Controversy among experts

Most countries outside of North America have not required or mandated booster doses for young healthy adults at universities,14 suggesting that, at a minimum, there is a diversity of expert views on whether the expected benefits of such policies outweigh their potential harms. In July 2021, the Centers for Disease Control and Prevention (CDC) released a joint statement with the Food and Drug Administration (FDA)15 reassuring the public that boosters were not necessary. Just 2 months later, in September 2021, a US FDA advisory committee overwhelmingly voted 16-2 against boosting healthy young adults.16 Yet, this recommendation was over-ruled by the White House and CDC leading to the resignation of two high-level FDA vaccine experts. These experts wrote in The Lancet about the ‘…need to identify specific circumstances in which the direct and indirect benefits of doing so are, on balance, clearly beneficial’.17 To date, the only risk-benefit assessment made public has narrowly focused on myo/pericarditis in the absence of sufficient safety data from an appropriately powered trial.18 In fact, the CDC’s own evidence-to-risk framework found no COVID-19 hospitalisation in either booster (three-dose) or placebo (two-dose) groups of the BNT162b2 booster trial.19

Because the mRNA vaccine third-dose booster trials were too small to measure important clinical endpoints, additional doses have been granted Emergency Use Authorization (EUA) based on observational data suggesting benefits in older populations.19 Prior to the emergence of the Omicron variant, the US CDC estimated19 that administering a booster dose to 8738 (BNT162b2) or 11 994 (mRNA-1273) 18–29 year-olds would prevent one COVID-19 hospitalisation over 6 months. As of August 2022, this estimate had not been updated to reflect increasing natural immunity or waning vaccine effectiveness. Data on booster vaccine effectiveness specific to young adults are scarce; reports typically either do not provide stratified data below a certain age (eg, 50 years20) or use younger adults as the baseline to assess effectiveness in older adults (in part because severe disease is already extremely rare in non-boosted young adults).21 In a recent CDC publication, which stratified for ages 18–49, a booster dose increased effectiveness against emergency department encounters and hospitalisations among immunocompetent adults during the Omicron wave, but the analysis did not adjust for comorbidities and excluded those with a history of prior infection ‘to reduce the influence of protection from previous infection’.22

Risk-benefit calculations for the primary series among younger children and adolescents are similarly limited. A cohort study conducted in Hong Kong estimated the number needed to harm (NNH) from myo/pericarditis for dose 2 of BNT162b2 was 2563 among adolescent males,23 yet there was no US-specific NNH published by the CDC, nor did the agency recommend shifting to a one-dose policy for adolescents as did the UK, Norway, Taiwan and Hong Kong.23 The CDC first presented a risk-benefit analysis of booster vaccination in September 2021, yet the harms focused strictly on myocarditis versus all SAEs and collapsed age strata with very disparate myocarditis risks.24 Moreover, the CDC’s outdated risk-benefit analysis for adolescents and young adults does not distinguish important subgroups such as or those who have recovered from previous infection or healthy young people (as opposed to those with comorbidities or immunocompromised status).24

Current data regarding COVID-19 vaccines

A thorough ethical evaluation of risks and benefits requires relevant empirical data, especially where risks and benefits can be quantified to a reasonable degree of certainty. Relevant data include those regarding average individual vaccine safety and effectiveness and age stratification of these data as well as the protective effect of prior infection and the effectiveness of vaccines against transmission.

Proponents of mandates have argued that current vaccines prevent transmission, which would support a standard ethical reason in favour of mandates: the protection of others. Yet it is increasingly clear that current vaccines provide, at most, partial and transient protection against infection, which decreases precipitously after a few months,25 26 with limited effects on secondary transmission.27 28 The CDC states: ‘anyone with Omicron infection, regardless of vaccination status or whether or not they have symptoms, can spread the virus to others.’29 It is therefore inaccurate in 2022 to infer a sustained or long-term reduction in transmission from a short-term reduction in infection.30

A second limitation is ignoring the protective effects of prior infection. In February 2022, the CDC estimated that 63.7% of adults aged 18–49 years had infection-induced SARS-CoV-2 antibodies, up from 30% in September 2021.13 By September 2022, the majority of young adults, both vaccinated and unvaccinated, are estimated to have been previously infected with COVID-19. Evidence increasingly shows that prior SARS-CoV-2 infection provides at least similar (and perhaps more durable) clinical protection to current vaccines,31–33 which current university policies fail to acknowledge (in addition to more general uncertainties about risks and benefits in relevant age groups34).

