How well do different COVID-19 vaccines protect against different viral variants? A systematic review and meta-analysis

Abstract

While the efficacy of coronavirus disease 2019 (COVID-19) vaccines has been evaluated in numerous trials, comprehensive evidence on how protection by different vaccines has varied over time remains limited. We aimed to compare protective effects of different vaccines against different viral variants. To achieve this, we searched Medline, Cochrane Library and Embase for randomized controlled trials assessing the efficacy of COVID-19 vaccines. Forest plots using Mantel–Haenszel and random-effects models were generated showing risk ratios (RRs) and 95% CIs by vaccines and variants. We included 36 studies with 90 variant-specific primary outcomes. We found a RR of 0.26 (95% CI 0.21 to 0.31) against all variants overall, with the highest protective effects against the wild-type (RR 0.13; 95% CI 0.10 to 0.18), followed by Alpha (RR 0.26; 95% CI 0.18 to 0.36), Gamma (RR 0.34; 95% CI 0.21 to 0.55), Delta (RR 0.39; 95% CI 0.28 to 0.56) and Beta (RR 0.49; 95% CI 0.40 to 0.62) variants. Nucleic acid vaccines showed the highest protection levels against all variants (RR 0.11; 95% CI 0.08 to 0.15), followed by protein subunit, inactivated virus and viral vector. In conclusion, we found high but heterogenous levels of protection for most COVID-19 vaccines, with decreasing protective effects for vaccines based on traditional technologies as SARS-CoV-2 variants emerged over time. Novel nucleic acid-based vaccines offered substantially higher and more consistent protection.

COVID-19 vaccine efficacy, meta-analysis, systematic review, viral variants

Introduction

The development of vaccines against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been crucial in controlling the spread of the pandemic and reducing its morbidity and mortality.^1–3 Several vaccines were approved early on for large-scale rollout alongside the continued evolution of viral variants with different clinical, epidemiological and immunological properties.⁴ The most commonly used coronavirus disease 2019 (COVID-19) vaccines can be classified into four types based on their technology: nucleic acid vaccines, adenovirus-based viral vector vaccines, protein subunit vaccines and inactivated virus vaccines.⁵ Nucleic acid vaccines, represented by mRNA-1273 (Moderna)⁶ and BNT162b2 (BioNTech),⁷ harness the potential of genetic engineering by encoding viral proteins through injected mRNA fragments, which allows for rapid vaccine development but requires stringent storage conditions.⁸ Adenoviral-based viral vector vaccines, such as Gam-COVID-Vac⁹ and AZD1222,¹⁰ use vectors derived from adenoviruses and the recombined spike gene of SARS-CoV-2 to stimulate an immune response.¹¹ Protein subunit vaccines, including NVX-CoV2373 (Novavax)¹² and ZF2001 (Longcom),¹³ are developed by expressing SARS-CoV-2 spike proteins in vitro, and elicit direct and rapid antibody production upon injection.¹⁴ Inactivated virus vaccines, exemplified by CoronaVac¹⁵ and BBIBP-CorV,¹⁶ employ traditional vaccine-making methods to eliminate the virus's pathogenicity while retaining immunogenicity, making them safe and suitable for large-scale vaccination programs.¹⁷

SARS-CoV-2 variants with mutations suspected to impact viral virulence, transmission and the efficacy of diagnostics, vaccines and antivirals have been labeled variants of concern (VOCs).⁴ By August 2023, the European Centre for Disease Prevention and Control and the WHO had identified six such VOCs^4,18: the Alpha variant (B.1.1.7 and Q lineage), initially detected in the UK in December 2020,¹⁸ was renowned for its heightened transmissibility and global spread.¹⁹ The Beta variant (B.1.351 and its descendent lineages) was first identified in South Africa in May 2020¹⁸ and raised concerns due to its E484K mutation, potentially impacting vaccine efficacy.²⁰ The Gamma variant (P.1 and its descendent lineages), first detected in Brazil in November 2020,¹⁸ exhibited resemblances to the Beta variant¹⁹ and led to concerns about reinfections and vaccine efficacy.²¹ The Delta variant (B.1.617.2 and its descendent lineages), as first identified in India in October 2020,¹⁸ demonstrated increased transmissibility and potentially higher clinical severity and also raised concerns about vaccine protection.¹⁹ The Epsilon variant (B.1.427/B.1.429), first identified in California in July 2020,¹⁸ carried the L452R mutation,²² which raised concerns about triggering the emergence of various further variants. The Omicron variant (B.1.1.529 and its descendant lineages), first identified in November 2021 simultaneously in multiple countries⁴ and characterized by higher infectiousness and lower clinical severity, has branched into many sub-lineages, while continuing to (re-)emerge and spread globally.²³

