Updated: Dec 16, 2020
Summary of key points:
We will soon have 2 approved vaccines providing enough doses for >40% of the nation, but given there is no guarantee more will be approved or how soon these will be distributed, we need to carefully prioritize distribution to save the most lives and move us towards herd immunity as fast as possible.
I posit that serology tests should be given to prospective vaccinated and people without antibodies prioritized in the first wave.
I present evidence-based reasoning that at least 20-30% of the USA has been exposed and infected by the end of November.
While reinfections have occurred, I make a simple statistical argument that demonstrates they are likely exceedingly rare, and along with recent growing literature it seems likely that a vast majority of infected and recovered individuals retain immunity for at least 3-6 months, and thus it is reasonable to prioritize them for later vaccination in a setting of limited supply.
While the mRNA vaccine studies do not demonstrate they prevent asymptomatic infection or viral spread, I make the case that it is very likely that these vaccines do suppress spread and so are a valuable tool in working towards herd immunity, and thus individuals at high risk of exposure and transmission such as essential workers should clearly be prioritized for early vaccination.
I posit that under limited supplies, it is likely that it would be more optimal to first vaccinate a large proportion of the country with a single dose before giving the second dose, but the lack of data will probably make this strategy infeasible, at least until it can be rigorously demonstrated.
With the emergency use authorization of the Pfizer/BioNTech SARS-CoV-2 vaccine last week, and likely EUA for the Moderna vaccine this week, distribution of the vaccine in the USA has commenced, starting with front line health care workers, and with vulnerable populations including nursing home residents and staff also included in the first pass.
Given the massive viral surge the USA is currently experiencing, the relatively limited vaccine supply of the Pfizer/BioNTech and Moderna vaccines, and the lack of guarantee that any other vaccine under development will have anywhere close to the efficacy of these first two vaccines, the only ones using a new mRNA-based technology, I think that careful thought must be given to the vaccine distribution plan to ensure both maximum protection of lives and maximum suppression of viral spread so we can bring the pandemic to an end as soon as possible.
Potential of Limited Supply of Vaccine
Given the uncertainty of which vaccines under development would pan out, Operation Warp Speed (OWS) only pre-purchased limited quantities of each of the top vaccines under development, with opportunities to purchase more. However, as certain vaccines are showing excellent performance, the demand for them internationally is spiking, and there is uncertainty about whether the USA will be able to quickly secure enough supply for the entire population.
For the Pfizer/BioNTech vaccine that was given EUA last week, the USA has a purchase order for 100 million doses, enough for 50 million Americans given the approved administration involves 2 doses. The USA has also already preordered 200 million doses of the Moderna vaccine likely to get EUA this week, and OWS is in negotiation with Pfizer to purchase an additional 100 million doses during 2021, which would provide vaccine for 50 million more Americans, although those additional doses might not be available until late into the year. Thus, the USA has currently secured enough doses for 150 million Americans, or 45% of the country, and if this negotiation is successful, for 200 million Americans, 60% of the country.
OWS has also pre-purchased 300 million doses of the Oxford/AstraZeneca Vaccine, also given in two doses, potentially providing enough for an additional 150 million Americans. However, there are scientific questions about the trials and results to date that make it unlikely to be approved by FDA for emergency use in the near future without further data, including the large phase 3 USA study that is carefully designed and should yield more definitive results on efficacy and safety. However, given it is using the originally planned dose with only estimated 62% efficacy for preventing symptomatic disease and 4% efficacy for preventing asymptomatic disease based on aggregating results across several partially complete phase 1-2-3 studies, there is considerably uncertainty about whether it will be a viable option given the stellar efficacy and safety results from the Moderna and Pfizer/BioNTech mRNA vaccines, and these results are not projected to be available until late January, so EUA is at least a couple months away.
OWS has also pre-purchased 100 million doses of each of the Novavax and the Johnson and Johnson/Janssen vaccines, whose large phase 3 studies are underway and may have results in the first half of 2021. Both vaccines are primarily testing single doses (although the Johnson and Johnson are also running a second trial looking at both single dose and two dose deliveries). Thus, if either of these are shown safe and efficacious, each could provide enough vaccine for 100 million more Americans.
Thus, among the Oxford/Astrazeneca, Novavax, and Johnson & Johnson Janssen vaccines, OWS has pre-purchased enough doses to vaccinate up to 350 million more Americans all together, so if even one of these pans out, it is likely we will have enough to vaccinate the entire country. This is encouraging, but there is still uncertainty that needs to be taken into account:
(1) There is no guarantee that any of these will work out -- so far only the two mRNA-based vaccines have rigorously demonstrated high efficacy. If for some reason the other vaccines using alternative approaches are significantly inferior, then this may contribute to some shortage, and none of the other vaccines under development are using the same mRNA-based technology. Early trial results are promising for these vaccines, but until we see the rigorous phase 3 results there will be some question. And in that scenario, the Moderna and Pfizer/BioNTech would be in even greater demand internationally, so the ability to rapidly secure more doses would be that much more difficult.
