thesis statement on covid 19 vaccine

• < Previous

Home > COMMUNITIES > STUDENTWORK > HONORSTHESIS > 201

Honor Scholar Theses

Attitudes towards covid-19 vaccination: literature review and attitudes of individuals who delayed vaccination.

Sydney Hornberger '22 , DePauw University

Date of Award

Document type, first advisor.

Dr. Ted Bitner

Second Advisor

Dr. Alicia Suarez

Third Advisor

Dr. Jeffrey Jones

This thesis examines attitudes towards and ethics of receiving one of the fastest vaccines ever developed— the COVID-19 vaccine. The Food and Drug Administrations (FDA) in the U.S. has granted either Emergency Use Authorization or full approval to three vaccines: the Pfizer-BioNTech, Johnson & Johnson, and Moderna-NIAID vaccines. However, although the FDA approved and the Center for Disease Control and Prevention (CDC) recommends getting the vaccines, that does not necessarily mean people have an ethical responsibility or a positive attitude towards getting vaccinated against COVID-19; this current paper explores both of these ideas as related to COVID-19 vaccination. First, it surveys sources highlighting the utility of vaccines to control infectious diseases and pandemics. Next, it questions whether getting vaccinated against any disease, and specifically COVID-19, is the ethical action to take. Then, there is a literature review of research into attitudes towards the COVID-19 vaccine, determining the most prevalent attitudes across all people and within specific demographics such as women, people belonging to certain political and religious groups, racial and ethnic minorities, and children. Finally, the results of a study conducted at DePauw University to investigate attitudes, attitude changes, and motivations of recently vaccinated individuals are reported in order to elucidate certain factors that may be useful to understand vaccine decision making.

Recommended Citation

Hornberger, Sydney '22, "Attitudes Towards COVID-19 Vaccination: Literature Review and Attitudes of Individuals Who Delayed Vaccination" (2022). Honor Scholar Theses . 201, Scholarly and Creative Work from DePauw University. https://scholarship.depauw.edu/studentresearch/201

Since June 07, 2022

Included in

Public Health Commons

Advanced Search

• Notify me via email or RSS
• Collections
• Disciplines

Author Corner

Library links.

• DePauw Libraries

Home | About | FAQ | My Account | Accessibility Statement

Privacy Copyright

The Legality and Ethics of Mandating COVID-19 Vaccination

• PMID: 34840152
• DOI: 10.1353/pbm.2021.0037

In light of ongoing concerns in the US that COVID-19 vaccine uptake is stagnating and that cases remain high amongst the unvaccinated, there is growing interest in increasing uptake by mandating vaccination. COVID-19 vaccine mandates must be understood and assessed in terms of who is requiring vaccination and who is required to be vaccinated. This essay considers the legal and ethical implications of states mandating vaccination for children and adults, as well as of employers mandating vaccines for employees. We conclude that COVID-19 vaccine mandates are legally and ethically permissible.

• COVID-19 Vaccines
• Vaccination

Evidence Review of the Adverse Effects of COVID-19 Vaccination and Intramuscular Vaccine Administration

Vaccines are a public health success story, as they have prevented or lessened the effects of many infectious diseases. To address concerns around potential vaccine injuries, the Health Resources and Services Administration (HRSA) administers the Vaccine Injury Compensation Program (VICP) and the Countermeasures Injury Compensation Program (CICP), which provide compensation to those who assert that they were injured by routine vaccines or medical countermeasures, respectively. The National Academies of Sciences, Engineering, and Medicine have contributed to the scientific basis for VICP compensation decisions for decades.

HRSA asked the National Academies to convene an expert committee to review the epidemiological, clinical, and biological evidence about the relationship between COVID-19 vaccines and specific adverse events, as well as intramuscular administration of vaccines and shoulder injuries. This report outlines the committee findings and conclusions.

Read Full Description

• Digital Resource: Evidence Review of the Adverse Effects of COVID-19 Vaccination
• Digital Resource: Evidence Review of Shoulder Injuries from Intramuscular Administration of Vaccines
• Press Release

Recent News

NAS Launches Science and Innovation Fund for Ukraine

Science Academies Issue Statements to Inform G7 Talks

Supporting Family Caregivers in STEMM

A Vision for High-Quality Preschool for All

• Load More...

Log in using your username and password

• Search More Search for this keyword Advanced search
• Latest content
• Current issue
• Browse by collection
• BMJ Journals More You are viewing from: Google Indexer

You are here

• Volume 110, Issue 9
• The role of COVID-19 vaccines in preventing post-COVID-19 thromboembolic and cardiovascular complications
• Article Text
• Article info
• Citation Tools
• Rapid Responses
• Article metrics

• Núria Mercadé-Besora 1 , 2 , 3 ,
• Xintong Li 1 ,
• Raivo Kolde 4 ,
• Nhung TH Trinh 5 ,
• Maria T Sanchez-Santos 1 ,
• Wai Yi Man 1 ,
• Elena Roel 3 ,
• Carlen Reyes 3 ,
• http://orcid.org/0000-0003-0388-3403 Antonella Delmestri 1 ,
• Hedvig M E Nordeng 6 , 7 ,
• http://orcid.org/0000-0002-4036-3856 Anneli Uusküla 8 ,
• http://orcid.org/0000-0002-8274-0357 Talita Duarte-Salles 3 , 9 ,
• Clara Prats 2 ,
• http://orcid.org/0000-0002-3950-6346 Daniel Prieto-Alhambra 1 , 9 ,
• http://orcid.org/0000-0002-0000-0110 Annika M Jödicke 1 ,
• Martí Català 1
• 1 Pharmaco- and Device Epidemiology Group, Health Data Sciences, Botnar Research Centre, NDORMS , University of Oxford , Oxford , UK
• 2 Department of Physics , Universitat Politècnica de Catalunya , Barcelona , Spain
• 3 Fundació Institut Universitari per a la recerca a l'Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol) , IDIAP Jordi Gol , Barcelona , Catalunya , Spain
• 4 Institute of Computer Science , University of Tartu , Tartu , Estonia
• 5 Pharmacoepidemiology and Drug Safety Research Group, Department of Pharmacy, Faculty of Mathematics and Natural Sciences , University of Oslo , Oslo , Norway
• 6 School of Pharmacy , University of Oslo , Oslo , Norway
• 7 Division of Mental Health , Norwegian Institute of Public Health , Oslo , Norway
• 8 Department of Family Medicine and Public Health , University of Tartu , Tartu , Estonia
• 9 Department of Medical Informatics, Erasmus University Medical Center , Erasmus University Rotterdam , Rotterdam , Zuid-Holland , Netherlands
• Correspondence to Prof Daniel Prieto-Alhambra, Pharmaco- and Device Epidemiology Group, Health Data Sciences, Botnar Research Centre, NDORMS, University of Oxford, Oxford, UK; daniel.prietoalhambra{at}ndorms.ox.ac.uk

Objective To study the association between COVID-19 vaccination and the risk of post-COVID-19 cardiac and thromboembolic complications.

Methods We conducted a staggered cohort study based on national vaccination campaigns using electronic health records from the UK, Spain and Estonia. Vaccine rollout was grouped into four stages with predefined enrolment periods. Each stage included all individuals eligible for vaccination, with no previous SARS-CoV-2 infection or COVID-19 vaccine at the start date. Vaccination status was used as a time-varying exposure. Outcomes included heart failure (HF), venous thromboembolism (VTE) and arterial thrombosis/thromboembolism (ATE) recorded in four time windows after SARS-CoV-2 infection: 0–30, 31–90, 91–180 and 181–365 days. Propensity score overlap weighting and empirical calibration were used to minimise observed and unobserved confounding, respectively.

Fine-Gray models estimated subdistribution hazard ratios (sHR). Random effect meta-analyses were conducted across staggered cohorts and databases.

Results The study included 10.17 million vaccinated and 10.39 million unvaccinated people. Vaccination was associated with reduced risks of acute (30-day) and post-acute COVID-19 VTE, ATE and HF: for example, meta-analytic sHR of 0.22 (95% CI 0.17 to 0.29), 0.53 (0.44 to 0.63) and 0.45 (0.38 to 0.53), respectively, for 0–30 days after SARS-CoV-2 infection, while in the 91–180 days sHR were 0.53 (0.40 to 0.70), 0.72 (0.58 to 0.88) and 0.61 (0.51 to 0.73), respectively.

Conclusions COVID-19 vaccination reduced the risk of post-COVID-19 cardiac and thromboembolic outcomes. These effects were more pronounced for acute COVID-19 outcomes, consistent with known reductions in disease severity following breakthrough versus unvaccinated SARS-CoV-2 infection.

• Epidemiology
• PUBLIC HEALTH
• Electronic Health Records

Data availability statement

Data may be obtained from a third party and are not publicly available. CPRD: CPRD data were obtained under the CPRD multi-study license held by the University of Oxford after Research Data Governance (RDG) approval. Direct data sharing is not allowed. SIDIAP: In accordance with current European and national law, the data used in this study is only available for the researchers participating in this study. Thus, we are not allowed to distribute or make publicly available the data to other parties. However, researchers from public institutions can request data from SIDIAP if they comply with certain requirements. Further information is available online ( https://www.sidiap.org/index.php/menu-solicitudesen/application-proccedure ) or by contacting SIDIAP ([email protected]). CORIVA: CORIVA data were obtained under the approval of Research Ethics Committee of the University of Tartu and the patient level data sharing is not allowed. All analyses in this study were conducted in a federated manner, where analytical code and aggregated (anonymised) results were shared, but no patient-level data was transferred across the collaborating institutions.

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See:  https://creativecommons.org/licenses/by/4.0/ .

https://doi.org/10.1136/heartjnl-2023-323483

Statistics from Altmetric.com

Request permissions.

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

COVID-19 vaccines proved to be highly effective in reducing the severity of acute SARS-CoV-2 infection.

While COVID-19 vaccines were associated with increased risk for cardiac and thromboembolic events, such as myocarditis and thrombosis, the risk of complications was substantially higher due to SARS-CoV-2 infection.

WHAT THIS STUDY ADDS

COVID-19 vaccination reduced the risk of heart failure, venous thromboembolism and arterial thrombosis/thromboembolism in the acute (30 days) and post-acute (31 to 365 days) phase following SARS-CoV-2 infection. This effect was stronger in the acute phase.

The overall additive effect of vaccination on the risk of post-vaccine and/or post-COVID thromboembolic and cardiac events needs further research.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

COVID-19 vaccines proved to be highly effective in reducing the risk of post-COVID cardiovascular and thromboembolic complications.

Introduction

COVID-19 vaccines were approved under emergency authorisation in December 2020 and showed high effectiveness against SARS-CoV-2 infection, COVID-19-related hospitalisation and death. 1 2 However, concerns were raised after spontaneous reports of unusual thromboembolic events following adenovirus-based COVID-19 vaccines, an association that was further assessed in observational studies. 3 4 More recently, mRNA-based vaccines were found to be associated with a risk of rare myocarditis events. 5 6

On the other hand, SARS-CoV-2 infection can trigger cardiac and thromboembolic complications. 7 8 Previous studies showed that, while slowly decreasing over time, the risk for serious complications remain high for up to a year after infection. 9 10 Although acute and post-acute cardiac and thromboembolic complications following COVID-19 are rare, they present a substantial burden to the affected patients, and the absolute number of cases globally could become substantial.

Recent studies suggest that COVID-19 vaccination could protect against cardiac and thromboembolic complications attributable to COVID-19. 11 12 However, most studies did not include long-term complications and were conducted among specific populations.

Evidence is still scarce as to whether the combined effects of COVID-19 vaccines protecting against SARS-CoV-2 infection and reducing post-COVID-19 cardiac and thromboembolic outcomes, outweigh any risks of these complications potentially associated with vaccination.

We therefore used large, representative data sources from three European countries to assess the overall effect of COVID-19 vaccines on the risk of acute and post-acute COVID-19 complications including venous thromboembolism (VTE), arterial thrombosis/thromboembolism (ATE) and other cardiac events. Additionally, we studied the comparative effects of ChAdOx1 versus BNT162b2 on the risk of these same outcomes.

Data sources

We used four routinely collected population-based healthcare datasets from three European countries: the UK, Spain and Estonia.

For the UK, we used data from two primary care databases—namely, Clinical Practice Research Datalink, CPRD Aurum 13 and CPRD Gold. 14 CPRD Aurum currently covers 13 million people from predominantly English practices, while CPRD Gold comprises 3.1 million active participants mostly from GP practices in Wales and Scotland. Spanish data were provided by the Information System for the Development of Research in Primary Care (SIDIAP), 15 which encompasses primary care records from 6 million active patients (around 75% of the population in the region of Catalonia) linked to hospital admissions data (Conjunt Mínim Bàsic de Dades d’Alta Hospitalària). Finally, the CORIVA dataset based on national health claims data from Estonia was used. It contains all COVID-19 cases from the first year of the pandemic and ~440 000 randomly selected controls. CORIVA was linked to the death registry and all COVID-19 testing from the national health information system.

Databases included sociodemographic information, diagnoses, measurements, prescriptions and secondary care referrals and were linked to vaccine registries, including records of all administered vaccines from all healthcare settings. Data availability for CPRD Gold ended in December 2021, CPRD Aurum in January 2022, SIDIAP in June 2022 and CORIVA in December 2022.

All databases were mapped to the Observational Medical Outcomes Partnership Common Data Model (OMOP CDM) 16 to facilitate federated analytics.

Multinational network staggered cohort study: study design and participants

The study design has been published in detail elsewhere. 17 Briefly, we used a staggered cohort design considering vaccination as a time-varying exposure. Four staggered cohorts were designed with each cohort representing a country-specific vaccination rollout phase (eg, dates when people became eligible for vaccination, and eligibility criteria).

The source population comprised all adults registered in the respective database for at least 180 days at the start of the study (4 January 2021 for CPRD Gold and Aurum, 20 February 2021 for SIDIAP and 28 January 2021 for CORIVA). Subsequently, each staggered cohort corresponded to an enrolment period: all people eligible for vaccination during this time were included in the cohort and people with a history of SARS-CoV-2 infection or COVID-19 vaccination before the start of the enrolment period were excluded. Across countries, cohort 1 comprised older age groups, whereas cohort 2 comprised individuals at risk for severe COVID-19. Cohort 3 included people aged ≥40 and cohort 4 enrolled people aged ≥18.

In each cohort, people receiving a first vaccine dose during the enrolment period were allocated to the vaccinated group, with their index date being the date of vaccination. Individuals who did not receive a vaccine dose comprised the unvaccinated group and their index date was assigned within the enrolment period, based on the distribution of index dates in the vaccinated group. People with COVID-19 before the index date were excluded.

Follow-up started from the index date until the earliest of end of available data, death, change in exposure status (first vaccine dose for those unvaccinated) or outcome of interest.

COVID-19 vaccination

All vaccines approved within the study period from January 2021 to July 2021—namely, ChAdOx1 (Oxford/AstraZeneca), BNT162b2 (BioNTech/Pfizer]) Ad26.COV2.S (Janssen) and mRNA-1273 (Moderna), were included for this study.

Post-COVID-19 outcomes of interest

Outcomes of interest were defined as SARS-CoV-2 infection followed by a predefined thromboembolic or cardiac event of interest within a year after infection, and with no record of the same clinical event in the 6 months before COVID-19. Outcome date was set as the corresponding SARS-CoV-2 infection date.

COVID-19 was identified from either a positive SARS-CoV-2 test (polymerase chain reaction (PCR) or antigen), or a clinical COVID-19 diagnosis, with no record of COVID-19 in the previous 6 weeks. This wash-out period was imposed to exclude re-recordings of the same COVID-19 episode.

Post-COVID-19 outcome events were selected based on previous studies. 11–13 Events comprised ischaemic stroke (IS), haemorrhagic stroke (HS), transient ischaemic attack (TIA), ventricular arrhythmia/cardiac arrest (VACA), myocarditis/pericarditis (MP), myocardial infarction (MI), heart failure (HF), pulmonary embolism (PE) and deep vein thrombosis (DVT). We used two composite outcomes: (1) VTE, as an aggregate of PE and DVT and (2) ATE, as a composite of IS, TIA and MI. To avoid re-recording of the same complication we imposed a wash-out period of 90 days between records. Phenotypes for these complications were based on previously published studies. 3 4 8 18

All outcomes were ascertained in four different time periods following SARS-CoV-2 infection: the first period described the acute infection phase—that is, 0–30 days after COVID-19, whereas the later periods - which are 31–90 days, 91–180 days and 181–365 days, illustrate the post-acute phase ( figure 1 ).

• Download figure
• Open in new tab
• Download powerpoint

Study outcome design. Study outcomes of interest are defined as a COVID-19 infection followed by one of the complications in the figure, within a year after infection. Outcomes were ascertained in four different time windows after SARS-CoV-2 infection: 0–30 days (namely the acute phase), 31–90 days, 91–180 days and 181–365 days (these last three comprise the post-acute phase).

Negative control outcomes

Negative control outcomes (NCOs) were used to detect residual confounding. NCOs are outcomes which are not believed to be causally associated with the exposure, but share the same bias structure with the exposure and outcome of interest. Therefore, no significant association between exposure and NCO is to be expected. Our study used 43 different NCOs from previous work assessing vaccine effectiveness. 19

Statistical analysis

Federated network analyses.

A template for an analytical script was developed and subsequently tailored to include the country-specific aspects (eg, dates, priority groups) for the vaccination rollout. Analyses were conducted locally for each database. Only aggregated data were shared and person counts <5 were clouded.

Propensity score weighting

Large-scale propensity scores (PS) were calculated to estimate the likelihood of a person receiving the vaccine based on their demographic and health-related characteristics (eg, conditions, medications) prior to the index date. PS were then used to minimise observed confounding by creating a weighted population (overlap weighting 20 ), in which individuals contributed with a different weight based on their PS and vaccination status.

Prespecified key variables included in the PS comprised age, sex, location, index date, prior observation time in the database, number of previous outpatient visits and previous SARS-CoV-2 PCR/antigen tests. Regional vaccination, testing and COVID-19 incidence rates were also forced into the PS equation for the UK databases 21 and SIDIAP. 22 In addition, least absolute shrinkage and selection operator (LASSO) regression, a technique for variable selection, was used to identify additional variables from all recorded conditions and prescriptions within 0–30 days, 31–180 days and 181-any time (conditions only) before the index date that had a prevalence of >0.5% in the study population.

PS were then separately estimated for each staggered cohort and analysis. We considered covariate balance to be achieved if absolute standardised mean differences (ASMDs) were ≤0.1 after weighting. Baseline characteristics such as demographics and comorbidities were reported.

Effect estimation

To account for the competing risk of death associated with COVID-19, Fine-and-Grey models 23 were used to calculate subdistribution hazard ratios (sHRs). Subsequently, sHRs and confidence intervals were empirically calibrated from NCO estimates 24 to account for unmeasured confounding. To calibrate the estimates, the empirical null distribution was derived from NCO estimates and was used to compute calibrated confidence intervals. For each outcome, sHRs from the four staggered cohorts were pooled using random-effect meta-analysis, both separately for each database and across all four databases.

Sensitivity analysis

Sensitivity analyses comprised 1) censoring follow-up for vaccinated people at the time when they received their second vaccine dose and 2) considering only the first post-COVID-19 outcome within the year after infection ( online supplemental figure S1 ). In addition, comparative effectiveness analyses were conducted for BNT162b2 versus ChAdOx1.

Supplemental material

Data and code availability.

All analytic code for the study is available in GitHub ( https://github.com/oxford-pharmacoepi/vaccineEffectOnPostCovidCardiacThromboembolicEvents ), including code lists for vaccines, COVID-19 tests and diagnoses, cardiac and thromboembolic events, NCO and health conditions to prioritise patients for vaccination in each country. We used R version 4.2.3 and statistical packages survival (3.5–3), Empirical Calibration (3.1.1), glmnet (4.1-7), and Hmisc (5.0–1).

Patient and public involvement

Owing to the nature of the study and the limitations regarding data privacy, the study design, analysis, interpretation of data and revision of the manuscript did not involve any patients or members of the public.

All aggregated results are available in a web application ( https://dpa-pde-oxford.shinyapps.io/PostCovidComplications/ ).

We included over 10.17 million vaccinated individuals (1 618 395 from CPRD Gold; 5 729 800 from CPRD Aurum; 2 744 821 from SIDIAP and 77 603 from CORIVA) and 10.39 million unvaccinated individuals (1 640 371; 5 860 564; 2 588 518 and 302 267, respectively). Online supplemental figures S2-5 illustrate study inclusion for each database.

Adequate covariate balance was achieved after PS weighting in most studies: CORIVA (all cohorts) and SIDIAP (cohorts 1 and 4) did not contribute to ChAdOx1 subanalyses owing to sample size and covariate imbalance. ASMD results are accessible in the web application.

NCO analyses suggested residual bias after PS weighting, with a majority of NCOs associated positively with vaccination. Therefore, calibrated estimates are reported in this manuscript. Uncalibrated effect estimates and NCO analyses are available in the web interface.

Population characteristics

Table 1 presents baseline characteristics for the weighted populations in CPRD Aurum, for illustrative purposes. Online supplemental tables S1-25 summarise baseline characteristics for weighted and unweighted populations for each database and comparison. Across databases and cohorts, populations followed similar patterns: cohort 1 represented an older subpopulation (around 80 years old) with a high proportion of women (57%). Median age was lowest in cohort 4 ranging between 30 and 40 years.

• View inline

Characteristics of weighted populations in CPRD Aurum database, stratified by staggered cohort and exposure status. Exposure is any COVID-19 vaccine

COVID-19 vaccination and post-COVID-19 complications

Table 2 shows the incidence of post-COVID-19 VTE, ATE and HF, the three most common post-COVID-19 conditions among the studied outcomes. Outcome counts are presented separately for 0–30, 31–90, 91–180 and 181–365 days after SARS-CoV-2 infection. Online supplemental tables S26-36 include all studied complications, also for the sensitivity and subanalyses. Similar pattern for incidences were observed across all databases: higher outcome rates in the older populations (cohort 1) and decreasing frequency with increasing time after infection in all cohorts.

Number of records (and risk per 10 000 individuals) for acute and post-acute COVID-19 cardiac and thromboembolic complications, across cohorts and databases for any COVID-19 vaccination

Forest plots for the effect of COVID-19 vaccines on post-COVID-19 cardiac and thromboembolic complications; meta-analysis across cohorts and databases. Dashed line represents a level of heterogeneity I 2 >0.4. ATE, arterial thrombosis/thromboembolism; CD+HS, cardiac diseases and haemorrhagic stroke; VTE, venous thromboembolism.

Results from calibrated estimates pooled in meta-analysis across cohorts and databases are shown in figure 2 .

Reduced risk associated with vaccination is observed for acute and post-acute VTE, DVT, and PE: acute meta-analytic sHR are 0.22 (95% CI, 0.17–0.29); 0.36 (0.28–0.45); and 0.19 (0.15–0.25), respectively. For VTE in the post-acute phase, sHR estimates are 0.43 (0.34–0.53), 0.53 (0.40–0.70) and 0.50 (0.36–0.70) for 31–90, 91–180, and 181–365 days post COVID-19, respectively. Reduced risk of VTE outcomes was observed in vaccinated across databases and cohorts, see online supplemental figures S14–22 .

Similarly, the risk of ATE, IS and MI in the acute phase after infection was reduced for the vaccinated group, sHR of 0.53 (0.44–0.63), 0.55 (0.43–0.70) and 0.49 (0.38–0.62), respectively. Reduced risk associated with vaccination persisted for post-acute ATE, with sHR of 0.74 (0.60–0.92), 0.72 (0.58–0.88) and 0.62 (0.48–0.80) for 31–90, 91–180 and 181–365 days post-COVID-19, respectively. Risk of post-acute MI remained lower for vaccinated in the 31–90 and 91–180 days after COVID-19, with sHR of 0.64 (0.46–0.87) and 0.64 (0.45–0.90), respectively. Vaccination effect on post-COVID-19 TIA was seen only in the 181–365 days, with sHR of 0.51 (0.31–0.82). Online supplemental figures S23-31 show database-specific and cohort-specific estimates for ATE-related complications.

Risk of post-COVID-19 cardiac complications was reduced in vaccinated individuals. Meta-analytic estimates in the acute phase showed sHR of 0.45 (0.38–0.53) for HF, 0.41 (0.26–0.66) for MP and 0.41 (0.27–0.63) for VACA. Reduced risk persisted for post-acute COVID-19 HF: sHR 0.61 (0.51–0.73) for 31–90 days, 0.61 (0.51–0.73) for 91–180 days and 0.52 (0.43–0.63) for 181–365 days. For post-acute MP, risk was only lowered in the first post-acute window (31–90 days), with sHR of 0.43 (0.21–0.85). Vaccination showed no association with post-COVID-19 HS. Database-specific and cohort-specific results for these cardiac diseases are shown in online supplemental figures S32-40 .

Stratified analyses by vaccine showed similar associations, except for ChAdOx1 which was not associated with reduced VTE and ATE risk in the last post-acute window. Sensitivity analyses were consistent with main results ( online supplemental figures S6-13 ).

Figure 3 shows the results of comparative effects of BNT162b2 versus ChAdOx1, based on UK data. Meta-analytic estimates favoured BNT162b2 (sHR of 0.66 (0.46–0.93)) for VTE in the 0–30 days after infection, but no differences were seen for post-acute VTE or for any of the other outcomes. Results from sensitivity analyses, database-specific and cohort-specific estimates were in line with the main findings ( online supplemental figures S41-51 ).

Forest plots for comparative vaccine effect (BNT162b2 vs ChAdOx1); meta-analysis across cohorts and databases. ATE, arterial thrombosis/thromboembolism; CD+HS, cardiac diseases and haemorrhagic stroke; VTE, venous thromboembolism.

Key findings

Our analyses showed a substantial reduction of risk (45–81%) for thromboembolic and cardiac events in the acute phase of COVID-19 associated with vaccination. This finding was consistent across four databases and three different European countries. Risks for post-acute COVID-19 VTE, ATE and HF were reduced to a lesser extent (24–58%), whereas a reduced risk for post-COVID-19 MP and VACA in vaccinated people was seen only in the acute phase.

Results in context

The relationship between SARS-CoV-2 infection, COVID-19 vaccines and thromboembolic and/or cardiac complications is tangled. Some large studies report an increased risk of VTE and ATE following both ChAdOx1 and BNT162b2 vaccination, 7 whereas other studies have not identified such a risk. 25 Elevated risk of VTE has also been reported among patients with COVID-19 and its occurrence can lead to poor prognosis and mortality. 26 27 Similarly, several observational studies have found an association between COVID-19 mRNA vaccination and a short-term increased risk of myocarditis, particularly among younger male individuals. 5 6 For instance, a self-controlled case series study conducted in England revealed about 30% increased risk of hospital admission due to myocarditis within 28 days following both ChAdOx1 and BNT162b2 vaccines. However, this same study also found a ninefold higher risk for myocarditis following a positive SARS-CoV-2 test, clearly offsetting the observed post-vaccine risk.

COVID-19 vaccines have demonstrated high efficacy and effectiveness in preventing infection and reducing the severity of acute-phase infection. However, with the emergence of newer variants of the virus, such as omicron, and the waning protective effect of the vaccine over time, there is a growing interest in understanding whether the vaccine can also reduce the risk of complications after breakthrough infections. Recent studies suggested that COVID-19 vaccination could potentially protect against acute post-COVID-19 cardiac and thromboembolic events. 11 12 A large prospective cohort study 11 reports risk of VTE after SARS-CoV-2 infection to be substantially reduced in fully vaccinated ambulatory patients. Likewise, Al-Aly et al 12 suggest a reduced risk for post-acute COVID-19 conditions in breakthrough infection versus SARS-CoV-2 infection without prior vaccination. However, the populations were limited to SARS-CoV-2 infected individuals and estimates did not include the effect of the vaccine to prevent COVID-19 in the first place. Other studies on post-acute COVID-19 conditions and symptoms have been conducted, 28 29 but there has been limited reporting on the condition-specific risks associated with COVID-19, even though the prognosis for different complications can vary significantly.

In line with previous studies, our findings suggest a potential benefit of vaccination in reducing the risk of post-COVID-19 thromboembolic and cardiac complications. We included broader populations, estimated the risk in both acute and post-acute infection phases and replicated these using four large independent observational databases. By pooling results across different settings, we provided the most up-to-date and robust evidence on this topic.

Strengths and limitations

The study has several strengths. Our multinational study covering different healthcare systems and settings showed consistent results across all databases, which highlights the robustness and replicability of our findings. All databases had complete recordings of vaccination status (date and vaccine) and are representative of the respective general population. Algorithms to identify study outcomes were used in previous published network studies, including regulatory-funded research. 3 4 8 18 Other strengths are the staggered cohort design which minimises confounding by indication and immortal time bias. PS overlap weighting and NCO empirical calibration have been shown to adequately minimise bias in vaccine effectiveness studies. 19 Furthermore, our estimates include the vaccine effectiveness against COVID-19, which is crucial in the pathway to experience post-COVID-19 complications.

Our study has some limitations. The use of real-world data comes with inherent limitations including data quality concerns and risk of confounding. To deal with these limitations, we employed state-of-the-art methods, including large-scale propensity score weighting and calibration of effect estimates using NCO. 19 24 A recent study 30 has demonstrated that methodologically sound observational studies based on routinely collected data can produce results similar to those of clinical trials. We acknowledge that results from NCO were positively associated with vaccination, and estimates might still be influenced by residual bias despite using calibration. Another limitation is potential under-reporting of post-COVID-19 complications: some asymptomatic and mild COVID-19 infections might have not been recorded. Additionally, post-COVID-19 outcomes of interest might be under-recorded in primary care databases (CPRD Aurum and Gold) without hospital linkage, which represent a large proportion of the data in the study. However, results in SIDIAP and CORIVA, which include secondary care data, were similar. Also, our study included a small number of young men and male teenagers, who were the main population concerned with increased risks of myocarditis/pericarditis following vaccination.

Conclusions

Vaccination against SARS-CoV-2 substantially reduced the risk of acute post-COVID-19 thromboembolic and cardiac complications, probably through a reduction in the risk of SARS-CoV-2 infection and the severity of COVID-19 disease due to vaccine-induced immunity. Reduced risk in vaccinated people lasted for up to 1 year for post-COVID-19 VTE, ATE and HF, but not clearly for other complications. Findings from this study highlight yet another benefit of COVID-19 vaccination. However, further research is needed on the possible waning of the risk reduction over time and on the impact of booster vaccination.

