In 2022, students at North American universities with third-dose COVID-19 vaccine mandates risk disenrolment if unvaccinated. To assess the appropriateness of booster mandates in this age group, we combine empirical risk-benefit assessment and ethical analysis. To prevent one COVID-19 hospitalisation over a 6-month period, we estimate that 31 207-42 836 young adults aged 18-29 years must receive a third mRNA vaccine. Booster mandates in young adults are expected to cause a net harm: per COVID-19 hospitalisation prevented, we anticipate at least 18.5 serious adverse events from mRNA vaccines, including 1.5-4.6 booster-associated myopericarditis cases in males (typically requiring hospitalisation). We also anticipate 1430-4626 cases of grade ≥3 reactogenicity interfering with daily activities (although typically not requiring hospitalisation). University booster mandates are unethical because they: (1) are not based on an updated (Omicron era) stratified risk-benefit assessment for this age group; (2) may result in a net harm to healthy young adults; (3) are not proportionate: expected harms are not outweighed by public health benefits given modest and transient effectiveness of vaccines against transmission; (4) violate the reciprocity principle because serious vaccine-related harms are not reliably compensated due to gaps in vaccine injury schemes; and (5) may result in wider social harms. We consider counterarguments including efforts to increase safety on campus but find these are fraught with limitations and little scientific support. Finally, we discuss the policy relevance of our analysis for primary series COVID-19 vaccine mandates.

For the first comparison, we weighed predicted hospitalisations prevented by one booster dose of BNT162b2 with vaccine-associated SAEs from the manufacturer's randomised trial (3/5055).3 We found that the rate of expected SAEs would outweigh the benefits of the booster against hospitalisation by at least 18-fold. Lam and Nichols suggest that this was an inappropriate comparison, as not all SAEs result in hospitalisation. However, the definition of SAE as used in the trial included death, hospitalisation, disability, permanent damage, life-threatening event or condition, which required medical or surgical intervention to prevent a serious outcome.4 While all comparisons include some degree of incommensurability, comparing these SAEs with hospitalisations prevented by the booster is more reasonable than Lam and Nichols' suggestion of comparing SAEs to infections prevented. The COVID-19 infection hospitalisation risk in this age group was <0.5% (or <1/200)5 even prior to widespread immunity, thus comparing SAEs to infection risk is entirely inappropriate. Furthermore, a booster dose will only offer transient (if any) protection against infection6 and cumulative infection rates in the boosted versus unboosted are expected to be approximately the same after several months.6 We disagree that an SAE from vaccination should be considered equivalent to or in any way comparable with postponing an infection for a few months.