The safety, immunogenicity and efficacy of different COVID-19 vaccines have been evaluated in numerous trials. However, many of the existing summaries of the evidence regarding their efficacy are: limited to certain types of vaccine²⁴ or specific vaccines²⁵; focused on certain at-risk populations,²⁶ regions²⁷ or transmission settings²⁸; only assessing selected clinical outcomes²⁹ or trial phases³⁰; restricted to³¹ or not disaggregating results by³² different viral variants; not presenting quantitative summary estimates from meta-analysis; or a combination of these limitations. In particular, evidence regarding the efficacy of different vaccines by viral variants remains scarce. The very limited evidence on this specific matter either offer only qualitative summaries³³ or no direct comparison of different vaccines against each other without meta-analysis.³⁴ In summary, no in-depth, comprehensive analysis exists to date that directly compares the protective effects of different vaccine brands for different viral variants. This systematic review and meta-analysis aims to close this gap.

As the development of COVID-19 vaccines with diverse modes of action and immunogenic properties continues alongside the ongoing evolution of viral variants with varying immune-evasive capacities, such detailed understanding of vaccine performance is crucial to inform future vaccine development and deployment.

Methods

This systematic review and meta-analysis were conducted following the guidelines outlined in the Cochrane Handbook for Systematic Reviews of Interventions³⁵ and adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations.³⁶ The protocol was registered in the International Prospective Register of Systematic Reviews prior to initiating data collection (protocol number CRD42021258455).

Inclusion and exclusion criteria

We focused on randomized controlled trials (RCTs) in humans evaluating the efficacy of SARS-CoV-2 vaccines against symptomatic or asymptomatic infection, or death. We did not impose any restrictions based on country, race, gender or age of participants. We excluded trials where vaccine efficacy (VE) was not a primary outcome and studies were published in languages other than English.

Search methods

A systematic search was conducted in Medline (via PubMed), Cochrane Library and EMBASE to identify relevant articles published prior to 22 January 2023. Detailed search statements for each database can be found in Supplement File 1.

Data collection and extraction

Extracted citations were imported into the Covidence software (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia),³⁷ where duplicate entries were identified and removed. Two researchers (TNAH and HLQ) independently screened the titles and abstracts of the remaining articles. The same researchers further examined the full texts of retained articles using a pre-established eligibility criteria checklist (see Supplement File 2). Discrepancies or disagreements between the two researchers during the screening process were discussed and, if necessary, resolved with the assistance of a third researcher (FV). Data from included studies were extracted by TNAH and AB using a predefined data extraction form (see Supplement File 3). To ensure data accuracy, a random sample of the extracted data was checked and verified by HLQ. Additionally, references from key systematic reviews and from all included articles were reviewed to identify any additional relevant studies.

Extracted data included: authors; methods (study design, blinding type, trial phase, allocation unit, randomization ratio); participants (countries of enrollment, number of participants analyzed); intervention and control (vaccine type and brand, control type); outcome type (symptomatic infection, asymptomatic infection, death); type of analysis (per protocol, intention to treat, modified intention to treat); follow-up duration (time between vaccine administration and confirmation of infection); and dominant viral variant. If the viral variant was not reported, we determined the most likely dominant variant based on enrollment dates and study locations. Separate records were created for each variant in the same trial.

Data analysis

We calculated the pooled risk ratios (RRs) and 95% CIs for the risk of infection in vaccinated vs unvaccinated groups, based on the raw data as reported in included studies, stratified by vaccine type (nucleic acid vaccines, viral vector vaccines, protein subunit vaccines and inactivated virus vaccines) and viral variant (wild-type, Alpha, Beta, Gamma, Delta and others). Forest plots using Mantel–Haenszel models within the fixed-effect model framework and random-effects models were generated showing pooled RRs with 95% CIs by vaccine type, vaccine brand and viral variant, and were visually inspected to evaluate the consistency of intervention effects across the included studies. Overlapping CIs suggested similar intervention effects, while weak overlap or the presence of outliers indicated statistical heterogeneity.³⁸ Furthermore, we calculated the I² and Cochrane's X² test statistics to quantify the degree of heterogeneity, with I² values >50% and p-values from Cochrane's X² tests of <0.10 indicating significant heterogeneity across studies.³⁸ In cases of heterogeneity, we used random-effects models instead of Mantel–Haenszel models within the fixed-effect model framework to generate RRs and 95% CIs.