Ethics statements

Patient consent for publication.

Not applicable.

Ethics approval

The study was approved by the CPRD’s Research Data Governance Process, Protocol No 21_000557 and the Clinical Research Ethics committee of Fundació Institut Universitari per a la recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol) (approval number 4R22/133) and the Research Ethics Committee of the University of Tartu (approval No. 330/T-10).

Acknowledgments

This study is based in part on data from the Clinical Practice Research Datalink (CPRD) obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. We thank the patients who provided these data, and the NHS who collected the data as part of their care and support. All interpretations, conclusions and views expressed in this publication are those of the authors alone and not necessarily those of CPRD. We would also like to thank the healthcare professionals in the Catalan healthcare system involved in the management of COVID-19 during these challenging times, from primary care to intensive care units; the Institut de Català de la Salut and the Program d’Analítica de Dades per a la Recerca i la Innovació en Salut for providing access to the different data sources accessible through The System for the Development of Research in Primary Care (SIDIAP).

• Pritchard E ,
• Matthews PC ,
• Stoesser N , et al
• Lauring AS ,
• Tenforde MW ,
• Chappell JD , et al
• Pistillo A , et al
• Duarte-Salles T , et al
• Hansen JV ,
• Fosbøl E , et al
• Chen A , et al
• Hippisley-Cox J ,
• Mei XW , et al
• Duarte-Salles T ,
• Fernandez-Bertolin S , et al
• Ip S , et al
• Bowe B , et al
• Prats-Uribe A ,
• Feng Q , et al
• Campbell J , et al
• Herrett E ,
• Gallagher AM ,
• Bhaskaran K , et al
• Raventós B ,
• Fernández-Bertolín S ,
• Aragón M , et al
• Makadia R ,
• Matcho A , et al
• Mercadé-Besora N ,
• Kolde R , et al
• Ostropolets A ,
• Makadia R , et al
• Rathod-Mistry T , et al
• Thomas LE ,
• ↵ Coronavirus (COVID-19) in the UK . 2022 . Available : https://coronavirus.data.gov.uk/
• Generalitat de Catalunya
• Schuemie MJ ,
• Hripcsak G ,
• Ryan PB , et al
• Houghton DE ,
• Wysokinski W ,
• Casanegra AI , et al
• Katsoularis I ,
• Fonseca-Rodríguez O ,
• Farrington P , et al
• Jehangir Q ,
• Li P , et al
• Byambasuren O ,
• Stehlik P ,
• Clark J , et al
• Brannock MD ,
• Preiss AJ , et al
• Schneeweiss S , RCT-DUPLICATE Initiative , et al

Supplementary materials

Supplementary data.

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

• Data supplement 1

AMJ and MC are joint senior authors.

Contributors DPA and AMJ led the conceptualisation of the study with contributions from MC and NM-B. AMJ, TD-S, ER, AU and NTHT adapted the study design with respect to the local vaccine rollouts. AD and WYM mapped and curated CPRD data. MC and NM-B developed code with methodological contributions advice from MTS-S and CP. DPA, MC, NTHT, TD-S, HMEN, XL, CR and AMJ clinically interpreted the results. NM-B, XL, AMJ and DPA wrote the first draft of the manuscript, and all authors read, revised and approved the final version. DPA and AMJ obtained the funding for this research. DPA is responsible for the overall content as guarantor: he accepts full responsibility for the work and the conduct of the study, had access to the data, and controlled the decision to publish.

Funding The research was supported by the National Institute for Health and Care Research (NIHR) Oxford Biomedical Research Centre (BRC). DPA is funded through a NIHR Senior Research Fellowship (Grant number SRF-2018–11-ST2-004). Funding to perform the study in the SIDIAP database was provided by the Real World Epidemiology (RWEpi) research group at IDIAPJGol. Costs of databases mapping to OMOP CDM were covered by the European Health Data and Evidence Network (EHDEN).

Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Read the full text or download the PDF:

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

Effects of COVID-19 vaccine safety framing on parental reactions

Contributed equally to this work with: Hao Tan, Jiayan Liu

Roles Conceptualization, Funding acquisition, Supervision, Writing – review & editing

Affiliations Lushan Lab, Hunan University, Changsha, China, School of Design, Hunan University, Changsha, China

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Roles Investigation

• Hao Tan, 
• Jiayan Liu, 
• Yingli Zhang

• Published: April 16, 2024
• https://doi.org/10.1371/journal.pone.0302233
• Reader Comments

As a major concern shared by parents globally, COVID-19 vaccine safety is typically being messaged to the public in a negative frame in many countries. However, whether the COVID-19 vaccine safety framing have an effect on parents when vaccinating their children is unclear. Here we implement an online survey with a convenience sample of 3,861 parents living in mainland China, all over 18 years old and with at least one child under 18. The parents were randomly assigned to receive information about COVID-19 vaccine safety in either a negative frame (incidence of side effects) or a positive frame (the inverse incidence of side effects), to compare parental reactions to a range of questions about communication, risk perception, trust, involvement and behavioral intention. We found that parents were more likely to regard vaccine safety as relevant to policy support and as a higher priority for government when receiving positively framed information (p = 0.002). For some specific subgroups, parents in positive framing group showed lower risk perception and higher trust (p<0.05). This suggests that positive framing of COVID-19 vaccine safety messages show more effective performance than negative framing in terms of involvement, as well as trust and risk perception in specific subgroups, which may lead to a reflection on whether to adjust the current widespread use of negative framing. Our findings inform how governments and health care workers strategically choose the framing design of COVID-19 vaccine safety information, and have important implications for promoting COVID-19 vaccination in children in the future.

Citation: Tan H, Liu J, Zhang Y (2024) Effects of COVID-19 vaccine safety framing on parental reactions. PLoS ONE 19(4): e0302233. https://doi.org/10.1371/journal.pone.0302233

Editor: Omar Enzo Santangelo, Regional Health Care and Social Agency of Lodi, ITALY

Received: August 1, 2023; Accepted: March 29, 2024; Published: April 16, 2024

Copyright: © 2024 Tan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript and its Supporting Information files.

Funding: This research was supported by the Natural Science Foundation of Hunan Province (2023JJ30149) and Research Foundation of Lushan Lab, which were got by Hao Tan. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Despite the benefits and worldwide approval use of COVID-19 vaccination for children [ 1 – 7 ], parents still remain a high level of vaccine hesitancy due to concerns about vaccine safety [ 8 – 15 ]. Growing evidence showing that behavioral nudges, which are usually shifts in how a message is framed, are desperately needed to boost COVID-19 vaccination [ 16 – 20 ]. However, public-health specialists and healthcare workers face a particular dilemma in communicating vaccine safety information to parents [ 21 ], because they are not provided with guidelines for presenting or framing the information. Public health agencies in many countries such as China, the UK and the US are using a negative frame (incidence of side effects) when explaining COVID-19 vaccine safety [ 22 – 26 ], but our understanding of the framing effect is limited. Understanding how the framing of COVID-19 vaccine safety information affects parental reactions helps examine the appropriateness of the frame currently used in most countries, and may help to address the challenge of risk communication regarding COVID-19 vaccination in the future when China rolls out COVID-19 vaccine for younger children.

Human choices and attitudes are influenced by the manner in which information are presented, referred to as framing effects [ 27 , 28 ], and such insights are beginning to be applied in the field of vaccination. Attribute framing, which manipulates an object’s quality or characteristics in a positive or negative frame [ 29 ], has been examined in many studies [ 30 – 32 ], and has been shown that subtle changes in the framing may have meaningful effects on readers’ understanding and reactions to information.

In previous research, scholars mainly focused on how different frames can affect participants’ risk perception or behavioral intention. Most of these studies found that positive framing led to significantly lower risk perceptions [ 33 , 34 ] and significantly higher behavioral intentions [ 26 , 35 , 36 ], but some studies did not demonstrate a significant effect on either risk perceptions or behavioral intentions [ 37 ]. Therefore, we hypothesized that compared with parents who received negatively framed information about COVID-19 vaccine safety, parents who received positively framed information had a significantly lower perceived risk of COVID-19 vaccine side effects (hypothesis 1) and a significantly higher intention to vaccinate their child when the vaccine was available (hypothesis 2). As evolving evidence suggests that regular or seasonal booster vaccinations against COVID-19 may be necessary [ 38 ], we also examined parents’ intentions to vaccinate their children regularly in the future. We hypothesized that parents in the positive framing group would show a higher intention to get their children vaccinated regularly (hypothesis 3).

Communication and trust are also issues that are often explored in attribute framing studies. Research evidence suggests that positively framed statements were more appealing to transmit [ 39 ], so we hypothesized that parents in the positive framing group were more likely to share vaccine safety information with family and friends (hypothesis 4). In a classic study of attribute framing, negative framing was found to weigh more in trust assessments [ 40 ], whereas Webster and Rubin showed no difference in the performance of trust between the two frames [ 34 ]. Therefore, we hypothesized that parents in the negative framing group would be significantly more trusting of the government’s reporting of information on the safety of the COVID-19 vaccine (hypothesis 5).

In addition, we also considered the effects of framing on involvement, including policy support and perceptions towards government priorities. Policy support is an important topic in framing studies, particularly in the field of environment and climate [ 41 – 44 ]. Results from attribute framing studies on vaccines suggested that participants exposed to positive framing would be more supportive of vaccine policy than those exposed to negative framing [ 45 ], so we hypothesized that parents in the positive framing group are more likely to be involved in vaccine policy support (hypothesis 6). The issue of judgments on government priorities has not been explored extensively and deeply in framing research, although a study on goal framing allowed participants to prioritized a list of potential government actions to test framing effects [ 46 ]. We hypothesized that parents in the positive framing group would give vaccine safety a significantly higher level of government priority (hypothesis 7).

In terms of framing research on COVID-19 vaccines, most studies have explored the effects of the goal framing by emphasizing the benefits of vaccinating or the losses of not vaccinating [ 47 – 58 ], and some studies have examined framing effects by emphasizing other conditions (e.g. individual-centered versus collective-centered) [ 59 – 62 ]. Only a few studies have focused on the message of COVID-19 vaccine safety, and investigated whether the framing of vaccine safety information has effects on the public reactions [ 26 , 35 , 63 ].

Although findings have shown that COVID-19 vaccine safety can influence vaccination of children [ 64 – 66 ], there is still insufficient research focusing on the effects of the COVID-19 vaccine safety framing on parents. Considering the importance of COVID-19 vaccine safety communication and the differences in its application to populations with different characteristics [ 67 ], it is essential to understand parental reactions to the COVID-19 vaccine safety framing and to present accurate vaccine safety information in an understandable and convincing form.

Using the Chinese parents as an example, this study compared the framing effects between the COVID-19 vaccine safety information in positive frame (the inverse adverse event rate) and negative frame (the adverse event rate) on multiple dimensions of COVID-19 vaccine belief and behavioral intentions. Specifically, we explored whether parents who received positively-framed information about COVID-19 vaccine safety were more likely to share vaccine safety information, support vaccine policies, and give vaccine safety a higher level of government priority, and exhibited higher intention to get their child vaccinated when the vaccine was available and to vaccine them on regular, while showed lower risk perception and trust. We also added sociodemographic characteristics and baseline COVID-19 vaccine mood as covariates for analysis and considered whether there was an interaction between framing and them. Finally, in addition to examining the effect of framing in the general parent population, we divided the sample according to socio-demographic factors and conducted subgroup analyses.

The remainder of this study is structured as follows: the “Results” section explains the t-test results, MANCOVA results, ANCOVA results and subgroup analysis results; the “Discussion” section offers a discussion of the empirical findings, practical implications, limitations, research contributions and protentional future research suggestions; and the “Materials and Methods” section provides details of the sample, experiment design, measurements and data analysis.

Ethics statement

This study was approved by the Research Ethics Committee of Hunan University (2019002). Written consent was obtained from respondents when they registered and completed the questionnaire on the online survey platform, and they were assured that all results would be disseminated in aggregate form to guarantee anonymity and confidentiality.

Our sample was derived from an online survey conducted across China from 18 January to 1 February 2022, that targeted Chinese parents whose children are under 18 years old. The survey was performed on Sojump ( www.sojump.com ), a professional online survey network with 52,000,000 users in China. Convenience sampling approach was adopted to recruit participants from the online survey network, which is an appropriate non-probability sampling method for researchers who need to recruit participants that meet specific criteria and widely used in exploratory research [ 68 ]. Furthermore, previous framing effect studies have effectively utilised convenience sampling [ 69 , 70 ]. Eligible parents had to be at least 18 years old, living in mainland China and at least one child under 18 years of age. Participants who complete the online survey will receive 120 rewards points from the survey company to redeem for money.

Among the 3861 participants, 1978 (51.2%) received COVID-19 vaccine information in the positive frame (here after ‘positively-framed information’ sample) and the remaining participants (1883, 48.8%) were asked questions after receiving COVID-19 vaccine information in the negative frame (here after ‘negatively-framed information’ sample) ( Fig 1 ). No differences were observed between the positively-framed information sample and the negatively-framed information in terms of gender distribution, age structure, education distribution, distribution of parents’ COVID-19 vaccination status, or distribution of their children’s influenza vaccination status.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

https://doi.org/10.1371/journal.pone.0302233.g001

Survey experiment design

The experiment began with the following question wording: ‘Here are a couple details about the COVID-19 vaccine, which was confirmed by experts in the immunization program of the Chinese Center for Disease Control and Prevention:’. In one treated group, respondents who received positively-framed information were told: “In every dose of COVID-19 vaccine, there is a more than 99.988% chance that an adverse event will not occur in vaccinated population. (According to the results of the analysis as of 30 April 2021).” Respondents in the other treated group who receiving negatively-framed information were told: “In every dose of COVID-19 vaccine, there is a less than 0.012% chance that an adverse event will occur in vaccinated populations. (According to the results of the analysis as of 30 April 2021).” And both of them were told that according to the current monitoring analysis, the probability of vaccine side effects in children and adolescents was no higher than that in adults over the age of 18. Then, all participants were asked the same questions in the same order, except for random assignment to a baseline of vaccine safety information.

Parental reaction indicators include communication, belief (risk perception and trust), involvement (policy support and government priority judgement) and behavioral intention (vaccination when available and on regular basis). The survey questions were adopted or revised versions of questions from relevant studies [ 26 , 35 – 37 , 39 , 45 , 71 – 73 ].

Communication.

We measured communication using two items with each scored on a 5-point scale (1  =  ‘very unlikely’ to 5  =  ‘very likely’). Items include, ‘How likely would you be to talk to your family members about issues related to COVID-19 vaccine side effects with your family members?’, and ‘How likely would you be to talk to your friends about issues related to COVID-19 vaccine side effects with your friends?’. The questions were revised from relevant studies [ 39 , 45 ], and the internal reliability was excellent (α  =  0.812).

Risk perception.

Three items assessing parents’ risk perception were taken from the Renner’s study of A/H1N1 influenza vaccination [ 71 ]. These were, (1) ‘How severely do you think the COVID-19 vaccine side effects could have harmed your children?’ (scored on a 5-point scale ranging from 1 ‘not serious’ to 5 ‘very serious’), and, (2) ‘How worried would you be about the impact of the COVID-19 vaccine side effects?’ (scored on a 5-point scale ranging from 1 ‘not worried’ to 5 ‘very worried’), and, (3) ‘How likely would your children be to experience an adverse event with COVID-19 vaccine?’ (scored on a 5-point scale ranging from 1 ‘very unlikely’ to 5 ‘very likely’). The internal reliability for this variable was excellent (α  =  0.758).

To measure parents’ trust, they were asked whether they believe that the CDC is faithfully reporting the risks of the COVID-19 vaccine [ 36 , 72 ], and responses were recorded on a 5-point scale from 1 ‘do not trust at all’ to 5 ‘completely trust’.

Involvement.

Similar to a climate change labelling effects research [ 73 ], we measured the question of involvement. Participants were asked, ‘Do you agree that the issue of an adverse event with COVID-19 vaccine is an important consideration for your decision regarding whether support COVID-19 vaccination policies for children?’ (scored on a 5-point scale ranging from 1 ‘strongly disagree’ to 5 ‘strongly agree’). Participants were also asked, ‘Do you think that safety of COVID-19 vaccines should be a low, medium, high, or very high priority for the government?’, and responses were recorded on a four-point Likert scale (1  =  low, 2  =  medium, 3  =  high, 4  = very high).

Behavioral intention.

Participants rate their likelihood of questions “How likely would you be to get your children a COVID-19 vaccine after the vaccine becomes available?” and “If regular COVID-19 vaccination is needed, how likely would you be to get your children vaccinated on a fairly regular basis?” Response options ranged from ‘very unlikely’ (1) to ‘very likely’ (5) with higher scores reflecting higher levels of vaccination intention [ 26 , 35 – 37 ].

To measure basic mood about COVID-19 vaccine, participants were asked, ‘How did you feel about COVID-19 vaccine?’ with response scale from 0 (‘extremely bad’) to 100 (‘extremely good’) [ 74 ]. We also collected information on gender (male and female), children’s age (under 3 years, 3–11 years and above 11 years), educational attainment (high school and below, junior college, undergraduate and postgraduate and above), and income (low, middle and high income). The Chinese-English translation was done by the second author and was reviewed by survey experts from the Sojump Research Website.

The survey data were managed using SPSS. We used two-sided independent-samples t-tests to examine the framing effects between the two groups. Table 1 provides the results of the t-tests that examined the framing effects on the full data set. We compared the parental reactions of different information framings using a multivariate analysis of covariance (covariates: parents’ gender, education, income, children age and basic mood for COVID-19 vaccine). Following the MANCOVA models, we also ran separate univariate analyses of covariance (ANCOVA) to test which of the dependent variables were statistically significant.

https://doi.org/10.1371/journal.pone.0302233.t001

The Box M test confirmed the homogeneity of the variance–covariance matrices. And we also checked the interaction between framing and covariates, which were not statistically significant (Wilk’s Lambda measures of 0.998–0.999, p > 0.05). The 7 dependent variables were not normally distributed; however, since our sample size was sufficient to obtain robust results (n > 1000 in both groups), we followed the standard parametric procedure for the univariate analysis.

In addition to examining the framing effects of the general parent population, we divided the sample according to several key characteristics, which are central to the discussion of parental reactions, as follows: parents’ gender, education, income and children age. Differences among these groups were tested by t-test, which could improve understanding of how sociodemographic factors influence framing effects.

As discussed above, we were aware that the 7 outcome variables were not normally distributed. However, the t-test has been found to be robust when data are non-normally distributed, particularly with a large sample size [ 75 , 76 ], so we kept the parametric method in our analysis.

Nevertheless, we performed Mann–Whitney tests for all analyses. We found no difference for full sample, and found that statistical significance was different in one case of subgroup analysis, between the two tests. The case was policy support for parents whose children were 3–11 years old, where in t-test it was statistically significant (p = 0.033, r = 0.048), but in Mann–Whitney test it was not significant (asymptotic significance p = 0.329, r = -0.019).

Statistical analysis was performed using SPSS. The statistical significance level was established at p < 0.05, however, we also reported a marginally significant effect at 0.06 > p > 0.05.

Our data were derived from a large online survey (N = 3861) conducted across China in 2022, during the period when children were receiving the COVID-19 vaccine. We first performed a series of two-sided independent samples t-tests to identify differences in framing effects between the negative frame sample and positive frame sample for the full sample. The 7 dependent variables were not normally distributed; however, the t-test has been found to be robust when data are non-normally distributed, particularly with a large sample size. We also performed the Mann-Whitney test and found no significant difference between two tests.

Significant difference in the involvement (government priority perception), and marginally significant difference in the involvement (policy support) were found between framing groups ( Table 1 ; Fig 2 ). Parents in negative framing group rated significantly lower level in priority for governments (negative frame M = 3.30 vs. positive frame M = 3.38, p = 0.002). Also, parents were less likely to perceive vaccine safety as relevant to vaccine policy support when receiving negatively-framed information than when receiving positively-framed information (negative frame M = 3.79 vs. positive frame M = 3.85, p = 0.058). This suggests that Chinese parents’ reactions to involvement were generally influenced by the COVID-19 vaccine safety information framing effect. However, no significant differences were found in communication, belief and behavioral intention (all p > 0.1, Table 1 ).

https://doi.org/10.1371/journal.pone.0302233.g002

The MANCOVA model show that the differences between positive and negative frames in parental reactions (Wilks λ  =  0.996, F (7,3848)  =  2.027, p  =  0.048) remained significant after controlling for parental gender, education, income, children age and basic mood of vaccine ( Table 2 ).

https://doi.org/10.1371/journal.pone.0302233.t002

Similar with the results of t-test, The results of the univariate analysis show a significant association between information framing and public support (F (1,3854)  =  4.06, p  =  0.044, η2  =  0.001) as well as government priority perception (F (1,3854)  =  10.275, p  =  0.001, η2  =  0.003). The full result of univariate analysis of MANCOVA models for parental reactions are reported in supplementary information.

We then considered the framing effects for sociodemographic factors. The results showed no difference in framing effects between the two frames in parents with children under three years of age, with low (high school and below degree) or high (postgraduate and above degree) education and high income level.

In contrast, several issues showed significant differences in parents with gender, children above three years of age, middle education level, and lower income levels ( Table 3 ; the full results are reported in supplementary document).

https://doi.org/10.1371/journal.pone.0302233.t003

In addition to involvement, we observed framing effects on parental trust and risk perception in some subgroups.

Specifically, Female respondents displayed less trust in the COVID-19 vaccine safety information when the negatively-framed information was received than when the positively-framed information was received (negative frame M = 4.02 vs. positive frame M = 4.09, p = 0.038), and the same effects were observed among parents with a junior college degree (negative frame M = 4.00 vs. positive frame M = 4.18, p = 0.042) and low income level (negative frame M = 3.94 vs. positive frame M = 4.09, p = 0.016).

Moreover, when receiving the negatively-framed information, parents with children above eleven years of age had a significantly higher risk perception than when receiving the positively-framed information (negative frame M = 2.42 vs. positive frame M = 2.24, p = 0.044).

Similar to some previous studies showing that subtle changes in framing had an impact on people’s reactions [ 45 , 77 , 78 ], our study revealed a significant framing effect for parents’ involvements. Specifically, parents exposed to positively framed COVID-19 vaccine safety messages were more likely to regard vaccine safety as relevant to policy support and as a higher priority for government than parents exposed to the same messages in the negative frame. In addition, more framing effects on trust and risk perception were observed among female participants, parents with children aged 11 years or older, parents with a junior college degree, and those on low incomes. The results suggest that the negative framing of COVID-19 vaccine safety information, which is widely used worldwide, should be used with particular caution, and that health professionals and policy makers need to carefully consider how to present information well.

The finding that parents were more likely to involve in vaccine policy support when receiving COVID-19 vaccine safety information in the positive frame than in the negative frame, are consistent with a study on human papillomavirus (HPV) vaccine that respondents exposed to positive framing were more supportive of vaccine mandate policy [ 45 ].

In addition, our research showed that presenting COVID-19 vaccine safety information to parents in the positive frame improved their perception of the government’s priorities on vaccine safety issues, more than presenting the same information in the negative frame. Based on a previous study on climate change showing that the respondents’ perceived susceptibility had a positive effect on the attitude towards government’s priority [ 79 ], the positively-framed information may lead to higher parental perceived susceptibility of COVID-19 vaccine side effects and therefore they believe that the government should give high priority to the safety of COVID-19 vaccines.

For framing effects in subgroup analysis, many backlash effects were observed. One backlash effect we found was that those mothers, parents with a junior college degree and low income level were less likely to trust the CDC-reported COVID-19 vaccine safety information when the negatively-framed information was received than when the positively-framed information was received.

This finding is different from the result observed for ground beef advertisement [ 80 ], which revealed negative frames are more influential for establishing trust. One possibility for the different results is that the research on ground beef advertising explored trust not in official government agencies but in individuals engaged in certain professions, such as merchants whose interests were perceived to be diametrically opposed to those of their clients. Thus, identifying the causes for this reaction requires further investigation.

We also found that compared with parents who received positively framed information about COVID-19 vaccine safety, those with children aged 11 years or older had significantly higher perceived risk of COVID-19 vaccine side effects when receiving negatively framed information. We have observed the impact of framing effects on risk perception in many studies [ 33 , 34 ], but explaining the reasons for different reactions towards perceived risk of COVID-19 vaccine among parents of adolescents under the framing effect needs further study.

Overall, our results revealed a significant framing effect on parents’ involvements, which would play an important role in policy development [ 81 , 82 ]. Also, we identified many negative, backlash framing effects for some specific subgroups, such as mothers. Understanding the framing effects for these groups are crucial to targeting audiences COVID-19 vaccine risk communication.

The current reporting of official data on vaccine safety information in many countries is based on negative framing of adverse event rates, and our study found negative framing effects in some specific populations under this frame. Our findings inform that the currently widely used negative framing needs to be seriously reconsidered, and we hope to provide public-health specialists and healthcare workers with some guidelines for presenting or framing the information when communicating the COVID-19 vaccine safety with parents. This study contributes to the understanding of under-investigated framing effects of parents vaccinating their children against COVID-19 and have important implications for promoting COVID-19 vaccination in children in the future.

The results of our study should be interpreted in light of its limitations. First, our study was only conducted in China, a non-Western cultural context. As cultural differences in vaccine-related reactions are common [ 83 – 86 ], the extent to which these results can be generalized to other countries is unknown. Second, we recruited the sample via the internet and the online survey respondents may have contributed to self-selection bias or the disproportionate youth of their samples. Third, there was a limitation in not employing quota sampling to ensure demographic representativeness, which may affect the generalizability of our findings to the broader population of Chinese parents. In addition, considering potential validity concerns related to participant engagement and the possibility of distractions in the task, enhanced validation techniques can be explored to further strengthen the reliability of online survey data in future work [ 87 ]. Finally, this is a cross-sectional study and cannot take into account the possible effects of time, while vaccine attitudes have been shown to be potentially dynamic and changing [ 88 – 90 ].

More work is needed to demonstrate the effects caused by these subtle framing changes. In addition to exploring the framing effects of COVID-19 vaccine safety information, as in this study, more studies should examine the effects of different information content and message delivery formats. Also, it may be worthwhile to consider examining the effects of presenting vaccine information in mixed (positive and negative) frames. Furthermore, future research could examine framing effects among people in different cultural contexts, given that framing effects are not specific to China and have been documented in other countries for other aspects of vaccines [ 60 , 91 , 92 ]. Finally, the study of framing effects involves legal, ethical, and political domains in future research, and larger and more comprehensive studies are needed.

Supporting information

https://doi.org/10.1371/journal.pone.0302233.s001

https://doi.org/10.1371/journal.pone.0302233.s002

Acknowledgments

We express our sincere thanks to all the participants of this study, whose involvement and dedication enabled us to explore and draw meaningful conclusions. We also thank Sojump for their invaluable support and the use of their online survey platform, which greatly facilitated our research.

• View Article
• Google Scholar
• PubMed/NCBI

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 13 April 2024

Identifying and overcoming COVID-19 vaccination impediments using Bayesian data mining techniques

• Bowen Lei 1 ,
• Arvind Mahajan 2 &
• Bani Mallick 1  

Scientific Reports volume  14 , Article number:  8595 ( 2024 ) Cite this article

218 Accesses

1 Altmetric

Metrics details

• Data mining
• Health care economics
• Health policy
• Statistical methods

The COVID-19 pandemic has profoundly reshaped human life. The development of COVID-19 vaccines has offered a semblance of normalcy. However, obstacles to vaccination have led to substantial loss of life and economic burdens. In this study, we analyze data from a prominent health insurance provider in the United States to uncover the underlying reasons behind the inability, refusal, or hesitancy to receive vaccinations. Our research proposes a methodology for pinpointing affected population groups and suggests strategies to mitigate vaccination barriers and hesitations. Furthermore, we estimate potential cost savings resulting from the implementation of these strategies. To achieve our objectives, we employed Bayesian data mining methods to streamline data dimensions and identify significant variables (features) influencing vaccination decisions. Comparative analysis reveals that the Bayesian method outperforms cutting-edge alternatives, demonstrating superior performance.

Similar content being viewed by others

A guide to vaccinology: from basic principles to new developments

Andrew J. Pollard & Else M. Bijker

An overview of clinical decision support systems: benefits, risks, and strategies for success

Reed T. Sutton, David Pincock, … Karen I. Kroeker

Genomic data in the All of Us Research Program

The All of Us Research Program Genomics Investigators

Introduction

The emergence of COVID-19 has greatly impacted people’s lives since 2020 and will continue to do so. The Center for Systems Science and Engineering (CSSE) at Johns Hopkins University 1 reports that there have been more than 676 million cases and 6.8 million deaths in the world. To combat COVID-19, there are a number of restrictive methods to inhibit the spread of the virus 2 , 3 , 4 , 5 . These include lockdowns, quarantine, etc. These methods are widely used in many countries but many studies raise concerns about the costs and side effects of their use 6 , 7 , 8 , 9 , 10 , such as loss of gross domestic product (GDP), educational opportunities, increased deaths, higher mental health risks, and other societal costs. In addition to these restrictive methods, vaccines are another potent way to tackle the pandemic 3 , 11 , 12 . Higher vaccination rates would bring many benefits. However, the facts show that many people are unable or hesitant to get vaccinated 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 . In our study, impediments to COVID-19 vaccination are defined as unwillingness or refusal to receive the COVID-19 vaccine, or inability to receive the COVID-19 vaccine due to lack of vaccine availability ( CDC provides a definition of vaccination hesitancy measure at the following link https://data.cdc.gov/stories/s/Vaccine-Hesitancy-for-COVID-19/cnd2-a6zw/ . However, this is a subset of our impediment measure since CDC hesitancy measure doesn’t consider the lack of availability.) We aim to predict vaccine impediment using Bayesian technique and to identify groups of important variables that contribute to impediments to vaccination. We then make policy recommendations to address impediments to vaccination. The World Health Organization (WHO) has recognized vaccine impediment as one of the top ten global health threats, as it can lead to low vaccination rates and the resurgence of preventable diseases. This impediment can stem from a variety of reasons.