(2) It is not clear when sufficient trial results will be available for EUA. As mentioned above, there are reasons to wonder whether the USA trial for Oxford/Astrazeneca will be successful, and the Novavax and Johnson & Johnson USA phase 3 trials are still early and it is not clear when they will be complete (although the Johnson & Johnson UK phase 3 study should have interim results in first quarter 2021). This means that even if they will become available, they may not become available until well into the year, which would leave us still dealing with significant viral surges throughout the country throughout the winter and spring.
This article discusses some of the manufacturing problems and delays that suggest that availability may be severely limited.
With this uncertainty, it is important for us to carefully plan the distribution of the available doses of the Pfizer/BioNTech and Moderna vaccines to protect the most vulnerable lives and also move as rapidly as possible towards herd immunity so we can prevent the mortality and morbidity of future infections as well as get society back closer to full operation.
From what I've seen, the current distribution plans are prioritizing front line health care workers, followed by the most vulnerable populations, including long term care residents, and other "essential workers" who provide the greatest exposure and transmission risks. This makes sense, but there are several factors relevant for distribution priority that I do not see being discussed much, and may be crucial in the event that supply is severely limited or constrained in the first half of 2021:
One crucial consideration that I have not seen discussed as much as I would like:
Should we vaccinate those who have already been infected and recovered?
I haven't seen much discussion of whether the initial distribution of the vaccine will focus on those who have not already been infected. It would be easy to give each prospective vaccine recipient a serology test to see if they already have anti-SARS-CoV-2 antibodies in their blood stream, and it might make sense to exclude those from initial vaccine distribution.
Likely 20-30% of Americans have been exposed to SARS-CoV-2
By this time, a substantial proportion of Americans have been exposed to SARS-CoV-2. As of December 14, 2020, the number of confirmed cases in the USA is almost 17 million, but the number exposed and infected is much higher because of various factors including asymptomatic disease, under testing, and limitations in assay sensitivity that depends on the quality and timing of the sampling. In early summer, the CDC director estimated that only 1/10 of infections were counted as cases, and a large serology study published in the Lancet estimated that 9.2% of the USA population had SARS-CoV-2 antibodies in early July, which would by about 10-fold more than the 3 million confirmed cases at that time, which represented 0.9% of the population.
As testing has ramped up in the summer and fall, the ratio of infections to confirmed cases has certainly fallen from this 10:1 ratio, but is still substantial. CDC scientists recently published a paper estimating that about 15% of Americans were exposed and infected through September 2020, at which time the 7 million confirmed cases comprised 2.2% of the population, with an estimated ratio of infected:confirmed cases of 7:1. Given the confirmed cases have increased from 7 million to 17 million since then, it is reasonable to think somewhere near 20-30% of Americans have been infected and exposed --- building on these results, if the ratio of infected:confirmed cases has been >5:1 since September, that would mean >100 million cumulative infections in the USA (30%), if the ratio is >3.3:1, then that would mean >80 million cumulative infections in the USA (25%), and if the ratio is at least 1.75:1, then that would mean >66 million cumulative infections in the USA (20%). If the CDC estimates of 15% at the end of September were roughly accurate, then it is safe to assume that at least 20% and likely more than 25% of Americans have been exposed and infected. This number is not enough for herd immunity to kick in, but is enough to start suppressing spread, and certainly comprises a substantial number who may not need to be immediately vaccinated if they are still immune.
It is likely that most of those still retain immunity and should not be first priority for vaccination.
Given the anecdotal evidence of some SARS-CoV-2 reinfections 3-4 months after recovery, it is clear that not all recovered individuals are immune from reinfection. However, it seems clear to me that the emphasis of these cases in the media has caused many to overestimate the reinfection risk and underestimate the durability of infection. Some basic statistical reasoning reveals that a vanishingly small proportion of recovered individuals loses immunity and are prone to reinfection within 3 months.
If we assume that everyone lost immunity at 3 months, then we would expect to have seen >130,000 reinfections through November in the USA alone, based on calculations using monthly confirmed case rates in the country (see table below). By the way, this rough estimate ignores any infections that are not confirmed by PCR test either at first infection or reinfection, so is extremely conservative.
Table 1: Confirmed cases by month in USA (covidtracking.com), corresponding percentage of Americans that represents, and expected reinfected confirmed cases by November 30 assuming everyone lost immunity at 3 months, all infections are confirmed cases, and everyone is equally likely to be infected.
