The Risks
Supposing you were to forcibly vaccinate every remaining unvaccinated person in the country, excepting those people with valid medical reasons for an exemption (such as a history of severe allergic reactions to vaccination, etc.). What would be the downside?
Vaccines are not perfectly safe, and vaccinating a large number of people will inevitably have side effects, up to and including the death of some people. Here, I am going to estimate the number of deaths only, and not attempt to quantify the amount of detriment to health that severe but non-lethal side effects will have. I will do the same thing for the health effects of Covid on the other side of the scales. The reasoning behind this is that a lot of different kinds of health problems tend to behave very similarly in this way: that the severity of outcomes occur in a natural distribution curve. That is, a small number of people will die, a larger number of people will have severe but non-lethal outcomes, and the largest number of people will have moderate to mild outcomes.
Given that distributions of outcomes tend to be shaped like a normal distribution curve of some type, it is often the case that comparing just the most easily measured of a possible range of outcomes will still give a valid comparison, rather than laboriously comparing most serious to most serious, least to least, and so forth. And for both vaccinations and disease, deaths are the least ambiguous outcome and one where we are most sure of the numbers. So that's why I'm going to compare just death rate to death rate, as a proxy for the entire health burden of vaccination versus disease.
So how many deaths would one expect from a massive vaccination campaign of about 124 million people (about the number of people remaining to be vaccinated)? Well, I know of three real and dangerous side-effects that have been seen from the vaccines: anaphylactic shock (a complication possible from any vaccine), myocarditis, mostly seen from the mRNA vaccines, and a rare type of blood clotting in women seen from the adenovirus vaccines.
Anaphylaxis
Incidence of anaphylaxis post-vaccination are pretty rare (about 3-5 per 1 million doses), and given that anaphylaxis is normally very treatable with on-site application of steroids, I have struggled to find any cases of deaths from anaphylaxis verifiably caused by the vaccine. I did find reports of one in India, one in England, and one in Vietnam, so I believe it *does* happen. I think it would be hard to justify a death rate of anything higher than 1 death per 10 million people vaccinated, though, so let's very pessimistically put this number at 12 people dead from anaphylactic shock.
Myocarditis
Incidences of myocarditis from the vaccines are rare enough, but since the myocarditis is also generally a pretty mild form when it does happen, deaths are even rarer. No person in the United States has been found to have died from myocarditis post-vaccination so far, out of about 200 million people vaccinated. Europe has seen 5 people die from myocarditis or pericarditis after about 200 million people were vaccinated, but it's unsure if all of those were actually related to vaccination or were coincidental. So a good number here is probably something like 1 per every 100 million people vaccinated. For our scenario, let's round that up to 2 people dead.
Blood Clots
For the unique form of blood clotting associated with the adenovirus vaccines, the incidence of deaths is higher. Three people in the United States have been verified to have died from blood clots caused by the Johnson & Johnson vaccine. Death rates in other places where usage of adenovirus vaccines are much more prevalent have been estimated at 7 per million people vaccinated. Thus far, only 8% of Americans have been vaccinated with the Johnson & Johnson vaccine . . . and I have to think that this percentage would *have* to decrease, especially among those women who are most susceptible to this kind of blood clotting. In fact, I would suspect that doctors would strongly urge women in the susceptible groups to get the mRNA vaccines instead, of which we have more than plenty supply for everybody.
So I'm going to assume that the J&J vaccine accounts for only 4% of the total vaccinations in our hypothetical campaign, which would then work out to it causing something like 35 deaths.
So I'm going to assume that the J&J vaccine accounts for only 4% of the total vaccinations in our hypothetical campaign, which would then work out to it causing something like 35 deaths.
Other Unknown Causes
It's difficult to account for the pure unknown. In this case, I find there is little need to do so. Statistics has methods of characterizing when a sample of a certain population is sufficient to give you statistical certainty that you are going to have a representative result from that sample. And those methods are all *entirely pointless here*, because the number of people to be vaccinated in this theoretical vaccination campaign is *smaller than the total number of people already vaccinated*--the "statistical sample" is larger than the rest of the population.
I don't know if there is a word for statistical "super significance", but the data we already have from vaccinating a majority of the population should classify for that if anything does.
There is no reason to believe at this point that any new surprises are going to arise from a new massive vaccination campaign. But hey! let's imagine something comes up anyway. We will arbitrarily double the number of projected deaths we have come up with so far, adding another 49 deaths from some mystery side effect--let's say something that effects small children only maybe.
I don't know if there is a word for statistical "super significance", but the data we already have from vaccinating a majority of the population should classify for that if anything does.