In this paper, we conducted an analysis of data sourced from a prominent health insurance provider in the United States. We briefly present how the vaccine impediment varies across insured populations, including gender, race, income level, and age, as shown in Fig. 1 . In terms of gender, similar to results of previous research 20 , we can see that women and men have almost the same percentage of vaccine impediments, with males having slightly higher impediments to vaccination. For different racial groups, Whites and Asians get relatively low impediment scores, while Hispanics, Blacks, and Native Americans have higher scores based on the data. A similar pattern has been found in existing works 21 . For each income level, people tend to be more willing to vaccinate as their income increases, from the lower to the upper middle class, which is also found in other studies 22 . However, the upper class is similar to the lower class, who are hampered in terms of vaccination. For different age groups, young and middle-aged people (from 20 to 50 years old) have very similar rates and are more hesitant to be vaccinated. In contrast, older people are more willing to vaccinate and the willingness increases with age (from 51 years and older). This is consistent with the findings of existing studies 22 , 23 , 24 .

Vaccine impediment grouped by variables ( a ) gender, ( b ) race, ( c ) income level, and ( d ) age.

Impediments to vaccination are influenced by a variety of factors, and our goal was to gain deeper insights into the obstacles preventing individuals from getting vaccinated, identify them at an early stage, and formulate data-driven policies to address these challenges. This paper makes significant contributions to the existing literature in two key aspects. Firstly, it leverages granular and objective data obtained from a major health insurance provider, enabling a more in-depth and comprehensive analysis. Secondly, we employ an advanced classification model to predict the likelihood of a member being hesitant to receive the vaccine, yielding more accurate results compared to other statistical methods. Although we have used COVID-19 vaccination data, most of the results will likely be applicable to other epidemic or pandemic vaccination situations.

In this study, we introduce a two-stage methodology. In the first stage, we employ Bayes factor 25 , 26 for preliminary screening, followed by the application of a Bayesian nonparametric regression technique known as Bayesian Multivariate Adaptive Regression Splines (BMARS) 27 , 28 , 29 in the second stage. This approach is applied to population characteristic data provided by a major health insurance provider, with the aim of identifying barriers to vaccination. The pre-screening step enables our approach to effectively handle high-dimensional feature spaces by selecting the key features, simplifying the complex problem within the Bayesian framework. Additionally, the BMARS regression method allows for the modeling of nonlinear relationships between these selected key features and the response variable.

In the following sections, we first describe our Bayes-factor-based pre-screening and BMARS-based classification modeling (B-BMARS) method and introduce the vaccine impediment dataset to identify vaccination impediments. We then compare the results of B-BMARS with other popular baseline methods and analyze which variables play a key role in impeding getting vaccinated. Next, based on the modeling results, we present analyses and policy implications from the business perspective. We also describe other alternative baseline forecasting methods in the Supplementary Information.

We propose a novel two-stage method to accurately and efficiently analyze people’s impediments to receiving the COVID-19 vaccine with proper selections of interpretable variables and their interactions. The first stage, pre-screening, is based on the Bayes factor, a widely used Bayesian method to quickly check the correlation between variables and response. Thus, we can effectively filter out apparently irrelevant variables and avoid unnecessary computational burdens and modeling challenges. In the second stage of BMARS-based classification, the unknown function is fitted by product-based spline basis functions, which can automatically fine-tune the selection of key variables and their interactions.

Stage I: Bayes-factor-based pre-screening

In our COVID-19 vaccination data analysis, the dimension of potential key variables is usually too high to use Bayesian nonparametric models directly. Therefore, it is necessary to reduce the dimensionality of the variable space. We propose to take advantage of the model comparison ability of the Bayes factor and use it as a screening step to reduce the dimensions. Since our goal is to predict vaccine impediments, it becomes a binary classification problem. Therefore, we chose a method widely used for classification tasks, the Probit model, in which the conditional probability of one of the two possible attitudes toward the vaccine is equal to a linear combination of the underlying variables, transformed by the cumulative distribution function of the standard Gaussian 30 , 31 . For classification tasks, a widely used approach is to combine the regression model with a probit model using auxiliary variables. Specifically, in the classification framework, we use z to denote the observed response, which is a binary variable and y as the auxiliary variable. We assume the binary z to be 1 if \(y>0\) and 0 otherwise. For the probabilistic model, it is defined as \(p(z=1|y)=\Phi (y)\) where \(\Phi\) is the standard Gaussian cumulative distribution function and y is defined as \(y \sim \mathcal {N}(\varvec{ \beta }{\textbf {x}}+\beta _0, \sigma ^2)\) where \(\varvec{ x}\) is the \(p^*\) dimensional explanatory variables (covariates), \(\varvec{ \beta }\) is the vector of regression parameters and \(\sigma ^2\) is the error variance.

High-dimensional data analysis is always a daunting task. When the dimension \(p^*\) is high, we run into a problem called “the curse of dimensionality” 32 . Though the high dimensional variables usually provide more information, they also lead to higher computational costs. The convergence of optimization algorithms or Bayesian sampling in a space of high dimensions is usually very slow. Also, it can harm the estimation accuracy, which is due to the difficult search in a space of high dimensions. Therefore, an effective and accurate variable selection is essential in high-dimensional modeling.

Pre-screening is a popular way to quickly filter out unimportant variables, making variable selection more efficient in a much lower-dimension space using a simpler model (like linear model), especially for ultrahigh-dimensional cases. In pre-screening methods, it is usually assumed that if one variable is important when predicting the response, it will be marginally associated with the response. Different measurements of the association are studied using, for example, p-value 32 , 33 , 34 . However, the pre-screening technique have not been fully explored in the Bayesian paradigm.

We use an off-the-shelf Bayesian method, Bayes factor 35 , 36 , for pre-screening. More specifically, the Bayes factor is a Bayesian alternative to classical hypothesis testing, which plays an important role in the model comparison and selection process. Essentially, the Bayes factor serves as a measure of how strongly data support a specific model compared to another. The Bayes factor is defined as a ratio of the marginal likelihood of two candidate models, typically regarded as a null and an alternative hypothesis. The general formula is as below.

where D denotes the available data and \(M_1\) and \(M_2\) denote two potential models. A larger value of this ratio indicates more support for \(M_1\) , and vice versa.

More specifically, to check the effect of the j th variable \(x_{j}\) with the corresponding regression parameter \(\beta _{j}\) , we calculate the Bayes factor ( \(\text {BF}_j\) ) via Probit regression model as below

where hypothesis \(\mathscr {H}_1\) assumes that \(y \sim \mathcal {N}(\beta _j x_j+\beta _0, \sigma _{j}^2)\) , hypothesis \(\mathscr {H}_0\) assumes that \(y \sim \mathcal {N}(\beta _0, \sigma ^2)\) , prior for \(\beta _j\) is Gaussian distribution \(p(\beta _j)\sim \mathcal {N}(0,\alpha )\) , and use conjugate prior for the variances.

To compute the intractable marginal likelihood \(p({\textbf {z}} | \mathscr {H}_1)\) (integrated over \(\varvec{ \beta }\) ), we choose to use Laplace Approximation 37 , 38 , 39 . Specifically, under \(\mathscr {H}_1\) , the posterior distribution of \(\beta _j\) is

Suppose \(\beta _j^*\) is a maximum of f , we can calculate the negative Hessian at \(\beta _j^*\)

Then, the approximate posterior can be written as \(Q(\beta _j) = \mathcal {N}(\beta _j | \beta _j^*, A^{-1})\) . Thus, we can approximate the marginal likelihood

A larger value of \(\text {BF}_j\) suggests our preference for the hypothesis \(\mathscr {H}_1\) to the hypothesis \(\mathscr {H}_0\) , implying a potential key role of \({\textbf {x}}_j\) when predicting \({\textbf {z}}\) . Then after calculating \(\{\text {BF}_j, j=1,\cdots ,p\}\) , we can choose the top ranked variables with respect to \(\text {BF}_j\) . Say we select p explantory variables out of \(p^*\) variables. Next, we use these p selected variables \(\varvec{ x}\) for the Bayesian nonparametric classification model.

Stage II: BMARS-based classification modeling

In stage 2, we use a flexible nonlinear method to relate the response z with the selected explanatory variables from step 1. More specifically, we use Bayesian multivariate adaptive regression splines (BMARS) 27 , 28 which is a Bayesian version of a flexible non-parametric regression and classification method named MARS 40 . We extend the previously defined linear probit model for nonlinear modeling using product spline basis functions. We use the probit model defined in the previous section, for the i th observation \(p(z_{i}=1|y_{i})=\Phi (y_{i}), (i=1,\cdots ,n)\) . Next we use BMARS to relate the auxilary variables y with the explanatory variables \({\textbf {x}}\) through a regression model. In BMARS, for regression tasks, the product-based spline basis functions are not only used to model the unknown function f , but also automatically select the nonlinear interactions among the variables. The mapping function between the selected variables \({\textbf {x}}_i \in \mathscr {R}^p\) and the auxiliary variable \(y_i\) as below

where m is the number of basis functions and \(\alpha _j\) denotes the coefficient for the basic function \(B_j\) which is designed as

where the \(s_{qj} \in \{-1,1\}\) , the v ( q ,  j ) denotes the index of the variables and the set \(\{v(q,j);q=1,\cdots ,Q_j\}\) are not repeated, the \(t_{qj}\) refers to the partition location, \((\cdot )_+ = \max (0,\cdot )\) , and \(Q_j\) is the polynomial degree of the basic function \(B_j\) and also indicates the number of variables involved in \(B_j\) .

For probit model, the posterior distribution is not available in explicit form so we use Markov Chain Monte Carlo (MCMC) algorithm to simulate from the posterior distribution. As the dimension of the model m is unknown, we use the reversible jump Metropolis-Hastings algorithm 41 . More specifically, the model parameters we are interested in within the Bayesian framework of BMARS 27 are assumed to include the number of basis functions m , as well as their degree of interaction \(Q_j\) , their coefficients \(\alpha _j\) , their associated split points \(t_{qj}\) , and the sign indicators \(s_{qj}\) . We can use \(\varvec{\theta }^{(m)} = \{ \mathscr {B}_1,\cdots ,\mathscr {B}_m \}\) where \(\mathscr {B}_j\) to denote the model parameters \((Q_j, \alpha _j, t_{1j}, \cdots , t_{Q_j,j}, s_{1j}, \cdots , s_{Q_j,j})\) for each basis function \(B_j\) . Then, the hierarchical model can be written as

and the joint posterior for parameters m and \(\varvec{\theta }^{(m)}\) can be written in the following factorized form

In this algorithm, we update the model randomly using one of three steps, including (a) changing a node position, (b) creating a basis function, or (c) deleting a basis function, and then correcting the proposed new sample by the Metropolis-Hastings step 42 , 43 . Under this sampling scheme, samples based on significant variables are more likely to be accepted, which enables automatic feature selection by the algorithm and is important for us to make policy implications.

Data description

To understand vaccine impediments, we analyze a dataset obtained from one of the major health insurance providers in the United States. Since the dataset comes from the insured population, our analysis of impediments to vaccination and potential policy implications focuses on the insured population. More specifically, the dataset includes a total of 974,842 observations, each presenting information about one member of the insurance provider, with 1 binary response and 368 variables. About 69% of the variables are numeric and the remaining 31% are categorical. We note that we use synthetic data based on real data, which maintains all relationships within the dataset but is not specific to any individual insured person. This minimizes the risks associated with privacy to share protected data.

Response measures

The data records whether an insurance member is vaccinated or not. We assume that if a member is not vaccinated, then that member has some sort of impediment to vaccination. We use a broad definition of impediments that includes various reasons such as not believing in the efficacy of the vaccine, barriers like lack of resources, inability, or ideological/political reasons, etc.

The data document a number of characteristics of insured members that are potential variables influencing their willingness and availability to receive vaccines. These variables can be categorized into eight groups of characteristics, including medical claims, pharmacy claims, laboratory claims, demographics, credit data, condition-related data, centers for Medicare & Medicaid services (CMS) features (original reasons for entry into Medicare), and other characteristics. In total, there are 253 numerical variables and 115 categorical variables. The detailed descriptions of each group are provided in Table  1 .

Data pre-processing

Before modeling the data, we use some pre-processing steps to make the data structure compatible with the model. First, for convenience, we transform each categorical variable into several dummy variables. Thus, when the data are put into the model, there are 898 variables. Second, to fairly compare the different models, we balance the two types of samples by using a sample of 10,000 vaccinated clients and a sample of 10,000 unvaccinated clients as training data. Similarly, we sample a balanced test data including 2,000 clients.

Classification analysis: accuracy

In this section, in terms of accuracy, we compare our two-stage method (B-BMARS) with several widely-used classification models, including extreme gradient boosting (XGBoost) 44 , Gaussian process classification (GP) 45 , random forest (RF) 46 , and multilayer-perceptrons-based deep neural network (DNN) 47 , which have all demonstrated good performance in various applications. We use 0.5 as the threshold to calculate the accuracy, which is widely used for binary classification. For the overall analysis with different thresholds, we further use the area under curve (AUC) values 48 for comparison (shown in the next Section). Specifically for our B-BMARS, in the first stage, we use Bayes factor to quickly examine the potential predictive power of each variable on the response. Then, in the second stage, we use B-BMARS to fit the unknown function between the key variables and the responses in a more refined manner. A detailed description can be found in the “ Methods ” section.

In the first stage of pre-screening, we experiment by keeping different numbers of the top variables where the pre-screening dimension \(p_{scr} =50\) which is a proper value found empirically. We also compare the scenario without using pre-screening which corresponds to using all the 898 variables. However, using all the 898 variables is not practical with limited computational resources. Table 2 shows the accuracy among pre-screening dimension \(p_{scr}=50\) and without the pre-screening step. Our proposed B-BMARS gives the highest accuracy 0.614 and beats other popular baseline alternatives. Random Forest’s best result is close to our B-BMARS result but always below it.

We also visualize accuracy comparisons under pre-screening \(p_{scr}=50\) and scenario without pre-screening step in Fig. 2 a and b. The slash bars represent our B-BMARS, and the star bars represent XGBoost, GP, RF, and DNN from left to right, respectively. As we can see, the green bars are the highest in Fig. 2 a, and also comparable to the highest blue columns in Fig. 2 b. This indicates that our B-BMARS can maintain the performance for different scenarios. However, other baselines are relatively more sensitive to different settings. Additionally, we can see that RF achieves the best performance when \(p_{scr}=898\) , which leads to a high computational burden and is not practical with limited computational resources.

Visualization of accuracy comparison with baseline methods under pre-screening dimension \(p_{scr}=50\) and without pre-screening step. We compare our B-BMARS with extreme gradient boosting (XGBoost), Gaussian process classification (GP), random forest (RF), and multilayer-perceptrons-based deep neural network (DNN). Our B-BMARS generally improves or maintains accuracy compared to other baseline methods.

Classification analysis using AUC values

Apart from the accuracy, we also choose AUC values 49 to measure model performance, where higher AUC values indicate a better classifier. The baselines considered are aligned with the previous accuracy comparison. Table 3 shows the best AUC value among different pre-screening dimensions of each model. Our proposed B-BMARS gives the highest AUC value 0.651, followed by RF.

Similar to accuracy comparison, we also show detailed AUC value comparisons under pre-screening \(p_{scr}=50\) and scenario without pre-screening step in Fig. 3 a and b. The slash bars represent our B-BMARS, and the star bars represent XGBoost, GP, RF, and DNN from left to right, respectively. As shown in the figures, the green bars are the highest in Fig. 2 a, and also comparable to the highest blue columns in Fig. 2 b. This indicates that the classification rule from our B-BMARS is consistently one of the best classification rules in different pre-screening dimensions \(p_{scr}\) . However, other popular baselines have more fluctuations in different scenarios, with a drop in AUC values when resources are limited and pre-screening has to be used.

Visualization of the AUC value comparison with baseline methods under pre-screening dimension \(p_{scr}=50\) and without pre-screening step. We compare our B-BMARS with extreme gradient boosting (XGBoost), Gaussian process classification (GP), random forest (RF), and multilayer-perceptrons-based deep neural network (DNN). Our B-BMARS generally improves or maintains AUC value compared to other baseline methods.

Variable selection

B-BMARS is effective in selecting the most important variables. We find that there are four main categories of variables playing a key role in influencing the vaccine impediments of insured members, i.e., low household assets, high health risks, highly uninsured areas, and physician-related information. As shown in Table 4 , we list ten interesting and important variables selected by our B-BMARS, along with their detailed descriptions and the categories to which they belong.

When trying to determine people’s willingness or ability to take the COVID-19 vaccine, it is helpful to look at their household asset status, and we find that people with low household assets will be hesitant to receive the vaccine, which is in line with existing research findings 50 . For example, among the significant variables listed, Supplemental Nutrition Assistance Program (SNAP) benefits per capita is selected 20 , which reflects whether people generally have a stable source of food and thus reflects their household asset status. It is also important to check the number of non-mortgage accounts that are more than 60 days past due 51 . If many non-mortgage accounts are chronically past due, it is likely that household assets are low. In addition, we can see that per capita income in Table  4 in the last 12 months is one of the key variables, which gives a direct indication of people’s economic situation.

Health risk is another important variable of COVID-19 vaccination propensity prediction, and people are more reluctant to get vaccinated if they already have a high health risk 52 , 53 . For example, the trend in the number of prescriptions per month is noteworthy. It represents a change in people’s health status and can indicate whether they are at high health risk. In addition, we need to look at the number of monthly prescriptions related to heart disease-heart failure medications, which also shows how often people are taking their medications and revealing their health status.

In addition to the categories mentioned above, the availability of better healthcare coverage in the area also affects people’s proclivity to get vaccinated, and populations living in highly uninsured areas are more unlikely to receive COVID-19 vaccination 54 . As listed in the key variables, the net monthly payments for behavioral health claims related to skilled nursing inpatient facilities have a significant impact. Also, trends in monthly prescription costs associated with vaccine drugs reflect health care coverage and indicate people’s attitudes to vaccinations, and the percentage of adults under age 65 without health insurance in the corresponding area is selected. The higher the percentage, the worse the health care coverage is.

Last but not least, there is a need to consider whether individuals trust their physicians and the public health system 55 , 56 ; if individuals are not willing to use their physicians as their primary source of medical information, they are unlikely to be vaccinated 57 . For instance, we can get some information about people’s beliefs from the percentage of physician evaluations and claims management related to outpatient visits in the past year. A relatively high percentage score means that people are more likely to use their public health system and trust their doctors.

Non-linear basic function selection

To provide more interpretation of the selected variables, we plot a nonlinear basis function curve including the selected variables. We take two important variables, namely the trend of the number of prescriptions and per capita income, as examples. As depicted in Fig. 4 a, the propensity to vaccinate begins to decline with increasing prescriptions when a small number of prescriptions is reached, indicating high health risks. In Fig. 4 b, people first become more willing to vaccinate and then gradually become less willing to vaccinate as their per capita income increases, consistent with what we observed in our data overview.

Curves of the non-linear basic functions selected by B-BMARS ( a ) trend of the number of prescriptions (Prescription Number Trend), and ( b ) per capita income (Per Capita Income).

Interaction selection

B-BMARS is not only effective in selecting the most important variables but also in identifying variables that have significant interactions. As shown in Table 5 , we list five important variable interactions selected by B-BMARS to predict COVID-19 vaccination propensity. Prescription count for cardiology is the most important and has many interactions with other variables 58 . Other variables are composed of health-related variables and financial conditions. Therefore, it is necessary to consider both the information related to the number of prescriptions and other important variables.

Business analysis and policy implication

Reducing vaccination impediments is important to slow the emergence of new virus variants. This will reduce the burden on patients and public health resources and will reduce costs incurred by insurance and healthcare providers. Thus, it is critical to develop targeted strategies to improve the ability to get vaccinated and reduce hesitancy. Vaccine impediment is a complex decision-making process influenced by a variety of contextual, individual and group, and vaccine-specific variables, including communication, socioeconomics, geographic barriers, vaccination experience, risk perception, and vaccination program design.

From our analysis in Section Variable Selection and Interaction Selection, it is clear that people with low assets, high health risks, low medical coverage, and distrust of doctors and the public health system are mostly reluctant to get vaccinated. Moreover, according to our analysis, these characteristics interact with each other. For example, people with low assets or low medical coverage have higher health risks. For these members, there are greater barriers than for other members. They may have fewer resources, more difficulty reaching vaccination sites, and less information about the nature of the pandemic.

We define the cause of COVID-19 vaccination impediments in these groups as due to physical barriers, psychological barriers, and health barriers. Physical barriers can be explained by having less access to the vaccine. Members with disabilities are likely to suffer from a lack of mobility. Psychological barriers can be explained by misunderstanding and mistrust. Members with few assets, low income, and high debt may live in communities where mistrust is prevalent or have fewer resources to obtain accurate information about the vaccine. Health barriers can be explained by high health risks, such as chronic diseases. The members may be older, living insecurely, or in poor health and worried about the side effects of vaccination.

Physical barrier

Potential policy implications to overcome challenges to access vaccination. People tend not to get vaccinated if it is difficult and cumbersome to obtain the vaccination. Difficulties often arise from limited mobility due to disability or age, availability of time, transportation, and low supply of vaccinations.

Implication 1: For people with limited mobility due to disability or age, we do not recommend that they visit a medical facility for vaccination, where there may be a high risk of cross-infection. We recommend that the health care provider provide home care services to help them get vaccinated at home.

Implication 2: For people with limited time, lack of transportation, and selected constraints, access to vaccines is limited for these and other reasons like poor financial status. The health care provider can provide them with travel assistance, such as language instruction and transportation help. The provider can also arrange some special activities including vaccination camps near their homes to help them get vaccinated.

Implication 3: For people in areas with low vaccine supply, it is more difficult for them to get vaccinated, even if they want to. Therefore, it is necessary to increase vaccine supply and reduce geographical inequalities. We recommend that the health care provider work with pharmacies to open more vaccination sites and productively send notifications to residents when vaccines are available.

Psychological barrier

Potential policy implications to overcome misunderstanding and mistrust of vaccine. People tend not to get vaccinated if they have misconceptions about the vaccine and think it will be harmful to them. The source of these misconceptions can be family, friends, social media, or social norms. Or they ignore the need for vaccination because they are currently in good health.

Implication 1: We recommend the health care provider get involved in community events and health activities to build stronger relationships with insured members. Then, it can select community leaders as vaccine ambassadors to deliver messages that allow vaccine recipients to share their reasons for vaccination, which will encourage people to reframe how they think about vaccines and build trust in the public health system.

Implication 2: For people who do not understand the necessity of vaccination, they may not be motivated enough to get vaccinated because they are in good health. We recommend emphasizing the age-independent health benefits and importance of vaccination, providing them instructional videos or organizing lectures.

Health barrier

Potential policy implications to overcome high health risk. People are more concerned about the side effects of vaccines if they are at high health risk. Specifically, they were concerned that the side effects of the vaccine would exacerbate their existing health problems.

Implication 1: We recommend obtaining more information about their health to understand if the vaccine can negatively interact with their current medications and existing problems. They should be educated if the vaccine is indeed safe for them.

Implication 2: We recommend using telemedicine to track their health after vaccination. This can prevent any unexpected health problems and make them feel more confident in the public health system.

Expected benefit analysis

In this section, we use an example to analyze the expected benefits of utilizing our methodology and the resulting policy implications to address vaccine impediments. We have actual data from a major U.S. healthcare provider. It is a publicly listed company that is committed to maximizing benefits for its stakeholders, particularly its shareholders. Given the data available to us, and utilizing publicly available data from other sources, we would like to estimate the preventable costs and incremental costs associated with our proposal to determine the impact on the savings of the healthcare provider.

More specifically, we use an all-vaccination rate (VRate) of 19.55% as of March 31st, 2021, which is derived by dividing the number of U.S. all-vaccinated persons by the U.S. population. The number of U.S. all-vaccinated persons is 64,852,669 from the Centers for Disease Control and Prevention’s (CDC) COVID data tracker 59 , and the U.S. population is 331,791,631 according to the United States Census Bureau’s U.S. and World Population Clock 60 . In addition, we use a number of Medicare Advantage members of one of the major health insurance providers in the United States (N) of 4,600,000 in 2020. Therefore, we can approximate the number of impedimented members as \(\text {N}_u\) = (1-VRate) \(\cdot\) N=3,700,700.

We then calculate the amount of preventable costs based on our B-BMARS method by encouraging more members to get vaccinated. From Centers for Medicare and Medicaid Services (CMS) report 61 , we obtain the average cost of medical services for COVID-19 hospitalizations and the number of medicare hospitalizations due to COVID-19 per 100,000 patients, which are $24,000 and 1,825, respectively. We also get the effectiveness of full vaccination in preventing hospitalization. According to the CDC’s August 2021 presentation in Morbidity and Mortality Weekly Report 62 , the effectiveness in adults 75 years or older is 91% for Pfizer-BioNTech, 96% for Moderna, and 85% for Janssen COVID-19 vaccines(CDC). Therefore, we choose 85% to approximate the lower bound of savings. Using our B-BMARS (as shown in Table  2 ), we are able to successfully identify 61.4% of impedimented members ( \(\text {R}_s\) ). Based on this information, we calculate the preventable costs for patients not vaccinated with COVID-19 via Equation ( 9 ). Specifically, the \(\text {Hospital}/100,000\) calculates the percentage of people who receive Medicare hospitalizations. We multiply this by \(\text {R}_s\) and \(\text {Ratio}\) to represent the approximate proportion of people spared hospitalization by vaccination. We then multiply this by \(\text {N}_u\) to get the approximate number of hospitalizations prevented by vaccination. Finally, we multiply this by \(\text {Fee}\) , which is the cost of patients preventable through vaccination. The results are shown in Table  6 , and we successfully prevent more than 845 million dollars in costs.

In addition, we approximate the extra cost of the incremental vaccination. We collect relevant information from Centers for Medicare and Medicaid Services (CMS) 63 . Specifically, for those without disabilities, the cost of the vaccination is $80 per person, assuming 2 doses of each vaccine and a single dose cost to Medicare of $40. For those with disabilities, the cost of the home vaccination has increased to $150 per person because of an additional $35 per dose. In our dataset, the percentage of people with disabilities is 25%. Therefore, we can estimate the cost of having impedimented members vaccinated following Equation ( 10 ). Specifically, we multiply the ratios \((1-\text {R}_d)\) and \(\text {R}_d\) by \(\text {N}_u\) to give the approximate numbers of people without and with disabilities, respectively. Next, we multiply these two numbers by \(\text {R}_s\) to get a rough estimate of the number of people in each impedimented group that we can successfully identify. This is then multiplied by the corresponding costs \(\text {Cost}_{nd}\) and \(\text {Cost}_{d}\) , respectively. Finally, we add the estimated costs of the two groups to arrive at the final total extra cost. The result is $221,542,406 as shown in Table  7 .

Based on all the above calculations, using Equation ( 11 ), we obtain a total savings of more than $ 624 million by addressing impediments to vaccination for the insured population. Specifically, we subtracted the total extra cost of vaccination from the total preventable costs due to vaccination to derive the total savings. We do not have a firm estimate of the marginal cost of implementing our policy recommendations. However, we are informed that it will be a small fraction of the $624 million savings calculated here. At a minimum, this number provides health insurance and healthcare providing organization guidance in developing a budget for implementing our policy recommendations. This example demonstrates how our methods can be transformed to a monetary value.

In this paper, we propose a flexible Bayesian method for predicting COVID-19 vaccination impediment scores under a Bayesian paradigm. Based on the accuracy of the results, we conclude that our proposed forecasting method performed better than the existing cutting-edge methods. The key findings of this study are:

The proposed method, B-BMARS performed better than XGBoost, Gaussian Process, and Random Forest in terms of classification accuracy.

Several important groups of variables are identified which could be the reasons for vaccine impediment, e.g., health risk and healthcare coverage.

We identified four main categories of variables playing a key role in influencing the attitude of the public towards vaccines, including low household assets, high health risks, highly uninsured areas, and infrequent use of physician information.

Interactions among some of these variables may play a crucial role in vaccine impediment, e.g. combining low medical coverage and low assets have more prediction power for vaccination impediment.

We define the cause of COVID-19 vaccination impediments in these groups as due to physical barriers, psychological barriers, and health barriers. We then provide policy recommendations to reduce barriers from the perspective of each of these three barriers.

Physical barriers can be explained by having less access to the vaccine, e.g., limited mobility, limited time, lack of transportation, and low vaccine supply. To overcome such barriers, we recommend that health care providers offer home care services, travel assistance, and arrange for special events, including vaccination camps.

Psychological barriers refer to misconceptions or neglect of vaccines and can come from family, friends, social media, social norms, or good health. To overcome these barriers, we recommend that health care providers engage in community events and wellness activities, build stronger relationships and trust with insured members, and provide them with instructional videos or organize lectures that emphasize the health benefits and importance of vaccination.

Health barriers are people’s existing health problems that make them more worried about the side effects of vaccines. To overcome such barriers, we recommend that health care providers obtain more information about people’s health, provide more specific advice to each individual, and use telemedicine to track their health after vaccination.

We estimated the dollar benefit based on actual data and publicly available information resulting from our potential policy implications.

To the best of our knowledge, this is the first research that uses these flexible methods to analyze the data and arrive at conclusions that will have a significant impact on corporate decision making. Our findings have broad implications for solving complex problems with large datasets that require forecasting. Finally, our framework can have a direct impact on corporate and public policy related to future pandemics.

For future research, it is of great importance to expand the study to uninsured members and barriers in other countries. Equally important is how other types of data (e.g., image and text data), if available, can be incorporated to further improve predictive accuracy, better address vaccine barriers, and provide additional benefits. Additionally, we plan to enhance the scalability of the algorithm by employing parallel Markov Chain Monte Carlo (MCMC) within a simulated annealing framework 64 . This approach aims to enable the implementation of the algorithm in a single stage.

Data availability

The data files of the customer’s vaccine intentions and characteristics are available upon reasonable request from Prof. Mahajan ([email protected]).

The center for systems science and engineering (csse) at johns hopkins university: Covid-19 content portal. www.systems.jhu.edu/research/public-health/ncov/ .

Haug, N. et al. Ranking the effectiveness of worldwide covid-19 government interventions. Nat. Hum. Behav. 4 , 1303–1312 (2020).