Mass vaccination had been proposed as a way to ‘end the pandemic’.35 However, elimination or eradication of the virus is not a tenable goal with vaccines that provide only temporary and incomplete reduction in infection risk, and the presence of multiple animal reservoirs. Because of this, nearly all human beings will eventually be infected with SARS-CoV-2, as with other endemic coronaviruses (and every pandemic influenza virus on record), many times in their life.36 Denmark has, for example, acknowledged vaccinating children was not effective at curbing spread of the virus and is no longer recommending vaccination against COVID-19 for most children.37 38 Taking population immunity into account with variant severity and projected coincident surges of influenza, SARS-CoV-2 and respiratory syncytial virus in the winter of 2022–2023, the UK’s Joint Committee on Vaccination and Immunisation (JCVI) currently recommends that high-risk groups be offered a booster.39

A fourth point relates to the burden of COVID-19 in young adults under 40. Using pre-COVID-19-vaccinera mortality data from 190 countries, the adjusted infection fatality ratio for 18–29 year-olds ranged from 100 per million (18 year-olds) to 500 per million (29 year-olds) with significant variation by country within each age stratum.40 A recent study from South Africa during the Omicron BA.1–BA.4/5 wave demonstrates that despite a high proportion of breakthrough infections, the risk of hospitalisation remains lowest among young adults.41

While both vaccination and prior infection can substantially reduce the likelihood of COVID-19 mortality,32 33 39 the protection against hospitalisation afforded by a booster wanes rapidly.41 The study from South Africa demonstrated that protection waned to less than 50% after 3–4 months.41 Protection against symptomatic disease can be initially restored but wanes approximately 10 weeks after a booster dose42; in the study from England, protection against severe disease could not be measured with the test-negative case–control design due to the few cases of severe disease during Omicron.42 Using a national population-wide data set in Qatar, both previous infection alone and vaccination alone were found to provide >70% protection against severe Omicron (BA.1 or BA.2) disease.43 However, the stratified data in Altarawneh, et al. supplemental table S5 show that prior infection alone was 91% effective against severe Omicron disease, whereas protection from two or three doses of vaccine alone was 66% and 83%, respectively.43

Finally, COVID-19 does cause acute illness, and may have long-term effects (Long COVID) for some, particularly those who develop critical illness, but vaccination may not entirely prevent longer term sequelae44 and the existing data are non-randomised, from variants that predate Omicron and with unclear relevance for adults under age 40. The existence of effective treatments for clinical management45 is also an argument against vaccine mandates, especially for groups not considered at risk for severe illness.

Hospitalisations prevented

To estimate the benefits of hospitalisations prevented by boosters, we updated the CDC’s estimated NNV19 for Omicron, which was found to be markedly less virulent than Delta.47 We selected Trobajo-Sanmartín et al because the analysis provides stepwise comparisons of Omicron BA.1 to Delta (adjusted OR (aOR)=0.28, 95% CI 0.16 to 0.47), as well as the more recent BA.2 to BA.1 (aOR=0.52, 95% CI 0.29 to 0.95). To be conservative, we used the BA.1 versus Delta aOR rather than attempting to estimate the combined BA.2 versus Delta risk reduction. Scaling the CDC’s NNV estimates of 8738 for BNT162b2 and 11 994 for mRNA-1273 by this reduced severity, we estimate that 31 207 (8738/0.28) to 42 836 (11 994/0.28) young adults would need to be boosted with BNT162b2 or mRNA-1273, respectively, to prevent one COVID-19 hospitalisation over a 6-month period. Hospitalisations prevented per million BNT162b2 and mRNA-1273 doses administered are 32.0 and 23.3, respectively (table 1).

