The article

In their article, the authors argue that it is ethically imperative that COVID-19 mandates exempt those with natural immunity. The authors argue that those who have become immune through infection are ‘potentially equivalent’ to those who were vaccinated, and refusing them an exemption is ethically unjustified—they call this the ‘unnaturalistic fallacy’. The authors convincingly conclude that existing data suggest that previous infection leads to durable immunity for at least several months and protects against reinfection (if not perfectly). The authors acknowledge that immunity from infection may decline, but point out that immunity from vaccines does, too. Although the authors acknowledge that Omicron challenges the picture, they point out that there is uncertainty regarding the effect of Omicron on both vaccine-derived and infection-derived immunity.

The authors’ case for allowing a natural immunity exemption is based on the argument that vaccine requirements have real costs for personal liberty—they limit liberty—and ‘there are non-trivial risks associated with vaccination’. The authors are also concerned about the effect of mandates on staffing of some professions—for example, nursing homes—and exacerbating existing social inequalities. The authors conclude that mandates need to be justified as necessary for achieving a public health benefit, and in cases of uncertainty they cannot be ethically imposed. Since the authors conclude there is no clear evidence vaccines reduce spread more than natural immunity, there is no ethical basis to refuse a natural immunity exemption.

The authors add that they envision the exemption to require ‘sufficient proof’ of natural immunity, either a recent positive PCR test or a serological test showing neutralising antibodies.

The authors see an approach that would place the burden of proof in uncertainty on those arguing for the exemption is flawed, because of the high costs of mandates to the individuals involved—in their words, it is ‘ethically fraught’ to ‘assume that avoiding an uncertain (but very likely low) risk of increased viral transmission should take precedence over avoiding the known and quantifiable harms of restricting individual liberties in this way’.

The ethical case for mandates

I agree with the authors that choosing to provide an exemption to those previously infected—as several countries did, as the authors point out—is reasonable. Policy makers could reasonably conclude that, in times of uncertainty, treating natural immunity as equivalent to vaccine-induced immunity is reasonable and can reduce implementation challenges. However, reasonable policy makers can also conclude, justifiably, that the risks and costs of such an exemption are too high.

The goal of a vaccine mandate is to increase vaccination rates, and there is extensive evidence that employment-related vaccine mandates do that.3 Increasing vaccination rates has social benefits in reducing disease and its sequelae in the society, and thus decreasing the burden of disease on hospitals, the need for support for those left with long-term harm and the risk to others. The authors emphasise the costs of vaccine mandates, but it is important to remember that even for the people subject to the mandate, there is a benefit—most of them, by getting the vaccine, will have their immunity boosted (even if they were previously infected), and their risk of disease, disability and death reduced. Although there is uncertainty, there is still evidence COVID-19 vaccines prevent transmission, and the comparative effectiveness of vaccines versus natural infection against Omicron is unclear—but substantial previous evidence suggests even the previously infected benefit from vaccination.4 5 A mandate on the previously infected would have some social benefits in reducing the burden they put on the health system and the potential risks to others, so it is reasonable to give officials the discretion on whether the practical considerations support exempting those with natural immunity or not.

While some unvaccinated and previously infected may resign, how many—and what impact it will have on institutions—is unclear. Experience does not show mass resignations as a result of a mandate; most people vaccinate. The institutional strain, therefore, is uncertain. Further, institutions may face strains if low vaccination rates lead to more severe disease or deaths.

Risks from COVID-19 mRNA vaccines are very low, both absolutely and in comparison with the disease, although the AstraZeneca vaccine may have additional risks. That said, benefits are lower for those previously infected. By itself, this may support the view that there are not enough benefits compared with the risks to justify a mandate—if not for the practical implementation challenges mandates face.

Pragmatic issues are real

What tips the balance towards official discretion, in my view, are the real practical challenges in implementing a previous infection exemption. Those practical challenges also undermine the authors’ point about inequalities: the exemption can exacerbate inequalities.

The authors suggest that the exemption needs to be based on valid testing. But the authors may be underestimating the availability and accessibility of valid testing. First, the evidence that we have strong serological testing to identify prior immunity is limited.6 In fact, we probably do not. This would suggest that PCR at the time of infection is the alternative. But not everyone would have such testing—early infections would not, and there may not be access in the relevant period. And both types of tests are expensive and may not be covered by insurance. That means that people from low socioeconomic status—the ones the authors suggest most need the exemption—will be less likely to have access to them. The authors suggest the person requesting the exemption should pay, which again could exacerbate inequalities.

The alternative is to accept other forms of evidence of prior infection, like a doctor certificate. This creates real concerns about policing abuse and fraud. There are doctors, for example, willing to sell treatments with no evidence behind them, like ivermectin and hydroxychloroquine, and those doctors may also be willing to sell fake immunity certificates. In the USA there has been evidence of widespread sale of fake vaccine cards. This, too, is likely to be more accessible to the well-off, making the prior infection exemption something that exacerbates, rather than ameliorates, inequalities.

These concerns need to be balanced with the practical effects of implementing vaccine mandates against those previously infected, and the authors set those out in helpful details, but that, too, supports giving officials the discretion to decide which is the lesser of the two evils.

Finally, the authors suggest that opponents may be exaggerating the number of people seeking to infect themselves to get such an exemption and underestimating the ‘large numbers of people’ whose liberty may be restricted by a mandate, although they were previously infected. But in both cases, we are estimating, and the authors’ estimate may be wrong in two ways. The authors may be underestimating the willingness of people to seek infections of a virus they have been told is not dangerous to avoid vaccines, and may be overestimating the number of people who are unvaccinated, previously infected and feel strongly enough about a mandate for it to be a meaningful limit on their freedom (levels of opposition vary; some people may not want to vaccinate unless mandated, but not all will perceive the mandate as a major interference with their freedom).

Policy makers on the ground are better placed to assess the relative prevalence of attitudes (even though they may not be very well placed) and can make the more fitting choice for their community.