There is no reason to believe at this point that any new surprises are going to arise from a new massive vaccination campaign. But hey! let's imagine something comes up anyway. We will arbitrarily double the number of projected deaths we have come up with so far, adding another 49 deaths from some mystery side effect--let's say something that effects small children only maybe.
Final Range
Now let's take the number we have come up with, and multiply it by 2 for a fudge factor--maybe all of the deaths we have looked at so far have only caught 50% of the true deaths caused by vaccination, for some reason I can't right now imagine. Then let's round up, and we have a total of 200 deaths caused by this vaccination campaign. This is probably an absurd number, as it is already arbitrarily quadrupled over a number that was already unrealistically pessimistic. But let's go with it for now as an upper bound, or the pessimistic side of our projection.
What's the optimistic side of this projection? Well, the total number of confirmed deaths from Covid vaccines so far in the United States is 3. Those were all from the J&J vaccine, and it's reasonable to think that if you are vaccinating a smaller group of people and you just avoid giving the J&J vaccine to vulnerable women, then nobody will die. Further, it is thought that now that we understand that this rare blood clotting is a risk, we understand also the warning symptoms that proceed the stroke it causes. It's possible that none of those three people would have died if doctors had known what to do and had taken preventative action that is now on the standard warning label for the J&J vaccine.
So it's quite reasonable to think that the total death toll of vaccinating 124 million people would be exactly zero.
Consequently, the final range for deaths caused by vaccination we can project to be something between 0 and 200 people.
So it's quite reasonable to think that the total death toll of vaccinating 124 million people would be exactly zero.
Consequently, the final range for deaths caused by vaccination we can project to be something between 0 and 200 people.
Benefits
Now, what are the benefits of this massive vaccination campaign? To answer that question, we would have to know how many people we expect to die if nothing special is done and people just continue to be vaccinated by choice, slowly (currently happening at the rate of about 500,000 people per day). Keeping in mind that vaccination takes 6 weeks to reach full effect, the current rate of vaccination will only reach about 15 million people before it's too late to do any good for this winter season--*if* the current rates hold up, which I find doubtful since we are currently "burning through" the only-somewhat-hesitant crowd. That's only about 10% of the total people you could vaccinate--probably not enough to make a substantial difference in the total illness and death.
This is a very difficult thing to estimate, full of different kinds of uncertainty. I'm going to try several different ways of estimating and compare the results of these different approaches.
Summer to Winter Comparison
In 2020, there was a summer surge in infections and deaths, followed by a winter surge in infections and death. Importantly, the social restrictions in place between those two time periods were largely the same, and the levels of vaccination were the same in both cases: effectively zero, as vaccinations didn't start having a real population-level effect until after the primary winter season was over.
In 2021, we have also had a summer surge, and it is inevitable that we will also have a winter surge. The social restrictions in place between the two surges will be the same, I assume: effectively nothing. Likewise, the total level of vaccinations will be roughly the same--new vaccination rates having dwindled to their lowest numbers ever just before the summer surge and having only just sluggishly risen to their current tepid numbers now as we are near the end of it. There will be about a 10-15% increase in meaningful vaccinations at the start of the winter surge compared to the start of the summer surge, but that will be only a second order effect on the total numbers (probably, more on this later).
So here is one reasonable approach to estimate the number of people who will die this winter from Covid: take the ratio of the number of people who died from Covid in the winter compared to the summer of 2020, and multiply the number of people who have died from Covid in the summer of 2021 by that ratio. Then, make some adjustment for the increased levels of vaccination: say, subtract 20% to account for the 15% increase in vaccination levels. The result will be the projected number of people who will die this winter.
I've done this in a spreadsheet here: Winter Death Estimates. The result that this gives you is about 450,000 deaths projected for this winter.
Percentage of the Remaining Vulnerable
I think 450,000 is extremely pessimistic, so let's try a different approach. Let's try thinking about how many people are even left to infect, and how many of those people might plausibly fall sick this coming winter season.
The reason I think the previous number is pessimistic is that if you start calculating up the total number of people who have been either previously infected or vaccinated, you start to come up with large percentages of the population--things like 88%. This is very close to levels where we could consider herd immunity in play. If we actually reach an effective herd immunity, then R levels could drop below 1.0 and we could have close to *zero* spread of the disease this winter--that is somewhat plausible.
Now, figuring how close we are to herd immunity this is an inexact science for several reasons, of which there are three primary ones.
The reason I think the previous number is pessimistic is that if you start calculating up the total number of people who have been either previously infected or vaccinated, you start to come up with large percentages of the population--things like 88%. This is very close to levels where we could consider herd immunity in play. If we actually reach an effective herd immunity, then R levels could drop below 1.0 and we could have close to *zero* spread of the disease this winter--that is somewhat plausible.