SAE rates reported from manufacturer-provided data

Of the 12 SAEs reported in the intervention arm of the randomised controlled trial (RCT) for BNT162b2 (n=5055), three were found by blinded investigators to be attributable to the vaccine, providing a rate of 1 in 1685 (3/5055).19 The three SAEs considered vaccine related included: moderate persistent tachycardia, moderate transient elevated hepatic enzymes and mild elevated hepatic enzymes.19 Based on 31 207 in this age group needing to receive the first BNT162b2 booster to prevent one hospitalisation over a 6-month period, the expected SAE rate is 18.5 (3/5055*31 207). (Table 1A) Per million doses administered, the SAE rate is 593.5. Although the safety populations were small, we also reviewed SAEs reported from these cohorts. Pfizer reported 1/306 but the event was not considered related to the vaccine (1/306=0.3%). Similarly, Moderna found that none of the five SAEs experienced by 4 of 344 participants50 in its safety population (4/344=1.2%)³ were attributable to the vaccine, thus our SAE estimates rely on the only available RCT data (BNT162b2).

Reactogenicity rates

According to self-report data, side effects from the booster dose prevent on average 28.3% of mRNA vaccine recipients from being able to carry out normal daily activities, typically the day after vaccination.55 Sponsor-reported rates from the safety studies for grade ≥3 reactogenicity are 1 in 22 (14/306)50 for the BNT162b2 booster to 1 in 9 (18/167)50 for the mRNA-1273 booster. Per million third doses, reactogenicity rates are therefore 45 751.6–107 784.4, respectively (table 1B). Per COVID-19 hospitalisation prevented over 6 months in adults aged 18–29 years, the expected number of grade ≥3 reactogenicity cases is therefore 1429.7 (45 751.6/32.0) to 4625.9 (107 784.4/23.3), respectively.

In those with a prior SARS-CoV-2 infection, postvaccination symptoms causing missed work or daily activities are reported twofold56 to threefold57 more often than those without a history of infection, a major concern given that seroprevalence among adults aged 18–49 years is now well above the February 2022 estimate of 63.7%.13 Conservatively assuming 63.7% as the proportion with a history of COVID-19 infection, and a twofold increased likelihood of systemic effects, expected grade ≥3 reactogenicity cases per single hospitalisation prevented would be at least 2340.5–7572.7 for BNT162b2 and mRNA-1273 boosters, respectively (table 1B). Even without taking into account prior infection, the proportion reporting to V-Safe being ‘unable to perform daily activities’ was between 20% and 40% depending on booster product, and higher among those receiving a heterologous booster.58

Booster vaccine-associated myocarditis rates in university-age males 18–29 years

The CDC estimated the rate of postbooster myocarditis during days 0–7 following BNT162b2 vaccine administration in males aged 16–17 years to be approximately 1 in 41 50051 using passive surveillance through the Vaccine Adverse Event Reporting System (VAERS), and approximately 1 in 500051 using active surveillance with the Vaccine Safety Datalink (VSD). In males aged 18–29 years, the postbooster myocarditis rate for both products combined using VAERS was reported to be 1 in 101 00052 (ages 18–24) to 1 in 208 00052 (ages 25–29) while the VSD rate was much higher at 1 in 14 20052 (mRNA-1273) to 1 in 21 00052 (BNT162b2). Two other population-based studies from the USA and Israel in males aged 18–39 years found the rate to be 1 in 7000 (147.0 per million third doses)53 to 9000 (126.6 per million third doses).54 In both of these studies, BNT162b2 was the vaccine administered prior to diagnosis. For our estimates, and assuming a precautionary stance, we have used active surveillance rates or population-based rates. For males aged 18–29 years we consider the rate 1 in 700053 to be the most reliable because the method relies on CDC definitions and databases.59 We also provide a 16–17 year-old rate because academic acceleration allows some older adolscents to attend college along with the freshman cohort, and in some cases students need to be vaccinated before their 18th birthday to enrol or be assigned to housing. For males aged 16–17 years, we use the VSD rate of 1 in 5000.51 In table 1C, we provide a range of myopericarditis estimates for consideration.