Month: Cases Proportion Expected Reinfected Cases
March 192,177 6,878
April 884,067 0.27% 29,394
May 718,221 0.22% 19,673
June 834,359 0.25% 19,130
July 1,922,730 0.59% 37,043
August 1,464,676 0.45% 19,187
September 1,201,822 0.37% NA*
October 1,915,046 0.58% NA*
November 4,408,082 1.34% NA*
* 3 months haven't passed yet by November 30.
Given the attention given to each confirmed case of reinfection and the fact that only a handful have been reported, one can reasonably infer that loss of immunity at 3 months is exceedingly rare, and a vast majority of recovered individuals are still immune after 3 months. Similar calculations done assuming immunity is lost at 4 months would suggest 88,971 infections, 5 months 50,577 infections, 6 months 31,085. It is likely that a vast majority of those who have recovered retain immunity 6 months later.
Additionally, there are numerous recent papers that suggest long-lasting immune response is common after recovery. One of them for which I did a blog post last month followed >2500 health care workers from NYC or Detroit who were infected in the early weeks of the pandemic and followed >6 months, the longest of any study to date. They found 95% of them had SARS-CoV-2 antibodies in their blood 6 months later, including 100% of those who were hospitalized and 90% of those who were asymptomatic. Also, as emphasized by immunologists, loss of antibodies in the blood does not indicate loss of immunity -- it is normal for antibody levels to wane over time, but B-cell and T-cell memory are able to regenerate a rapid, vigorous immune response once re-exposed. While there are still many unknown details about durability of immune response, there is strong evidence that the durability is substantial, and likely enough to support an annual vaccination regimen.
Given that it is not clear exactly how long immunity lasts for a given individual after recovery, I agree that all individuals should have the opportunity to be vaccinated. However, in an environment of limited initial supply and urgency to distributed widely to reach herd immunity, it should be evident that those who have already been infected, recovered, and have SARS-CoV-2 antibodies should not be in the first priority group. Thus, I propose that in these early days of vaccine distribution, serology tests should be given to any vaccine candidate and the vaccine only given to those who are seronegative, and any seropositive be moved to the end of the line.
There is a second question I have seen discussed, but perhaps not deeply enough:
What groups should be prioritized first, the most vulnerable, or those most likely to be exposed and transmit to others?
From what I've seen, the prioritization seems to be 1. Front line HCW, 2. Nursing home workers and residents, and 3. Essential workers with high exposure and high likelihood to transmit to others, including meat packing plant workers, police officers, teachers, grocery workers, etc. 4. High risk adults who are older and/or have pre-existing conditions putting them at risk for poor outcomes. 5. Other adults. 6. Children (for whom studies are commencing and presumably efficacy and safety determined by that time).
This looks reasonable to me, but depending on how limited the supply is at first, some more thought may need to be considered about exposure and transmission. If we have very limited supplies this winter and focus on populations that are vulnerable but also well-isolated from the rest of society and not likely to be exposed or transmit to others, this might delay the indirect benefit of vaccination -- the reduction of societal transmission levels and moving towards herd immunity -- which is an exponential benefit rather than a linear one. Thus, while the goal of saving lives and thus vaccinating those at most risk of poor outcomes if infected is of primary importance, so is the vaccination of those who unavoidably will be at high risk of infection and transmission to others because they serve essential societal roles. The vaccination of such a person does not only prevent disease in them, but also any other person to whom they would have spread the virus, as well as the people they would have spread to, etc. Hopefully, we have enough supply to prioritize both, but we need to think carefully about the order of priority based on how much supply is available and the need to build towards herd immunity.
What if the vaccine only prevents symptoms and severe disease, not infection and transmission?
Given that the Pfizer/BioNTech and Moderna phase 3 studies only focused on efficacy vs. symptomatic disease, only giving PCR tests to confirm infections for those reporting COVID-19 related symptoms, it is possible that the vaccines do not actually prevent infection, but just reduce symptoms and convert symptomatic cases to asymptomatic ones. Given that much of the societal viral spread is from asymptomatic infected, this would mean that the vaccines would not prevent transmission of the virus or lead to herd immunity.
Because of this uncertainty, clearly people after vaccination should still wear masks and practice physical distancing just in case they can still become symptomatically infected. But does this mean that we should worry that the vaccines aren't preventing infection, and thus focus early distribution only on the vulnerable and not those who are less vulnerable but in a position to infect many others? I say absolutely not.
We need to be careful in our thinking here -- the "absence of evidence of prevention of asymptomatic disease" is not the same thing as "the evidence of absence of prevention of asymptomatic disease", as we statisticians know well. It would be very unusual indeed, based on study results, if the vaccine did not prevent infection but only prevented severe disease.