Because the Mantel–Haenszel model and the random-effects model would exclude studies with zero events in both groups, we adjusted the analysis to ensure the inclusion of eight studies with zero events in the intervention group. We did so by changing one observation in the intervention group from ‘no event’ to ‘event’ as recommended.³⁹ For example, in a study with 1005 vaccinated participants where none were infected with SARS-CoV-2, we changed the outcome status of one participant in the intervention group from ‘not infected’ to ‘infected’ to ensure the study was included in the pooled analysis.

Furthermore, percentages and median RRs including IQRs were used to present outcomes by trial characteristics, and p-values from Pearson's χ² test were used to test for statistically significant differences with 5% alpha error levels.

All statistical analyses were performed using STATA version 17.0.

Risk of bias assessment

TNAH and HLQ conducted risk of bias assessments independently for the included studies using the Cochrane Risk of Bias for Randomized Controlled Trials Tool version 2,³⁵ generating a summary risk score as low risk (low risk of bias in all domains), high risk (high risk of bias in at least one domain or some concerns in multiple domains in a way that substantially lowers confidence in the result) or some concerns (some concerns in at least one domain, but not sufficient to be classified as high risk of bias in any domain).

Results

Study selection

The initial literature search yielded a total of 6032 results. After removing 922 duplicates, the titles and abstracts of the remaining 5110 entries were screened for relevance, which resulted in the exclusion of 4826 articles. The full text assessment of the remaining 284 articles led to the inclusion of 36 studies. Nineteen out of the 36 included studies reported VE on more than one viral variant, resulting in a total of 90 variant-specific primary VE outcomes (Figure 1).

Figure 1.

Selection of included studies and outcomes from search results. RCT, randomized controlled trial; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

Descriptive results

The majority of outcomes was obtained from studies using viral vector vaccine technology (46%), followed by protein subunit (24%), nucleic acid (21%) and inactivated virus vaccines (9%). The Johnson & Johnson Viral Vector vaccine (Ad26.COV2.S vaccine) accounted for 26 (29%) outcomes, followed by the AstraZeneca Viral Vector Vaccine (ChAdOx1 nCoV-19) with 13 (14%) outcomes and the Pfizer-BioNTech Nucleic Acid Vaccine (BNT162b2) with 11 (12%) outcomes. The Clover Biopharmaceuticals Protein Subunit Vaccine (SCB-2019) and the Covifenz Protein Subunit Vaccine (CoVLP+AS03) contributed 10 (11%) and five (6%) outcomes, respectively, while the other vaccines accounted for 25 (28%) outcomes.

More than one-quarter of the outcomes assessed VE against wild types (26%), with the shares for Alpha, Beta, Delta and Gamma variants being 12%, 8%, 12% and 10%, respectively.

Double-blinded studies and trials using per protocol analysis made up more than one-half of the outcomes (57% and 61%, respectively). Twenty-three (26%) outcomes were based on sample sizes of <10 000 participants, while 32 (36%) outcomes had sample sizes of >25 000 participants. Outcomes from the majority of studies used a parallel group comparison design and had follow-up periods of at least 14 or 15 d (79% and 74%, respectively). The randomization ratio for 80 (89%) outcomes was 1:1.

Four (4%) outcomes were assessed to be at a low risk of bias, 31 (34%) had some concerns and 55 (61%) outcomes were classified to be at a high risk of bias (Table 1 and Supplement File 4).

Table 1.

Median risk ratios of vaccine efficacy by characteristics for the 90 primary outcomes of the 36 included studies (n=36)