Risk-benefit estimates

The figures display benefits and harms per million third doses administered: SAEs (figure 1A), grade ≥3 reactogenicity (figure 1B) and myopericarditis (figure 1C). At this scale, and as shown in figure 1A, boosting young adults with BNT162b2 could cause 18.5 times more SAEs per million (593.5) than COVID-19 hospitalisations averted (32.0).

To prevent one hospitalisation over 6 months by boosting 31 207–42 836 students, a large university campus may also expect 1429.7–4625.9 young adults to experience grade ≥3 reactogenicity disrupting daily activities or requiring medical care when vaccinated with a third dose of BNT162b2 or mRNA-1273, respectively. Per million third-doses of mRNA vaccine administered, between 45 751.6 and 107 784.4 cases of grade ≥3 reactogenicity may be created (figure 1B). Given that prior SARS-CoV-2 infection increases the rate of systemic reactions by twofold to threefold,56 57 the number of young adults expected to experience disruptions in their school and daily activities is likely to exceed 74 895.4 with BNT162b2 and 176 443.1 with mRNA-1273 (figure 1B).

Per million third doses of mRNA vaccine administered, 23.3–32.0 hospitalisations may be averted while 47.6–147.0 cases of myo/pericarditis may be caused among young males aged 18–29 years (figure 1C). Thus, to prevent a single hospitalisation among young males aged 18–29 years, we estimate between 1.5 and 4.6 occurrences of myo/pericarditis (rates up to 1 in 700053) among males aged 18–29 years (figure 1C). For adolescents aged 16–17 years and using available data from CDC’s VSD,51 we expect 6.3 cases of myo/pericarditis among males and 1.4 among females. Thus, per single hospitalisation averted by boosting 31 207–42 836 young males in this age group, approximately 1.5–6.3 cases of myopericarditis may result.

Most media reports, as well as a recent systematic review60 and expert opinion from the American College of Cardiology (ACC),61 present vaccination-associated myo/pericarditis as rare, (typically) ‘mild’ and followed by rapid recovery with anti-inflammatory treatment. The reviews have not framed vaccine-associated risks versus infection-associated risks using compatible denominators based on exposure (vaccination) and infection (seroprevalence), thus the infection-associated risks may have been overstated by at least a factor of 4 according to CDC estimates of the burden of COVID-19 illness.62 However, vaccine-associated myocarditis has been found to occur in as many as 1 in 2652 males aged 12–17 years and 1 in 1862 males aged 18–24 years after the second dose59 (and as high as 1/1300 after the second dose in a BNT162b2–mRNA-1273 combination).63 An Israeli study described one in five cases among 16–29 year-olds to be of intermediate severity, meaning these cases had persistent new/worsening abnormalities in left ventricular function, or persistent ECG anomalies, or frequent non-sustained ventricular arrhythmias without syncope.64 The CDC reported that 1200 of the 1314 verified myocarditis cases with known hospitalisation status following the primary series or booster had been hospitalised.65 Among adolescents, 69%66–80%67 of those diagnosed with vaccine-associated myo/pericarditis had findings consistent with cardiac inflammation on MRI testing 3–8 months after the second dose. The potential long-term impact of scar tissue on heart conduction remains unknown.66 67 Postvaccination myocarditis has been found to be equivalent to or exceed the risk of post-COVID myocarditis in males less than 40 years old despite the lack of seroprevalence-based estimates of COVID-associated myocarditis.68 Rare incidences of death in young males attributed to mRNA vaccine-induced myocarditis have also been reported.69 70

