefer the lpng-term over the short term.
The current vaccines were approved on the basis of blocking disease symptoms, not blocking transmission. We don't know if they block transmission and so therefore it's not prudent to distribute them based primarily on transmission-blocking benefits
How do you identify the key vertices? I saw a recent article about using the Friend's Paradox (on average, your friends have more friends that you), but all of this seems impractical and fraught with (as Zeynep says) fraught with lobbying, etc. I like the approach of just doing it (after the first round) by age. Maybe not optimal, but it can be done faster/quicker. But I do recognize there is a lot of power in your approach.
Under an assumption (reasonable, but to my knowledge no studies asked this) that the vaccine is effective at making one not contagious in addition to being effective at preventing infection, the greatest effect in reducing spread would be among those who have the most contact with the most other people. If half (time/closeness weighted) of one's interpersonal contacts are vaccinated, that would reasonably be expected to reduce one's chance of being infected by half, i.e. one'd get half the effect of the vaccine without being vaccinated.
Combine with the longer-term effects of not dying from CV19 being uncertain (there definitely seems to be some risk of long-lasting consequences), and I think the argument for going for R-reduction (which has the side effect of reducing death) is stronger than trying to minimize deaths.
Granted, the process of identifying who are the key vertices in contact graphs (and thus those for whom vaccination would have the greatest effect on spread) is difficult to impossible to determine exactly, and in only some cases matches the lobbying (I would think that flight attendants and restaurant/hospitality workers would be high up: I'm not sure that firefighters would qualify). There's also perhaps a defensible argument that, say, the 5 million "most key" vertices are also the most likely to have had it already, though the evidence against any herd immunity having been reached argues against that.
I agree with Thaddeus and Levi here; if transmission is significantly reduced (thinking here mainly about in-house transmission but also about lack of adequate protection/conditions for some essential workers) then the question of who should get in it a nursing home (the worker or the resident?) becomes more complicated.
In a nursing home, it should be both.
Except that not great evidence that the vaccine stops spread, so those transmitters would still be transmitting (haters gotta hate). Moderna has better data for stopping trans than Phizer.
Yeah, there are unknown there and officially speaking, we don't have evidence that it stops transmission (in that it cannot be an official claim). Unofficially, I'd be surprised that if both vaccines do not dampen transmission to a great degree, when we get around to more precisely measuring it. (Which goes a bit against my argument for prioritizing the elderly, but I still think the severity of the disease is so exponentially dependent on age that we should do that. Plus, the elderly definitely seem to transmit more--probably because they shed more due to higher viral loads).
When the UK began to vaccinate according to risk I asked myself a similar question. Is there more power in using the vaccine as a tool to break chains of transmission between the young and socially mobile, or should it be wielded as a preventer of death and suffering in the elderly and at risk? In my mind, as the son of a firefighter, it seems at first sensible to throw water on the raging flames so to speak; and vaccinate those with high degrees of social contact. However, like you and countless other folks mention, the rate of death in long term care and hospitals in the elderly age groups in stunning. Care homes are dangerous places to be. I’m glad that my country is taking the risk based approach, and I can only hope that the US manages to take enough of the same path that the behaviour of a strategy like that becomes emergent. Do you think that a mix of the two in differing municipalities depending on varying population age factors and trends in viral behaviour might show some interesting results? Or will it all be terrible as we’re becoming accustomed to?
Indeed. The shortage is real, and targeting transmitter population is a great idea I think once we get the mortality/hospitalization under control and that clearly points at age as the key factor to take into consideration.
Targeting transmission at some point in the future would be great, but first would't we need a vaccine that stops transmission? The current vaccine phase 3 studies looked at their ability to stop disease symptoms, not stop transmission . . .
If one were developing an algorithm for prioritizing different groups, two things would be required: clearly defining the outcome measure one wishes to maximize, roughly "potential spreader" vs. "medical vulnerability of the individual", and then the factors predicitive of same.