Characteristics	Number of outcomes (%)	Median risk ratioa  (IQR)	p-value✝
Overall	90 (100)	0.32 (0.15, 0.45)
Vaccine type			<0.001
Viral vector	41 (46)	0.38 (0.32, 0.64)
Inactivated virus	8 (9)	0.30 (0.19, 0.37)
Protein subunit	22 (24)	0.20 (0.12, 0.30)
Nucleic acid	19 (21)	0.07 (0.06, 0.30)
Vaccine brand☼			<0.001
Johnson & Johnson	26 (29)	0.48 (0.34, 0.66)
AstraZeneca	13 (14)	0.38 (0.28, 0.46)
Pfizer-BioNTech	11 (12)	0.06 (0.05, 0.10)
Clover Biopharmaceuticals	10 (11)	0.13 (0.08, 0.25)
Covifenz	5 (6)	0.25 (0.17, 0.27)
Others	25 (28)	0.27 (0.10, 0.43)
Viral variant			<0.001
Wild	23 (26)	0.09 (0.06, 0.27)
Alpha	11 (12)	0.28 (0.07, 0.33)
Beta	7 (8)	0.50 (0.30, 0.64)
Delta	11 (12)	0.33 (0.21, 0.64)
Gamma	9 (10)	0.38 (0.13, 0.65)
Other	28 (31)	0.37 (0.20, 0.49)
Blinding type			<0.001
Open-label or not reported	20 (22)	0.15 (0.06, 0.30)
Single	16 (18)	0.08 (0.06, 0.20)
Double	51 (57)	0.38 (0.31, 0.51)
Triple	3 (3)	0.21 (0.08, 0.41)
Type of analysis			0.005
Intention to treat	6 (7)	0.31 (0.19, 0.54)
Per protocol	55 (61)	0.34 (0.21, 0.50)
Not reported	29 (32)	0.12 (0.06, 0.32)
Sample size			0.003
<10 000	23 (26)	0.32 (0.06, 0.39)
10 000–25 000	35 (39)	0.25 (0.12, 0.33)
>25 000	32 (36)	0.43 (0.25, 0.66)
Study design			<0.001
Crossover	20 (22)	0.50 (0.37, 0.66)
Parallel analysis	70 (78)	0.25 (0.08, 0.36)
Follow-up days (at least)			<0.001
7	20 (22)	0.10 (0.06, 0.22)
14 or 15	67 (74)	0.35 (0.21, 0.50)
21 or 28	3 (3)	0.33 (0.08, 0.43)
Randomization ratio			0.383
1:1	80 (89)	0.31 (0.10, 0.44)
>1:1	10 (11)	0.35 (0.15, 0.48)
Risk of bias			0.552
High	55 (61)	0.31 (0.11, 0.50)
Low	4 (4)	0.25 (0.14, 0.32)
Unclear	31 (34)	0.32 (0.10, 0.41)

a

Median risk ratios of vaccine efficacy.

✝

Calculated using Pearson's χ² test.

☼

Five most frequently reported vaccine brands among included studies.

Meta-analysis results

Data from 90 outcomes were used to estimate RR by variants, employing random-effects models to pool RRs due to detecting >50% heterogeneity among the outcomes (Figure 2). The overall RR was 0.26 (95% CI 0.22 to 0.31), indicating that the vaccines examined in the included studies are effective against symptomatic or asymptomatic infection or death. Conducting a subgroup analysis by viral variant, the pooled RR for SARS-CoV-2 infection was lowest against the wild-type (RR=0.13; 95% CI 0.09 to 0.18), followed by Alpha (RR=0.26; 95% CI 0.18 to 0.36), Gamma (RR=0.33; 95% CI 0.18 to 0.58), Delta (RR=0.39; 95% CI 0.27 to 0.56) and Beta (RR=0.53; 95% CI 0.43 to 0.65) (Figure 2).

Figure 2.

Risk ratios (RRs) of all included outcomes (n=90) by viral variant. Diamonds represent overall pooled estimates. Weights and between-subgroup heterogeneity were derived from random-effects models. Note: the common effect model is the Mantel–Haenszel model.

Analysis by vaccine type showed that nucleic acid vaccines had the lowest RR (RR=0.11; 95% CI 0.07 to 0.16), followed by protein subunit vaccines (RR=0.22; 95% CI 0.16 to 0.31), inactivated virus vaccines (RR=0.30; 95% CI 0.20 to 0.43) and viral vector vaccines (RR=0.41; 95% CI 0.35 to 0.49) (Figure 3).

Figure 3.

Risk ratios (RRs) of all included outcomes (n=90) by vaccine type. Diamonds represent overall pooled estimates. Weights and between-subgroup heterogeneity were derived from random-effects models. Note: the common effect model is the Mantel–Haenszel model.

The meta-analysis of SARS-CoV-2 infection for different vaccine brands by variant is presented in Figure 4A-F. Both the Moderna COVID-19 Vaccine and Pfizer-BioNTech Nucleic Acid Vaccine demonstrated the highest protection against wild-type variants, showing similar RR values (RR=0.07; 95% CI 0.06 to 0.09 and RR=0.08; 95% CI 0.05 to 0.12, respectively). Sputnik V showed the second highest protective effects against the wild-type variant, with a RR of 0.08 (95% CI 0.05 to 0.15), followed by AstraZeneca Viral Vector Vaccine and Sinovac CoronaVac COVID-19 Vaccine, both with similar RR values of 0.23 (95% CI 0.13 to 0.39 and 95% CI 0.09 to 0.60, respectively) (Figure 4A).