Limitations of analysis

These estimates have a number of limitations. First, our estimates rely on sponsor-reported and CDC summaries of AEs; we cannot account for failures to report small sample sizes, poor quality evidence subject to serious bias or loss to follow-up during the clinical trials. Second, our SAE estimate does not distinguish between specific types or the clinical significance of SAEs because of scarce data. The BNT162b2 RCT found more SAEs in the placebo group (24/5020) than the booster group (16/5055). However, blinded investigators attributed as vaccine-related three SAEs in the vaccine group (moderate persistent tachycardia, moderate transient elevated hepatic enzymes and mild elevated hepatic enzymes) and two SAEs in the placebo group (myocardial infarction and chest pain of unknown origin).19 Per million doses, the SAEs were therefore 593.5/million in the vaccine group vs 398.4/million in the placebo group, resulting in a risk difference of 195.1/million doses. The phase II/III BNT162b2 booster trial participants were of median age 42.0 and the company’s adolescent booster trial, for example, included only 78 individuals aged 16–17 years randomised to receive booster or placebo.71 Nevertheless, one male in this age group was hospitalised with myopericarditis after receiving a third dose of BNT162b2.71 It is possible that multiple severe side effects were reported by the same participant in the RCT trials and that the number of people impacted by such reactions is lower than our estimate. Hence, the causal relationship between our estimated SAEs and the COVID-19 vaccines needs to be approached with caution. We are extrapolating SAE data to young adults (18–29 years old) that were originally generated in clinical trials involving all age groups. However, studies have shown that younger people have a greater likelihood of vaccine-related AEs.72

More generally, data limitations affect the CDC’s ability to evaluate both BNT162b2 and mRNA-1273. For example, the CDC’s Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) review50 noted ‘serious’ risk of bias for SAEs and ‘very low’ certainty of evidence (type 4) for all measures. While Pfizer conducted an RCT among 10 000 participants assigned 1:1 to booster or placebo, a sample size of 5000 is not sufficient to detect SAEs occurring at a rate of 1 in 7000 (such as vaccine-associated myo/pericarditis) among a subset of the population aged 18–29 years at highest risk. However, the trial data do suggest that the rate of AEs was higher among the intervention group than the placebo group (25.2% vs 6.8%).73 Moderna conducted a small, non-randomised safety study among 344 participants who elected to get a booster, and the reported SAEs were subject to serious risk of bias.50 ,⁴

Despite these limitations, we believe that the data suggest caution is warranted. Haas et al 74 suggested that many systemic AEs in the RCTs (76% of systemic and 24% of local reactogenicity) may have been due to a nocebo effect—anxiety, expectations and background symptoms. It is very likely, however, that real-world severe or serious AEs may be greater than those reported in the RCT data because standard trials are underpowered to detect rare AEs and there may also be selection bias: those who had a reaction during the primary series may have a greater expectation of harmful side effects to the booster and are less likely to enrol in a trial. In fact, these data are usually collected after a drug has been approved and is on the market (phase IV clinical trial data). Such limitations show the need for more robust postmarketing data and ideally large, controlled trials to determine risks and benefits for any future booster doses, especially in younger age groups.

Universities have not published cumulative AE rates on their COVID-19 dashboards, thus there is no current way to validate these estimates with real-world data. Even with the residual uncertainties, our risk-benefit assessment shows that it is at least plausible that expected individual harms outweigh benefits for young healthy people (ie, most young adults), and it is implausible that individual benefits significantly outweigh risks. Pfizer’s own booster data support this inference.71 In requesting the EUA for boosting adolescent males, the BNT162b2 risk-benefit analysis estimated 23–69 cases of myocarditis per 1 million booster doses administered and 29–69 COVID-19 hospitalisations averted,71 yet this estimate of 23–69 cases of myocarditis per million third BNT162b2 doses administered is now known to be an order of magnitude below the 200.3 per million reported by the US CDC among adolescents aged 16–17 years.51 Finally, our NNV with a booster dose to prevent one hospitalisation likely errs on the side of overestimating the effectiveness of the booster. We do not incorporate the protective effects of prior infection, for example. Recent studies have found rapid waning of effectiveness against hospitalisation during Omicron to <50% by 3–4 months,41 with some studies failing to detect any significant benefit against hospitalisation of a booster dose among those <40.21 If accurate, these data would render our booster risk-benefit analysis even less favourable.

Transparency

Risk-benefit assessment is essential to the ethical acceptability of public health policy, and transparent, peer-reviewed assessments help maintain trust in public health, especially in the context of controversial policies. There is an even stronger rationale for thorough and transparent risk-benefit assessment when interventions are mandated or when (given uncertainty or relevant population differences) some people might face harms not outweighed by individual benefits. In such cases, risk-benefit assessments should be stratified by demographic factors and updated as new data become available to reduce uncertainty. At a minimum, if an intervention is implemented despite significant uncertainty (especially if it is mandated), there is a strong ethical rationale to collect (controlled) data to resolve relevant uncertainties.

