COVID-19 Vaccine passport systems should largely ignore recovery from COVID-19

Over the last few weeks, people have been asking me for my opinions regarding vaccine passports and some have been outraged by the fact that many such systems do not consider infection-acquired immunity and that I as a proponent of science-based policy should voice dissent with this apparently unscientific policy. I’m about to disappoint those people profoundly though- these policies are completely correct to ignore infection-acquired immunity, and I would go as far as saying in all settings in which vaccine supply is not an issue, no deference should be given to the immunity of survivors of COVID-19 as far as accessing certain social privileges.

As a starting point, we should define vaccine passports and whether these are permissible. These are certifications that permit vaccinated individuals to access certain social privileges that are constrained or prohibited for the unvaccinated as public health measures. I will say in advance that I will limit this discussion to the moral aspects of the passports rather than the legal ones (though both are covered here). To start with, I think that such a system is justified both as a safety measure and to aid in incentivizing vaccination. Usually the concept of passports brings up the notion of mandates and whether or not vaccine mandates are morally permissible. Lawrence Gostin, an expert on the subject, reviews the history and value of vaccine mandates here but the short version is that there is extensive legal precedent in the US to justify them and they do indeed work. My perspective on mandates is that they are a natural extension of risk management and not a violation of autonomy/respect for persons. Forcing vaccination (which is generally considered to be vaccinating someone despite their objection/refusal through the use of physical or chemical restraints) is not appropriate provided the patient has decision-making capacity, in that they abrogate choice, but mandates do not (though I think it may be fair to describe them as permitted coercion and this distinction is not semantical). Mandates may disincentivize people to remain unvaccinated by restricting access to certain social privileges e.g. public school for the unvaccinated, but this does not mean that parents cannot e.g. homeschool their children. Is it more inconvenient or difficult? Certainly, and not everyone will be able to do it. But the parents seem to have decided that it is more important that their child remain unvaccinated than that they receive an education in public school. Similarly, vaccine mandates for employment in healthcare are not new, despite the recent kerfuffle they seem to be causing in New York. Some may decide it is more important for them to remain unvaccinated than it is to be employed. So be it- that is a choice they made. Anti-vaccine activists have often resorted to the phrase “Where there is risk there must be choice” as a rallying cry against mandates. Setting aside that there are no risk-free options in life and thus this amounts to little more than a hyper-libertarian tautology, choice is not being taken away with a mandate. The fundamental question is about risk management. If you deem the vaccination to be too risky for your wellbeing, you are within your rights to refuse it. In turn, society is within its rights to manage the risk that you as an unvaccinated person now pose to its members. Society at large is a relevant stakeholder in any individual’s choice to vaccinate and thus is permitted to change the structure of incentives in accordance with its needs. I’d much prefer people would make the choice to take the vaccine without having to alter the landscape of incentives and I’m more than content to take as much time as it needs to answer any questions about the vaccines that are asked in good faith. However, I am absolutely certain that a healthcare worker who remained unvaccinated against COVID-19 (or influenza) for anything other than a true medical reason is not one I would trust to manage the care of my loved one, so this sudden mass exodus of refusers isn’t something I view as particularly harmful. For the US to have a vaccination rate just slightly above 50% despite the absolute glut of vaccine it has procured is unconscionable. Now then, having established the appropriateness of mandates and vaccine passports, let’s discuss the evidence for immunity following recovery.

Proponents of acknowledging infection-acquired immunity argue that because it seems comparable in protectiveness to vaccine-acquired immunity, it is unscientific to ignore it, but there are several issues with that.

Firstly, reinfections, though rare, are indeed possible, and though usually milder, they can even be fatal. At the public health level, one key question is how well these reinfections transmit which is a loaded question and I have not seen literature that thoroughly explores this. It is reasonable however to presume that this depends on the immunological competence of the reinfected person in question though and thus the answer is likely not simple (and indeed who precisely is susceptible to reinfection is an open question but data from Denmark does implicate the elderly are at much higher risk suggesting that immunological competence is an important factor). There are furthermore challenging questions raised when findings emerge regarding the changes to the function of B cells following COVID-19 as well as innate lymphoid cells. These concerns are mitigated by the simple observation that reinfection appears relatively rare, but, suffice it to say, it would be preferable not to have COVID-19 multiple times. Fortunately, there’s a pretty simple way to help prevent that: vaccination.