Figure 4.

(A–F) Risk ratios of all included outcomes (n=90) by vaccine brand for each viral variant. Red diamonds represent overall pooled estimates. The missing p-value is because no pooled analysis was conducted due to the presence of only one study in this group.

The Clover Biopharmaceuticals Protein Subunit Vaccine had the highest protective effects against the Alpha variant (RR=0.05; 95% CI 0.01 to 0.39), followed by Zifivax or ZF-UZ-VAC-2001 (RR=0.07; 95% CI 0.02 to 0.31) and the Novavax COVID-19 Vaccine (RR=0.14; 95% CI 0.07 to 0.29). The AstraZeneca Viral Vector Vaccine and the Johnson & Johnson Viral Vector Vaccine also demonstrated protection against the Alpha variant, with RR values of 0.29 (95% CI 0.22 to 0.38) and 0.25 (95% CI 0.15 to 0.42), respectively (Figure 4B).

While the Pfizer-BioNTech Nucleic Acid Vaccine, Clover Biopharmaceuticals Protein Subunit Vaccine, Johnson & Johnson Viral Vector Vaccine and Novavax COVID-19 Vaccine continued to show important protection against infection with the Beta variant, this was not the case for AstraZeneca Viral Vector Vaccine (RR=0.79; 95% CI 0.43 to 1.44) (Figure 4C).

Similarly, the Johnson & Johnson Viral Vector Vaccine did not seem to reduce infection risks against the Delta variant (RR=1.09; 95% CI 0.60 to 2.01). The most protective vaccines against the Delta variant were Zifivax or ZF-UZ-VAC-2001 (RR=0.19; 95% CI 0.12 to 0.31), Clover Biopharmaceuticals Protein Subunit Vaccine (RR=0.24; 95% CI 0.16 to 0.36) and Covifenz (RR=0.27; 95% CI 0.14 to 0.52) (Figure 4D).

Additionally, the Clover Biopharmaceuticals Protein Subunit Vaccine demonstrated the highest protective effects against the Gamma variant (RR=0.07; 95% CI 0.03 to 0.20), followed by the Covifenz Protein Subunit Vaccine (RR=0.13; 95% CI 0.05 to 0.30), CureVac N.V. and the CureVac Nucleic Acid Vaccine (RR=0.33; 95% CI 0.16 to 0.71) and the Johnson & Johnson Viral Vector Vaccine (RR=0.66; 95% CI 0.53 to 0.81). AstraZeneca COVID-19 Vaccine also showed reduced protection against the Gamma variant (RR=0.44; 95% CI 0.22 to 0.88) (Figure 4E).

For other variants, Moderna COVID-19 Vaccine demonstrated the highest infection risk reduction (RR=0.07; 95% CI 0.06 to 0.09), followed by Braha Biotech COVID-19 Vaccine (RR=0.10; 95% CI 0.01 to 0.78) and Clover Biopharmaceuticals Protein Subunit Vaccine (RR=0.17; 95% CI 0.06 to 0.47) (Figure 4F).

Protective effects of the Johnson & Johnson Viral Vector Vaccine declined with the emergence of subsequent variants, with the lowest RR observed for wild-type (RR=0.27, 95% CI 0.21 to 0.34) and the Alpha variant (RR=0.27, 95% CI: 0.15 to 0.49) (Supplement 5A). The AstraZeneca Viral Vector Vaccine also showed reduced protection over time, with the lowest RR for wild-type (RR=0.23, 95% CI 0.13 to 0.39), Alpha variant (RR=0.29, 95% CI 0.22 to 0.38) and Gamma variant (RR=0.44, 95% CI 0.22 to 0.88) (Supplement 5B). The Pfizer-BioNTech Nucleic Acid Vaccine had a low RR for both wild-type and the Beta variant (RR=0.08, 95% CI 0.05 to 0.12 and RR=0.11, 95% CI 0.01 to 0.83, respectively) (Supplement 5C). The Clover Biopharmaceuticals Protein Subunit Vaccine demonstrated strong protection against all variants (Supplement 5D). Covifenz had a low RR against Delta and Gamma variants (RR=0.27, 95% CI 0.14 to 0.52 and RR=0.13, 95% CI 0.05 to 0.30, respectively), but not against the Alpha variant and other variants (Supplement 5E).