Article   PubMed   Google Scholar  

Mavragani, A. & Gkillas, K. Exploring the role of non-pharmaceutical interventions (npis) in flattening the greek covid-19 epidemic curve. Sci. Rep. 11 , 11741 (2021).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Giordano, G. et al. Modelling the covid-19 epidemic and implementation of population-wide interventions in Italy. Nat. Med. 26 , 855–860 (2020).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Kwame, A., Makarova, V., Hudu, F. & Petrucka, P. M. The covid-19 pandemic in ghana: Exploring the discourse strategies in president nana addo’s speeches. Humanit. Soc. Sci. Commun. 10 , 1–10 (2023).

Article   Google Scholar  

Allen, D. W. Covid-19 lockdown cost/benefits: A critical assessment of the literature. Int. J. Econ. Bus. 29 , 1–32 (2022).

Matzinger, P. & Skinner, J. Strong impact of closing schools, closing bars and wearing masks during the covid-19 pandemic: Results from a simple and revealing analysis. medRxiv (2020).

Saqib, M. A. N. et al. Effect of covid-19 lockdown on patients with chronic diseases. Diabetes Metab. Syndr. Clin. Res. Rev. 14 , 1621–1623 (2020).

Orgilés, M., Morales, A., Delvecchio, E., Mazzeschi, C. & Espada, J. P. Immediate psychological effects of the covid-19 quarantine in youth from Italy and Spain. Front. Psychol. 11 , 2986 (2020).

Krekel, C., Swanke, S., De Neve, J.-E. & Fancourt, D. Happiness predicts compliance with preventive health behaviours during Covid-19 lockdowns. Sci. Rep. 13 , 7989 (2023).

Principi, N. & Esposito, S. Why it is important to develop an effective and safe pediatric covid-19 vaccine. Vaccines 9 , 127 (2021).

Sallam, M. Covid-19 vaccine hesitancy worldwide: A concise systematic review of vaccine acceptance rates. Vaccines 9 , 160 (2021).

Biswas, N., Mustapha, T., Khubchandani, J. & Price, J. H. The nature and extent of covid-19 vaccination hesitancy in healthcare workers. J. Community Health 46 , 1244–1251 (2021).

Article   PubMed   PubMed Central   Google Scholar  

Machingaidze, S. & Wiysonge, C. S. Understanding covid-19 vaccine hesitancy. Nat. Med. 27 , 1338–1339 (2021).

Article   CAS   PubMed   Google Scholar  

Solís Arce, J. S. et al. Covid-19 vaccine acceptance and hesitancy in low-and middle-income countries. Nat. Med. 27 , 1385–1394 (2021).

Dubé, E. & MacDonald, N. E. Covid-19 vaccine hesitancy. Nat. Rev. Nephrol. 18 , 409–410 (2022).

Lazarus, J. V. et al. Revisiting covid-19 vaccine hesitancy around the world using data from 23 countries in 2021. Nat. Commun. 13 , 3801 (2022).

Macharia, J. M. et al. An empirical assessment of the factors influencing acceptance of covid-19 vaccine uptake between Kenyan and Hungarian residing populations: A cross-sectional study. Sci. Rep. 12 , 22262 (2022).

Article   ADS   PubMed   PubMed Central   Google Scholar  

Lazarus, J. V. et al. A survey of covid-19 vaccine acceptance across 23 countries in 2022. Nat. Med. 29 , 366–375 (2023).

Acharya, B. & Dhakal, C. Implementation of state vaccine incentive lottery programs and uptake of covid-19 vaccinations in the united states. JAMA Netw. Open 4 , e2138238–e2138238 (2021).

Nguyen, L. H. et al . Racial and ethnic differences in covid-19 vaccine hesitancy and uptake. medrxiv (2021).

Nguyen, K. H. et al. Disparities in national and state estimates of covid-19 vaccination receipt and intent to vaccinate by race/ethnicity, income, and age group among adults ≥ 18 years, United States. Vaccine 40 , 107–113 (2022).

Kini, A. et al. Differences and disparities in seasonal influenza vaccine, acceptance, adverse reactions, and coverage by age, sex, gender, and race. Vaccine 40 , 1643–1654 (2022).

Sypsa, V. et al. Trends in covid-19 vaccination intent, determinants and reasons for vaccine hesitancy: Results from repeated cross-sectional surveys in the adult general population of Greece during November 2020-June 2021. Vaccines 10 , 470 (2022).

Hoijtink, H., Mulder, J., van Lissa, C. & Gu, X. A tutorial on testing hypotheses using the Bayes factor. Psychol. Methods 24 , 539 (2019).

Keysers, C., Gazzola, V. & Wagenmakers, E.-J. Using Bayes factor hypothesis testing in neuroscience to establish evidence of absence. Nat. Neurosci. 23 , 788–799 (2020).

Denison, D. G., Mallick, B. K. & Smith, A. F. Bayesian mars. Stat. Comput. 8 , 337–346 (1998).

Lei, B. et al. Bayesian optimization with adaptive surrogate models for automated experimental design. Npj Comput. Mater. 7 , 194 (2021).

Article   ADS   Google Scholar  

Holmes, C. C., Denison, D. G. T. & Mallick, B. K. Accounting for model uncertainty in seemingly unrelated regressions. J. Comput. Graph. Stat. 11 , 533–551, 3 (2002).

McCullagh, P. Generalized Linear Models (Routledge, 2019).

Albert, J. H. & Chib, S. Bayesian analysis of binary and polychotomous response data. J. Am. Stat. Assoc. 88 , 669–679 (1993).

Article   MathSciNet   Google Scholar  

Hong, S., Kim, Y. & Park, T. Practical issues in screening and variable selection in genome-wide association analysis. Cancer Inform. 13 , CIN–S16350 (2014).

Kirpich, A. et al. Variable selection in omics data: A practical evaluation of small sample sizes. PloS One 13 , e0197910 (2018).

Ghosh, A. & Thoresen, M. A robust variable screening procedure for ultra-high dimensional data. Stat. Methods Med. Res. 30 , 1816–1832 (2021).

Article   MathSciNet   PubMed   Google Scholar  

Schönbrodt, F. D. & Wagenmakers, E.-J. Bayes factor design analysis: Planning for compelling evidence. Psychon. Bull. Rev. 25 , 128–142 (2018).

Liang, F., Paulo, R., Molina, G., Clyde, M. A. & Berger, J. O. Mixtures of g priors for Bayesian variable selection. J. Am. Stat. Assoc. 103 , 410–423 (2008).

Article   MathSciNet   CAS   Google Scholar  

Daxberger, E. et al. Laplace redux-effortless Bayesian deep learning. Adv. Neural Inf. Process. Syst. 34 , 20089–20103 (2021).

Google Scholar  

Girolami, M. & Rogers, S. Variational Bayesian multinomial probit regression with gaussian process priors. Neural Comput. 18 , 1790–1817 (2006).

Kuss, M., Rasmussen, C. E. & Herbrich, R. Assessing approximate inference for binary Gaussian process classification. J. Mach. Learn. Res. 6 (2005).

Friedman, J. H. Multivariate adaptive regression splines. Ann. Stat. 1–67 (1991).

Green, P. J. Reversible jump Markov chain monte Carlo computation and Bayesian model determination. Biometrika 82 , 711–732 (1995).

Denison, D. G., Holmes, C. C., Mallick, B. K. & Smith, A. F. Bayesian methods for nonlinear classification and regression, vol. 386 (John Wiley & Sons, 2002).

Holmes, C. & Mallick, B. Generalized nonlinear modeling with multivariate free-knot regression splines. J. Am. Stat. Assoc. 98 , 352–368 (2003).

Chen, T. et al . Xgboost: Extreme gradient boosting. R package version 0.4-2 1 , 1–4 (2015).

Nickisch, H. & Rasmussen, C. E. Approximations for binary Gaussian process classification. J. Mach. Learn. Res. 9 , 2035–2078 (2008).

MathSciNet   Google Scholar  

Breiman, L. Random forests. Mach. Learn. 45 , 5–32 (2001).

Almeida, L. B. Multilayer perceptrons. In Handbook of Neural Computation , C1–2 (CRC Press, 2020).

Huang, J. & Ling, C. X. Using auc and accuracy in evaluating learning algorithms. IEEE Trans. Knowl. Data Eng. 17 , 299–310 (2005).

Article   CAS   Google Scholar  

Ling, C. X., Huang, J. & Zhang, H. Auc: a better measure than accuracy in comparing learning algorithms. In Conference of the Canadian Society for Computational Studies of Intelligence , 329–341 (Springer, 2003).

Kadoya, Y. et al. Willing or hesitant? A socioeconomic study on the potential acceptance of covid-19 vaccine in Japan. Int. J. Environ. Res. Public Health 18 , 4864 (2021).

Biddle, N., Edwards, B., Gray, M. & Sollis, K. Change in vaccine willingness in Australia: August 2020 to January 2021. MedRxiv 2021–02 (2021).

Warren, A. M., Perrin, P. B., Elliott, T. R. & Powers, M. B. Reasons for covid-19 vaccine hesitancy in individuals with chronic health conditions. Health Sci. Rep. 5 (2022).

Ku, L. The association of social factors and health insurance coverage with covid-19 vaccinations and hesitancy, July 2021. J. Gen. Intern. Med. 37 , 409–414 (2022).

Allen, J. D., Abuelezam, N. N., Rose, R. & Fontenot, H. B. Factors associated with the intention to obtain a covid-19 vaccine among a racially/ethnically diverse sample of women in the usa. Transl. Behav. Med. 11 , 785–792 (2021).

Paul, E., Steptoe, A. & Fancourt, D. Attitudes towards vaccines and intention to vaccinate against covid-19: Implications for public health communications. The Lancet Regional Health–Europe 1 (2021).

Dhalaria, P., Arora, H., Singh, A. K. & Mathur, M. Covid-19 vaccine hesitancy and vaccination coverage in India: An exploratory analysis. Vaccines 10 , 739 (2022).

Holroyd, T. A. et al. Development of a scale to measure trust in public health authorities: Prevalence of trust and association with vaccination. J. Health Commun. 26 , 272–280 (2021).

Bogart, L. M. et al . Covid-19 related medical mistrust, health impacts, and potential vaccine hesitancy among black Americans living with HIV. J. Acquir. Immune Defic. Syndr. (1999) 86 , 200 (2021).

Centers for disease control and prevention report: Trends in number of covid-19 vaccinations in the us. www.covid.cdc.gov/covid-data-tracker/#vaccination-trends .

United states census bureau: U.S. and world population clock. www.census.gov/popclock/ .

Centers for medicare and medicaid services: Medicare covid-19 data snapshot. www.cms.gov/files/document/medicare-covid-19-data-snapshot-services-through-2021-03-20.pdf .

Centers for disease control and prevention report: Morbidity and mortality weekly report. www.cdc.gov/mmwr/volumes/70/wr/mm7032e3.htm?s_cid=mm7032e3_w .

Centers for medicare and medicaid services: Medicare covid-19 vaccine shot payment. www.cms.gov/medicare/payment/covid-19/medicare-covid-19-vaccine-shot-payment .

Payne, R. D., Guha, N. & Mallick, B. K. A Bayesian survival tree partition model using latent Gaussian processes. Biometrics, To appear (2024).

Download references

Acknowledgements

We thank Vinay Chiguluri, Sravya Etlapur, Andrew J. Fieldhouse, and Geoffrey Monsees for valuable comments, and Xiaan Zhou for capable research assistance.

Author information

Authors and affiliations.

Department of Statistics, Texas A&M University, College Station, TX, USA

Bowen Lei & Bani Mallick

Department of Finance, Texas A&M University, College Station, TX, USA

Arvind Mahajan

You can also search for this author in PubMed   Google Scholar

Contributions

B.L. and B.M. conceived the concept of B-BMARS. B.L. implemented the algorithm and performed the experiments. AM provided data files on the vaccine intentions of the customers and provided business and financial background. All authors analyzed the results, contributed to the manuscript, and edited it. All authors reviewed the final version of the manuscript.

Corresponding author

Correspondence to Bani Mallick .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary information., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Lei, B., Mahajan, A. & Mallick, B. Identifying and overcoming COVID-19 vaccination impediments using Bayesian data mining techniques. Sci Rep 14 , 8595 (2024). https://doi.org/10.1038/s41598-024-58902-1

Download citation

Received : 23 November 2023

Accepted : 04 April 2024

Published : 13 April 2024

DOI : https://doi.org/10.1038/s41598-024-58902-1

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Assessment of Risk for Sudden Cardiac Death Among Adolescents and Young Adults After Receipt of COVID-19 Vaccine — Oregon, June 2021–December 2022

Weekly / April 11, 2024 / 73(14);317–320

Juventila Liko, MD 1 ; Paul R. Cieslak, MD 1 ( View author affiliations )

What is already known about this topic?

In April 2021, cases of myocarditis after COVID-19 vaccination, particularly among young male vaccine recipients, were reported to the Vaccine Adverse Event Reporting System.

What is added by this report?

To determine risk for sudden cardiac death among adolescents and young adults after COVID-19 vaccination, investigators examined June 2021–December 2022 Oregon death certificate data for decedents aged 16–30 years. Of 40 deaths that occurred among persons who had received an mRNA COVID-19 vaccine dose, three occurred ≤100 days after vaccination. Among these, two occurred in persons with underlying illness, and one decedent had an undetermined cause of death.

What are the implications for public health practice?

The data do not support an association of COVID-19 vaccination with sudden cardiac death among previously healthy young persons. COVID-19 vaccination is recommended for all persons aged ≥6 months to prevent COVID-19 and complications, including death.

• Article PDF
• Full Issue PDF

COVID-19 vaccination has been associated with myocarditis in adolescents and young adults, and concerns have been raised about possible vaccine-related cardiac fatalities in this age group. In April 2021, cases of myocarditis after COVID-19 vaccination, particularly among young male vaccine recipients, were reported to the Vaccine Adverse Event Reporting System. To assess this possibility, investigators searched death certificates for Oregon residents aged 16–30 years who died during June 2021–December 2022 for cardiac or undetermined causes of death. For identified decedents, records in Oregon’s immunization information system were reviewed for documentation of mRNA COVID-19 vaccination received ≤100 days before death. Among 1,292 identified deaths, COVID-19 was cited as the cause for 30. For 101 others, a cardiac cause of death could not be excluded; among these decedents, immunization information system records were available for 88, three of whom had received an mRNA COVID-19 vaccination within 100 days of death. Of 40 deaths that occurred among persons who had received an mRNA COVID-19 vaccine dose, three occurred ≤100 days after vaccination. Two of these deaths were attributed to chronic underlying conditions; the cause was undetermined for one. No death certificate attributed death to vaccination. These data do not support an association between receipt of mRNA COVID-19 vaccine and sudden cardiac death among previously healthy young persons. COVID-19 vaccination is recommended for all persons aged ≥6 months to prevent COVID-19 and complications, including death.

Introduction

In December 2020, the Food and Drug Administration authorized two COVID-19 mRNA vaccines for use in the United States. Early vaccine supplies were prioritized for health care personnel and long-term care facility residents, with phased vaccination of other persons, beginning with those who were older or had high-risk medical conditions, and concluding with healthy younger persons ( 1 ). In Oregon, healthy persons aged ≥16 years became eligible for COVID-19 vaccination on April 19, 2021. In April 2021, reports of myocarditis after COVID-19 vaccination, particularly among young male vaccine recipients, began to appear.* , † Investigators in Israel estimated that the risk for myocarditis associated with receipt of mRNA COVID-19 vaccine was 2.13 per 100,000 among vaccine recipients, and was highest among adolescents and young adult males (10.69 per 100,000) ( 2 ). Published accounts suggest that postvaccination myocarditis is typically mild and associated with good outcomes after brief hospitalization ( 3 , 4 ). As of July 17, 2023, no fatal cases of myocarditis in Oregon had been reported to the federal Vaccine Adverse Event Reporting System (VAERS); however, because VAERS is a passive reporting system, adverse events after vaccination are likely underestimated. In late 2022, reports of sudden deaths among previously healthy young athletes, with suggested attribution to COVID-19 vaccination, appeared in the lay press § and then in the medical literature ( 5 , 6 ). To ascertain whether young persons in Oregon might be dying from cardiac causes shortly after having received a COVID-19 vaccine dose, Oregon death certificate data were reviewed.

Data Sources

Oregon law requires that a certificate of death be completed for each death in Oregon. Oregon’s vital records system abides by CDC’s National Center for Health Statistics’ data-quality standards ¶ , including extensive quality-assurance review. An independent source of data for assessing the completeness of death certificate reporting is not available. Data on Oregon resident deaths occurring outside the state are also collected through interstate exchange agreements. The ALERT Immunization Information System (IIS) is Oregon’s statewide and lifespan immunization registry. During the COVID-19 pandemic, reporting of all COVID-19 vaccinations to ALERT IIS was mandated in Oregon.

Data Analysis

To ascertain the occurrence of sudden cardiac deaths among adolescents or young adults that might plausibly be attributed to recent COVID-19 vaccination, investigators searched the Oregon death certificate database to identify persons aged 16–30 years who died during June 1, 2021–December 31, 2022 with “sudden death,” “arrhythmia,” “dysrhythmia,” “asystole,” “cardiac arrest,” “myocarditis,” “congestive heart failure,” “unknown,” “undetermined,” or “pending” cited among the immediate or four possible entries for underlying causes of death and other significant conditions contributing to death. Among the subset of decedents for whom death from a cardiac cause could not be ruled out by accompanying information in the death certificate database, records of mRNA COVID-19 vaccination within 100 days ( 7 ) before the date of death were retrieved from ALERT(IIS. Findings were stratified by sex. This activity was reviewed by the Oregon Health Authority, deemed not research, and was conducted consistent with applicable federal law and Oregon Health Authority policy.**

In Oregon, during June 2021–December 2022, a total of 1,292 deaths among persons aged 16–30 years were identified. These decedents included 925 (72%) males and 367 (28%) females ( Figure ).

Male Decedents

Among the 925 male decedents, no death certificate listed vaccination either as the immediate or as a contributing cause of death. Overall, 17 (2%) deaths among males were attributed to COVID-19. Death certificates cited noncardiac causes of death or other conditions contributing to death for 842 (91%) of the male decedents. Among the remaining 66 (7%) male decedents, excluding a cardiac cause of death based on the death certificate was not possible. Among these 66 decedents, IIS vaccination records were available for 58 (88%); receipt of at least one mRNA COVID-19 vaccination was recorded for 24 (41%).

Among the 24 male decedents with an mRNA COVID-19 vaccination record in IIS, two (8%) died within 100 days of having received the vaccine. The first death was recorded as having occurred in a natural manner 21 days after COVID-19 vaccination. The immediate cause of death noted on the death certificate was congestive heart failure attributed to hypertension; other significant conditions included morbid obesity, type 2 diabetes, and obstructive sleep apnea. The second decedent had received a COVID-19 vaccine dose 45 days before the date of death; the cause of death was recorded as “undetermined natural cause.” Toxicology results were negative for alcohol, cannabinoids, methamphetamine, and opiates; aripiprazole, ritalinic acid, and trazodone were detected. Follow-up with the medical examiner could neither confirm nor exclude a vaccine-associated adverse event as a cause of death for this decedent.

Female Decedents

Among the 367 female decedents, no death certificate listed vaccination as either the immediate or a contributing cause of death. Thirteen (4%) deaths were attributed to COVID-19. Noncardiac causes were recorded on the death certificates for 319 (87%) decedents. Among the remaining 35 (10%) female decedents, IIS records for 30 (86%) were identified, 16 (53%) of whom had documentation of receipt of at least 1 mRNA COVID-19 vaccine dose. Only one of these deaths occurred within 100 days of having received an mRNA COVID-19 vaccine dose; the decedent died 4 days after COVID-19 vaccination. The manner of death was recorded as natural, and the immediate cause was listed as undetermined but as a consequence of chronic respiratory failure with hypoxia attributed to mitral stenosis.

Electronic health record data from 40 U.S. health care systems during January 2021–January 2022, showed that the risk for cardiac complications was significantly higher after COVID-19 infection than after mRNA COVID-19 vaccination among persons aged ≥5 years ( 8 ). Data from CDC’s National Center for Health Statistics show a background mortality rate from diseases of the heart among Oregonians aged 15–34 years of 2.9 and 4.1 deaths per 100,000, during 2019 and 2021, respectively. Although the rate was higher during the pandemic year of 2021, myocarditis remained an infrequent cause of death among persons in this age group. †† Detection of a small difference in mortality rate from myocarditis would require a larger sample size.

In this study of 1,292 deaths among Oregon residents aged 16–30 years during June 2021–December 2022, none could definitively be attributed to cardiac causes within 100 days of receipt of an mRNA COVID-19 vaccine dose; one male died from undetermined causes 45 days after receipt of a COVID-19 vaccine. During May 1, 2021–December 31, 2022, a total of 979,289 doses of COVID-19 vaccines were administered to Oregonians aged 16–30 years (unpublished data, ALERT IIS, 2024.)

During the same period, COVID-19 was cited as the cause of death for 30 Oregon residents in this age group. Among these 30 decedents, ALERT IIS had records for 22 (73%), only three of whom had received any COVID-19 vaccination. Studies have shown significant reductions in COVID-19–related mortality among vaccinated persons; during the first 2 years of COVID-19 vaccine availability in the United States, vaccination prevented an estimated 18.5 million hospitalizations and 3.2 million deaths ( 9 ).

Limitations

The findings in this report are subject to at least two limitations. First, this report cannot exclude the possibility of vaccine-associated cardiac deaths >100 days after COVID-19 vaccine administration. However, published data indicate that potential adverse events associated with vaccinations tend to occur within 42 days of vaccine receipt ( 10 ). Second, small population size made it less likely that Oregon would see a rare event such as sudden cardiac death among adolescents and young adults.

Implications for Public Health Practice

These data do not support an association between receipt of mRNA COVID-19 vaccine and sudden cardiac death among previously healthy young persons. COVID-19 vaccination is recommended for all persons aged ≥6 months to prevent COVID-19 and complications, including death.

Acknowledgments

Michael Day, Tasha Martin, Anne Vancuren, Center for Health Statistics, Oregon Public Health Division; Rebecca Millius, Office of the Chief Medical Examiner, Medical Examiner Division, Oregon State Police.

Corresponding author: Juventila Liko, [email protected] .

1 Public Health Division, Oregon Health Authority, Portland, Oregon

All authors have completed and submitted the International Committee of Medical Journal Editors form for disclosure of potential conflicts of interest. No potential conflicts of interest were disclosed.

* www.cdc.gov/vaccines/acip/work-groups-vast/report-2021-05-17.html

† https://www.cdc.gov/vaccines/acip/meetings/downloads/slides-2021-06/04-COVID-Lee-508.pdf

§ https://www.nytimes.com/2022/01/28/technology/covid-vaccines-misinformation.html

¶ https://www.oregon.gov/oha/PH/BirthDeathCertificates/VitalStatistics/death/Pages/index.aspx

** 45 C.F.R. part 46.102(l)(2), 21 C.F.R. part 56; 42 U.S.C. Sect. 241(d); 5 U.S.C. Sect. 552a; 44 U.S.C. Sect. 3501 et seq.

†† https://wonder.cdc.gov/ucd-icd10-expanded.html (Accessed February 12, 2024).

• Dooling K, Marin M, Wallace M, et al. The Advisory Committee on Immunization Practices’ updated interim recommendation for allocation of COVID-19 vaccine—United States, December 2020. MMWR Morb Mortal Wkly Rep 2021;69:1657–60. https://doi.org/10.15585/mmwr.mm695152e2 PMID:33382671
• Witberg G, Barda N, Hoss S, et al. Myocarditis after COVID-19 vaccination in a large health care organization. N Engl J Med 2021;385:2132–9. https://doi.org/10.1056/NEJMoa2110737 PMID:34614329
• Power JR, Keyt LK, Adler ED. Myocarditis following COVID-19 vaccination: incidence, mechanisms, and clinical considerations. Expert Rev Cardiovasc Ther 2022;20:241–51. https://doi.org/10.1080/14779072.2022.2066522 PMID:35414326
• Behers BJ, Patrick GA, Jones JM, et al. Myocarditis following COVID-19 vaccination: a systematic review of case reports. Yale J Biol Med 2022;95:237–47. PMID:35782472
• Polykretis P, McCullough PA. Rational harm-benefit assessments by age group are required for continued COVID-19 vaccination. Scand J Immunol 2022;98:e13242. https://doi.org/10.1111/sji.13242 PMID:38441161
• Sun CLF, Jaffe E, Levi R. Increased emergency cardiovascular events among under-40 population in Israel during vaccine rollout and third COVID-19 wave. Sci Rep 2022;12:6978. https://doi.org/10.1038/s41598-022-10928-z PMID:35484304
• Sexson Tejtel SK, Munoz FM, Al-Ammouri I, et al. Myocarditis and pericarditis: case definition and guidelines for data collection, analysis, and presentation of immunization safety data. Vaccine 2022;40:1499–511. https://doi.org/10.1016/j.vaccine.2021.11.074 PMID:35105494
• Block JP, Boehmer TK, Forrest CB, et al. Cardiac complications after SARS-CoV-2 infection and mRNA COVID-19 vaccination—PCORnet, United States, January 2021–January 2022. MMWR Morb Mortal Wkly Rep 2022;71:517–23. https://doi.org/10.15585/mmwr.mm7114e1 PMID:35389977
• Fitzpatrick M, Moghadas S, Pandey A, Galvani A. Two years of U.S. COVID-19 vaccines have prevented millions of hospitalizations and deaths. New York, NY: The Commonwealth Fund; 2022. https://www.commonwealthfund.org/blog/2022/two-years-covid-vaccines-prevented-millions-deaths-hospitalizations https://doi.org/10.26099/whsf-fp90
• CDC. Update: vaccine side effects, adverse reactions, contraindications, and precautions. Recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep 1996;45(No. RR-12):1–35. PMID:8801442

FIGURE . Deaths* among persons aged 16–30 years, by sex, cause of death, † and mRNA COVID-19 vaccination status §,¶, ** (N = 1,292) — Oregon, June 2021–December 2022

* Coded on the death certificate as sudden death, arrhythmia, dysrhythmia, asystole, cardiac arrest, myocarditis, congestive heart failure, unknown, undetermined, or pending.

† Cardiac versus noncardiac.

§ Six of the 34 males who did not receive mRNA COVID-19 vaccine received Janssen (Johnson & Johnson) vaccine.

¶ An alternative plausible cause of death was identified for one of the males who had been vaccinated ≤100 days before death. After review of death certificate and medical examiner findings, an adverse event from COVID‐19 vaccination could neither be confirmed nor excluded as the cause for the other decedent.

** The only female decedent vaccinated ≤100 days before death was vaccinated 4 days before death. The manner of death was recorded as natural, and the immediate cause was “undetermined” as a consequence of chronic respiratory failure with hypoxia due to mitral stenosis.

Suggested citation for this article: Liko J, Cieslak PR. Assessment of Risk for Sudden Cardiac Death Among Adolescents and Young Adults After Receipt of COVID-19 Vaccine — Oregon, June 2021–December 2022. MMWR Morb Mortal Wkly Rep 2024;73:317–320. DOI: http://dx.doi.org/10.15585/mmwr.mm7314a5 .

MMWR and Morbidity and Mortality Weekly Report are service marks of the U.S. Department of Health and Human Services. Use of trade names and commercial sources is for identification only and does not imply endorsement by the U.S. Department of Health and Human Services. References to non-CDC sites on the Internet are provided as a service to MMWR readers and do not constitute or imply endorsement of these organizations or their programs by CDC or the U.S. Department of Health and Human Services. CDC is not responsible for the content of pages found at these sites. URL addresses listed in MMWR were current as of the date of publication.

All HTML versions of MMWR articles are generated from final proofs through an automated process. This conversion might result in character translation or format errors in the HTML version. Users are referred to the electronic PDF version ( https://www.cdc.gov/mmwr ) and/or the original MMWR paper copy for printable versions of official text, figures, and tables.

Exit Notification / Disclaimer Policy

• The Centers for Disease Control and Prevention (CDC) cannot attest to the accuracy of a non-federal website.
• Linking to a non-federal website does not constitute an endorsement by CDC or any of its employees of the sponsors or the information and products presented on the website.
• You will be subject to the destination website's privacy policy when you follow the link.
• CDC is not responsible for Section 508 compliance (accessibility) on other federal or private website.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Oxford University Press - PMC COVID-19 Collection

In Defense of Vaccine Mandates: An Argument from Consent Rights

Daniel a wilkenfeld.

Department of Acute and Tertiary Care, University of Pittsburgh School of Nursing, USA

Christa M Johnson

Department of Philosophy, University of Dayton, United States of America

This article will focus on the ethical issues of vaccine mandates and stake claim to the relatively extreme position that outright requirements for people to receive the vaccine are ethically correct at both the governmental and institutional levels. One novel strategy employed here will be to argue that deontological considerations pertaining to consent rights cut as much in favor of mandating vaccines as against them. The presumption seems to be that arguments from consent speak semi-definitively against forcing people to inject something into their bodies, and so any argument in favor of mandates must produce different and overriding logical and ethical considerations. Our central claim will be that the same logic that might seem to prohibit vaccine mandates as violations of consent actually supports such mandates when viewed from the perspective of the potential bystander who might otherwise be exposed to COVID-19.

Introduction

Recently, the question has arisen of the ethics of pressuring people into getting one of the COVID-19 vaccines. Debates exist along several crisscrossing axes, including:

• The acceptable form of any potential mandate: incentives vs. outright requirement.
• The acceptable locus of any potential mandate: governmental vs. institutional.
• Legal vs. ethical vs. policy considerations with respect to any potential mandate.

This article will focus on the ethical issues and stake claim to the relatively extreme position that outright requirements for people to receive the COVID-19 vaccine are ethically correct at both the governmental and institutional levels. By ‘outright requirement’, we do not mean to suggest that people will be forcibly vaccinated, but rather that some penalty will be assessed for most of those who choose to forgo a vaccine. One novel strategy employed here will be to argue that deontological considerations–and consent rights in particular—cut as much in favor of mandating vaccines as against them. To make allowances for a (narrow) realm of vaccine refusal, we do carve out an exception for those who are willing to take what we call Maximal Preventive Measures (MPMs): doing all of masking, social distancing and providing evidence of a negative test whenever they go into a public space; this carve-out would be sufficiently onerous for most people that it would act as another form of mandate, while allowing for certain legitimate exceptions. Note that our thesis is specifically applicable only to COVID-19 vaccines; we will however explore to what extent our argument might generalize to other vaccines. Even more precisely, our argument was originally formed in the context of variants of COVID-19 through delta. We comment below on how circumstances have changed with omicron (though not in a way that invalidates our argument), but of course given the likelihood that facts on the ground will continue to evolve, it is possible that some of our arguments might prove outdated. Even should that be the case, we maintain that the following ethical analysis of vaccine mandates in the era through delta (and to a lesser extent omicron) still has value in manifesting general argumentative positions that will likely apply to future variants or other viruses altogether.