In the Moderna study, the 94% efficacy came from the fact that the vaccine seems to have prevented 165/174 of the symptomatic cases that are expected to have happened sans vaccine, and in the Pfizer/BioNTech, the 95% efficacy meant 154/162 of the symptomatic cases expected to have happened sans vaccine were prevented. Is it possible that some of these 165+154=319 prevented symptomatic cases were still asymptomatic infections? Sure -- and likely there are some. But it seems exceptionally unlikely that most of them were - it would be strange to be so effective that it prevented 95% of symptomatic cases, preventing even mild symptoms like fatigue, fever, cough or sore throat, without also preventing the viral infection levels needed for the person to spread to other people. Until we have data on asymptomatic spread, we do not know for sure, but it is likely that these vaccines are also preventing infection and transmission, and so will contribute to mitigating viral spread in the community.
Thus, while we should recommend people still practice basic precautions like mask wearing and distancing just in case, we should not spend much energy worrying that these vaccines will not be effective in preventing infection or spread, and we certainly should not let this worry change our distribution prioritization strategies. Thus, the essential workers with high risk of exposure and transmission to others clearly should be highly prioritized.
Would it be better to give twice as many people a single dose of the vaccine rather than giving fewer people both doses?
First, on the basis of science and regulatory approvals, both the Pfizer/BioNTech and Moderna vaccines have only been studied as two-dose regimens, so it is unlikely that officials would consider only giving a single dose for now and waiting to give the second dose until a higher proportion of society has been given the first dose. However, it might be the case that in fact this would be the optimal strategy to get the pandemic quickly under control, and would save many lives as well as prevent collateral societal damage that would be caused by mitigation strategies extending far into 2021.
The MMR vaccine has 97% efficacy against measles and 88% efficacy against mumps when given in the recommended two doses, but has been shown to have 93% efficacy against measles and 78% efficacy against mumps after just a single dose. With the observed 94-95% efficacy for the Pfizer/BioNTech and Moderna vaccines after two doses, it seems likely that the efficacy of these vaccines might be quite high after the first dose. If at least 80-85%, it would CLEARLY be a much better strategy to wait to give second doses until after a high proportion of society is given the first dose.
Unfortunately, single-dose efficacy was not prioritized in the phase 3 studies for either of these vaccines. Had they known they would get such high efficacy on the two-dose regimen, it is likely that they would have looked at single-dose efficacy as well to see if such a strategy could be tenable. Indeed, the Johnson&Johnson/Janssen vaccine is being studied in both single- and two-dose regimen.
There is some evidence of potential efficacy, however. The Moderna full phase 3 report to the FDA was posted online today. In table 15, they show results for a subset of individuals who developed infections after the first dose but before the second dose. In counting cases >14 days after first dose giving time for the vaccine to take effect, there were 2 cases in 983 individuals in the vaccine arm and 28 out of 1059 individuals in the placebo arm, which would be an efficacy of 92.1% (CI 68.8%-99.1%), which provides some evidence of efficacy for the single dose, but given the limited follow up after the single dose without having a second dose, this is not strong enough evidence to conclude long term protection. Hopefully there is a way to get additional evidence to validate this result, since if it held up it would clearly justify the strategy I am suggesting.
I clarify that I am NOT proposing officials use this strategy be implemented without proper studies -- my point is that I think it is likely that this strategy would be optimal, and am pining the lack of data to justify it, and hoping that it can be quickly studied.
In conclusion, it appears that we have secured 300-400 million doses of the two mRNA-based vaccines that have demonstrated outstanding efficacy and safety, and for which we have EUA as of the end of this week. Given as two doses, this is enough to vaccinate 150-200 million, or 45%-60% of the country. Hopefully, the Oxford/AstraZeneca, Johnson&Johnson/Janssen, Novavax, or some other vaccines are approved in the next 3-6 months, but given the uncertainty, I think it is important for us not to presume that will be the case. We need to carefully prioritize the distribution of the vaccine doses we have in hand.
We need to simultaneously account for vulnerability and likelihood of exposure/transmission in the prioritization, as seems to be the case.
In spite of the lack of firm evidence that the vaccines prevent infection, we should design our distribution strategies to move as quickly towards herd immunity as possible.
Until we are certain we have sufficient doses for the entire country, think we should use serology tests to ensure that limited first doses do not go to individuals who already have evidence of anti-SARS-CoV-2 antibodies.
The key to bringing the epidemic phase of the pandemic to a close is herd immunity, trying to remove >70% of society from the susceptible population via immunity from vaccine or natural immune response. By focusing on those who we know are still susceptible, we can accelerate our progress towards that endpoint.
If there is a way to quickly ascertain single-dose efficacy of the mRNA vaccines, then we could double the rate at which we move towards herd immunity, which could shave months off the time in which we need to have high levels of societal restrictions to mitigate viral spread. It may not be possible to get these data in time to implement such a strategy this winter, but perhaps if prioritized we could get confirmation of single-dose efficacy by the Spring.