CoChat
Sign Up
Cited by 590SARS-CoV-2 and COVID-19 ResearchCOVID-19 epidemiological studiesCOVID-19 Clinical Research Studies

Neutralising antibody titres as predictors of protection against SARS-CoV-2 variants and the impact of boosting: a meta-analysis

Open access
Cromer, Deborah, Steain, Megan, Reynaldi, Arnold, Schlub, Timothy E, Wheatley, Adam K, Juno, Jennifer A, Kent, Stephen J, Triccas, James A, Khoury, David S, Davenport, Miles P·2022·The Lancet Microbe
DOI 10.1016/s2666-5247(21)00267-6PMC PMC8592563

Published version of a preprint

Open this paper in CoChat — free accountOpen PDF

Abstract

BACKGROUND: Several SARS-CoV-2 variants of concern have been identified that partly escape serum neutralisation elicited by current vaccines. Studies have also shown that vaccines demonstrate reduced protection against symptomatic infection with SARS-CoV-2 variants. We explored whether in-vitro neutralisation titres remain predictive of vaccine protection from infection with SARS-CoV-2 variants. METHODS: In this meta-analysis, we analysed published data from 24 identified studies on in-vitro neutralisation and clinical protection to understand the loss of neutralisation to existing SARS-CoV-2 variants of concern. We integrated the results of this analysis into our existing statistical model relating in-vitro neutralisation to protection (parameterised on data from ancestral virus infection) to estimate vaccine efficacy against SARS-CoV-2 variants. We also analysed data on boosting of vaccine responses and use the model to predict the impact of booster vaccination on protection against SARS-CoV-2 variants. FINDINGS: =0·81, p=0·0005) and the existing model remained predictive of vaccine efficacy against variants of concern once decreases in neutralisation to the variants of concern were incorporated. Modelling of predicted vaccine efficacy against variants over time suggested that protection against symptomatic infection might decrease below 50% within the first year after vaccination for some vaccines. Boosting of previously infected individuals with existing vaccines (which target ancestral virus) is predicted to provide a higher degree of protection from infection with variants of concern than primary vaccination schedules alone. INTERPRETATION: In-vitro neutralisation titres remain a correlate of protection from SARS-CoV-2 variants and modelling of the effects of waning immunity predicts a loss of protection to the variants after vaccination. However, booster vaccination with current vaccines should enable higher neutralisation to SARS-CoV-2 variants than is achieved with primary vaccination, which is predicted to provide robust protection from severe infection outcomes with the current SARS-CoV-2 variants of concern, at least in the medium term. FUNDING: The National Health and Medical Research Council (Australia), the Medical Research Future Fund (Australia), and the Victorian Government.

AI snapshot

Written from the paper’s full text.

TL;DR

In a meta-analysis of published SARS-CoV-2 neutralisation and vaccine-efficacy data, in-vitro neutralising antibody titres remained predictive of protection against variants of concern, and the authors’ prior model predicted waning protection over the first year after vaccination. Boosting previously infected or vaccinated individuals with existing ancestral-spike vaccines was predicted to raise neutralisation and restore robust protection against current variants.

Key findings

  • Neutralisation against ancestral SARS-CoV-2 strongly predicted neutralisation of variants; geometric mean decreases in titre versus ancestral virus were 1·6-fold for alpha (95% CI 1·5–1·7; 9 studies), 8·8-fold for beta (8·0–9·7; 9 studies), 3·5-fold for gamma (3·1–4·0; 3 studies), and 3·9-fold for delta (3·5–4·4; 3 studies). Loss of neutralisation did not differ significantly by vaccine or prior infection after adjusting for variant and laboratory effects (p=0·26).
  • Neutralisation remained correlated with protection from symptomatic infection with variants of concern (rS=0·81, p=0·0005), and 13 (93%) of 14 efficacy/effectiveness studies fell within the 95% CI of the model prediction.
  • Modelling suggested protection against symptomatic infection might fall below 50% within the first year after vaccination for some vaccines; for example, previous infection was predicted to provide about 90% protection against ancestral virus early, falling by 6 months to 41% against delta and 24% against beta.
  • Boosting convalescent individuals with mRNA vaccines led to 2–10-fold higher neutralisation titres than primary vaccination of naive individuals, with a geometric mean 12-fold (range 6–29) higher than early convalescence; predicted BNT162b2 efficacy against delta increased from 75% (95% CI 53–89) in naive individuals to 95% (85–98) in previously infected individuals.