We take it that one of the strongest arguments against requiring vaccines is that people generally have a right to refuse consent to any infringement on bodily integrity. We frequently hear vaccine opponents invoking the language of medical choice or informed consent. The presumption seems to be that arguments from consent speak semi-definitively against forcing people to inject something into their bodies, and so any argument in favor of mandates must produce different and overriding logical and ethical considerations.

Other defenses of vaccine mandates in the literature have generally taken this structure of rights vs. some other good. For example, they have focused on the unvaccinated’s contribution to a collective harm ( Brennan, 2018 ), the protection of the general public ( Flanigan, 2014 ; Savulescu, 2021 ), the protection of the otherwise unvaccinated ( Giubilini and Savulescu, 2019 ), herd immunity ( Giubilini, 2020 ) and considerations of fairness ( Giubilini, 2020 ). In most of these arguments, the general idea is that the considerations in favor of the mandate outweigh whatever claims or rights an unvaccinated person has to remain unvaccinated. The exception to this strategy is Brennan’s (2018) libertarian argument. Since the libertarian does not allow that considerations of the wellbeing of others can ever outweigh one’s right to liberty, a mandate will only be viable if the mandate falls outside the scope of one’s liberty rights. As such, Brennan argues that the unvaccinated contributes to a collective harm that the government is justified in preventing via vaccine mandate.

Our argument comes closest to Brennan’s in that we also do not focus on benefits of the vaccine mandate outweighing any harms or rights to the otherwise unvaccinated. Where it differs, however, is that our focus is not on a contribution to a collective harm that a government may protect against, but rather on a conflict of rights between the would-be unvaccinated and individuals with an interest in people with whom they interact being vaccinated. Indeed, our central claim is that this is a conflict of the very same right, i.e. one’s purported right to remain unvaccinated is undergirded by the same deontological logic of consent rights that we contend motivate the right of a potential bystander to not be unnecessarily exposed to COVID-19. As a catch-all term, we speak of a general right to engage in the world free of harms imposed on one’s body without consent; we mean by this construal to pick out whichever right or rights vaccine refusers and bystanders alike seek to invoke. The crux of the argument is that when one defends one’s right to remain unvaccinated, one inevitably also accepts infringement upon a bystander’s right to not be exposed to COVID-19 by the unvaccinated. What justifies a vaccine mandate, or so we will argue, is how this conflict of rights in kind gets settled.

We will be focusing our arguments exclusively on the justification of mandates for vaccines in Western cultures. We suspect that the strongest case against individual mandates can be made for a Western culture such as that of the USA, for two reasons. First, given their emphasis on individualism and individual rights, one would suspect a stronger cultural norm against essentially overriding individual decision making. Second—and we will return to this in the objections and replies—given the emphasis in at least the more liberal corners of Western culture on bodily integrity in support of reproductive rights, we might expect similar logic to speak in favor of preserving bodily integrity in the case of vaccine refusal.

We will also largely be omitting one standard argument in favor of vaccine mandates, as hinted to above. One might think that the sheer scope of the COVID-19 pandemic would justify overriding what would normally be ethical rights for the sake of avoiding catastrophe. While the scope of the pandemic will play a role in our argument, we do not intend to argue that otherwise unethical action is justified in this case on purely utilitarian (i.e. outcome-based) grounds. Rather, we argue that—properly thought of—mandating COVID-19 vaccines is not unethical in the first place. While others (especially Brennan, 2018 ) have argued that vaccine mandates do not violate general constraints against government restriction, we take our advance to be framing the defense of mandates in the very language of rights and consent most commonly used by their opponents.

For purposes of this paper, we begin with three assumptions. First, we assume the extreme safety and relative efficacy of the COVID vaccines. More specifically, we assume that there are very low odds of unforeseen serious side-effects ( Blumenthal et al. , 2021 , Klein et al. , 2021 ), that being vaccinated reduces one’s odds of acquiring COVID-19 ( Thompson et al. , 2021 ) and that being vaccinated greatly reduces the odds of transmitting COVID-19 to other people. This could be true because it reduces viral load ( Petter et al. , 2021 ; Vitiello et al. , 2021 ) or because it reduces the length of one’s contagion ( Thompson et al. , 2021 ), but if nothing else it follows as at least plausible from the fact that vaccinated individuals are less likely to be infected in the first place ( Centers for Disease Control and Prevention, 2021b ). Obviously many people who oppose vaccines on any level are likely to dispute this assumption, arguing either that vaccines are unsafe or that they are ineffective. However, the fact that people argue something does not in itself imply that it is a plausible position, and so—given the overwhelming empirical evidence—in this case it is reasonable to simply set aside for the purposes of ethical analysis claims that deny the vaccine’s (general) safety and effectiveness, so long as there is support for those who suffer side effects, as well as a very limited set of exceptions for those with legitimate medical reasons not to get the vaccine. That said, while the safety of the vaccines is unlikely to change, the soundness of the assumption that they reduce infection and transmission might wax and wane as new variants become dominant. For example, when this paper was drafted the dominant variant was delta (see citations above), but during revision the omicron strain became responsible for almost all infections in the USA ( Centers for Disease Control and Prevention, 2022 ). While the omicron variant exhibits significantly more vaccine escape than earlier variants ( Lyngse et al. , 2022 ), preliminary results indicate that vaccines are still somewhat effective at reducing infection and transmission ( Lyngse et al. , 2022 ) Perhaps by the time this paper is published or read the situation will have changed sufficiently that our underlying assumptions are no longer sound. If so, we present our arguments as pre-emptive considerations for how to treat vaccine mandates in the face of future variants or pandemics where these assumptions do apply.

Second, we assume that in the absence of a mandate there will be a large number of people who do not get the vaccine and that in the presence of a mandate this number will go down. The former claim is undeniably empirically accurate—there are as of this writing large swaths of the population who refuse to get the vaccine. The latter point is more speculative—it is possible that a mandate would somehow backfire and lead to fewer total vaccinations. However, evidence suggests that at least at the institutional level when mandates are enacted people become more likely to get the vaccine ( Greyson et al. , 2019 ; Gostin, 2021 ); so this seems like a reasonable assumption. (We do not however commit to how many more people would be likely to get it.) Finally, we assume that the vaccine is readily available—obviously it would be unjust to mandate someone get something that there is no way for them to get.

Note also that the exception for people taking MPMs discussed above (masking + distancing + testing whenever entering a shared public space) also entails that our discussion only applies to people who enter shared public spaces. If one really lives one’s entire life in a wholly insular fashion (as in a hermit in the woods), then one can trivially satisfy the mandate by doing nothing new (since one is not entering publicly shared spaces). In addition, there is likely a gray area where one has such minimal contact with outsiders—perhaps in a rural farming community—that taking MPMs is sufficiently doable that it represents a reasonable alternate path such that the vaccine is in some sense no longer ‘mandated’. Given how interconnected most people are (even those who spend most time in isolated locations generally have some need of interacting with the broader population) we thus focus our discussion on those who have some interest in regularly being in shared public spaces.

The structure of this paper is as follows. The next section provides the major arguments in the paper, showing how the very same deontological considerations that might speak against mandating vaccines in support of the consent rights of the recipient also speak in favor of mandating vaccines in support of the consent rights of those who might potentially be exposed. We will then discuss how to address these competing rights-claims and argue that the best resolution is to favor the rights of potential victims of COVID-19. In the following section, we will build on well-known analogies from clinical/medical/nursing ethics—this is intended merely to be illustrative. We then expand the argument from governmental mandates also to defend on similar grounds institutional mandates at effectively any sort of institution. We conclude with objections and replies.

Rights-Based Arguments for and against Mandates

Vaccines and individual rights.

When someone makes a decision not to get a COVID-19 vaccine, they are of course making a decision pertaining to their own healthcare. However, what is sometimes overlooked is that they are making a decision pertaining to others’ healthcare as well. Though there is no guarantee that anyone who is not vaccinated will be exposed to COVID-19 and will pass it on, not being vaccinated makes it more likely that they will do so (see above). In that way, they affect the rights of future people with whom they interact. By analogy, it is clearly wrong to put toxic chemicals in someone else’s water; we can then consider a person or company that allows potentially (but not certainly) dangerous chemicals into a local water supply. They might not know that they are putting anyone in danger and certainly would not be able to point to a specific individual who will be harmed. However, the potentially more diffuse nature of the wrong in not knowing whom will be hit seems on the surface almost exactly counterbalanced by the very real possibility that they will harm multiple people. In the same way, while we cannot point to a specific individual Y who will get COVID-19 as a result of person X’s decision not to get vaccinated, there are a whole host of individuals Y 1 – Y n who suddenly find themselves in unwanted harm’s way. It is also worth noting that some of the Y i s might not be able to get the vaccine themselves, either due to overriding medical reasons or the simple fact that (as of this writing) it is not yet approved for or available to all populations. In the case of pollutants, the right being violated is Y’s ability to live their life and assume a certain level of safety in their water supply, or more generally the right to engage in the world free of harms imposed on one’s body without consent.

What right then is being violated with respect to a person who is forced to get a vaccine? The clearest answer is the same as above, i.e. one’s right to engage in the world free of harms imposed on their bodies without their consent. This suggests we look at the literature on consent to ascertain whether their rights are more sacrosanct than the victims of any potential COVID-19 exposures. In the remainder of this section, we speak as if there is one particular person who will be exposed to COVID-19 as a result of an individual’s vaccine refusal—as pursuant to the previous paragraph, we take this to be morally similar to the more realistic scenario where there are multiple people with massively increased potential exposure.

Sources of Consent Rights

In this section, we will go over some common justifications for people’s right to refuse interventions on their bodies and argue that those same justifications provide at least some prima facie reason to think that they in most cases do not have a right to refrain from getting a COVID-19 vaccine. The key to this strategy is arguing not that there are conflicting kinds of rights, but rather that the very same kind of right that would justify vaccine refusal also justifies vaccine mandates.

For example, the right to refuse interventions is most frequently grounded in autonomy, which is literally the right to make laws for oneself. Spelling out precisely why this is the case is complicated by the fact that philosophers have no clear unified conception of autonomy ( Buss and Westlund, 2018 ). We do not need a full account of autonomy however to note that one necessary condition for autonomy is the liberty to decide for oneself how to live one’s life free of unnecessary externally imposed impediments. It is this liberty criterion that vaccine mandates are often thought to violate, but we will argue that the absence of mandates is responsible for violations of that same liberty criterion. For approximately as long as philosophers have discussed anything like autonomy or liberty there has been a general recognition that liberty rights can conflict in such a way as to make it impossible for everyone to have maximal liberty all of the time. Hobbes (2016 /1651, Chapter 13) famously observed that if everyone were free to do as they would, life for everyone would be ‘nasty, brutish, and short’, and even John Stuart Mill’s (2011 /1859) most famed statement of maximal individual freedoms in ‘On Liberty’ acknowledged that one’s liberty always needs be curtailed when its exercise would infringe upon the liberty of others. Yet—given our assumptions about the effectiveness of the vaccine and the need for common areas—this seems like a paradigmatic example of where one person’s liberty would limit another’s. My liberty to be able to engage in society without being ‘assaulted’ by a vaccine is no more obviously sacrosanct than your liberty to be able to engage in society without being ‘assaulted’ unnecessarily by deadly virus. I cannot govern myself as I will when I am willfully exposed to COVID-19.

Two notes are in order. First, some would argue that the battery of having a needle puncture your body violates one’s rights in a way that an increased risk of contracting COVID-19 does not (for one such argument, see Kowalik, 2021 ). For the most part we simply reject the reasonableness of this distinction on several grounds. First, on the actuality of the harm the needle causes vs. the mere possibility of contracting COVID-19, we note that the harm of the needle itself is quite minimal and that is the only harm 100% guaranteed. The reception of the vaccine itself is not a harm, unless there are adverse side effects, which are simply an added risk—not unlike the added risk of contracting COVID-19. Thus, if we step back and look at the overall expected utility of the actual needle jab and the possibility of adverse side effects of the vaccine with the overall expected utility of the increased risk of COVID-19 exposure, we contend that the latter is much worse. The risks of COVID-19 include unpleasant symptoms [ Ma et al. (2021) recently provided a headline-generating result that 40% of cases are asymptomatic, but that suggests that 60% might not be], ‘Long COVID’ ( Crook et al. , 2021 ), hospitalization ( Scobie et al. , 2021 , especially Figure 2) and death. The seeming similarity in kind and relative seriousness of potential harms from the virus as compared to the risks of the vaccine ( Blumenthal et al. , 2021 ; Klein et al. , 2021 ) + the actual harm of the jab make it seem like the rights violation are minimally of a piece (leading to our discussion in the next section of how they should be adjudicated). Second, focusing on the unwanted foreign agent itself, whether one receives an unwanted vaccine or an unwanted infection, the issue is that an unwanted foreign agent is entering one’s body without one’s consent–drawing a sharp distinction based on the foreign agent’s mode of entry would suggest that a vaccine mandate would be ethically worse than making the vaccine airborne and spreading it throughout the country. We suspect that most of those we have encountered who argue against mandates on the grounds that they do not consent to the intervention of a shot would be unlikely to accept the intervention being thrust on them via a different and more pervasive mechanism such as being omnipresent in the atmosphere. Finally, one might argue that there are different levels of consent violations–an unapproved cheek swab is an ethical problem, but clearly a smaller one than an unapproved surgery. Precisely what makes one violation worse than another is beyond the scope of this paper, but presumably one vector of evaluation is the expected harm done (as measured in the severity of possible outcomes multiplied by the likelihood of those outcomes obtaining). As just discussed, the calculus of expected harms speaks in favor of mandating a vaccine—the point here is that this same calculus might well also speak to the severity of a rights violation in exposing someone to that harm without consent relative to the consent violation of being mandated to get an unwanted vaccine.

Of course, one might argue that we are underestimating vaccine risks and overestimating how severe COVID-19 is to everyone. After all, there have been cases of reactions to COVID-19 vaccines (Centers for Disease Control and Prevention, 2021a) and there are populations for whom severe cases of COVID-19 are rare ( American Academy of Pediatrics, 2021 ). On the first issue, (of underestimating vaccine risks) we make three points. First, we began with the assumption that the vaccines are safe. To that end, it may be that certain vaccines, e.g. Johnson and Johnson or AstraZeneca, may not be ethically mandated due to their increased safety risks and lower efficacy ( Centers for Disease Control and Prevention, 2021c ). Second, we remind the reader that the mandate we propose does include MPMs as an alternative to receiving the vaccine. Those unwilling to receive the jab may choose N95 masking, distancing and testing as an alternate route to avoid violating the consent rights of bystanders. Third, we would agree that we can set aside the relatively rare instances of vaccine side effects, so long as there are accommodations for those who have side effects, as well as an exemption for legitimate medical reasons. The idea here is that when there are indeed side effects from the vaccine received due to a mandate, the ethical mandate will include provisions for compensation. On the latter issue of overestimating the severity of COVID-19, we again make three points. First, there are cases of severe COVID-19 across all age-groups, even if prevalence of cases is lower in certain age-groups ( American Academy of Pediatrics, 2021 ). Indeed, the prevalence of severe COVID-19 across groups is higher than the prevalence of severe reactions to COVID-19 vaccines (compare Centers for Disease Control and Prevention, 2021d to Delahoy et al ., 2021 for cases of adverse reactions to the vaccine to COVID-19 hospitalizations). Second, unless the unvaccinated can be sure only to interact with individuals from those groups who do not regularly suffer from severe COVID-19, it will not matter that some individuals fall into that camp. The unvaccinated will inevitably interact with those for whom severe COVID-19 has a higher prevalence. Finally, while it is possible to offer compensation and accommodation to those few who react poorly to the vaccine, a parallel proposal for those who ultimately suffer from severe COVID-19 is untenable. That is, it seems much more plausible to make whole those who have bad side effects from the vaccine mandate than to make whole those who suffer severe COVID-19 due to the lack of a vaccine mandate.

As a second note on the liberty argument, Brennan (2018) has already argued that variants of Mill’s harm principle are sufficient to justify vaccine mandates. Our approach is subtly different in that Mill’s harm principle is characterized as a general limit on person X’s liberty whereas we are grounding our argument in the very same rights justifying vaccine refusal (e.g. liberty). This has an advantage that it defends against those who might think that unwanted medical interventions are a different kind of consent violation that cannot be overridden by Brennan’s ‘clean hands principle’—we argue that those the very same principles that support the vaccine refuser’s argument also undermine it. (Brennan’s approach has other advantages in engaging with specific libertarian concerns–as such we consider the two complementary rather than in competition.)

Similar strategies of looking at the question of rights from the potential of the prospective victim of COVID-19 exposure suffices to defray many other concerns with other intrusions on bodily integrity without consent. For example, some people ground the right to refuse intrusions in the fact that we own our own bodies (Eyal, 2012: 14). But just as my ownership right to a field gives me a claim against a neighbor whose conduct polluting risks dropping soot on my crops, so my ownership of my body gives me a claim against someone whose conduct risks dropping unnecessary SARS-CoV-2 droplets in my breathing area. Likewise, while your bodily integrity is undermined by receiving an unwanted shot, mine is undermined by receiving an unwanted COVID exposure.

One final worry worthy of special mention is that allowing the right to refuse bodily infringements is necessary to prevent abuse at the hands of authority figures ( Manson and O’Neill, 2007 ). In this case, one might worry that allowing the government the authority to mandate one shot will open the door for allowing future governments to mandate shots for more nefarious purposes. Another version of this concern might be a ‘slippery slope’ objection, which acknowledges that a vaccine mandate might be justified in this case but that allowing one would open the door to instances where such a mandate would be unjustified. However, the proper response to this is perhaps the standard one to most slippery slope arguments, which is that if the current action is justified but a future later one might not be then we need a mechanism in place that pulls the brakes right at the juncture between the justified and the unjustified. The way to prevent unjustified behavior is not to ban justified behavior, but rather to be vigilant regarding when one might cross the relevant boundary. This objection reasonably speaks against giving the government carte blanche authority to institute vaccine mandates but does not speak against allowing it to mandate this specific one. We would in effect require a new analysis to be done for each prospective vaccine. For example, current flu vaccines might not be amenable to mandates, as they violate the assumptions of strong effectiveness and high likelihood of spread and conceivably alter the calculation of expected harm that might be relevant for weighing consent violations against each other. As the effectiveness of flu vaccines increases and if the contagiousness and severity of flu infections increase the case will approach to COVID-19; our current situation provides a clear case against which other vaccine mandates could be compared—if the harms of the virus and safety/effectiveness of the vaccine are at least as great as they are for COVID-19, then a mandate is justified. Anything less must be evaluated on a case-by-case basis.

Competing Rights Claims

Suppose one accepts as above that there are competing rights claims—of the same kind—between potential unwilling vaccine recipients and potential unwilling victims of COVID-19 exposure. The next question is how we adjudicate between such conflicting rights claims. One move that would be reasonable here would be to reinvoke the societal costs of COVID-19 and argue the default should be the permissibility of a vaccine mandate unless there is a rights-based argument against having one. If the rights-based arguments all turn out to counterbalance, that would leave in place the default need to protect society of a rampaging pandemic. We think this would be a perfectly reasonable argument; however, as the ‘consequentialism vs. deontology’ argument (basically an argument between achieving positive outcomes at the cost of violating ethical ‘rules’) is well-trod ground, we table that line of reasoning in favor of arguing that a consideration of rights on their own terms favors vaccine mandates.

It is of course well beyond the scope of this paper to consider every way in which one might resolve conflicts among different people’s rights. We will thus argue from a framework inspired by Rawls’ landmark A Theory of Justice (1971/1999), widely considered to be the dominant work of political philosophy of the last century. We believe that the choice of this framework is not necessary for our ultimate conclusion and that virtually any system for trading off rights would get the same result—however, we obviously save proving this claim for future work. We will however entertain the possibility that this whole approach is wrong-headed and that a proper deontological (i.e. rule-based rather than outcome based) perspective demands that rights cannot really be weighed against each other or traded off in the first place.

Rawls’ central innovation is the ‘Veil of Ignorance’, wherein people in an ‘Original Position’ determine what is just by what one would agree to if one did not know exactly who one was. The basic idea is to imagine a group of people setting the rules for a new society, in particular the allocation of primary goods [including (at least in our version) such ‘goods’ as rights]. However, no one in that room has any idea who they are in the society; they do not know their race, gender, economic status, or any other identifying feature. Since they do not know who they are, anyone can be reasonably expected to represent all of humankind. Rawls, for his part, concludes that two principles of justice fall out of this setup. However, there is a wealth of literature debating whether Rawls is correct about what principles would fall out of the Original Position as well as how and to what those principles should apply, if correct. We do not wish to get bogged down in Rawlsian interpretation here. For our purposes, we instead turn to a Rawlsian lesson: the contractarian under-pinning of moral principles.

In envisioning the social contract, we need to discern what we would all agree to if we were fully rational and free of prejudice. This is what the Original Position and Veil of Ignorance are meant to establish. Though individual public health issues go beyond the scope of Rawls’s vision, we can use his thought experiment to develop one way of thinking through how a society ought to trade off rights when they conflict. We maintain that when setting up a society, if you do not know who you will be in that society, it is in your interest to protect those worst-off, in case you are one of those people. As such, when an issue arises in which not everyone’s rights can be met, one way of thinking through how to resolve the conflict of rights is to focus on protecting the rights of whoever would be worse off for the violation. Getting back to COVID-19 vaccine mandates, we contend that this reasoning speaks fairly clearly in favor of mandates. Given that we carve out exceptions for those with legitimate medical needs, the person who gets a vaccine they did not want is significantly better positioned than the person who gets COVID-19 exposure they did not want.

Given our use of Rawls’s setup, it is worth considering some of the push back it has received. First, some ( MacIntyre, 1981 ; Sandel, 1982 ) have argued that it is problematic to deny people in the original position all knowledge about their identity. How can I make a rational choice if I have no knowledge about my values or aims? If what is rational is whatever is in my best interest, I need to know what interests I have. Minimally, one should be offered their probability of belonging to a particular group that has particular interests. For example, if one knew that there was only a 1/7,000,000,000 chance of being a single person picked out for human sacrifice in a world where everyone else is obscenely rich, one might reasonably choose to take one’s chances. However, providing knowledge of probabilities would only make the case for mandating vaccines that much stronger, since one is much more likely to be harmed by exposure to COVID-19 from an unvaccinated individual than to receive any harm from the vaccine (see previous section). Others ( Harsanyi, 1975 ) have worried that even in the absence of probabilities Rawls (and in turn we) overestimates how risk averse people either are or should be. Psychologically speaking, perhaps people would be willing to risk a low well-being floor in the hopes of achieving a high well-being ceiling. This may be true, but notice that in this case, since one’s well-being floor and ceiling both go up if there are vaccine mandates (with suitable narrow medical exemptions), for each individual person in the population one’s odds of harm are greater if there is no mandate than if there is a mandate. Thus—whomever one thinks one might be—one is better off with the mandate. And the same math works for average utility. Given that the question of rights was a wash, this suggests that anyone in the Original Position should opt for a mandate.

One might at this point object that this entire section is based on a faulty assumption that rights claims can be traded off at all. One might think that certain rights are inviolate, even if respecting them involves a greater infringement on the rights of others ( Thomson, 1990 ; Kamm, 1996 ). There are countless cases used to show that one may not harm an individual to prevent harm to others. For instance, many argue that one may not push a hiker off a footbridge to stop an out-of-control trolley from killing five others ( Thomson, 1976 ). Likewise, it is argued that one may not kidnap an innocent person and harvest their organs to save the lives of five people in need of organ transplants ( Foot, 1967 ). To generalize the point, if there is an existing threat to some group of people, it is wrong to introduce a new threat to a third party to protect the group already under threat (or so the argument goes). In the case of COVID-19, one might argue that those who might get COVID-19 are already under threat and that the vaccine mandate introduces a new threat to the unvaccinated to protect the group already under threat. However, there is a clear disanalogy here insofar as the unvaccinated individuals are the threat. There is a morally important difference between putting an individual at risk when an out-of-control trolley will possibly cost lives and putting an individual at risk when that very individual will possibly cost lives.

There is another way to see the case, however. We have been arguing that there is a conflict of rights in the vaccine mandate case. Yet, the trolley and surgeon cases above are not necessarily conflicts of rights. These cases involve violating a right to save people, and few actually argue that individuals have a genuine right to be saved from harm. Many do argue, however, that one may not infringe a right to prevent others from having their rights violated ( Kamm, 1989 ; Heuer, 2011 ; Johnson, 2019 ). Indeed, one may not even do so when the same right is at issue. That is, I am not permitted to kill one even if it would stop five others from being killed. In the literature, this particular case has been dubbed the ‘paradox of deontology’. After all, it seems a bit odd that one would think killing is bad, yet not try to minimize them ( Nozick, 1974 ; Scheffler, 1988 ). However, deontologists (i.e. ethicists who focus on rules rather than outcomes) have argued at length, and in many ways, that we are not permitted to treat an individual as a mere means to an end. In these cases, violating that one right would be akin to using that individual as a mere means to the end of preventing other rights violations. Bringing this back to the vaccine mandate, it seems that we have a case of violating an individual’s liberty/consent rights (as characterized above) to prevent the violation of the liberty/consent rights of others, a clearly impermissible action according to these deontologists.

In response, again we can see a disanalogy. In ordinary cases discussed in the literature, there are a number of people whose rights will be violated, unless the rights of a neutral third party are violated. In the vaccine mandate case, however, the unvaccinated individual is not a neutral third party. Rather, the unvaccinated individual is the one who, if their rights are not violated, will violate the rights of the masses. To summarize, we have two parties at issue (the potential COVID-19 getter and the unvaccinated) and two possible situations (mandate or no mandate). Both parties have the potential to have their rights violated, depending on the situation. However, it is only the unvaccinated that would become a rights violator (in the no mandate situation). As such, this is not an ordinary conflict of rights. Rather, we have an innocent party at risk from a potential guilty party. And, although one might argue that seen this way, the mandate constitutes a sort of Pre-Crime preventative justice measure, a safe and effective vaccine can hardly be seen as a punishment, and prior to vaccination, we would argue that the unvaccinated is already violating the rights of potential COVID-19 getters. As such, it is not merely preventative. This marks a key area where our argument reaches farther than Brennan’s (2018) —since that piece was not addressed to deontologists, it (quite reasonably) does not consider the position of those who would take certain specific rights to be inviolable even to protect the rights of others. Our argument does so. (In a sense, our task has been made much easier by the sheer virulence of COVID-19 allowing us to assign individual culpability rather than rely upon concerns relating to collective action.)

There is another disanalogy worth mentioning before moving on. In the cases deontologists normally discuss, it is an ordinary bystander that we imagine either initiating the new threat or else violating the rights of the individual. Deontologists then argue that a bystander is not morally permitted to perform such acts to prevent harm or rights violations. However, in the vaccine mandate case, we do not have a mere bystander, we are considering government and institutional mandates. A bystander has no special obligation to the persons whom they would protect. Governments do have such special obligations, and some institutions might as well. So, not only do the stakes change insofar as the unvaccinated individual is the threat or potential rights violator, but the Government or institution who would infringe the rights of the unvaccinated via a mandate also have a special obligation to all parties involved to do what is necessary to protect them.

While one might seem to have a liberty/consent right not to be forced to get a vaccine, refraining from getting a vaccine makes one a perpetrator violating the liberty/consent rights of others. As such, it is legitimate for the government to prohibit one from doing so.

Analogies: Rights Violations and the Protection of Others

In this section, we point out that not only are there other circumstances (even in the medical domain) where we think that it is acceptable to infringe on what seem to be the rights of someone to protect the rights of others, but that (again) the same logic applies even more forcefully in the case of mandating COVID-19 vaccines. Some of the claims in this section will be controversial, so we note that our central argument in the previous section can (and should) be accepted independently of the analogies presented here. However, we believe the present analogies are still instructive regarding when it might be acceptable to infringe on what seem to be the rights of X for the sake of protecting the rights of Y.

To take perhaps the most obvious example, psychiatrists are required (legally and presumably also ethically) to break what is otherwise a strong right of confidentiality if not doing so would endanger the health and safety of a potential victim of violence ( Kahn, 2020 ). That case on the surface is fairly analogous to the present one, where mere potential harm to someone else suffices to override someone’s rights. Nor do we think the ethical calculus changes dramatically if—instead of threatening a specific individual—a psychiatric patient ‘just’ threatens to put potentially toxic chemicals into a shared water reserve—a diffuse risk of harm to a large number of anonymous people seems just as ethically relevant as a more specific risk of harm to a named individual.

However, one might believe that confidentiality rights are somehow more contingent or defeasible than consent rights, and so we turn to a second analogy perhaps more closely aligned with vaccine mandates. Parents generally have a right to decide for their children whether or not they will receive a medical intervention ( Wilkinson and Savulescu, 2018 ). However, the default view of ethicists in the relevant domains is that there are generally some (limited) circumstances where it is acceptable to override those rights for the sake of protecting someone else’s—in this case the child’s. For example, it is generally believed (e.g. Conti et al. , 2018 ) that it is acceptable to provide blood transfusions for the children of Jehovah’s Witnesses, even if the parents believe that doing so will cost the child their soul. If this position is correct (and we think that it is), then by itself it shows that we can override X’s rights for the sake of Y’s health. One might object that in this case the parental right is really just the child’s right by proxy, and hence, the cases are not relevantly analogous. However, there is still a conflict of autonomous individuals even in this case. While parents have default decisional authority on behalf of their children, the child still has a liberty interest of their own, which the parent is potentially violating by making a decision that has the potential to harm the child. (For more on the distinction between decisional authority and children’s liberty/autonomy, see Wilkenfeld and McCarthy, 2020 ). Seen as a potential conflict of liberty rights, we argue that a recent look at the best logic behind overriding parental rights also suggests overriding the apparent right to refuse a vaccine.