Discussion

To the best of our knowledge, this is the first study to systematically compile and comprehensively assess the global literature to compare the protective effects of COVID-19 vaccines at brand level detail across different viral variants that emerged over a prolonged period during the pandemic.

Overall, we found vaccines to be protective against all SARS-CoV-2 variants, consistent with previous research. The highest protective effects were observed against the original (wild-type) strain, with decreasing magnitudes of protection against the Alpha, Gamma, Delta and Beta variants, in that order. Across all variants, nucleic acid vaccines exhibited the highest level of protection, followed by protein subunit vaccines, inactivated virus vaccines and viral vector vaccines.

The development of COVID-19 vaccines represents a remarkable achievement in the field of science. In just 1 y since the initial outbreak and isolation of the SARS-CoV-2 virus, multiple vaccines have been developed worldwide.⁴⁰ Over the last 3 y, numerous effective vaccines have been introduced.⁴¹ While these vaccines have varied in platforms and efficacies, the global vaccination goal has been largely achieved, albeit with regional disparities. As a result, the COVID-19 pandemic is now relatively under control.⁴² However, the ongoing emergence of new variants remains a cause for general concern and the vaccine efficacy against these different variants has become a subject of ongoing research.^33,34,43,44 While the initial clinical trials confirmed good protection against the early variants, the emergence of subsequent variants showed that these vaccines do not provide the same level of protection.³³

While in our study nucleic acid vaccines showed by far the highest protection against any variant for which data were available, it was observed that other vaccine platforms had heterogeneous levels of protection across variants. For example, the Johnson & Johnson Viral Vector Vaccine exhibited a high protective effect against the wild-type and the Alpha variant, but its protection decreased notably against Delta and other variants. A similar pattern was observed for the AstraZeneca viral vector vaccine, which displayed the highest protection against the wild-type and the Alpha variant, but reduced protection against the Beta and Gamma variants.

While some limited evidence of varying vaccine efficacy against different variants existed prior to our study,^33,34 those reviews, contrary to ours, did not provide a head-to-head comparison of different vaccines at brand level by variants, did not cover multiple variants over a prolonged period of the pandemic, did not follow PRISMA reporting guidelines, lacked a risk of bias assessment, mixed results from RCTs and observational studies in the same forest plots without presenting disaggregated estimates, did not provide a quantitative assessment of heterogeneity between studies and did not provide summary estimates in forest plots. Our study covers all these points.

We acknowledge some limitations. For some regions, particularly in developing countries, there was a lack of information presented on circulating variants at the time of the study, which affected our ability to assess vaccine effects against different variants in these populations comprehensively. Limited data availability further did not allow us to cover differences between the effects of a two-dose course and booster vaccination, and between homologous and heterologous vaccine schedules. Lastly, this review covers SARS-CoV-2 vaccine trials only until the emergence of the Omicron variant. This was a deliberate choice due to the rapidly evolving, incomplete evidence with regards to Omicron in the peer-reviewed literature at this time.

Our findings highlight that none of the vaccines based on traditional technologies were able to maintain stable levels of protection as SARS-CoV-2 variants continued to evolve. The virus's ability to produce new variants in the future poses a continuous challenge for vaccine development,⁴⁵ and while vaccines significantly reduce the risk of severe illness and death, breakthrough infections with new variants are still possible.^46,47 Innovations in vaccine development are essential to stay ahead of the virus and maximize protection for the global population. Our study provides invaluable baseline comparison data for such ongoing and future work.

Conclusions

We found high but heterogenous levels of protection for most COVID-19 vaccines, with overall decreasing protective effects as SARS-CoV-2 variants continued to evolve for vaccines based on traditional technologies. Novel nucleic acid-based vaccines offered substantially higher and consistent protection. Our analyses present important comparison data for ongoing and future COVID-19 vaccine development efforts.

Authors’ contributions

Substantial contributions to the conception or design of the work; or the acquisition, analysis or interpretation of data for the work: TNAH, AB, HLQ, MBT and FV. Drafting the manuscript of the work or reviewing it critically for important intellectual content: TNAH, AB, HLQ, MBT and FV. Final approval of the version to be published: TNAH, AB, HLQ, MBT and FV. All authors attest they meet the ICMJE criteria for authorship.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Competing interests

All authors declare that they have no conflicts of interest.

Ethical approval

Not required.

Data Availability

Data will be made available upon request to the corresponding author.