Now, figuring how close we are to herd immunity this is an inexact science for several reasons, of which there are three primary ones.
The first primary reasons is that there is always a large body of people who are infected but not diagnosed--the asymptomatic majority. What the proportion is between the known infections and these unknown infections, however, is still a matter of some conjecture. Something like 50% to 75% of the total Covid infections are these "occult" infections, and since you don't know that number, you don't know how much immunity is really out there.
The second major source of uncertainty here is that we don't know for sure how much existing immunity--either from previous infections or from older vaccinations--is going to prevent spread of the virus this coming winter. Both older vaccinations and older infections may hold up well to prevent serious illness and death in those who have them--but as for holding up for preventing spread of the disease, that is far less certain.
Finally, herd immunity doesn't work as well in a fragmented society. If there are large populations of people who are less vaccinated than other people *and* they spend a lot of time in close proximity to each other (say, for example, school children), then it won't matter if the population as a whole has a high rate of immunization, the disease will still be able to spread robustly through that proportion of the population which is less vaccinated. So, overall vaccination numbers don't tell you the whole story as far as limits on disease spread.
*But* let's attempt to do this calculation anyway. The population of the United States and the number of vaccinated people are easy numbers to obtain, as is the number of individuals who have recovered from diagnosed Covid. To calculate the remaining vulnerable people, you have to assume a certain ratio of diagnosed Covid to non-diagnosed Covid, and the lower that ratio, the more vulnerable people are still around.
Beyond that, you have to determine what you think the attack rate this coming winter will be: what percentage of vulnerable people will come down with the disease. This is an extremely difficult question to settle. Given that the Delta variant is about twice as infective as original covid, with an R0 of 5-6, traditional models for infectious diseases gives you a very high mark for what the final attack rate will be--anything above 3 and the final attack rate is typically reckoned in the mid-to-high 90's. But you can't just take Delta's R0 since one assumes that, first of all, not all people who could possibly get sick from Covid eventually will actually get sick precisely this winter, and secondly, that the interference of large amounts of immune people will have *some* oppressive effect on spread, and hence final attack rate.
So by taking 75% of the remaining vulnerable population as a worst-case estimate, and maximizing the size of the remaining population by *minimizing* the number of people who are already immune, I come to a maximum death rate for this winter of 672,000 people.
Beyond that, you have to determine what you think the attack rate this coming winter will be: what percentage of vulnerable people will come down with the disease. This is an extremely difficult question to settle. Given that the Delta variant is about twice as infective as original covid, with an R0 of 5-6, traditional models for infectious diseases gives you a very high mark for what the final attack rate will be--anything above 3 and the final attack rate is typically reckoned in the mid-to-high 90's. But you can't just take Delta's R0 since one assumes that, first of all, not all people who could possibly get sick from Covid eventually will actually get sick precisely this winter, and secondly, that the interference of large amounts of immune people will have *some* oppressive effect on spread, and hence final attack rate.
The Range
In the absence of far too much information, I have put as an upper-end-of-plausibility an attack rate of 75%. For reference, the attack rate of just the 2020 winter season was about 3% of the total population, so this rate is *remarkably* higher than what we have ever seen in a season before. The fact that we will be seeing a holiday season for the first time, however, both with a very infectious version of Covid very much about *and* no social distancing to speak of, makes this attack rate maybe not entirely fantastical.So by taking 75% of the remaining vulnerable population as a worst-case estimate, and maximizing the size of the remaining population by *minimizing* the number of people who are already immune, I come to a maximum death rate for this winter of 672,000 people.
On the lower end, I assumed a higher number of existing immune people, and assumed an attack rate of 1.5%, which is assuming that the disease spreads half as vigorously among the remaining vulnerable than it did among the vulnerable from last winter (probably an extremely questionable assumption given how robustly it has spread during this summer, but let's go with it). These assumptions lead to a projected maximum death rate of only about 3000 people.
This is, to say the least, an extremely broad range of possibilities.
Equalizing Death Rates
The final method of coming up with a winter death rate is as follows:
We know that Covid is capable of killing up to 3297 people for every million persons in a population. We know this, because that is currently the total death rate for the State in the union that currently has the highest number (Mississippi). Now, when people look to explain why Mississippi and other states have had their death rate totals be so high, they normally have been looking to its comparatively low vaccination rates compared to other states. That, however, is an over simplification. In fact, while Mississippi *has* had a lower vaccination rate than other states, it's not all *that* much lower. Meanwhile, Florida *also* had an extremely bad summer and has ended with a very high death rate, and *its* vaccination rates overall weren't bad.
The real reason behind these high death rates is not the overall vaccination rates. Rather, because Covid spreads among interconnected network of vulnerable people, the high death rates in the states that have suffered them this summer has been primarily due to large numbers of unvaccinated people *living in close communities* with each other. Primarily we are talking here about close-knit black and Hispanic communities with low vaccination rates.