A recent article by Brummett (2021) makes the point that despite ethicists’ best efforts, it is not really plausible to ground the acceptability of overriding parental refusal in terms of neutral criteria like ‘minimizing harm’ ( Salter, 2012 ) or demanding internal consistency ( Bester, 2018 ). Brummett’s insight is that if one really took seriously the prospect that receiving a blood transfusion might cost a child their soul, then one could not reasonably maintain that doing so minimizes harm or in some way enforces consistency. Rather, we override the parent’s judgment not based on neutral procedural grounds, but based on our firm conviction that they believe a metaphysical claim that is simply false. If Brummett has correctly identified the justification for overriding parental rights, then it applies one thousand-fold to the question of vaccine mandates. The reason is that while we might believe that Jehovah’s Witnesses are wrong about blood transfusions costing children’s souls, it is hard to reasonably claim that we could possibly know it, and impossible to reasonably claim that we could ever prove it. However, per our assumptions, we do know that beliefs about the dangers of vaccines are simply incorrect and we have already proven it. Thus, if X’s endangering Y being based on a false belief is reason to override X’s rights, then the case is significantly stronger here than it is in the case of blood transfusions. Lest one worry that this logic could prove too much by allowing clinicians to paternalistically override patients’ wishes whenever those wishes are based on a provably false belief, note that when X’s decision only endangers himself there is no competing rights claim and the issue never arises in the first place.

Institutional Mandates

If the case has been successfully made that government vaccine mandates are ethically acceptable, then most of the logic applies doubly to institutional mandates, such as a university requiring vaccination as a condition of enrollment (subject to legitimate medical exemptions and corresponding precautions for those cases). The concern with government mandated vaccines is that they infringe on someone’s rights; however, if we are correct that doing so is part of the best system of overall rights protection then it is just as legitimate for institutions to respect potential victims’ rights in the same way.

In addition, there is the obvious point that groups of people are—with various exceptions—ethically free to associate as they see fit, and so they are likely entitled to demand people waive certain genuine rights as a condition of association. Presumably people have a right against being tackled by others, yet it is reasonable for professional sports associations such as the National Football League to demand that athletes waive that right to participate in on-field activities. It is their game, so they get to set the rules—if one does not want to waive that right, one always has the options not to play or to start one’s own group.

There are several lines of resistance one could put up to this argument. First, one might argue that some institutions (e.g. hospitals) have an ethical obligation to be open to the public, and so logic gleaned from a football organization does not apply. One might also point out that if every institution instituted a mandate then there would be nowhere else for people who did not want vaccines to go. However, in both cases the answer is the same—at the limit, the most restrictive institutional mandates can be is akin to government mandates, depriving individuals of a choice regardless of their own decisions to associate. If we have already established that government mandates are acceptable, at most these arguments show that there are no additional reasons in support of institutional mandates.

Another objection might be that similar logic to that used to defend institutional mandates above (i.e. freedom to associate) has historically been used for pernicious ends such as refusing minorities service (e.g. by refusing to make wedding cakes for gay marriages). For the most part, the ethics of allowing refusal of service based on minority status are complex and beyond the scope of this paper. However, there are two clear disanalogies between requiring that (for example) students receive vaccines and requiring that wedding cake customers be heterosexual. First, in the bakery case there would be a concern that if all bakeries had similar policies, then it would be impossible for gay couples to get wedding cakes at all. However, in this case, one can acquire the services simply by getting the vaccine, so there is no risk of being shut out simply in virtue of one’s identity. (We do assume that a gay person cannot just choose to be heterosexual, but even if they somehow could, it would be metaphysically impossible for this gay couple qua gay couple to somehow be heterosexual.) Second, we suspect (though will not here defend) that part of the issue with the bakery example is that refusing service on the grounds of sexual orientation is a capricious reason to do so—it seems exclusionary for no legitimate reason. Since there are clearly strong legitimate reasons that an institution would want its students/workers/customers/etc. to be vaccinated, there is no worry about capriciousness here.

Objections and Replies

Objection 1: If one can be required to waive bodily rights for the sake of another person, that will be used as a reason to limit abortion rights. That functions as a reductio against the original argument.

Reply 1: First, let us grant for the sake of argument that the fetus is a fully rights-bearing person. Note that if it is anything less than fully rights-bearing then there is no conflict of rights among equals, and the arguments above never get off the ground. But in any event the argument still does not go through, because the translation of our original premise that the vaccine is safe is simply false ( Kazemi et al. , 2017 ). Many pregnancies go relatively smoothly, but even then the woman is severely restricted for roughly nine months. And quite a lot of pregnancies do not go smoothly. Women can develop wrenching and dangerous nausea ( Bustos et al. , 2017 ), heart problems ( Iftikhar and Biswas, 2019 ), blood clots ( Devis and Knuttinen, 2017 ), etc. So there is simply no analogy between mandating a vaccine and mandating a continued pregnancy. One might get the result that if a fetus is fully rights-bearing and if a woman can do so without cost or danger and if no one else can do so then she might have some minimal obligation to aid the safe extraction of a post-viability fetus. But such a triply conditional conclusion does not seem like an obvious reductio. Arguably it is just a restatement of famed abortion rights philosopher Judith Jarvis Thomson’s (1976) concession that the right to an abortion is the right to end a pregnancy rather than a right to a dead baby. Note also that the limited conclusion might not allow for an enforcement mechanism as readily as would a vaccine mandate—knowing in the first place who is pregnant and what they are doing for their fetus would require a level of invasion of every woman’s privacy (even those not actually pregnant) that has no analogy in the case where everyone is required to get vaccinated (or show evidence of MPMs) to enter public spaces.

Objection 2: Institutional mandates risk unintended consequences. For example, if a hospital mandates that nurses get the vaccine, then nurses might quit and go work at a less well-regulated care facility where still more vulnerable people will be exposed to the virus.

Reply 2: We consider this a very real concern, though note that it has only limited application. While unvaccinated nurses congregating at less well-regulated nursing homes might be a risk, there is no reason to expect (for example) unvaccinated students would gather anywhere vulnerable and less well-regulated. This is also more of a policy question than an ethical one, where what really needs be resolved is not whether institutional mandates are ethical but rather how we can make sure that the absence of mandates are not disproportionately burdensome on particular populations. Interestingly this very objection strengthens the case for a government mandate, as one of the points of government action is to make sure that we avoid a race-to-the-bottom where some institutions see advantages in refusing to enact vaccine mandates.

Objection 3: We do (and presumably should?) let people take all sorts of actions that pose risks to others, such as driving. Similarly, we should let people walk around unvaccinated.

Reply 3: This objection is potentially more potent in the wake of omicron than it was upon drafting this paper. As mentioned at the start, vaccines are potentially less effective against the omicron variant than past variants. If the vaccine is not as effective, then one might to tempted to think that we might as well allow people to walk around unvaccinated at this stage in the pandemic. Unless we are endorsing a strict lockdown, people’s rights to not be assaulted by COVID-19 will be infringed, vaccine or not. To reiterate the objection, we allow people to take all sorts of actions that post risks to other, so why not the act of walking around unvaccinated? There are several disanalogies between cases like being allowed to drive and being allowed to refuse a vaccine. First, there are legitimate societal reasons for wanting people to be able to drive. Even if sometimes people drive for no discernible reason, it is still at least potentially in everyone’s interests for people to be able to drive generally. Returning to the Original Position, if no one were allowed to drive that would severely hamper one’s unknown self’s potential well-being in a way that being forced to receive a particular vaccine would not. Second, as Giubilini et al. (2021) argues, even in the case of driving, there is massive government regulation regarding how precisely it must be done. We cannot (and should not be able to) just drive as we see fit—if one wants to enter the sphere of drivers, there are certain rules. In fact, to even enter the sphere of drivers at all one needs to meet a certain government-imposed requirement (getting a license)—in the same way, to enter the sphere of societal interaction one might need to meet another condition. One might argue that one could simply refuse to drive, but the foregoing is still sufficient to address the issue that we simply allow people to risk the lives of others. We can also see from this example why general lockdowns are less ethically justifiable than vaccine mandates, even in the face of a more transmissible variant such as omicron. The vast majority of people would be significantly harmed by being barred from public spaces altogether, so it is unlikely people would choose such an option from our version of the original position. As with driving, allowing and regulating the valuable activity is significantly more justifiable than simply banning it outright.

Objection 4: One reviewer notes that we generally countenance communities running risks of spreading the common cold or the flu, so we have no principled reason to deny localities the right to run the risk of spreading COVID-19.

Reply 4: As noted above, there are several disanologies between COVID-19 and the flu (and a fortiori even more disanlogies with the common cold). The flu is not analogous to COVID-19 in terms of either virulence or severity, and the vaccines are not analogous in terms of effectiveness (even in the era of omicron). As such, the diseases/vaccines are different in kind and a reasonable individual within a community with high disease risk tolerance could more justly complain of their neighbors’ actions with regard to COVID-19 than the flu. We remain neutral on where the line is at which point an individual’s objectively defensible claim to a rights violation become decisive, but COVID-19 is clearly on one side of it. Note that this is particularly true where high risk tolerance of a particular disease is based on false empirical beliefs about its severity (e.g. that COVID-19 is no worse than the flu), as this undermines the validity of everyone’s consent to take the risk.

Objection 5: Once herd immunity nears or is reached, the risk of contracting COVID-19 in public spaces is reduced to the point that the conflict of rights ought to favor the unvaccinated, i.e. mandates are no longer permissible ( Giubilini, 2020 ; Williams, 2021 ).

Reply 5: This objection is interesting insofar as it may grant our argument up to a point. What our argument gets thus far is that when the risk of COVID-19 (or some other infectious disease) is sufficiently high, the consent rights of the bystander trump the consent rights of the would-be unvaccinated. One goal of vaccination is to achieve herd immunity, such that a disease is unable to find a host, and eventually the spread peters out. This is especially important in protecting those that cannot be vaccinated due to age or medical conditions. Our argument largely set herd immunity aside, insofar as we were not defending a mandate as a way to achieve herd immunity. Here, however, it is important to acknowledge that herd immunity is indeed a hopeful and likely result of a successful vaccine mandate. Yet, once herd immunity is reached, and the risk of COVID-19 exposure diminishes, it seems that the bystander’s right can no longer be said to trump the right of the would-be unvaccinated individual, such that the mandate is no longer ethical based on our argument.

A number of points are worth noting in response. First, as of this writing, herd immunity with respect to COVID-19 is far from becoming a reality. As new variants continue to emerge, the prospect of reaching herd immunity anytime soon continues to dwindle. As such, our argument stands strong for a COVID-19 vaccine mandate for the immediate and likely protracted future, even if not for all times. Second, removing a vaccine mandate once herd immunity has been reached invites new outbreaks and a general breakdown of the herd immunity. That is, it remains plausible that the risks of being unvaccinated, even once herd immunity is reached, continue to be high, insofar as herd immunity can easily be lost. We are seeing this occur presently with measles outbreaks and the prediction of many more to come in 2022 ( Center for Disease Control and Prevention, 2020 ; World Health Organization, 2021 ). Finally, if herd immunity is reached in such a way that a disease is eliminated entirely, with no clear risk of reemergence, then we concede our argument for a vaccine mandate has concluded—as we are making specific claims about the applied ethics of a particular policy in a particular context, the fact that it would no longer be applicable in a radically different context is no objection.

In summation, we think the case is extremely strong for requiring everyone who is able to receive a COVID-19 vaccine, ideally at the level of governmental mandate and also at the level of individual institutions. This case is strong even without looking at the utilitarian arguments that allowing the virus to spread and mutate can have catastrophic consequences, which arguments seem fairly impressive on their own. Rather, we argue that the same logic of a deontological right to consent or not to bodily infringements that speaks in favor of not requiring people to be injected with a vaccine also speaks in favor of not requiring people to be unnecessarily exposed to COVID-19, and so a full reckoning will involve a tradeoff of rights that will speak in favor of vaccine mandates.

Acknowledgments

We would like to thank Allison McCarthy for extensive comments and feedback and Dean Jacqueline Dunbar-Smith for the original impetus for this project.

This work was not supported by any particular funding mechanism.

Conflict of Interest

None declared.

Contributor Information

Daniel A Wilkenfeld, Department of Acute and Tertiary Care, University of Pittsburgh School of Nursing, USA.

Christa M Johnson, Department of Philosophy, University of Dayton, United States of America.

• American Academy of Pediatrics (2021). Children and Covid-19: State-Level Data Report , available from: https://www.aap.org/en/pages/2019-novel-coronavirus-covid-19-infections/children-and-covid-19-state-level-data-report [accessed 30 December 2021].
• Bester J. C. (2018). The Harm Principle Cannot Replace the Best Interest Standard: Problems with Using the Harm Principle for Medical Decision Making for Children . The American Journal of Bioethics , 18 , 9–19. [ PubMed ] [ Google Scholar ]
• Blumenthal K. G., Phadke N. A., Bates D. W. (2021). Safety Surveillance of COVID-19 mRNA Vaccines through the Vaccine Safety Datalink . JAMA , 326 , 1375. [ PubMed ] [ Google Scholar ]
• Brennan J. (2018). A Libertarian Case for Mandatory Vaccination . Journal of Medical Ethics , 44 , 37–43. [ PubMed ] [ Google Scholar ]
• Brummett A. L. (2021). Defending, Improving, Expanding, and Applying a Moral-Metaphysical Proceduralism for Secular Clinical Ethics . The American Journal of Bioethics , 21 , W6–W9. [ PubMed ] [ Google Scholar ]
• Buss, S., and Westlund, A. (2018). Personal Autonomy. In E. N. Zalta (ed.), Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University. https://plato.stanford.edu/archives/spr2018/entries/personal-autonomy/ [accessed 25 March 2022]. , [ Google Scholar ]
• Bustos M., Venkataramanan R., Caritis S. (2017). Nausea and Vomiting of Pregnancy—What’s New? Autonomic Neuroscience , 202 , 62–72. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Centers for Disease Control and Prevention (2020). Measles (Rubeola) , available from: https://www.cdc.gov/measles/index.html [accessed 30 December 2021].
• Centers for Disease Control and Prevention (2021a). Selected Adverse Events Reported after COVID-19 Vaccination , available from: https://www.cdc.gov/coronavirus/2019-ncov/vaccines/safety/adverse-events.html [accessed 30 December 2021].
• Centers for Disease Control and Prevention (2021b). Allergic Reactions Including Anaphylaxis After Receipt of the First Dose of Pfizer-Biontech COVID-19 Vaccine—United States, December 14–23, 2020 , available from: https://www.cdc.gov/mmwr/volumes/70/wr/mm7002e1.htm?s_cid=mm7002e1_w [accessed 30 December 2021].
• Centers for Disease Control and Prevention (2021c). CDC COVID-19 Study Shows MRNA Vaccines Reduce Risk of Infection by 91 Percent for Fully Vaccinated People , available from: https://www.cdc.gov/media/releases/2021/p0607-mrna-reduce-risks.html [accessed 30 December 2021].
• Centers for Disease Control and Prevention (2021d). Updates on Thrombosis with Thrombocytopenia Thrombocytopenia Syndrome (TTS) , available from: https://www.cdc.gov/vaccines/acip/meetings/downloads/slides-2021-12-16/02-COVID-See-508.pdf [accessed 30 December 2021].
• Centers for Disease Control and Prevention (2022). COVID Data Tracker Weekly Review, 25 February , available from: https://www.cdc.gov/coronavirus/2019-ncov/covid-data/covidview/index.html [accessed 1 March 2022].
• Conti A., Capasso E., Casella C., Fedeli P., Salzano F. A., Policino F., Terracciano L., Delbon P. (2018). Blood Transfusion in Children: The Refusal of Jehovah’s Witness Parents . Open Medicine , 13 , 101–104. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Crook H., Raza S., Nowell J., Young M., Edison P. (2021). Long Covid—Mechanisms, Risk Factors, and Management . BMJ 374:n1648 http://dx.doi.org/10.1136/bmj.n1648 [ PubMed ] [ Google Scholar ]
• Delahoy M. J., , UjamaaD., , WhitakerM., , O'HalloranA., , AnglinO., , BurnsE., , CummingsC., , HolsteinR., , KambhampatiA. K., , MiluckyJ., , PatelK., , PhamH., , TaylorC. A., , ChaiS. J., , ReingoldA., , AldenN. B., , KawasakiB., , MeekJ., , Yousey-HindesK., , AndersonE. J., , OpenoK. P., , TenoK., , WeigelA., , KimS., , LeegwaterL., , ByeE., , Como-SabettiK., , RoppS., , RudinD., , MuseA., , SpinaN., , BennettN. M., , PophamK., , BillingL. M., , ShiltzE., , SuttonM., , ThomasA., , SchaffnerW., , TalbotH. K., , CrosslandM. T., , McCaffreyK., , HallA. J., , FryA. M., , McMorrowM., , ReedC., , GargS., and , Havers F. P. (2021). Hospitalizations Associated with COVID-19 AMONG Children and Adolescents- COVID-NET, 14 States, March 1, 2020-August 14, 2021 . MMWR. Morbidity and Mortality Weekly Report , 70 , 1255–1260. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Devis P., Knuttinen M. G. (2017). Deep Venous Thrombosis in Pregnancy: Incidence, Pathogenesis and Endovascular Management . Cardiovascular Diagnosis and Therapy , 7 (Suppl 3 ), S309. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Flanigan J. (2014). A Defense of Compulsory Vaccination . In HEC Forum , Vol. 26 . Springer, pp. 5–25. [ PubMed ] [ Google Scholar ]
• Foot P. (1967). The Problem of Abortion and the Doctrine of the Double Effect . Oxford Review , 5–15. [ Google Scholar ]
• Giubilini A. (2020). An Argument for Compulsory Vaccination: The Taxation Analogy . Journal of Applied Philosophy , 37 , 446–466. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Giubilini A., Minerva F., Schuklenk U., Savulescu J. (2021). The ‘Ethical’ COVID-19 Vaccine Is the One That Preserves Lives: Religious and Moral Beliefs on the COVID-19 Vaccine . Public Health Ethics , 14 , 242–255. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Giubilini A., Savulescu J. (2019). Vaccination, Risks, and Freedom: The Seat Belt Analogy . Public Health Ethics , 12 , 237–249. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Gostin L. O. (2021). Vaccine Mandates Are Lawful, Effective and Based on Rock-Solid Science , Scientific American , available from: https://www.scientificamerican.com/article/vaccine-mandates-are-lawful-effective-and-based-on-rock-solid-science/ [accessed 13 September 2021].
• Greyson D., Vriesema-Magnuson C., Bettinger J. A. (2019). Impact of School Vaccination Mandates on Pediatric Vaccination Coverage: A Systematic Review . CMAJ Open , 7 , E524–E536. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Harsanyi J. C. (1975). Can the Maximin Principle Serve as a Basis for Morality? A Critique of John Rawls’s Theory . American Political Science Review , 69 , 594–606. [ Google Scholar ]
• Heuer U. (2011) The Paradox of Deontology, Revisited. In Oxford Studies of Normative Ethics . New York: Oxford University Press. [ Google Scholar ]
• Hobbes T. (2016). Thomas Hobbes: Leviathan (Longman Library of Primary Sources in Philosophy) . New York: Routledge. [ Google Scholar ]
• Iftikhar S. F., Biswas M. Cardiac Disease In Pregnancy. In: StatPearls. StatPearls Publishing, Treasure Island (FL) PMID: 30725946. [ PubMed ]
• Johnson C. M. (2019). The Intrapersonal Paradox of Deontology . Journal of Moral Philosophy , 16 , 279–301. [ Google Scholar ]
• Kahn J. H. (2020). Confidentiality and Capacity . Emergency Medicine Clinics of North America , 38 , 283–296. [ PubMed ] [ Google Scholar ]
• Kamm F. M. (1989). Harming Some to save Others . Philosophical Studies , 57 , 227–260. [ Google Scholar ]
• Kamm F. M. (1996). Morality Mortality . New York: Oxford University Press. [ PubMed ] [ Google Scholar ]
• Kazemi F., Nahidi F., Kariman N. (2017). Disorders Affecting Quality of Life during Pregnancy: A Qualitative Study . Journal of Clinical and Diagnostic Research , 11 , QC06–QC10. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Klein N. P., Lewis N., Goddard K., Fireman B., Zerbo O., Hanson K. E., Donahue J. G., Kharbanda E. O., Naleway A., Nelson J. C., Xu S., Yih W. K., Glanz J. M., Williams J. T. B., Hambidge S. J., Lewin B. J., Shimabukuro T. T., DeStefano F., Weintraub E. S. (2021). Surveillance for Adverse Events after COVID-19 mRNA Vaccination . JAMA , 326 , 1390. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Kowalik M. (2022). Ethics of vaccine refusal. Journal of Medical Ethics , 48 , 240. https://doi.org/10.1136/medethics-2020-107026. [ PubMed ] [ Google Scholar ]
• Lyngse, F. P., Kirkeby, C. T., Denwood, M., Christiansen, L. E., Mølbak, K., Møller, C. H., Skov, R. L., Krause, T. G., Rasmussen, M., Sieber, R. N., Johannesen, T. B., Lillebaek, T., Fonager, J., Fomsgaard, A., Møller, F. T., Stegger, M., Overvad, M., Spiess, K., and Mortensen, L. H. (2022). Transmission of SARS-CoV-2 Omicron VOC subvariants BA.1 and BA.2: Evidence from Danish Households. MedRxiv , 2022.01.28.22270044. https://doi.org/10.1101/2022.01.28.22270044 [ Google Scholar ]
• Ma Q., Liu J., Liu Q., Kang L., Liu R., Jing W., Wu Y., Liu M. (2021). Global Percentage of Asymptomatic SARS-CoV-2 Infections among the Tested Population and Individuals with Confirmed COVID-19 Diagnosis: A Systematic Review and Meta-Analysis . JAMA Network Open , 4 , e2137257. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• MacIntyre A. (1981). After Virtue. Notre Dame: Notre Dame.
• Manson N. C., O’Neill O. (2007). Rethinking Informed Consent in Bioethics . New York: Cambridge University Press. [ Google Scholar ]
• Mill J. S. (2011). On Liberty. Project Guttenberg. https://www.gutenberg.org/files/34901/34901-h/34901-h.htm . [accessed 25 March 2022]. [ Google Scholar ]
• Nozick R. (1974). Anarchy, State, and Utopia . New York: Basic Books. [ Google Scholar ]
• Petter E., Mor, O., Zuckerman, N., Oz-Levi, D., Younger, A., Aran, D., and Erlich, Y. (2021). Initial Real World Evidence for Lower Viral Load of Individuals Who Have Been Vaccinated by BNT162b2 . medRxiv . [ Google Scholar ]
• Rawls J. (1999). A Theory of Justice: Revised Edition . New York: Harvard University Press. [ Google Scholar ]
• Salter E. K. (2012). Deciding for a Child: A Comprehensive Analysis of the Best Interest Standard . Theoretical Medicine and Bioethics , 33 , 179–198. [ PubMed ] [ Google Scholar ]
• Sandel M. (1982). Liberalism and the Limits of Justice . Cambridge: Cambridge University Press. [ Google Scholar ]
• Savulescu J. (2021). Good Reasons to Vaccinate: Mandatory or Payment for Risk? Journal of Medical Ethics , 47 , 78–85. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Scheffler S. (1988). Agent-Centered Restrictions, Rationality, and the Virtues. In Scheffler S. (ed.) Consequentialism and Its Critics . New York: Oxford University Press, ; pp. 243–260. [ Google Scholar ]
• Scobie H. M., Johnson A. G., Suthar A. B., Severson R., Alden N. B., Balter S., Bertolino D., Blythe D., Brady S., Cadwell B., Cheng I., Davidson S., Delgadillo J., Devinney K., Duchin J., Duwell M., Fisher R., Fleischauer A., Grant A., Griffin J., Haddix M., Hand J., Hanson M., Hawkins E., Herlihy R. K., Hicks L., Holtzman C., Hoskins M., Hyun J., Kaur R., Kay M., Kidrowski H., Kim C., Komatsu K., Kugeler K., Lewis M., Lyons B. C., Lyons S., Lynfield R., McCaffrey K., McMullen C., Milroy L., Meyer S., Nolen L., Patel M. R., Pogosjans S., Reese H. E., Saupe A., Sell J., Sokol T., Sosin D., Stanislawski E., Stevens K., Vest H., White K., Wilson E., MacNeil A., Ritchey M. D., Silk B. J. (2021). Monitoring Incidence of COVID-19 Cases, Hospitalizations, and Deaths, by Vaccination Status—13 US Jurisdictions, April 4–July 17, 2021 . Morbidity and Mortality Weekly Report , 70 , 1284–1290. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Thompson M. G., Burgess J. L., Naleway A. L., Tyner H. L., Yoon S. K., Meece J., Olsho L. E. W., Caban-Martinez A. J., Fowlkes A., Lutrick K., Kuntz J. L., Dunnigan K., Odean M. J., Hegmann K. T., Stefanski E., Edwards L. J., Schaefer-Solle N., Grant L., Ellingson K., Groom H. C., Zunie T., Thiese M. S., Ivacic L., Wesley M. G., Lamberte J. M., Sun X., Smith M. E., Phillips A. L., Groover K. D., Yoo Y. M., Gerald J., Brown R. T., Herring M. K., Joseph G., Beitel S., Morrill T. C., Mak J., Rivers P., Harris K. M., Hunt D. R., Arvay M. L., Kutty P., Fry A. M., Gaglani M. (2021). Interim Estimates of Vaccine Effectiveness of BNT162b2 and mRNA-1273 COVID-19 Vaccines in Preventing SARS-CoV-2 Infection among Health Care Personnel, First Responders, and Other Essential and Frontline Workers—Eight US Locations, December 2020–March . Morbidity and Mortality Weekly Report , 70 , 495–500. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Thompson M. G., Burgess J. L., Naleway A., Tyner, H. L., Yoon, S. K., Meece, J., Olsho, L. E. W., Caban-Martinez, A. J., Fowlkes, A. L., Lutrick, K., Groom, H. C., Dunnigan, K., Odean, M. J., Hegmann, K. T., Stefanski, E., Edwards, L. J., Schaefer-Solle, N., Grant, L., Ellingson, K., Kuntz J. L., Zunie, T., Thiese, M. S., Ivacic, L., Wesley, M. G., Lamberte, J. M., Sun, X., Smith, M. E., Phillips, A. L., Groover, K. D., Yoo, Y. M., Gerald, J., Brown, R. T., Herring, M. K., Joseph, G., Beitel, S., Morrill, T. C., Mak, J., Rivers, P., Lynch, B., Zhou, Y., Zhang, J., Kelleher, A., Li, Y., Dickerson, M., Hanson, E., Guenther, K., Tong, S., Bateman, A., Reisdorf, E., Barnes, J., Azziz-Baumgartner, E., Hunt, D. R., Arvay, M. L., Kutty, P., Fry, A. M., and Gaglani, M. (2021). Prevention and Attenuation of COVID-19 by BNT162b2 and mRNA-1273 Vaccines . medRxiv . [ Google Scholar ]
• Thomson J. J. (1976). Killing, Letting Die, and the Trolley Problem . The Monist , 59 , 204–217. [ PubMed ] [ Google Scholar ]
• Thomson J. J. (1990). The Realm of Rights . Cambridge, MA: Harvard University Press. [ Google Scholar ]
• Vitiello A., Ferrara F., Troiano V., La Porta R. (2021). COVID-19 Vaccines and Decreased Transmission of SARS-CoV-2 . Inflammopharmacology , 29 , 1357–1354. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Wilkenfeld D. A., McCarthy A. (2020). Ethical Concerns with Applied Behavior Analysis for Autism Spectrum Disorder . Kennedy Institute of Ethics Journal , 30 , 31–69. [ PubMed ] [ Google Scholar ]
• Wilkinson D., Savulescu J. (2018). Ethics, Conflict and Medical Treatment for Children: From Disagreement to Dissensus . Poland: Elsevier. [ PubMed ] [ Google Scholar ]
• Williams B. M. (2021). The Ethics of Selective Mandatory Vaccination for COVID-19 . Public Health Ethics . [ PMC free article ] [ PubMed ] [ Google Scholar ]
• World Health Organization (2021). Global Progress against Measles Threatened Amidst COVID-19 Pandemic , available from: https://www.who.int/news/item/10-11-2021-global-progress-against-measles-threatened-amidst-covid-19-pandemic [accessed 30 December 2021].

ORIGINAL RESEARCH article

Immunogenicity of covid-19 booster vaccination in iei patients and their one year clinical follow-up after start of the covid-19 vaccination program.

• 1 Department of Viroscience, Erasmus MC University Medical Center Rotterdam, Rotterdam, Netherlands
• 2 Travel Clinic, Erasmus MC University Medical Center Rotterdam, Rotterdam, Netherlands
• 3 Department of Medical Microbiology and Infection Prevention, Amsterdam Institute for Infection and Immunity, Amsterdam UMC, University of Amsterdam, Amsterdam, Netherlands
• 4 Department of Internal Medicine, Infectious Diseases, University Medical Center Utrecht, Utrecht, Netherlands
• 5 Department of Infectious Diseases, Amsterdam UMC, Amsterdam, Netherlands
• 6 Department of Internal Medicine, Division Clinical Immunology, Maastricht UMC, Maastricht, Netherlands
• 7 Department of Rheumatology and Clinical Immunology, UMC Groningen, Groningen, Netherlands
• 8 Department of Infectious Diseases, Leiden University Medical Center, Leiden, Netherlands
• 9 Department of Internal Medicine, Radboud University Medical Center Nijmegen, Nijmegen, Netherlands
• 10 Department of Internal Medicine, Division of Allergy & Clinical Immunology, Erasmus MC University Medical Center Rotterdam, Rotterdam, Netherlands
• 11 Department of Immunology, Erasmus MC University Medical Center Rotterdam, Rotterdam, Netherlands

Purpose: Previous studies have demonstrated that the majority of patients with an inborn error of immunity (IEI) develop a spike (S)-specific IgG antibody and T-cell response after two doses of the mRNA-1273 COVID-19 vaccine, but little is known about the response to a booster vaccination. We studied the immune responses 8 weeks after booster vaccination with mRNA-based COVID-19 vaccines in 171 IEI patients. Moreover, we evaluated the clinical outcomes in these patients one year after the start of the Dutch COVID-19 vaccination campaign.