Some proponents of infection-acquired immunity contend that it is incumbent that vaccine passports consider prior infections because the protection is superior to that from vaccination. There is currently no good evidence that COVID-19 confers superior immunity to vaccination despite the impassioned outcries of those individuals. Proponents who advance this hypothesis often turn to this preprint which demonstrated that indeed prior infection was superior in preventing infection and hospitalization than was vaccination with the Bnt162b2 vaccine (which you may also know as the Pfizer/BioNTech vaccine/Comirnaty/Tozinameran, and yes, these are all the same). The Science magazine article also included quotes by many scientists who work on vaccines and immunity who were impressed by the results and found them compelling. However, I was… less impressed. To understand why, take a look at table 1a. The vaccinated groups tended to have significantly more comorbidities. For model 1, more than twice as many of the vaccinated are immunocompromised, significantly more also have diabetes mellitus, COPD, and cancer. The disparities are similar but less substantial for model 2 (which also finds less dramatic differences). This kind of difference suggests immortal time bias of the survivorship flavor mixed in: you can’t get a second case of COVID-19 if the first one kills you. By the nature of this model, you are selecting a sample of patients who have already proven that they can, to some extent (i.e. they don’t die), handle a case of COVID-19. On the other hand, it’s reasonable to presume that had some of the vaccinated people encountered COVID-19 before they were vaccinated, they would have died. Hence the comparison isn’t entirely parallel. Beyond this, the actual number of events was pretty small to have much confidence in the findings (9 total hospitalizations isn’t particularly informative). Furthermore, Dr. Diego Bassani shared a thread discussing the issues with the preprint on Twitter, and also provided this handy discussion of collider biases in observational studies undermining our ability to accurately interpret their findings. Note additionally that this study did not have random testing and is retrospective, which are weaknesses UK’s ONS study did not have and found that vaccines are at least non-inferior to prior infection (though it did not attempt to match participants for the timing of vaccination vs. infection). I also wish to point out that I have an issue with the immunologic rationalization offered for the results of the study by the preprint authors who suggest that the immune response to COVID-19 is “more extensive” than the vaccines. This remark is somewhat vague, but the rest of the sentence suggests to me that they are referring to the fact that mRNA vaccines contain only the spike protein and thus are at a disadvantage compared with infection as far as the immune response (rather than the fact that the inflammation elicited by COVID-19 is far greater than that from the mRNA vaccines and likely helps to explain the profound devastation to the body in many cases). While this is not wholly without merit, I feel obligated out point out that even a quarter-dose version of Moderna’s vaccine which contains ONLY spike elicits T cell responses comparable to that of infection, and antibody responses targeting non-spike proteins (likely occurring because of priming seemingly from other coronavirus infections) are not protective in animals. If an advantage exists for infection over vaccination, the immunological explanation almost certainly would have to be the ability of infection to induce local mucosal immunity in the upper airway that could rapidly arrest infection whereas immunity elicited by vaccination focuses on the systemic compartment- though emerging literature suggests that the vaccines can even induce secretory IgA in the saliva, indicating the need for more data (furthermore, per Dr. Gommerman, an author on the study, this secretory IgA is persistent). Admittedly, profiling of mucosal T cells in vaccinees, which may be quite important in protection after recovery from COVID-19, remain an unknown currently and it would be valuable to compare them to those of recovered COVID-19 patients to design better vaccines (and as far as I am concerned this is the only worthwhile reason to do comparisons of the immunity conferred by vaccination vs. the immunity conferred by infection as the costs of infection-acquired immunity will always be unacceptably high). In short, the superiority of one form of protection against COVID-19 against the other is presently best regarded as contentious (and we should certainly expect differences based on the specific vaccine). Except, not really, because getting a highly contagious infectious disease to avoid getting a highly contagious infectious disease is not a public health strategy.

The next matter that creates an issue is how one proves immunity. I cannot tell you how many times I have heard people suggest that they had COVID-19 in 2019 or very early 2020 but have absolutely no proof. Certainly, even in a circumstance in which we created a parity between vaccine and recovery from infection, we could not reliably count these. However, many people do have evidence of a positive PCR test for SARS-CoV-2 that demonstrates an infection occurred. But the issue here is we do not know whether that infection conferred immunity. The immunity an asymptomatic, healthy 20-year-old generates will differ greatly from that of an 80-year-old who develops pneumonia and requires hospitalization. For instance, we know from nonhuman primate models that a relatively high dose of virus is required to produce seroconversion (the development of antibodies). In humans, the Ct value on PCR is inversely related to the probability of seroconversion. Did the asymptomatic 20-year-old just encounter a lower dose of virus than the hospitalized 80-year-old? Whose immunity can we trust? In vaccinology, a correlate of protection describes a metric that can be evaluated to make predictions about whether or not a particular disease phenomenon is likely (and thus they must be defined according to the specific phenomenon e.g. infection, disease, severe disease), and an absolute correlate is a threshold at which a judgment can be made about who is or is not protected. Usually, this is an antibody because it is really easy to evaluate antibody levels, but there are T cell correlates of protection as well. We lack this for SARS-CoV-2 (and actually most vaccine-preventable diseases. For a really complex story, try looking at the history of attempting to establish correlates of protection for pertussis). Evidence that serum neutralizing anti-SARS-CoV-2 spike antibody levels correlate with protection against infection is accumulating but correlates are harder to define for other disease phenomena. The critical role of B cells also seems to be implied by mouse models where they can augment viral clearance, but do appear dispensable for recovery. For example, patients who are on B cell-depleting agents that render them incapable of normal antibody responses usually fare well against SARS-CoV-2 provided that their T cell responses are intact. That said, for public health purposes, the key parameter of interest is transmission. Though infection is required for transmission, infection alone may not be enough for it to occur. Thus it cannot be presumed that a correlate against infection simply translates to transmission- though that’s also academic because we don’t have either. The evidence that vaccines reduce transmission, however, is substantial. While both these studies are pre-delta variant, vaccines do clearly work against it too (this case report shows that the parents never tested positive on PCR, indicating infection likely never occurred and thus they could not transmit). See also, this which explains that unvaccinated and post-vaccine infection cases are not equally likely to transmit. I will add the note that Ct values on PCR cannot be presumed to reflect replication-competent virus, and thus claims made that viral loads (and more relevantly, inoculum sizes) are equal between the two groups are not justified.