If focusing on medical vulnerability, some comments implicitly argue for, approximately: maximizing "years of quality of life" vs. "lives saved". (Setting aside for the moment the fraught ethical issue of whether anyone can ever assess someone else's quality of life... )
Counterintuitively, maximizing years of quality of life may argue *for* continued weighting of age.
As people get older, on average, there are more and more things for which *in person* medical appointments are necessary on a fairly regular basis. For some conditions, there may be irreversible cost to additional postponement of those appointments another half year or so, e.g. incipient glaucoma, heart disease.
In this case, age functions as a very imperfect, population-level proxy for the time-sensitive nature of such care. Imperfect, objective, population-level proxies are of high value when there's a need to quickly prioritize on the "wholesale vs retail" level, which is the nature of public health.
However, there are several other large groups for whom continued postponement of in-person appointments has a high cost:
+People with chronic and/or progressive illnesses which do not fall into the narrow definitions of the few "additional conditions" which decrease the survival rate if Covid were contracted, e.g., auto-immune disorders.
+People with a strong family history of certain cancers or progressive neurological illness
+ People in the middle of cancer treatments.
Age isn't a great proxy for these. Taking them into account would either require an inevitably incomplete list of such conditions, or else input from individual clinicians. The latter might be seen as too subject to favoritism (and is biased against those w/o a relationship w/ a particular clinician/clinic) ; the former might lead to protracted disputes about inclusion/exclusion of specific conditions. But it would still be worth trying to consider these groups in assigning priority.
Perhaps the biggest demographic variable that ought to bump people forward in the line is being a person of color, and/or, poor. These highly correlate with all three major outcome measures: having a public-facing job which increases the chance of being a spreader; and at the individual-harm level, both the likelihood of getting Covid in the first place, as well as decreased survival rates once illness is contracted.
It's difficult to imagine having a sensible public discussion of race as a priority factor at this point in time. But it's past time.
Do you have a way to get this argument to some of the relevant CDC people?
I don't know. It's one of those harder to get through topics but I'm thinking about what else to do.
They have gone to a lot of effort in their APIC documents to create analyses output slides which purport to show that prioritizing the 65+ population before essential workers doesn't make a difference (though their analysis does not actually model this prioritization and therefore does not prove this), so you'll be fighting an uphill battle with CDC officials who only have a cursory knowledge of the work that was done and will just say "our analysis showed there isn't a benefit from prioritizing people who are 65+" -- and those officials may really think there's a high quality analysis that shows that, even though there isn't.
Germany is also going to vaccinate high-risk groups first: https://www.rki.de/DE/Content/Infekt/EpidBull/Archiv/2021/Ausgaben/02_21.pdf?__blob=publicationFile
Essential workers outside healthcare are only getting the vaccine in the 5th group, along with people aged 60-64.
Very interesting, thank you!
The question of whether to initially prioritize those who are categorically at higher risk of death vs. those who are more likely to spread illness invites thoughtful inquiry.
But the whole, careful structure of how to achieve the greatest reduction in morbidity and mortality depends on widespread trust that patiently waiting your turn won't amount to self-harm, if your turn continues to recede while the estimates of vaccine availability decrease. Only if the rollout is perceived as fair and predictable, is it rational to be patient.
All this disquiet about who should go first-- either because of a different opinions re weighting of variables or because of fear for oneself and one's cohort-- was made much worse when 45 & co inexplicably cut the initially available supply by half: "as recently as October, federal officials had turned down [Pfizer's] entreaties to lock in another 100 million doses", only to beg for those doses after they were committed elsewhere. It was the easiest way possible to achieve more widespread vaccination more quickly, but they didn't do it. If you have twice as many doses, the questions about priority order are immediately less fraught.
Today, we read that states had the number of doses promised in the current wave cut by as much as half, and "States report confusion as government reduces vaccine shipments, while Pfizer says it has ‘millions’ of unclaimed doses". It's hard to know which is more disturbing: attributing this to malice or incompetence on the administration's part.
https://www.washingtonpost.com/health/2020/12/17/pfizer-vaccine-supply-states/
We are back to the same crucial point: none of this works without trust in government. And that's been deliberately besmirched in this country.