Methods

  • Meta-analysis of published data from 24 identified studies, including 17 studies directly comparing neutralisation titres against ancestral virus and variants of concern, amounting to 1305 individual neutralisation titres; live-virus assays were prioritised, except for NVX-CoV2373 where pseudovirus data were used.
  • Multiple linear regression accounted for variant-, vaccine-, and laboratory-specific effects, with censoring for titres below assay detection limits; the authors integrated fold-change neutralisation estimates into their previously published model relating neutralisation to protection, parameterised on ancestral-virus data.
  • Boosting analyses used studies comparing previously infected or vaccinated individuals with naive vaccinated individuals from the same laboratory/assay; decay modelling assumed a neutralisation half-life of 108 days and the same decay rate before and after boosting.

Limitations

  • The paper states that the main limitations centre on aggregating neutralisation data from multiple laboratories using different assays, few studies of vaccine efficacy/effectiveness against variants of concern especially severe outcomes, and wide CIs; the largest contributor to uncertainty was the estimate of mean neutralisation titre for each vaccine after normalisation to convalescent individuals.
  • The paper states that standardised neutralisation assays and consistent definitions of severe disease would improve comparability; it also notes that severe-disease predictions have the lowest confidence, partly because fewer than 100 severe infections were reported across studies combined in the original model development.
  • The text makes evident that all 10 observational test-negative case-control studies fell above the model prediction line, suggesting possible overestimation of vaccine efficacy in that design; the model also assumes neutralisation alone drives protection against severe disease, so it might underestimate efficacy against severe infection.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/s2666-5247(21)00267-6

Cites

63 of 63 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 47 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Neutralising activity against ancestral SARS-CoV-2 is highly predictive of neutralisation of variants of concern.

    “The neutralising activity against the ancestral SARS-CoV-2 was highly predictive of neutralisation of variants of concern.”
    Quote in the paperSubstantial claim
  • Findingown data

    The geometric mean decrease in neutralisation titre to the beta variant compared with ancestral virus is 8.8-fold (95% CI 8.0-9.7; 9 studies).

    “The geometric mean of the decrease in neutralisation titres was found to be 1·6-fold for alpha (95% CI 1·5–1·7; 9 studies), 8·8-fold for beta (8·0–9·7; 9 studies), 3·5-fold for gamma (3·1–4·0; 3 studies), and 3·9-fold for delta (3·5–4·4; 3 studies) compared with the ancestral virus (appendix p 23).”
    Quote in the paperSubstantial claim
  • Findingown data

    The geometric mean decrease in neutralisation titre to the delta variant compared with ancestral virus is 3.9-fold (95% CI 3.5-4.4; 3 studies).

    “The geometric mean of the decrease in neutralisation titres was found to be 1·6-fold for alpha (95% CI 1·5–1·7; 9 studies), 8·8-fold for beta (8·0–9·7; 9 studies), 3·5-fold for gamma (3·1–4·0; 3 studies), and 3·9-fold for delta (3·5–4·4; 3 studies) compared with the ancestral virus (appendix p 23).”
    Quote in the paperSubstantial claim
  • Findingown data

    Neutralisation remains strongly correlated with protection from symptomatic infection with SARS-CoV-2 variants of concern (rS=0.81, p=0.0005).

    “Neutralisation remained strongly correlated with protection from symptomatic infection with SARS-CoV-2 variants of concern (rS=0·81, p=0·0005)”
    • Also stated in: § Results · plain
      “Despite the variability in study design, we found that predicted serological neutralisation activity against each variant elicited by vaccines was significantly correlated with protection from symptomatic SARS-CoV-2 infection (r S=0·81, p=0·0005; appendix p 24).”
    Quote in the paperSubstantial claim
  • Findingown data

    Decreases in neutralisation titre to the alpha, beta, gamma, and delta variants compared with ancestral virus are not significantly different between different vaccines.

    “Decreases in neutralisation titre to the alpha (1·6-fold), beta (8·8-fold), gamma (3·5-fold), and delta (3·9-fold) variants (compared to the ancestral virus) were not significantly different between different vaccines.”
    Quote in the paperSubstantial claim
  • May combine two claims. 13
  • Classified as probably not a substantial claim. 9

  • May combine two claims.

    Findingown data

    Within a given study, the observed decrease in neutralisation titre for a given variant was very similar for both convalescent and vaccinee serum, suggesting a strong study-specific or assay-specific effect.