An Omicron-era risk-benefit assessment published by the CDC and FDA could provide additional insight into the appropriateness of university booster mandates. However, such a risk-benefit assessment has not been published to date. Without such a formal analysis, professional associations (such as the ACC expert panel61) have been forced to infer from the literature and CDC’s own analyses. For example, the ACC expert panel produced a graphic displaying a favourable harms versus benefits ratio for the second dose among young adults aged 12–29 years.61 The ACC’s widely promoted graphic is tied to data presented by the CDC75 and relies on four key assumptions which bias the findings in favour of vaccination: (1) vaccine effectiveness of 95% over 120 days to prevent COVID-19 cases and hospitalisations; (2) myocarditis rates were derived from passive surveillance in VAERS instead of active surveillance available to the CDC (VSD) resulting in harms being underestimated by a factor of 1051 52; (3) harms and benefits were averaged across ages 12–29 when the risk may be highest among those aged 16–19 years51 52; and (4) hospitalisation rates were tied to May 2021 data, more than a year prior to the ACC’s review and pre-Omicron. Nevertheless, for adolescent males aged 12–17 years, the CDC estimated 56–69 myocarditis cases would be expected while 71 intensive care unit admissions could be averted.75

It was foreseeable that the decision to recommend boosters for all adults (against the advice of the FDA panel) would be followed by booster mandates since pandemic vaccine mandates were already in place in many universities and colleges throughout the USA at the time.13 Universities rely on public health agencies such as the CDC for guidance. Thus, we maintain that if mandates remain then it is critically important to update public risk-benefit estimates for boosters among adults younger than 40, stratified by sex, comorbidity status and history of infection to provide evidence that the intervention confers an expected net benefit to younger individuals in the context of the prevailing SARS-CoV-2 variants and pre-existing immunity. Without this, it is problematic to repeatedly and emphatically claim that COVID-19 vaccines are ‘safe and effective’ without specific risk-benefit analyses for different age categories and with consideration for individual health status, including evidence of prior infection, because risks of both disease and vaccination are highly variable according to these factors.9 10

Since there has not been any RCT specific to evaluating boosters in young adults, the CDC relied on data from an older cohort with a median age of 42.0–51.771 73 and incorrectly assumed that the benefits would also outweigh risks for younger age groups. As we have shown, it is likely that this assumption is incorrect. Under such uncertainties, ethical vaccine policymaking arguably requires transparency about scientific knowledge and uncertainties regarding vaccine risks and benefits (ie, even more transparency than where certainty is high), and at the very least allows for shared decision-making aligned with an appreciation of stratified risks instead of placing the emphasis on simplistic messaging.

Transparent policymaking can encounter a ‘trust paradox’ in providing information about vaccine risks to the public. As noted by Petersen et al,76 governments have a perverse incentive to withhold negative information about vaccines since they are actively promoting such products and negative information about vaccines reduces vaccination uptake. And yet transparent disclosure about negative information (eg, side effects) helps sustain trust in health officials and reduces the politicisation of vaccines.77 Transparency may reduce the uptake of vaccination in the short term but will uphold trust in health authorities and vaccines in the longer term—just as open disclosure regarding clinical harms promotes trust in medicine.78 To address the ‘trust paradox’ in regulatory politics, and to maintain trust in government and scientific institutions, greater data accountability (in this case, a risk-benefit analysis) should precede any policy debate about mandates. Given concerns about pharmaceutical influence on the political process78 79 this should be facilitated by mechanisms to ensure independent scrutiny of regulatory science.79