Thus far I have essentially pointed to uncertainties in the immunity to SARS-CoV-2 following recovery and one may astutely point out that vaccination may also not guarantee immunity. This is technically true, although there is far greater consistency in the immune responses to vaccination than for SARS-CoV-2 infection. But this is also not the winning argument that post-infection immunity proponents seem to think it is because policy on this is not founded solely on immunological questions. A vaccine passport system would help to keep people safe as a public health measure, but it will affect vaccination incentives. Before there were chickenpox vaccines, chickenpox parties were in vogue, based on the observation that varicella in adulthood is much more dangerous than in childhood and the condition could not avoided indefinitely and also a deep underappreciation of the seriousness for the spectrum of disease varicella could cause. Today I would readily consider this child abuse, but the point of this is there are reasons that individuals could be compelled to deliberately contract an infectious disease. Consider for a moment some of the rhetoric around children and COVID-19 (which for the record, as I have collaborated to detail here and here has been extremely fallacious). COVID-19 has been described as being essentially benign for children, with some countries essentially seeming to want kids to contribute towards herd immunity by infection. Add in the highly injurious rhetoric regarding post-vaccine myocarditis (C-VAM), despite the available evidence that even with this significant risk the overall benefit is still extremely favorable for pediatric vaccines, and you get a COVID-19 party. It doesn’t even have to be children. This isn’t theoretical- there are already people being admitted to ICUs because of COVID-19 parties. Beyond the strain this places on the healthcare system, there is another public health risk, which I really can’t state better than the authors of this commentary:

As natural immunity builds in the population, SARS-CoV-2 variants may be increasingly selected as immune escape variants. Vaccination, on the other hand, even when partial, is unlikely to contribute significantly to the emergence of escape variants owing to the vaccines’ ability to strongly restrict the evolutionary and antigenic escape pathways accessible to SARS-CoV-2, reducing the emergence of such variants. It has also been shown that intra-host SARS-CoV-2 genetic diversity remains limited during acute infections in healthy hosts. Thus, the sudden emergence of new SARS-CoV-2 variants, often with several key mutations, is hypothesized to be rare during transmission among primarily healthy individuals (Braun et al., 2021).

Ergo, it would be an absolutely insane public health policy to consider an equivalency between vaccination and immunity post-recovery from SARS-CoV-2.

Now, after the presented arguments, I think it’s appropriate to explain why this is subject to the condition that there is adequate vaccine supply and discuss where I think things get gray. A key component of my argument is the fact that recovered patients can and per guidelines still should be vaccinated (for details on this I wrote about them here). But if there is a true shortage of vaccine, it does not make sense to give equal priority to patients who are known to have recovered and those who have never been exposed because we know those who have recovered likely do have fairly robust immunity that at a minimum could help prevent them from straining healthcare resources as they are less likely to need high-acuity care if reinfected compared with a naive patient. In a similar vein, questions exist about how many doses to give recovered patients and at what intervals. In this MMWR a clinical benefit of vaccination against reinfection is confirmed only with both doses of vaccination- though there is a trend towards a lower risk even with one dose (but it does not attain statistical significance likely due to the smaller number of cases). But immunologically it seems that there is a ceiling effect where a second dose 3 weeks after the first fails to confer a benefit as far as antibodies. Ultimately, clinical data should always be held paramount but because of small numbers, it’s a bit gray there too. A single dose strategy for recovered patients can potentially help cover more people in settings of relative scarcity for maximal public health benefit.