I will continue to wait patiently, because I deeply believe in the common good. And I certainly never expected something of this magnitude and novelty to go completely smoothly. But I must admit that after the recent reports of 45&co actively making it worse, my "reptile brain"-- which I will overridge-- is clamoring for me to get out there and try to hustle a place in line.
Yes. Lack of trust in institutions and the administration certainly makes it harder.
Edit: Over-ride
The problem is that it is not just age that leads to increased risk, in some counties in the US a 60 year old Black or Latino man may have more risk of dying than a 75 year old white woman. And by targeting just those who are older you might end up disproportionately giving the vaccine to wealthier white people first. I do agree that those who live in LTCFs should be first on the list as we have seen they are at much greater risk than other groups.
I think LTCR are a clear priority, and everyone seems to easily agree on that given the horrific numbers. There are some gender differences, but the research I've seen says race differences are coming down not to biology but circumstances and comorbidities—which are also often tied up with circumstances. (Less like Tay–Sachs). People with co-morbidities are at higher risk, for sure but the age factor still overwhelms comorbidities. However, as we go down younger ages, comorbidities would/should certainly get priority. Just by age, someone who is above 75 has about 30 times the risk of dying compared the risk of someone in the 50-64 group (rough calculation will look up numbers). That's why I like the UK list.
It's less clear to me that death should be the only factor. I think we're crippling a sizable chunk of our population with "long COVID". If you try to take into account quality of life loss, then I think we just don't know enough to say what the best solution is. I think nursing homes should be a priority (COVID is very dangerous there, and logistically easy to do as well, and should include residents and workers), but past that it's less clear. So a police officer will get COVID in March so that an 80 year old person getting groceries delivered can get vaccinated first? I suspect we'll end up with "a little of everything", and split vaccines between essential workers and elderly, doing both at once, thereby slowing down completing each group. And each state will do it slightly differently.
I completely agree on studying long COVID and the aftermath of this virus, but those 80 year-olds have been severely isolated for almost a year now, and I think there's a quality-of-life argument there, too. There is no option here without trade-offs. Also, hospitals are under strain because of age (it's mostly 65+ people who get the severe disease), even though younger cases definitely exist (and somewhat understandably get a lot more attention).
A really good Twitter thread on who should be vaccinated first: It is both about age as well as risk of infection https://twitter.com/WhitneyEpi/status/1340854715068891136?s=2
She's a colleague! And she's excellent.
I took a scan of the effects of prioritization paper and I did not see mention if they excluded care residents from their analysis. This seems relevant because most of the elder deaths have been in care homes (eg in BC 80% of the deaths over 70 were in care facilities). Since, as you note, care facility residents are being prioritized by everyone, it’s important to exclude their statistics from further analysis.
Apologies if this is already accounted for.
Indeed, that is a good point and we have models that do account for it, but as you note, it is not as straightforward.
A peripheral point: I don't think it is quite accurate to say that the Marshmallow effects disappeared when controlled for various factors. It is true that a Psych Science study claimed this, but a follow up commentary was quick to point out that if one controls for everything, one can make an effect go away (cf. bad controls). See https://medicalxpress.com/news/2019-12-reanalysis-famous-marshmallow.html A snippet: "They tried to account for so many effects that it becomes impossible to interpret what these effects are telling us about the real relation between early self-control and later success." Falk, Kosse and Pinger have now performed a similar analysis. Crucially, however, they controlled only for confounding factors that could be clearly interpreted as such. Their re-examination of the data suggests that the replication study actually reveals a relatively strong correlation between readiness to delay gratification and subsequent scholastic success."
But yes, surely true that the effect is interconnected with society / the person who offers the Marshmallow.
That's interesting, thank you. Readiness as fueled/made easier by the context is probably the better way of phrasing it.