Potential net expected individual harm

The reasonable possibility of a net harm to individuals (as presented in our risk-benefit assessment) should provide a strong basis to argue for the ethical case against booster mandates for young adults. Mandates at institutions of higher education serve the age group with one of the lowest public health burdens from COVID-19. Hence, boosters provide a low and transient impact on transmission and hospitalisation for an age group with a vague and unquantified prospect of benefit. Arguably, this has been considered by most universities and colleges and is the reason why most do not have booster mandates for the fall of 2022. In fact, this is also likely why European countries, including the UK, France, Germany, Norway, Sweden and Denmark (to our knowledge), never had university-implemented mandates.14 When the European Centre for Disease Prevention and Control (ECDC), a body serving some 300 million European residents, recommended boosters in November 2021, priority was focused on those over age 40.80 Taking a different view of the data, in the fall of 2021 the US CDC recommended boosters for all adults and recommended a second booster for all Americans aged 50 years or more for fall 2022.81 The ECDC, in contrast, recommended that first boosters be ‘offered’ with prioritisation for those over 40 years, and now recommends second boosters for the 2022 autumn campaign only for those over age 60 and those with an immunocompromised status or high-risk medical conditions.82

Reflecting again to fall of 2021, the UK’s JCVI provides an example of using the potential for net harm to advise against the primary vaccination series for 12–15 year-olds.83 The JCVI argued that the potential benefit of vaccination in this age group was only ‘marginally greater than the potential known harms’, since healthy 12–15 year-olds are at very low risk of serious outcomes from COVID-19. Although it may be the case that the JCVI adopted worst-case estimates,84 such an approach reinforces the need to act judiciously under conditions of uncertainty where the clear benefits of an intervention are not confidently above the potential harms. Note also that they mention ‘potential known harms’ without taking into consideration potential long-term effects. The UK Health Ministers subsequently voted to offer a single dose of vaccination to adolescents aged 12–15 years in consideration of: ‘…the health and wider social benefits to this cohort’.85 A second dose was offered to those with underlying health conditions. There are important parallels between the JCVI decision and the outcome of the FDA panel that recommended against universal booster recommendations for adults in the USA in the fall of 2021: in both cases, the US and UK governments disregarded these recommendations. A key ethical difference is that the UK has not implemented any COVID-19 vaccine mandates at schools or universities, and the mandate proposed for care home and healthcare workers was withdrawn.86

As noted above, blanket mandates ignore widely available critical data, such as the benefits of prior infection and data on adverse effects. These factors make an expected net harm now even more likely than when mandates began and make it more urgent to update COVID-19 vaccine policy. Policies for other vaccines have been updated following the accumulation of new data. For example, adult boosters for tetanus and diphtheria vaccines (though previously widely administered) have been shown to provide no benefit.87 Vaccines for influenza, dengue and rotavirus have been withdrawn or had strict limitations placed on their use in children due to unexpected harms.88 Adenovirus-vectored COVID-19 vaccines have been limited in their use due to thrombosis (especially in younger women).89 Uncertainties remain regarding mRNA vaccines, for example, related to their effects on menstruation and fertility,90 shingles91 or the overall safety of current formulations in younger adults and children as well as evidence in support of booster vaccination.92

There are two other theoretical problems that could be factored into mandatory programmes from a precautionary standpoint: original antigenic sin and the non-specific effects of vaccines. Original antigenic sin refers to the decreased ability of an individual to respond to a new viral variant because the immune system has been ‘locked’ onto the original immunogen.93 While data have not shown this to occur with COVID-19 vaccine, it cannot yet be ruled out as an important side effect of repeat vaccination, including with the new bivalent booster. Non-specific effects of vaccination refer to the effects of a vaccine on overall health and all-cause mortality, which have been shown to differ based on the type of vaccine (eg, live vs non-live) and age/sex.94 95 Both of these theoretical issues are at the frontiers of our current knowledge of vaccinology and are rarely considered in the media and by the lay public. We cite these examples to support our main point: proportionality of mandates should account for uncertainty regarding evidence that benefits outweigh harms, especially as the marginal benefits of vaccination and boosting for young adults become vanishingly small with increased population immunity.