Methods: This study was embedded in a large prospective multicenter study investigating the immunogenicity of COVID-19 mRNA-based vaccines in IEI (VACOPID study). Blood samples were taken from 244 participants 8 weeks after booster vaccination. These participants included 171 IEI patients (X-linked agammaglobulinemia (XLA;N=11), combined immunodeficiency (CID;N=4), common variable immunodeficiency (CVID;N=45), isolated or undefined antibody deficiencies (N=108) and phagocyte defects (N=3)) and 73 controls. SARS-CoV-2-specific IgG titers, neutralizing antibodies, and T-cell responses were evaluated. One year after the start of the COVID-19 vaccination program, 334 study participants (239 IEI patients and 95 controls) completed a questionnaire to supplement their clinical data focusing on SARS-CoV-2 infections.

Results: After booster vaccination, S-specific IgG titers increased in all COVID-19 naive IEI cohorts and controls, when compared to titers at 6 months after the priming regimen. The fold-increases did not differ between controls and IEI cohorts. SARS-CoV-2-specific T-cell responses also increased equally in all cohorts after booster vaccination compared to 6 months after the priming regimen. Most SARS-CoV-2 infections during the study period occurred in the period when the Omicron variant had become dominant. The clinical course of these infections was mild, although IEI patients experienced more frequent fever and dyspnea compared to controls and their symptoms persisted longer.

Conclusion: Our study demonstrates that mRNA-based booster vaccination induces robust recall of memory B-cell and T-cell responses in most IEI patients. One-year clinical follow-up demonstrated that SARS-CoV-2 infections in IEI patients were mild. Given our results, we support booster campaigns with newer variant-specific COVID-19 booster vaccines to IEI patients with milder phenotypes.

Introduction

Inborn errors of immunity (IEI), commonly referred to as primary immunodeficiencies (PID), are a diverse group of congenital disorders affecting single or multiple components of the immune system. IEI result in increased susceptibility to infections, and sometimes autoimmune complications, autoinflammatory diseases, allergies and an increased risk for malignancies. In many IEI disturbed or absent responses to vaccination are found. During the COVID-19 pandemic, patients with IEI were prioritized in the Dutch COVID-19 vaccination program to receive 2 doses of an mRNA-based COVID-19 vaccine (mRNA-1273). Multiple studies have investigated the immunogenicity of COVID-19 vaccines in these patients. We and others found that in patients with primary antibody deficiencies an overall serologic response of 72% was observed, ranging from 0% in X-linked agammaglobulinemia (XLA) patients, 52-81% in common variable immunodeficiency (CVID) patients, to 100% in specific polysaccharide antibody deficiency (SPAD) patients ( 1 – 3 ). In patients with combined immunodeficiencies (CID), variable serological responses have been described, ranging from 0 to 100%, although the numbers of studied patients were low and clinical phenotypes heterogeneous ( 1 , 2 , 4 , 5 ). In addition, SARS-CoV-2 specific T-cell responses in IEI patients are reported to be robust and comparable to those in controls ( 1 , 6 ). Although response rates after vaccination were promising, lower levels of neutralizing antibodies were detected in IEI patients when compared to controls, which raised questions about the long-term protection and the need for booster vaccinations ( 1 – 3 ).

Recently, we reported the six-month immunogenicity of the mRNA-1273 COVID-19 vaccine in our cohort of Dutch IEI patients ( 7 ). Binding and functional antibody titers significantly declined at six months after the second vaccination in both IEI patients and controls, with no differences in decay rates. However, antibody titers at 28 days after vaccination in patients with CID and CVID were lower when compared to controls, and antibody titers dropped below the responder cut-off in these patients more frequently at six months after completion of the priming regimen. Moreover, most CVID patients that did not respond to the initial regimen of two mRNA-1273 COVID-19 vaccines, did not respond to a third vaccination either ( 7 , 8 ).

In addition to declining antibody titers after the priming regimen, the Omicron variant, which emerged in late 2021, showed a sharp reduction in sensitivity to neutralizing antibodies, leading to reduced or absent neutralization of this variant in healthy individuals ( 9 ). Booster vaccination partially restored this neutralizing capacity against Omicron ( 9 – 13 ). As a consequence, adults, including IEI patients, were advised to receive booster vaccinations. Although boosters enhance vaccine effectiveness, their effects wane over time, leading to more breakthrough infections ( 14 ). A Danish study found a correlation between higher Spike (S)-specific antibody titers and a reduced risk of breakthrough infections for the Delta variant, but this correlation was not demonstrated for the Omicron variant ( 15 ).

Various studies have described the effects of boosters in heterogeneous cohorts of IEI patients, showing an increase in antibody titers and/or neutralizing capacities ( 8 , 16 ). However, no study has specifically analyzed the effects of boosters in different subgroups of IEI patients, which is crucial to determine which specific patients would benefit most from additional boosters. This knowledge is essential for policy-making to allow for optimal (booster) vaccination strategies in these different cohorts. While clinical data on breakthrough infections in IEI patients after the initial vaccination regimen is available, limited data exists on breakthrough infections following booster vaccination ( 17 , 18 ). A large cohort study conducted in the United Kingdom demonstrated an increased risk of hospitalization and mortality after breakthrough infections following a mRNA booster dose in individuals with an immunodeficiency compared to healthy subjects. However, the study did not specifically examine this risk in IEI patients ( 19 ). Therefore, in the present study, we investigated SARS-CoV-2-specific antibody and T-cell responses after the administration of mRNA-based boosters in 171 IEI patients and 73 controls. Patients were categorized into cohorts based on different types of immunodeficiencies. These cohorts include XLA, CVID, CID, isolated antibody deficiencies (IgG subclass deficiency ± IgA deficiency (IgG), SPAD, phagocyte defects, and undefined antibody deficiencies. Additionally, we measured neutralizing antibodies 8 weeks after booster vaccination in participants who experienced breakthrough infections after booster vaccination and compared them to those who did not have breakthrough infections. Finally, we examined the clinical outcomes of our IEI cohort for up to one year after the start of the Dutch national vaccination campaign.

Ethical statement

The Vaccination Against COvid in Primary Immune Deficiencies (VACOPID) study is a prospective, controlled, multicenter research initiative carried out in seven academic hospitals in the Netherlands, involving patients with IEI. The study adheres to the principles of the Declaration of Helsinki and has received approval from the Dutch Central Committee on Research Involving Human Subjects (CCMO, NL7647.078.21, EudraCT number 2021-000515-24), the Medical Research Ethics Committee from Erasmus University Medical Center (MEC-2021-0050) and the local review boards of all other participating centers. Written informed consent was provided by all participants before inclusion.

Study participants and design

A total of 505 patients with IEI and 192 non-IEI controls were initially included and stratified into different cohorts ( Table 1 , Figure 1 ). The study design and criteria for inclusion/exclusion are described in detail in the Supplementary Material . The XLA, CID and CVID cohorts are characterized by more severe clinical phenotypes. Thereby, within the CVID cohort, patients may also experience auto-immune, granulomatous, lymphoproliferative and/or oncological complications. Isolated IgG subclass deficiency ± IgA deficiency and SPAD are clinically similar cohorts with a milder clinical phenotype, and were studied as one group. The undefined antibody deficiency cohort included patients with primary hypogammaglobulinaemia and preserved T-cellular immunity who do not meet the diagnostic criteria for any other primary antibody deficiencies. SARS-CoV-2 infections reported by the participants were also tracked throughout the study. At the beginning of the study, all study participants received two doses of the mRNA-1273 COVID-19 vaccine, administered with a 28-day interval, in accordance with the Dutch COVID-19 vaccination program. Results of the immune responses at 28 days- and six months following the second mRNA-1273 COVID-19 vaccine have been previously published ( 1 , 7 ).

Table 1 Baseline characteristics of VACOPID participants that donated a blood sample after booster vaccination.

Figure 1 Study design. Study design including study visits, number of participants per study visit and median intervals between study visits or between vaccination and study visits. Study visit 5 is divided into part A and B. Part A are participants receiving a third vaccination. In our study, these participants are CVID patients who use immunosuppressive drugs, or in specific individual cases when a medical specialist had reasonable arguments to make an exception to the aforementioned indications, based on proven or assumed non-response. Part B are participants who received a regular booster vaccination. Results of study visits 1, 2 and 3, 4 and 5A were published previously (blue and green panels). Results of study visit 5B and 6 (pink panels) are included in this study.*The mRNA-1273 (Moderna)- and the BNT162b2 (Pfizer) vaccines were used as third vaccination or booster vaccination based on local availability and personal preferences and were administered at public vaccination sites.

A third mRNA-based COVID-19 vaccine was recommended by the Dutch healthcare authorities to patients with CID or CVID who use immunosuppressive drugs, or in specific individual cases, based on proven or assumed non-response ( 20 ). These so-called third vaccinations were mainly administered in October 2021 ( Figure 1 , visit 5A). Alongside this third dose, a booster vaccination campaign was launched, recommending mRNA-based boosters based on the ancestral SARS-Cov-2 strain for adults who received their last COVID-19 vaccination and/or had a SARS-CoV2 infection more than three months ago. Both controls and IEI patients who were not eligible for a third vaccination, and IEI patients who received a third vaccination more than three months ago were eligible for this booster vaccination. Most participants received their booster vaccination in December 2021 and January 2022. The mRNA-1273 (Moderna) and the BNT162b2 (Pfizer) vaccines were used as boosters based on local availability and personal preferences and were administered at public vaccination sites ( Figure 1 , visit 5B).

An amendment to the original study protocol was approved by the Medical Research Ethics Committee from Erasmus University Medical Center. With this amendment the immune responses of mRNA-based COVID-19 booster vaccination in our patient cohort could be studied. Participants of six academic centers were invited to donate additional blood samples four to ten weeks after booster vaccination. 171 IEI patients and 73 controls responded to this invitation. Three months after the third vaccination or booster, IEI patients could receive another booster vaccination according to the Dutch vaccination program. Blood was collected from 36 IEI patients four to ten weeks after receiving this 4th dose ( Figure 1 , visit 6). The study period ran until April 30 th 2022. At the end of the study, all participants that were initially included were asked to complete a questionnaire to supplement their clinical data, regardless of whether they had completed all study visits. This request was fulfilled by 338 participants (241 IEI patients and 97 controls).

Measurement of humoral and cellular immune responses

The assays to evaluate humoral and cellular immune responses are described more extensively in the Supplementary Material . Briefly, the quantitative Luminex assay was used to measure S-specific IgG and nucleocapsid (N) specific IgG, with results expressed as international Binding Antibody Units per mL (BAU/ml) ( 21 , 22 ). Participants with S-specific IgG above 44.8 BAU/ml were considered seropositive ( 23 ). Nucleocapsid (N)-specific IgG antibodies were analyzed to identify participants who contracted COVID-19 before or during the study. We selected a cutoff value with 100% specificity, measured in pre-pandemic sera from healthy donors, to minimize the possibility of false-positive results due to administration of (low levels of) N-specific antibodies in IGRT ( 23 , 24 ). Therefore, an infection with SARS-CoV-2 was defined as a history of a positive PCR test, positive antigen test, and/or an N-specific IgG titer above 42.2 BAU/ml. We tested samples for the presence of neutralizing antibodies from 104 participants (32 controls, 19 CVID, 53 IgG/SPAD) whose clinical information was completed and who had not experienced COVID-19 until blood collection after booster vaccination. We compared serum neutralization capacity between participants who contracted COVID-19 after blood sample collection and those who remained uninfected, using a pseudovirus system targeting ancestral SARS-CoV-2 and Omicron subvariants BA.1 and BA.2. Neutralization titers were expressed as International Units per mL (IU/ml) ( 25 , 26 ). The WHO International Standard for anti-SARS-CoV-2 immunoglobulin (NIBSC 20/136) was used to normalize the IgG and neutralizing antibody titers. T-cell responses were assessed in samples obtained from the university hospitals of Leiden and Rotterdam using an Interferon-gamma (IFN-ɣ) release assay (IGRA, QIAGEN) including a peptide pool that covers the S protein (Ag2). The results of the IGRA were quantified in IU IFN-ɣ/ml after subtraction of the negative control value.

Statistical analysis

Categorical variables are displayed as numbers and percentages and analyzed with Fisher’s exact test. Continuous variables are presented as median ± interquartile range (IQR). The time interval between booster administration and blood collection was corrected for outliers. Outliers were defined as below the first quartile minus 1.5 times the IQR or above the third quartile plus 1.5 times the IQR. Fold changes are displayed as the median of individual fold changes. Results of the immunological assays were displayed in figures and text with geometric means, medians, and (interquartile) ranges. The Wilcoxon rank-sum test was used to analyze continuous variables and the Wilcoxon signed rank test was used to analyze paired data. Log-rank tests were used to compare breakthrough infections after booster vaccination. P-values below 0.05 were considered statistically significant.

Study data were collected in an online electronic data capture system (Castor © , Amsterdam, the Netherlands), compliant to the General Data Protection Regulation (GDPR). R studio was used for statistical analyses. Graphs were prepared with GraphPad PRISM, version 9.1.2 (San Diego, CA, USA).

Blood samples were collected from a total of 244 participants (171 IEI and 73 controls) who received a booster vaccination ( Figure 1 ). The median time between the second vaccination of the priming regimen and the booster vaccination was 258 days (IQR 244-267), the median time between the blood sample taken 6 months after second vaccination and booster vaccination was 74 days (IQR 61-85) and the median time between booster administration and blood sample collection was 59 days (IQR 49-67). None of these participants had previously received a third vaccine. Eight participants (2 controls, 6 IEI) were excluded from analysis due to outlying intervals between booster administration and blood sample collection (>94 days). Forty IEI patients and 17 controls who had contracted COVID-19 before- or during the study were analyzed separately. COVID-19 was diagnosed by PRC or antigen test in 50 participants (13 controls, 37 IEI), while 7 (4 controls, 3 IEI) were diagnosed by anti-N antibodies above 42.2 BAU/ml. After excluding the participants who had contracted COVID-19, 125 IEI patients and 54 controls remained eligible for the main analysis of the booster vaccination study. Their baseline characteristics are displayed in Table 1 . As the numbers of patients with phagocyte defects and undefined antibody deficiencies were low, findings regarding these cohorts are reported in detail in the Supplementary Material ( Supplementary Figure 1 ).

Booster vaccination increases SARS-CoV-2 S-specific IgG titers in IEI patients

To determine to what extent SARS-CoV-2 S-specific antibody titers increase after booster vaccination, S-specific IgG titers at 6 months after the second vaccination were compared with S-specific IgG titers 6-10 weeks after booster vaccination. No differences were observed in responses following either a mRNA-1273 (Moderna) or BNT162b2 (Pfizer) booster ( Supplementary Figure 2 ); the results hereafter represent the pooled outcomes. Booster vaccination increased SARS-CoV-2 S-specific IgG titers in all cohorts when compared to the titers six months after the second vaccination. The geometric mean titer (GMT) of S-specific IgG in the control cohort increased 9.7-fold, from 533 to 5050 BAU/ml (P<.0001) ( Figure 2 ). The GMT’s of the CVID and IgG/SPAD cohorts (in total) also increased significantly (CVID: 7.8 fold from 322 BAU/ml to 2670 BAU/ml (P<.0001), IgG/SPAD: 13.4-fold from 367 to 4950 BAU/ml (P<.0001)). When stratifying these cohorts based on the use of immunoglobulin replacement therapy (IGRT), S-specific IgG still increased significantly in CVID patients on IGRT (8.7 fold from 296 BAU/ml to 2545 BAU.ml (P<.0001)) and IgG/SPAD patients with and without IGRT use (with IGRT: 17.9 fold from 249 BAU/ml to 4985 BAU/ml (P<.0001), without IGRT: 9.9 fold from 563 BAU/ml to 4919 BAU/ml (P<.0001)). No significant increase was observed in the smaller cohorts, consisting of the CVID cohort without use of IGRT (N=3) and the CVID cohort (N=4)(7.3 fold from 647 BAU/ml to 4070 BAU/ml (P=.50), and 14.8-fold from 181 to 2350 BAU/ml (P=.13), respectively) ( Figure 2 ). Although the geometric mean titer (GMT) of S-specific IgG in the XLA cohort increased significantly, this titer is still very low (6.3-fold from 24 to 94 BAU/ml (P=.03)) and this effect should be contributed to IGRT. The GMT’s of S-specific IgG after booster vaccination did not differ between patients in the CVID and IgG/SPAD cohort who received IGRT and those who did not (P=.65 and P=.77, respectively) ( Figure 2A ). However, fold changes of the IgG/SPAD patients who use IGRT were significantly higher compared to the fold changes of IgG/SPAD patients who did not use IGRT (P=.03) ( Figure 2B ). Fold changes did not differ between (total) IEI cohorts and controls, except for the XLA cohort (P=.02). Trajectories of the GMTs of the different cohorts are shown in Figure 2C . This figure also includes previously reported results of CVID participants that received a third vaccination (instead of a booster vaccination) ( 7 ).

Figure 2 S-specific IgG pre- (28 days and six months after second vaccination) and post-boost (+/- 8 weeks after booster vaccination). (A) S-specific IgG measured by Luminex for controls and different cohorts of inborn errors of immunity (IEI) patients 28 days after second vaccination, six months after the second vaccination, and eight weeks after booster vaccination. The number of participants per cohort correspond to Table 1 . The CVID and IgG cohort were stratified based on IGRT use. Results are expressed in binding antibody units per milliliter (BAU/mL). The dotted line is the pre-defined responder cut-off (resp). Data in panel A are presented in box-and-whisker plots. The horizontal lines of the box-and-whisker plots indicate the median, the bounds of the boxes indicate the interquartile range, and the whiskers indicate the range. All datapoints are shown. The numbers below the box-and-whisker plots indicate the geometric mean titers (GMT) per time point. Participants not using immunoglobulin replacement therapy (IGRT) are shown as circles, participants using IGRT are shown as squares. IgG titers were compared per cohort using the Wilcoxon paired signed rank test. The SPAD cohort is indicated with white symbols (with orange borders) while the IgG cohort is indicated with orange symbols. (B) Fold changes of IgG antibodies post-boost (+/- 8 weeks after booster vaccination) compared to 6 months after primary regimen of each cohort in total plus the fold changes the CVID and IgG/SPAD cohort stratified by IGRT use. Data in panel B are presented in scatter dot plots. The horizontal lines indicate the median, the whiskers indicate the interquartile range. All data points are shown. The dashed line represents a fold change of 1, where the titer at 6 months after primary regimen is equal to the titer after booster. All data points above the dashed line represents a fold increase, all data points below the dashed line a fold decrease. The numbers below indicate the median fold change. Fold changes were compared per cohort using the Wilcoxon rank-sum test. (C) Trajectories of the GMTs after third vaccination (CVID) and booster vaccination (all cohorts). The results of CVID participants receiving a third vaccination have also been published previously. A subset of these CVID participants received a booster after their third vaccination (fourth dose in total, named post-boost after 3 rd vaccination on the x-axis). The results after booster vaccination of the CVID cohort and the other cohorts were computed only on those who donated blood after booster vaccination. Color coding is the same in all figures. S, Spike; XLA, X-linked agammaglobulinemia; CID, Combined Immunodeficiency; CVID, Common Variable Immunodeficiency; IgG, Isolated IgG subclass deficiency ± IgA deficiency; SPAD, Specific polysaccharide antibody deficiency; * = P<.05, ** = P<.01, *** = P<.001, **** = P<.0001.

COVID-19 does not result in higher SARS-CoV-2 S-specific IgG titers after booster vaccination

In total, 15 controls and 40 IEI patients who donated a blood sample after receiving a booster vaccination contracted COVID-19 before this blood donation. The majority of infections occurred before the first vaccination (N=15, 27.3%) or between the booster vaccination and the collection of the blood sample (N=30, 55%) ( Table 1 ). GMT of participants who contracted COVID-19 before or during the study were similar to those who did not contract COVID-19 ( Supplementary Figure 3 ).

Neutralizing antibodies in the IgG/SPAD cohort as correlate of protection against breakthrough infections

Neutralizing antibodies were compared between participants who contracted COVID-19 after blood sample collection and those who remained uninfected, focusing on our largest cohorts (controls (N=32), CVID (N=19) and IgG/SPAD (N=53)). The follow-up duration after blood sample collection was comparable between participants who did and who did not contract COVID-19 (median 67 days (IQR 62-72) and 66 days (IQR 58-72) respectively, P=.18). Significantly lower neutralizing antibody titers were found in IgG/SPAD patients who developed COVID-19 shortly after blood sample collection (ancestral: P=.0021, BA.1: P=.0012, BA.2: P=.0041) ( Supplementary Figure 4 ). Similarly, IgG/SPAD participants with relatively low titers of neutralizing antibodies against the ancestral virus, BA.1, and BA.2 (below the median of 283 IU/ml, 82.5 IU/ml, and 136 IU/ml respectively) had a significantly higher risk of contracting a breakthrough infection during the initial months after blood sample collection (Ancestral: hazard ratio 3.5, 95% confidence interval [1.02-12.17], P=.047. BA.1 and BA.2: hazard ratio 5.2, 95% confidence interval [1.50-17.92], P=.0093). The same trend was seen when comparing binding antibodies, with anti-S antibodies being significantly lower in IgG/SPAD participants who developed COVID-19 shortly after blood sample collection (IgG/SPAD: P=0.006, controls: P=.30, CVID: P=.18).

Booster vaccination increases SARS-CoV-2 specific T-cell responses

SARS-CoV-2-specific T-cell responses after booster vaccination were assessed in samples collected from two study sites using the QIAGEN assay ( Figure 3A ). In this assay, IFN-γ levels after booster vaccination increased significantly from 0.45 IU/ml to 0.61 IU/ml in the control cohort (P<.001) and from 0.30 IU/ml to 0.71 IU/ml in the IgG/SPAD cohort (P<.0001) compared to the IFN-γ levels 6 months after the primary regimen. No significant increase was observed in the other (smaller) cohorts (XLA: 0.34 to 0.81 IU/ml (P=.88), CID: 0.09 to 0.20 IU/ml (P=.50), CVID: 0.36 to 0.51 IU/ml (P=.08). IFN-γ levels did not increase after booster vaccination compared to the initial response, and IFN-γ levels after booster were not statistically different between controls and IEI cohorts. Of the 101 participants analyzed, 15 (9%) did not have IFN-γ levels above the responder cut-off of 0.15 IU/ml (4 controls, 1 CID, 4 CVID and 6 IgG/SPAD), but all these T-cell non-responders had SARS-CoV-2 S-specific IgG titers above the cut-off level (44.8 IU/ml) ( Figure 3B ).

Figure 3 SARS-CoV-2-specific T-cell responses pre-boost (28 days and six months after second vaccination) and post-boost (+/- 8 weeks after booster vaccination). (A) SARS-CoV-2-specific T-cell responses measured by an interferon γ (IFN-γ) release assay (QIAGEN) after stimulation of whole blood 28 days and six months after second vaccination and eight weeks after booster vaccination. Lower limit of detection (LLOD) is.01 IU/ml and responder cut off (resp) is.15 IU/ml. Results are expressed as international units/milliliter (IU/mL). The dotted line is the pre-defined responder cut-off (resp). Data is presented in box-and-whisker plots. The horizontal lines of the box-and-whisker plots indicate the median, the bounds of the boxes indicate the interquartile range, and the whiskers indicate the range. All datapoints are shown. The numbers above the box-and-whisker plots indicate the geometric mean titer (GMT). Within each cohort, IFN-γ levels at 28 days and six months were compared using Wilcoxon paired signed rank test. The SPAD cohort is indicated with white symbols while the IgG cohort is indicated with orange symbols. (B) Correlation between IgG titers and T-cell responses 8 weeks after booster vaccination. The dotted horizontal line is the responder cut-off of the QIAGEN interferon-gamma release assay (0.15 IU/mL). The dotted vertical line is the responder cut-off of the Luminex assay (44.8 BAU/ml). Color coding is the same in all figures. XLA, X-linked agammaglobulinemia; CID, Combined Immunodeficiency; CVID, Common Variable Immunodeficiency; IgG, Isolated IgG subclass deficiency ± IgA deficiency; SPAD, Specific polysaccharide antibody deficiency; *** = P<.001, **** = P<.0001.

CVID cohort

Clinically, CVID patients are known to experience recurrent (bacterial) infections with varying severity. A number of CVID patients show additional autoimmune, granulomatous, lymphoproliferative and/or oncological complications ( 27 ). As reported previously, fifty CVID patients of our cohort received a third vaccination ( 7 ). Fifteen of them also donated blood samples after receiving a fourth dose (administered at least three months after the third vaccination). After this fourth dose, the GMT of anti-S IgG increased from 37 BAU/ml to 100 BAU/ml (P=.0006) ( Figures 2C , 4A ). Although this was a significant increase, the GMT is still considered low compared to the trajectories of the other cohorts. Fourteen of these participants received IGRT, which may also explain the increase due to increasing levels of anti-S antibodies in those preparations at the time ( 24 ). Previously, we found that the presence of a noninfectious complication with or without the use of immunosuppressive medication was negatively associated with response after two mRNA-1273 vaccinations. This association persisted after administration of a third dose of a mRNA COVID-19 vaccine (third vaccination or booster) ( Figure 4B ) ( Supplementary Table 1 ).

Figure 4 SARS-CoV-2-specific IgG in CVID participants. (A) S-specific IgG measured by Luminex for CVID patients that received a third dose and a (fourth) booster dose after the initial regimen of 2 mRNA-1273 COVID-19 vaccines. S-specific IgG was measured 28 days after second vaccination, six months after the second vaccination, five weeks after third vaccination and eight weeks after booster vaccination. (B) S-specific IgG measured by Luminex for CVID patients after a third dose of a mRNA COVID-19 vaccine (either a third vaccination of a booster dose) stratified by the presence of non-infectious complications. The following non-infectious complications were defined: Autoimmune cytopenia, organ specific autoimmunity, systemic autoimmunity, enteropathy, malignancy, lymphoproliferative diseases, granulomatous lymphocytic interstitial lung disease (GLILD), and other granulomatous diseases. (A, B) Results are expressed in binding antibody units per milliliter (BAU/mL). The dotted line is the pre-defined responder cut-off. Data in panels A and B are presented in box-and-whisker plots. The horizontal lines of the box-and-whisker plots indicate the median, the bounds of the boxes indicate the interquartile range, and the whiskers indicate the range. All datapoints are shown. The numbers below the box-and-whisker plots indicate the geometric mean titer (GMT) per timepoint. IgG titers were compared per cohort using the Wilcoxon paired signed rank test. CVID, Common Variable Immunodeficiency; *** = P<.001, **** = P<.0001; ns = not significant.

Clinical outcomes one year after primary vaccination regimen

After one year of follow-up, 243 study participants had contracted at least one SARS-CoV-2 infection (66 controls, 178 IEI patients) ( Table 2 ). Reinfections with SARS-CoV-2 were reported in 5 controls (7%) and 20 IEI patients (10%). Eleven participants were admitted to a hospital because of COVID-19. Main reasons for admission included symptomatic illness by COVID-19 and/or admission to day clinics for (experimental) monoclonal antibody treatment ( Table 3 ). The binding antibody titers, measured 28 days after second vaccination, were not different between CVID and IgG/SPAD patients who were hospitalized due to COVID-19 related illness and patients from the same cohorts who were not hospitalized (P=.48 and P=.45 respectively). The number of hospitalized patients in which T-cell response was evaluated was too low to correlate T-cell responses with risk of hospitalization due to COVID-19 (data not shown). Clinical data on the occurrence of COVID-19 has been completed up to the study endpoint in 338 participants (241 patients with IEI and 97 controls) ( Table 2 ). Of these 338 participants, two CVID patients and one CID patient died during the study. Their causes of death were unrelated to COVID-19. Of the remaining 335 participants, 130 infections occurred in the IEI patients and 50 in the control cohort, resulting in similar infection rates (54% and 52% respectively). Most participants were infected after the Omicron lineages became dominant. COVID-19 related symptoms including cough, dyspnea, fever, nasal congestion, throat pain, and the duration of symptoms were evaluated by a COVID-19 questionnaire. IEI patients experienced significantly more frequent fever and dyspnea compared to controls (40% and 37% in IEI patients, 16% and 8% in controls respectively, P=.024 and P=.002). Additionally, symptom duration was significantly longer (IEI median 8 days [IQR 5-14], controls median 5 days [IQR 3-7], P<.0001) ( Table 2 ). Subsequent regression analyses revealed that variables including gender, age, underlying disease, use of IGRT and/or immunosuppressive medication were not able to explain the observed difference in symptom duration (data not shown).

Table 2 Total SARS-CoV-2 infections and detailed information on SARS-CoV-2 infections.

Table 3 Hospital admissions of IEI patients related to COVID-19.

This study investigated the effect of mRNA-based COVID-19 booster vaccination in a group of 171 IEI patients. Booster vaccination increased S-specific IgG titers in all cohorts when compared to the titers six months after the priming vaccination regimen. In addition, we found that lower neutralizing antibody titers after booster vaccination in the IgG/SPAD cohort were predictive of developing COVID-19 shortly thereafter. Clinical evaluation of the IEI cohort one year after primary vaccination showed that incidence of COVID-19 was comparable between IEI patients (54%) and controls (52%) and only 11 out of 177 IEI patients that contracted COVID-19 were admitted to hospital, none of whom required admission to an intensive care unit.

The increase in antibody titers after booster vaccination in patients with IEI suggests that the first two mRNA-1273 vaccinations induced immunological memory, which is also known from previous studies in non-IEI cohorts ( 28 ). Specifically, the IEI cohorts with milder clinical phenotypes, such as the IgG subclass deficiency ± IgA deficiency (IgG) and specific polysaccharide antibody deficiency (SPAD), show similar trajectories of antibody titers compared to the control cohort. The fold increase of S-specific IgG titers following booster vaccination, in comparison to titers 6 months after the primary vaccination regimen, were similar between control- and IEI cohorts (except XLA). However, fold changes were significantly higher in IgG/SPAD participants that received IGRT compared to IgG/SPAD participants who did not. IgG/SPAD patients on IGRT generally have a more profound immunodeficiency and, as a result, exhibited significantly lower antibody titers 28 days and 6 months after the primary regimen compared to IgG/SPAD participant who did not receive IGRT. The findings of the higher fold changes after booster vaccination in these participants remain to be explained. Outliers in our dataset could contribute to this result, and the use of IGRT may explain some of the additional increase. However, we believe that the higher fold change cannot be fully attributed to the effect of IGRT, given the modest absolute increase in the XLA patients and the considerably lower fold change in the CVID patients receiving IGRT.