So then the next realization you can have is that, when Thanksgiving and Christmas come around *everyone* becomes a close connected community. This is a major--perhaps *the* major--reason why these seasons are replete with illness. Consequently, one could expect that when the holiday season comes, all of the other States will be vulnerable to disease spread among the unvaccinated in the same way that the other States already had disease spread among the unvaccinated. Unvaccinated people who were previously unconnected because there were not living in close community with other unvaccinated people will all of a sudden be spending time, cheek by jowl, with each other over the holidays.
So then the next realization you can have is that, when Thanksgiving and Christmas come around *everyone* becomes a close connected community. This is a major--perhaps *the* major--reason why these seasons are replete with illness. Consequently, one could expect that when the holiday season comes, all of the other States will be vulnerable to disease spread among the unvaccinated in the same way that the other States already had disease spread among the unvaccinated. Unvaccinated people who were previously unconnected because there were not living in close community with other unvaccinated people will all of a sudden be spending time, cheek by jowl, with each other over the holidays.
The hypothesis then is, take the current highest death rate among the States (3297 people per million of population), and assume that all of the other States will rise to equal that level over the holidays as unconnected unvaccinated become connected.
This will probably *underestimate* the death rates because it will assume that nobody will rise *above* that existing death rate (and I'm sure that more people will die from Mississippi, making that assumption untrue). But it also probably *overestimates* the death rates because there are a number of states that are *way* beneath that level and seem unlikely to go that far in a season. But let's assume that the over and under estimates somehow balance each other out and see what we get.
This calculation is also on the spreadsheet, and it comes out to about 365,000 deaths.
This will probably *underestimate* the death rates because it will assume that nobody will rise *above* that existing death rate (and I'm sure that more people will die from Mississippi, making that assumption untrue). But it also probably *overestimates* the death rates because there are a number of states that are *way* beneath that level and seem unlikely to go that far in a season. But let's assume that the over and under estimates somehow balance each other out and see what we get.
This calculation is also on the spreadsheet, and it comes out to about 365,000 deaths.
Final Calculation of Benefits
The final calculation of the benefits of a massive vaccination campaign, in terms of lives saved, is therefore a number of extremely broad range. The vaccines have been holding up in terms of preventing death from Covid at something around the 95% rate, so you only need to adjust those final death numbers down a little bit to get total lives saved. In reality, if you did somehow manage to vaccinate *all* the eligible people, you'd save more than 95% of the potential lives, since you'd *seriously* cut into infections as well as prevent deaths. Actual prevention of death would probably rise to something closer to 99% of the people who would have otherwise died. But let's be pessimistic again at call it just 90% of people saved.
In calculating the final range of people who might possibly die, I also refuse to consider the 672,000 figure from the "Remaining Vulnerable" method. I don't know what's actually wrong with that calculation, but instinct tells me that it is not plausible. So I'm only going to use 3000 for the bottom range and 450,000 for the top range. Taking 90% of that tells me that the projected benefit of an extremely aggressive vaccination campaign getting close to 100% of eligible people before winter could result in between 2,700 and 400,000 lives saved.
In calculating the final range of people who might possibly die, I also refuse to consider the 672,000 figure from the "Remaining Vulnerable" method. I don't know what's actually wrong with that calculation, but instinct tells me that it is not plausible. So I'm only going to use 3000 for the bottom range and 450,000 for the top range. Taking 90% of that tells me that the projected benefit of an extremely aggressive vaccination campaign getting close to 100% of eligible people before winter could result in between 2,700 and 400,000 lives saved.
Risks vs. Benefits
Is this analysis worth anything? With such a wide range of possible outcomes, is there any take-away action we can recommend? That is beyond the scope of this post, but I will point out two things:
- The risks of extremely aggressive vaccination at its most pessimistic is more than an order of magnitude less than the benefits at *its* most pessimistic. There is therefore no realistic scenario, at all, in which an extremely aggressive vaccination program will not provide more benefit than harm, at least at the level of physical health.
- The range of possible outcomes is for projected deaths this winter is extremely broad, but I think definitely realistic. I would be greatly shocked if deaths this winter were less than 3000 people, and I would be shocked if they exceeded 450,000 people.
And this means, I think, that numbers in the middle range are very appropriate projections for a State to plan for. I think that 100-200,000 is a very realistic possibility, though I'm very hopeful that the number will be closer to 20-50,000. In general, if there are very few downsides (which we have shown is true, comparatively), the State *ought* to plan for the worst. So I would advocate for a way of thinking in which we assume that 100-200,000 deaths will happen this winter if we do nothing.