Lack of proportionate public health benefit

Proportionality, a key principle in public health ethics, requires that the benefits of a public health policy must be expected to outweigh harms, including harms arising from the restriction of individual liberty and basic human rights such as access to education and employment.1 5–8 86 Where mass vaccination involves harm to a minority of individuals, or coercion or undue inducements are used to increase vaccine uptake, proportionality requires that these considerations be outweighed by public health benefits, typically in the form of reduced transmission from vaccinated individuals to others.96

COVID-19 booster mandates often involve a degree of coercion, including the threat of loss of access to education and free choice of occupation, disproportionately affecting disenfranchised groups.96 Contrary to those who restrict the concept of coercion to situations of a direct threat to something people should have access to as a matter of right,97 we endorse here a broader concept of coercion that includes situations of structural pressure that deprive people of reasonable options.98 ,99 To be ethically acceptable, such severe restrictions of individual liberty need to be justified by an individual benefit and by the expectation that vaccination reduces harm to others. Booster doses of COVID-19 vaccines provide limited lasting reduction in the probability of infection or transmission,27–29 hospitalisation41 and limited expected benefits to young healthy individuals, especially those who have already been infected.31–33 100–102 The net expected harms to individuals and the harms of coercive mandates themselves are not counterbalanced by a large public health benefit (and in fact may harm the public health through the attrition of healthcare workers); such harms and restrictions of liberty are therefore disproportionate and ethically unjustifiable.

Failure of reciprocity

The use of booster mandates raises an additional ethical problem of reciprocity for institutions of higher education and public health authorities.103 104 Most vaccines are covered in the USA105 and Canada106 by an injury compensation programme based on fair (reciprocal) compensation for those who experience a vaccine-related harm. Mandatory vaccines arguably require even stronger protections for individuals who experience consequences that lead to permanent harm107 because their free choice regarding vaccination has been limited. While institutions of higher education are mandating boosters, the US and Canadian compensation programmes have failed to uphold their social justice responsibility to injured individuals. In the USA, COVID-19 vaccines and therapeutics are processed by the Countermeasures Injury Compensation Program (CICP) which is designed to cover epidemics, pandemics and security threats as designated by the Secretary of Health and Human Services and as authorised by the Public Readiness and Emergency Preparedness (PREP) Act.105 As of 1 August 2022, thirty-seven claims have been denied compensation because ‘the standard of proof for causation was not met’ or ‘a covered injury was not sustained’.108 No claims have been paid out by the US CICP but one claim for anaphylaxis has been approved for compensation and payout is currently pending assessment of eligible expenses.108

The federal US vaccine injury programme has failed to compensate but one COVID-19 vaccine-injured individual in the context of booster mandates in place at hundreds of US universities.108 It is also important to note that boosters have been granted an EUA by the FDA, but are still not fully approved.109 Thus, universities and colleges that mandate COVID-19 boosters are pressuring young adults to receive a vaccine that, in case of injury, has no transparent legal route to adequate compensation. In sum, one core precondition for vaccine mandates is a functioning and fair compensation programme, which has not been achieved for COVID-19 vaccines.

Wider social harms

Strong coercion may create significant social harms. COVID-19 vaccine mandates have generally involved a high degree of coercion, effectively ostracising unvaccinated individuals from society. University mandates involve significant coercion in that they exclude unvaccinated people from the benefits of university education (or employment) and thereby entail major infringements to free choice of occupation and freedom of association. When such mandates are not supported by a compelling public health justification and where exemptions are not easily available, the likelihood of reactance and negative social effects are increased.1 The social harms of university COVID-19 mandates have not been formally studied, but there is reason to think that they will be significant.1 Policies can have wide-ranging consequences for non-compliance, such as loss of employment, loss of internet use, restriction to off-campus versus on-campus housing, delays or refusal to process student housing requests, loss of enrolment, a hold placed on grades, inability to use recreational facilities to train or compete in sports, access to scholarships for competitive sports, registration for class and delays in ability to repay student loans after graduation. A number of young adults and professors affected by mandates have outlined publicly their perspectives and the social harms of these policies, such as loss of access to schooling and social services,110 psychosocial stress, reputational damage and lost income and threats of being disenrolled or deported.111 This punitive public health approach may also provoke reactance in young adults,1 with long-term negative consequences on trust in society and institutions and vaccine confidence in general, including vaccine hesitancy for routine paediatric and adult vaccines, a problem which predated the pandemic and is considered one of the WHO’s top 10 threats to global health.112