We found that IgG/SPAD patients with lower neutralizing antibody titers after booster vaccination had a higher risk of developing breakthrough infection during the following months. This is consistent with previous research investigating the importance of neutralizing antibodies as a correlate of protection against symptomatic SARS-CoV-2 infection ( 29 , 30 ). In both the CVID and the control cohort, we did not find differences in titers between participants who did or did not experience breakthrough infections. This was possibly due to the lower sample sizes of these cohorts, but possible confounders such as differences in living situations or (protective) behavior could not be excluded.

Results from the CVID cohort, which constitutes the largest cohort in this study, are more difficult to interpret. This cohort encompasses a heterogeneous group of patients, and the vaccination regimens within this group varied. Some individuals received a third vaccination in addition to the priming vaccination regimen. These were primarily CVID patients who were using immunosuppressive drugs and/or exhibited proven or assumed non-response to the priming COVID-19 vaccination regimen ( 7 ). In addition to the limited impact of this third vaccination on S-specific IgG titers ( 7 ), our study revealed that an extra booster vaccination in this group (resulting in a total of four doses) did not result in a substantial increase of S-specific IgG antibodies either. The observed small increase in S-specific IgG antibodies likely reflected the general rise of these antibodies in IGRT preparations, as evidenced by the parallel trajectory of the increase of antibodies in the XLA cohort ( Figure 2B ). This finding is consistent with other studies that showed increasing titers of SARS-CoV-2-specific antibodies in IGRT preparations during our study period ( 16 , 31 – 33 ). CVID patients who received a booster vaccination instead of a third dose generally exhibited a milder clinical disease and demonstrated an antibody trajectory comparable to that of controls.

The GMTs of the S-specific T-cell responses, as determined by measuring IFN-γ levels after stimulation of whole blood with peptides covering the S protein, increased after the booster vaccination compared with the levels observed at six months after the second vaccination. This confirms the presence of immunological memory, although the increase in T-cell responses was not statistically significant in all cohorts. This could be explained by the timing of blood collection. Previous studies demonstrated that T-cell responses after booster vaccination demonstrate the greatest increase within the first week post-vaccination, followed by a slight decline in the subsequent month ( 28 , 34 ). Given that our assessment of T-cell responses was performed on average eight weeks after booster vaccination, it is plausible that our measured GMTs already started decreasing. In previous studies, cross-reactive CD4+ and CD8+ T-cells were detected after booster vaccination with monovalent mRNA-based vaccines, and these T-cells cross-reacted with the Omicron BA.1 variant ( 9 , 35 , 36 ). Notably, IFN-γ levels did not show significant differences between the control and IEI cohorts in our study. This suggests that IEI cohorts may also have developed cross-reactive T-cells following vaccination, which potentially protect them against severe disease after infection with an emerging SARS-CoV-2 variant. The advantageous role of COVID-19 mRNA vaccination in patients with IEIs was also recently demonstrated in another study that examined the breadth and epitope specificity of SARS-CoV-2-specific T cell receptor clonotypes ( 37 ).

In our study, IEI patients and controls had similar incidence rates of COVID-19. However, a Danish study with 313 CVID patients and 2504 controls found a higher incidence of SARS-CoV-2 among CVID patients. Despite this disparity, several similarities were observed between their study and ours. The majority of infections occurred after the emergence of the Omicron variant, and while the CVID group faced an increased risk of hospitalization due to COVID-19, their risk for mechanical ventilation and mortality did not differ significantly from the general population ( 38 ). On the other hand, other studies examining SARS-CoV-2 infections in IEI patients did identify an elevated case fatality rate, particularly in younger age groups, as well as an increased risk of intensive care admission ( 39 ). The prolonged duration of symptoms observed in IEI patients compared to controls in our study aligns with several other studies ( 39 , 40 ).

This study has several limitations. Despite having 505 IEI patients and 192 controls at the beginning of this prospective cohort study, only 171 IEI and 73 controls donated blood following booster vaccination. Several reasons can be identified. One center did not complete the study. Moreover, despite being approached, many participants remained unresponsive. This could be due to lack of urgency because of the milder disease caused by the Omicron variant, which emerged around the time of the booster immunizations. However, it is impossible to rule out response bias, and the limited participants per cohort reduced the power of this study. Nevertheless, we were able to examine the immunogenicity of the booster vaccination in a relatively large number of IEI patients, in order to get a robust understanding of the “boostability” of IEI patients compared to controls. Furthermore, the assays we used in this study, for example the Luminex assay and the IGRA assay, are not able to identify whether the respective S-specific IgG antibodies or IFN-γ levels were derived from a memory B- and T-cells (recall response) or a de novo immune response. We assume that booster vaccination induces a recall response based on previous studies which have shown that S-specific IgG rapidly increases within 7 days after booster vaccination ( 28 , 34 ). In addition, other studies demonstrated B- and T-cell memory by using assays that can specifically measure memory markers ( 9 , 35 , 41 ). In addition, the IGRA test used in this study focuses on the (S-specific) production of INF-γ and does not provide data on the quantity of SARS-CoV-2 specific T-cells. Finally, increasing concentrations of anti-S antibodies in IGRT could potentially have an effect on our results ( 32 , 33 ). However, in our studies we found very low concentrations of antibodies in patients with XLA and we therefore consider the effects of IGRT to be limited at the time of our studies.

Our study implies that mRNA-based booster vaccination induces robust recall of memory B-cell and T-cell responses in IEI patients with a milder phenotype (CVID without non-infectious complications, SPAD, isolated antibody deficiencies, phagocyte defects, undefined antibody deficiencies). Although many participants were infected with the Omicron variant during the final months of the study period, the clinical outcomes remained favorable. The mild clinical presentation may potentially be explained by the broad cross-reactivity of T-cells, considering the limited neutralizing antibody response against Omicron following booster vaccination. Recently, a monovalent mRNA vaccine against the omicron XBB variant resulted in higher antibody titers compared to a bivalent variant (against the omicron XBB and omicron BA.4/BA.5 variants) in a non-IEI cohort and was recommended by the U.S. FDA for use in 2023-2024 immunizations ( 42 ). To our knowledge, there is currently no data available on variant-specific booster vaccination in IEI patients. However, we recommend these monovalent omicron XBB boosters to IEI patients with milder clinical phenotypes, given the favorable results observed thus far in our cohort ( 1 , 7 ). It is more difficult to make a recommendation for severe IEI patient such as XLA and CVID with complicated disease. These patients are likely to have a persistently poor antibody after (variant-specific) booster vaccinations. Besides, the usefulness of giving virus-specific boosters for the aim of boosting (cross-reactive) T-cell responses is not clear yet. Rising levels of (neutralizing) antibodies in IGRT preparations offer perspective in these patients, although these preparations may have limited effect against newer variants because of the delay between plasma donations and their clinical use in IGRT preparations ( 24 ). Lastly, we argue that continued research into other therapies, such as antivirals, is needed to protect IEI patients from severe illness after contracting COVID-19 as well as for other endemic viruses and future virus outbreaks.

Data availability statement

The original contributions presented in the study are included in the article/ Supplementary Material , further inquiries can be directed to the corresponding author/s.

Ethics statement

The studies involving humans were approved by the Dutch Central Committee on Research Involving Human Subjects (CCMO, NL7647.078.21), the Medical Research Ethics Committee from Erasmus University Medical Center (MEC-2021-0050) and the local review boards of all other participating centers. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

LvL: Conceptualization, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing. MG: Conceptualization, Investigation, Methodology, Validation, Writing – original draft, Writing – review & editing. CG: Conceptualization, Investigation, Methodology, Writing – review & editing. PE: Conceptualization, Investigation, Methodology, Writing – review & editing. GdB: Conceptualization, Investigation, Methodology, Writing – review & editing. JP: Conceptualization, Investigation, Methodology, Writing – review & editing. AR: Conceptualization, Investigation, Methodology, Writing – review & editing. HJ: Conceptualization, Investigation, Methodology, Writing – review & editing. FV: Conceptualization, Investigation, Methodology, Writing – review & editing. MvG: Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Writing – original draft, Writing – review & editing. RdV: Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Writing – original draft, Writing – review & editing. VD: Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Visualization, Writing – original draft, Writing – review & editing.

Group members of VACOPID research group

Eric C.M. van Gorp 1,2 MD PhD, Faye de Wilt 1 BSc, Susanne Bogers 1 MSc, Lennert Gommers 1 BSc, Daryl Geers 1 Marianne W. van der Ent 10 MSc, P. Martin van Hagen 10,11 MD PhD, Jelle W. van Haga 5 BSc, Bregtje A. Lemkes 5 MD PhD, Annelou van der Veen 5 BSc, Rogier W. Sanders 3 PhD, Karlijn van der Straten 3 MD, Judith A. Burger 3 , BSc, Jacqueline van Rijswijk 3 BSc, Khadija Tejjani 3 BSc, Joey H. Bouhuijs 3 BSc, Karina de Leeuw 7 MD PhD, Annick A.J.M. van de Ven 7 MD PhD, S.F.J. de Kruijf-Bazen 6 MSc, Pieter van Paassen 6 MD PhD, Lotte Wieten 6 PhD, Petra H. Verbeek-Menken 8 BSc, Annelies van Wengen 8 MSc, Anke H.W. Bruns 4 MD PhD, Helen L. Leavis 4 MD PhD, Stefan Nierkens 12,13 PhD.

12 Center for Translational Immunology, UMC Utrecht, The Netherlands

13 Princess Máxima Center for Pediatric Oncology, Utrecht, The Netherlands

The author(s) declare financial support was received for the research, authorship, and/or publication of this article. Funded by ZonMw (10430072010006).

Conflict of interest

JP received a grant from GlaxoSmithKline for an improvement of clinical care project and received support from Prothva Biosolutions for attending meetings and cover of travel expenses. JP participates in an Advisory Board for Janssen. FV received a grant from ZonMW for a study on lanadelumab in COVID-19, and consulting fees from GSK made to his department. VD received consulting fees from GlaxoSmithKline, Pharming NV for Advisory board meetings and honoraria for lectures from Takeda Pharmaceutical Company, Kedrion, AstraZeneca.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author(s) declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2024.1390022/full#supplementary-material

Abbreviations

BAU/mL, Binding antibody units per milliliter; CID, combined B- and T-cell immunodeficiency; COVID-19, coronavirus disease-2019; CVID, common variable immunodeficiency; DMARD, disease modifying anti-rheumatic drugs; GMT, geometric mean titer; IEI, inborn errors of immunity; IGRA, interferon gamma release assay; IGRT, immunoglobulin replacement therapy; IFN-ɣ, interferon gamma; IU/mL, International Units per milliliter; LLoD, lower limit of detection; N-(protein), nucleocapsid-protein; PID, primary immunodeficiencies; RBD, receptor binding domain; SARS-CoV-2, severe acute respiratory syndrome coronavirus-2; SPAD, specific polysaccharide antibody deficiency; S(-protein), spike-protein; TNF, tumor necrosis factor; VNT, virus neutralization test; XLA, X-linked agammaglobulinemia.

1. van Leeuwen LPM, GeurtsvanKessel CH, Ellerbroek PM, de Bree GJ, Potjewijd J, Rutgers A, et al. Immunogenicity of the mRNA-1273 COVID-19 vaccine in adult patients with inborn errors of immunity. J Allergy Clin Immunol . (2022) 149(6):1949–57. doi: 10.2139/ssrn.4001461

PubMed Abstract | CrossRef Full Text | Google Scholar

2. Shields AM, Faustini SE, Hill HJ, Al-Taei S, Tanner C, Ashford F, et al. SARS-CoV-2 vaccine responses in individuals with antibody deficiency: findings from the COV-AD study. J Clin Immunol . (2022) 42:923–34. doi: 10.1007/s10875-022-01231-7

3. Durkee-Shock JR, Keller MD. Immunizing the imperfect immune system: Coronavirus disease 2019 vaccination in patients with inborn errors of immunity. Ann Allergy Asthma Immunol . (2022) 129(5):562–71. doi: 10.1016/j.anai.2022.06.009

4. Ponsford MJ, Evans K, Carne EM, Jolles S, Immunodeficiency Centre for W, Division of Population M. COVID-19 vaccine uptake and efficacy in a national immunodeficiency cohort. J Clin Immunol . (2022) 42:728–31. doi: 10.1007/s10875-022-01223-7

5. Delmonte OM, Bergerson JRE, Burbelo PD, Durkee-Shock JR, Dobbs K, Bosticardo M, et al. Antibody responses to the SARS-CoV-2 vaccine in individuals with various inborn errors of immunity. J Allergy Clin Immunol . (2021) 148(5):1192–7. doi: 10.1016/j.jaci.2021.08.016

6. Steiner S, Schwarz T, Corman VM, Jeworowski LM, Bauer S, Drosten C, et al. Impaired B cell recall memory and reduced antibody avidity but robust T cell response in CVID patients after COVID-19 vaccination. J Clin Immunol . (2023) 43(5):869–81. doi: 10.21203/rs.3.rs-2409532/v1

7. van Leeuwen LPM, Grobben M, GeurtsvanKessel CH, Ellerbroek PM, de Bree GJ, Potjewijd J, et al. Immune responses 6 months after mRNA-1273 COVID-19 vaccination and the effect of a third vaccination in patients with inborn errors of immunity. J Clin Immunol . (2023) 43(6):1104–17. doi: 10.1007/s10875-023-01514-7

8. Shields AM, Faustini SE, Hill HJ, Al-Taei S, Tanner C, Ashford F, et al. Increased seroprevalence and improved antibody responses following third primary SARS-CoV-2 immunisation: an update from the COV-AD study. Front Immunol . (2022) 13:912571. doi: 10.3389/fimmu.2022.912571

9. GeurtsvanKessel CH, Geers D, Schmitz KS, Mykytyn AZ, Lamers MM, Bogers S, et al. Divergent SARS-CoV-2 Omicron-reactive T and B cell responses in COVID-19 vaccine recipients. Sci Immunol . (2022) 7:eabo2202. doi: 10.1126/sciimmunol.abo2202

10. Nemet I, Kliker L, Lustig Y, Zuckerman N, Erster O, Cohen C, et al. Third BNT162b2 vaccination neutralization of SARS-CoV-2 omicron infection. New Engl J Med . (2021) 386:492–4. doi: 10.1056/NEJMc2119358

11. Sablerolles RSG, Rietdijk WJR, Goorhuis A, Postma DF, Visser LG, Geers D, et al. Immunogenicity and reactogenicity of vaccine boosters after Ad26.COV2.S priming. N Engl J Med . (2022) 386:951–63. doi: 10.1056/NEJMoa2116747

12. Vietri MT, D’Elia G, Caliendo G, Passariello L, Albanese L, Molinari AM, et al. Antibody levels after BNT162b2 vaccine booster and SARS-CoV-2 Omicron infection. Vaccine . (2022) 40:5726–31. doi: 10.1016/j.vaccine.2022.08.045

13. Thompson MG, Natarajan K, Irving SA, Rowley EA, Griggs EP, Gaglani M, et al. Effectiveness of a third dose of mRNA vaccines against COVID-19-associated emergency department and urgent care encounters and hospitalizations among adults during periods of delta and omicron variant predominance - VISION network, 10 states, August 2021-January 2022. MMWR Morb Mortal Wkly Rep . (2022) 71:139–45. doi: 10.15585/mmwr.mm7104e3

14. Andrews N, Stowe J, Kirsebom F, Toffa S, Rickeard T, Gallagher E, et al. Covid-19 vaccine effectiveness against the omicron (B.1.1.529) variant. New Engl J Med . (2022) 386:1532–46. doi: 10.1056/NEJMoa2119451

15. Stærke NB, Reekie J, Nielsen H, Benfield T, Wiese L, Knudsen LS, et al. Levels of SARS-CoV-2 antibodies among fully vaccinated individuals with Delta or Omicron variant breakthrough infections. Nat Commun . (2022) 13:4466. doi: 10.1038/s41467-022-32254-8

16. Zimmerman O, Altman Doss AM, Kaplonek P, Liang CY, VanBlargan LA, Chen RE, et al. mRNA vaccine boosting enhances antibody responses against SARS-CoV-2 Omicron variant in individuals with antibody deficiency syndromes. Cell Rep Med . (2022) 3:100653. doi: 10.1016/j.xcrm.2022.100653

17. Shields AM, Tadros S, Al-Hakim A, Nell JM, Lin MMN, Chan M, et al. Impact of vaccination on hospitalization and mortality from COVID-19 in patients with primary and secondary immunodeficiency: The United Kingdom experience. Front Immunol . (2022) 13:984376. doi: 10.3389/fimmu.2022.984376

18. Di Fusco M, Moran MM, Cane A, Curcio D, Khan F, Malhotra D, et al. Evaluation of COVID-19 vaccine breakthrough infections among immunocompromised patients fully vaccinated with BNT162b2. J Med Econ . (2021) 24:1248–60. doi: 10.1080/13696998.2021.2002063

19. Bedston S, Almaghrabi F, Patterson L, Agrawal U, Woolford L, Anand SN, et al. Risk of severe COVID-19 outcomes after autumn 2022 COVID-19 booster vaccinations: a pooled analysis of national prospective cohort studies involving 7.4 million adults in England, Northern Ireland, Scotland and Wales. Lancet Reg Health Eur . (2024) 37:100816. doi: 10.1016/j.lanepe.2023.100816

20. RIVM. Vaccinatie bij afweerstoornis: Rijksinstituut voor Volksgezondheid en Milieu (2021). Available online at: https://www.rivm.nl/covid-19-vaccinatie/vragen-achtergronden/afweerstoornis .

Google Scholar

21. Grobben M, van der Straten K, Brouwer PJ, Brinkkemper M, Maisonnasse P, Dereuddre-Bosquet N, et al. Cross-reactive antibodies after SARS-CoV-2 infection and vaccination. Elife . (2021) 10. doi: 10.7554/eLife.70330

22. Keuning MW, Grobben M, de Groen AC, Berman-de Jong EP, Bijlsma MW, Cohen S, et al. Saliva SARS-CoV-2 antibody prevalence in children. Microbiol Spectr . (2021) 9:e0073121. doi: 10.1128/Spectrum.00731-21

23. Keuning MW, Grobben M, Bijlsma MW, Anker B, Berman-de Jong EP, Cohen S, et al. Differences in systemic and mucosal SARS-CoV-2 antibody prevalence in a prospective cohort of Dutch children. Front Immunol . (2022) 13. doi: 10.3389/fimmu.2022.976382

24. Lindahl H, Chen P, Åberg M, Ljunggren HG, Buggert M, Aleman S, et al. SARS-CoV-2 antibodies in commercial immunoglobulin products show markedly reduced cross-reactivities against omicron variants. J Clin Immunol . (2023) 43:1075–82. doi: 10.1007/s10875-023-01486-8

25. Schmidt F, Weisblum Y, Muecksch F, Hoffmann HH, Michailidis E, Lorenzi JCC, et al. Measuring SARS-CoV-2 neutralizing antibody activity using pseudotyped and chimeric viruses. J Exp Med . (2020) 217. doi: 10.1084/jem.20201181

26. Brouwer PJM, Brinkkemper M, Maisonnasse P, Dereuddre-Bosquet N, Grobben M, Claireaux M, et al. Two-component spike nanoparticle vaccine protects macaques from SARS-CoV-2 infection. Cell . (2021) 184:1188–200 e19. doi: 10.1016/j.cell.2021.01.035

27. Picard C, Bobby Gaspar H, Al-Herz W, Bousfiha A, Casanova JL, Chatila T, et al. International union of immunological societies: 2017 primary immunodeficiency diseases committee report on inborn errors of immunity. J Clin Immunol . (2018) 38:96–128. doi: 10.1007/s10875-017-0464-9

28. Tan NH, Geers D, Sablerolles RSG, Rietdijk WJR, Goorhuis A, Postma DF, et al. Immunogenicity of bivalent omicron (BA.1) booster vaccination after different priming regimens in health-care workers in the Netherlands (SWITCH ON): results from the direct boost group of an open-label, multicentre, randomised controlled trial. Lancet Infect Dis . (2023) 23(8):901–13. doi: 10.1016/S1473-3099(23)00140-8

29. Khoury DS, Cromer D, Reynaldi A, Schlub TE, Wheatley AK, Juno JA, et al. Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection. Nat Med . (2021) 27:1205–11. doi: 10.1038/s41591-021-01377-8

30. Cromer D, Steain M, Reynaldi A, Schlub TE, Khan SR, Sasson SC, et al. Predicting vaccine effectiveness against severe COVID-19 over time and against variants: a meta-analysis. Nat Commun . (2023) 14:1633. doi: 10.1038/s41467-023-37176-7

31. Raphael A, Shamriz O, Tvito A, Magen S, Goldberg S, Megged O, et al. SARS-CoV-2 spike antibody concentration in gamma globulin products from high-prevalence COVID-19 countries are transmitted to X-linked agammaglobulinemia patients. Front Immunol . (2023) 14:1156823. doi: 10.3389/fimmu.2023.1156823

32. Lindahl H, Klingström J, Da Silva Rodrigues R, Christ W, Chen P, Ljunggren HG, et al. Neutralizing SARS-CoV-2 antibodies in commercial immunoglobulin products give patients with X-linked agammaglobulinemia limited passive immunity to the omicron variant. J Clin Immunol . (2022) 42(6):1–7. doi: 10.1007/s10875-022-01283-9

CrossRef Full Text | Google Scholar

33. Nielsen BU, Drabe CH, Barnkob MB, Johansen IS, Hansen AKK, Nilsson AC, et al. Antibody response following the third and fourth SARS-CoV-2 vaccine dose in individuals with common variable immunodeficiency. Front Immunol . (2022) 13:934476. doi: 10.3389/fimmu.2022.934476

34. Luca MZ, Ngoc HT, Wim JRR, Daryl G, Roos SGS, Susanne B, et al. Distinct COVID-19 vaccine combinations result in divergent immune responses. medRxiv . (2023). 2023.08.25.23294606. doi: 10.1101/2023.08.25.23294606

35. Geers D, Shamier MC, Bogers S, den Hartog G, Gommers L, Nieuwkoop NN, et al. SARS-CoV-2 variants of concern partially escape humoral but not T-cell responses in COVID-19 convalescent donors and vaccinees. Sci Immunol . (2021) 6. doi: 10.1126/sciimmunol.abj1750

36. Maringer Y, Nelde A, Schroeder SM, Schuhmacher J, Hörber S, Peter A, et al. Durable spike-specific T cell responses after different COVID-19 vaccination regimens are not further enhanced by booster vaccination. Sci Immunol . (2022) 7:eadd3899. doi: 10.1126/sciimmunol.add3899

37. Delmonte OM, Oguz C, Dobbs K, Myint-Hpu K, Palterer B, Abers MS, et al. Perturbations of the T-cell receptor repertoire in response to SARS-CoV-2 in immunocompetent and immunocompromised individuals. J Allergy Clin Immunol . (2023). doi: 10.1016/j.jaci.2023.12.011

38. Katzenstein TL, Rasmussen LD, Drabe CH, Larsen CS, Hansen AE, Stærkind M, et al. Outcome of SARS-CoV-2 infection among patients with common variable immunodeficiency and a matched control group: A Danish nationwide cohort study. Front Immunol . (2022) 13:994253. doi: 10.3389/fimmu.2022.994253

39. Tangye SG, consortium CHGE. Impact of SARS-CoV-2 infection and COVID-19 on patients with inborn errors of immunity. J Allergy Clin Immunol . (2023) 151:818–31. doi: 10.1016/j.jaci.2022.11.010

40. Giardino G, Milito C, Lougaris V, Punziano A, Carrabba M, Cinetto F, et al. The impact of SARS-CoV-2 infection in patients with inborn errors of immunity: the experience of the Italian primary immunodeficiencies network (IPINet). J Clin Immunol . (2022) 42:935–46. doi: 10.1007/s10875-022-01264-y

41. Goel RR, Painter MM, Lundgreen KA, Apostolidis SA, Baxter AE, Giles JR, et al. Efficient recall of Omicron-reactive B cell memory after a third dose of SARS-CoV-2 mRNA vaccine. Cell . (2022) 185:1875–87 e8. doi: 10.1016/j.cell.2022.04.009

42. Spyros C, Nichole M, Jordan LW, Brandon E, Adam B, Joanne ET, et al. Safety and immunogenicity of XBB.1.5-containing mRNA vaccines. medRxiv . (2023). 2023.08.22.23293434. doi: 10.1101/2023.08.22.23293434

Keywords: inborn errors of immunity, primary immunodeficiency disorders, SARS-CoV-2, mRNA-1273 COVID-19 vaccine, booster vaccination, immunogenicity, antibody response, T-cell response

Citation: van Leeuwen LPM, Grobben M, GeurtsvanKessel CH, Ellerbroek PM, de Bree GJ, Potjewijd J, Rutgers A, Jolink H, van de Veerdonk FL, van Gils MJ, de Vries RD, Dalm VASH and VACOPID Research Group (2024) Immunogenicity of COVID-19 booster vaccination in IEI patients and their one year clinical follow-up after start of the COVID-19 vaccination program. Front. Immunol. 15:1390022. doi: 10.3389/fimmu.2024.1390022

Received: 22 February 2024; Accepted: 05 April 2024; Published: 18 April 2024.

Reviewed by:

Copyright © 2024 van Leeuwen, Grobben, GeurtsvanKessel, Ellerbroek, de Bree, Potjewijd, Rutgers, Jolink, van de Veerdonk, van Gils, de Vries, Dalm and VACOPID Research Group. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Virgil A. S. H. Dalm, [email protected]

† These authors have contributed equally to this work and share last authorship

‡ These authors equally contributed as second author

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

How to get vaccines to remote areas? In Sierra Leone they’re delivered by foot, boat or motorbike

Postdoctoral Research Fellow, University of Oxford

Disclosure statement

Niccolò Francesco Meriggi receives funding from Weiss Asset Management, UKRI and the International Growth Centre.

University of Oxford provides funding as a member of The Conversation UK.

View all partners

In Sierra Leone almost 59% of the population live in remote, rural areas. Roads may be non-existent or in bad condition, making it very difficult for rural dwellers to access healthcare. This is one of the key reasons why COVID-19 vaccination rates in the country are low.

During an innovative vaccine programme mobile vaccine clinics were dispatched to the furthest parts of the country, sometimes on motorcycles and boats. The results showed COVID-19 vaccination rates tripled in three days.

Development Economist Niccolò F. Meriggi tells Nadine Dreyer about the programme’s potential to become a blueprint for future healthcare delivery in the country and other remote regions in Africa.

Why are vaccination rates low in Sierra Leone?

By 10 March 2022, more than a year after COVID-19 vaccines arrived on the market, 80% of people living in high-income countries had received at least one dose. In stark contrast, only 15% of people had been vaccinated in low-income countries.

Fast-forward to November 2023 and still only 33% of the population in Africa had received at least their first dose of a COVID-19 vaccine.

The hardships Sierra Leoneans face are typical of the obstacles people in low-income countries have to overcome to access healthcare.

In the early days of the COVID-19 vaccination campaign in Sierra Leone, it took the average Sierra Leonean living in a rural community three-and-a-half hours each way to the nearest vaccination centre.

Things improved as more clinics started offering the vaccine, but the cost of reaching clinics remained high and, in many cases, prohibitive. In Sierra Leone 60% of the rural population live on less than US$1.25 a day. Getting to a clinic would cost more than one week’s wages.

How did this vaccine drive tackle the problem?

A team of researchers designed a COVID-19 vaccination drive that was implemented in March and April  2022 by the Ministry of Health and Sanitation and their technical partner Concern World Wide, an international humanitarian agency.

The primary aim of this intervention was to take vaccine doses and nurses to administer vaccines to remote, rural communities, preceded by seeking permission and community mobilisation.

At the time, only 6% to 9% of the adults who took part in the programme were already immunised.

Just over 20,000 Sierra Leoneans, living in 150 rural towns outside the country’s national clinic network, took part in the vaccination campaign.

The first step was to approach village leaders including the chief and the mammy queen, the most important woman in the village. Youth and religious leaders were also consulted. They were briefed about the purpose of the visit and the vaccination team answered questions about the available vaccines.

The leaders were asked for their cooperation in encouraging eligible community members to take the COVID-19 vaccine.

That evening, when labourers returned home from farms, the health team talked directly to all villagers about vaccine efficacy and safety and the importance of getting vaccinated. They also addressed villagers’ questions and concerns.

Finally, vaccine doses and healthcare workers arrived at the villages to administer the doses. Some travelled on motorbikes or on boats because of the lack of any road access.

This last-mile vaccine intervention tripled vaccination rates within three days in treated communities.

Large numbers of people from neighbouring communities also showed up to receive vaccines at the temporary vaccination sites.

Looking forward

These results suggest that people who live far from clinics are less likely to seek healthcare and that last mile delivery is a cost-effective intervention capable of overcoming that problem.

The intervention cost in this campaign was US$33 per person vaccinated. This approach proved 76% more cost-effective than other vaccination campaigns.

Transport accounted for a large share of the costs, so the cost-effectiveness of last mile delivery can be increased by offering a “bundle” of health products. The bundle could include routine child immunisation, as well as human papillomavirus and malaria vaccines, combined with other important health supplies such as deworming tablets, vitamin A supplements, oral rehydration solutions and chlorine for drinking water.

The World Health Organization reported that between 2020 and 2021, 5.42 million people died of COVID-19.

Other estimates put the death toll for the same period at 14.83 million , which is 2.74 times higher.

Developing cost-effective strategies to make vaccines easily accessible to everyone, everywhere, is the most promising solution to prevent future pandemics. This blueprint could also be used to address obstacles to other life-saving medical care.

The research team has since been awarded funds from the International Growth Centre and the Social Science Research Council through its Mercury Project. These grants will be used to expand the model in this paper to a bundle of health products and services, including additional vaccines (HPV and Malaria) and maternal and child health interventions, and further explore its feasibility and cost-effectiveness

• Sierra Leone
• COVID-19 vaccination
• Health challenges Africa

Sydney Horizon Educators (Identified)

Senior Disability Services Advisor

Deputy Social Media Producer

Associate Professor, Occupational Therapy

GRAINS RESEARCH AND DEVELOPMENT CORPORATION CHAIRPERSON