Hi Zeynep, as always thanks for you excellent writing and analyses. Your intro on hope and your experiences in Turkey reminded me of a Jim Collin's passage in his "Good to Great" on what he called the Stockdale Paradox (https://innovationandcreativityinstitute.com/stockdale-paradox/):
“You must never confuse faith that you will prevail in the end —which you can never afford to lose —with the discipline to confront the most brutal facts of your current reality,
whatever they might be.” - James Stockdale
Stockdale said that the people who did not make it out of the prisons were the optimists, those who thought they were getting out soon. I have seen so many parallels today the past 9 months with the pandemic and people losing faith/getting angry/getting depressed.
But it appears now we will prevail.
If you look at slides 31-33 of the ACIP presentation, it is clear that the ACIP recommendations are mostly based on purely subjective determinations that they classify as “Ethics.” I see little scientific or technical justification for their proposed course of action, which seems mostly designed to appeal to a perceived audience.
Is it true that 10% of people w COVID end up as "long haulers"? The determination of who goes first is a difficult one and I'm not advocating for a different priority system. However, I do think that we should consider the impact, both personally and societally, of long haulers in our thinking. The amount of chronic disease and disability that will follow in the wake of this virus is distressing.
Isn't there a way around these ideas by creating a surplus of vaccine? As Stephen Beranyi suggested in the NYT yesterday (The Covid-19 Vaccine Doesn't Mean Big Pharma Is Your Savior https://nyti.ms/37o6zq1) can't we force the opening up of Covid vaccine creation to other companies and deliver a large enough quantity to vaccinate everyone - elderly and essential workers within less than 100 days? I'd love to see an elected leader like Biden act for the mass of people instead of weighing lobbyist-led insights for various populations.
I just co-wrote a piece for the NYT, based on that single-dose trial idea. https://www.nytimes.com/2020/12/18/opinion/coronavirus-vaccine-doses.html Yes, solving the vaccine shortage would obviously the best way forward, but we'd still face the question of what to do this month. We have 50 million doses now.
I might be wrong but it looks like from the CDC APIC website that they're basing their recommendation off of old (and now faulty) assumptions in an Oct. 30th analysis that assumed that the vaccine (a) would be only 35% effective in those 65+; and that (b) the vaccine would stop transmission [which as I understand it may or may not be the case, we don't know yet]. The 11/23 CDC APIC slide deck incorporates its quantitative analysis from a 10/30 epidemiological analysis that lists these as its assumptions on pages 7 and 8. On the CDC website at https://www.cdc.gov/vaccines/acip/meetings/downloads/slides-2020-10/COVID-Biggerstaff.pdf
Edit: There's a statement in the 11/23 CDC APIC slide deck saying they updated the vaccine efficacy assumption to 90% . . . though looking at the 10/30 document, I can't shake the feeling that something is very off about the analysis in it . . . it's the same feeling I get at work when I first realize someone's trying to use Powerpoint to hide something from me
I hope they provide proper numbers to their models.
I do too but it might not matter, because their 10/30 document scenario assumptions barely prioritize any group, and the outputs are used in later APIC docs to misleadingly argue that group prioritization doesn't have an impact on infections or deaths. Specifically, the models they run in the 10/30 doc do 20mm healthcare workers first, then either (a) 20 mm 65+; or (b) 20 mm essential workers; or (c) 20 mm high risk people. In each scenario they then give the next 160mm vaccinations broadly with nor prioritization between 80mm essential workers, 55 mm 65+, etc. Despite this scenario (a) leads to double the decline in deaths in their own analysis, but they use bar charts that obscure the magnitude of difference, and bury that page towards the back of the slide deck. They never show an analysis that totally prioritizes all 55m 65+, and if they did, even leaving the rest of their model numbers/assumptions the same, I suspect it would show a massive decrease in potential deaths.
Edit: They also (egregiously!) in their analysis emphasized scenarios where the vaccine is "infection-blocking" even though what we have are "disease-blocking" vaccines. In the "disease-blocking" scenario, age-prioritization saved more lives . . .