    “However, within a given study the observed decrease in neutralisation titre for a given variant was very similar for both convalescent and vaccinee serum, suggesting a strong study-specific (or assay-specific) effect (appendix p 21).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    The four randomised controlled trials are placed relatively symmetrically around the predictive line, with one above and three below the mean prediction line.

    “We note that the four randomised controlled trials are placed relatively symmetrically around the predictive line (one above and three below the mean prediction line; figure 2A).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The design of study might have a role in the observed efficacy, with all 10 observational test-negative case-control studies placed above the line.

    “The results also hint that the design of study might have a role in the observed efficacy, with all 10 observational test-negative case-control studies placed above the line.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Previous work has shown a correlation between titres of neutralising antibodies, normalised for assay differences, and protection from SARS-CoV-2 infection, and has derived a model for predicting vaccine efficacy from mean neutralisation titres.

    “Our previous work has shown a correlation between titres of neutralising antibodies (normalised for differences in the assays used) and protection from SARS-CoV-2 infection and has derived a model for predicting vaccine efficacy from mean neutralisation titres.8”
    • Also stated in: § Introduction · boosted
      “Previous work has identified that neutralising antibodies to ancestral SARS-CoV-2 virus are a strong predictor of protection from symptomatic infection with a similar virus.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [8] Khoury, 2021, Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection, Nat Med source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    The association between neutralisation titres and vaccine efficacy against variants of concern is unclear.

    “The association between neutralisation titres and vaccine efficacy against variants of concern is unclear.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Current vaccines might be less protective against symptomatic infection with some SARS-CoV-2 variants, and several studies have now shown reduced efficacy of vaccination against infection with SARS-CoV-2 variants.

    “Initial reports from clinical trials or from breakthrough community infections suggest that current vaccines might be less protective against symptomatic infection with some SARS-CoV-2 variants.5, 6”
    • Also stated in: § Results · boosted
      “Several studies have now shown reduced efficacy of vaccination against infection with SARS-CoV-2 variants (appendix p 17).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [5] Madhi, 2021, Efficacy of the ChAdOx1 nCoV-19 Covid-19 vaccine against the B.1.351 variant, N Engl J Med source

      Not checked in the free run.

    • cites [6] Sheikh, 2021, SARS-CoV-2 Delta VOC in Scotland: demographics, risk of hospital admission, and vaccine effectiveness, Lancet source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Vaccination with a single dose of mRNA vaccine is sufficient to boost neutralising antibody responses in previously infected individuals, and a second dose has minimal additional effect.

    “These studies have found that vaccination with a single dose of mRNA vaccine is sufficient to boost responses in previously infected individuals (and indeed a second dose has minimal effect).28, 29”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [28] Stamatatos, 2021, mRNA vaccination boosts cross-variant neutralizing antibodies elicited by SARS-CoV-2 infection, Science source

      Not checked in the free run.

    • cites [29] Goel, 2021, Distinct antibody and memory B cell responses in SARS-CoV-2 naive and recovered individuals following mRNA vaccination, Sci Immunol source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Our work predicts a major loss of protection from symptomatic infection with current variants of concern during the first year after vaccination, which might be reversed by a third booster vaccination.

    “Our work predicts a major loss of protection from symptomatic infection with current variants of concern during the first year after vaccination, which might be reversed by a third booster vaccination.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: Boosting increases the normalised geometric mean neutralisation titre to the mean reported in the boosting studies.

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    The study focused primarily on assays using live SARS-CoV-2 virus to reduce potential variability from different pseudoviral constructs, with the exception of NVX-CoV2373 for which pseudovirus data were used.

    “We focused primarily on assays using live SARS-CoV-2 virus (to reduce the potential variability that might arise from different pseudoviral constructs23), with the exception of the NVX-CoV2373 vaccine, for which only data from neutralisation assays using a spike-expressing pseudovirus were available.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [23] Khoury, 2020, Measuring immunity to SARS-CoV-2 infection: comparing assays and animal models, Nat Rev Immunol source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Study design and conductown data

    The study performed a multiple linear regression accounting for variant-specific, vaccine-specific, and laboratory-specific effects.

    “To assess the impact of vaccine type and laboratory on cross-reactivity in the aggregated data, we performed a multiple linear regression accounting for variant-specific, vaccine-specific, and laboratory-specific effects.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The lower bound on efficacy estimates from the model was determined using a bootstrapping approach.

    “The lower bound on efficacy estimates from the model was determined using a bootstrapping approach (appendix pp 7–14).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    All 10 observational test-negative case-control studies are placed above the line, indicating a potential role for the design of study in the observed efficacy.

    “However, all 10 observational test-negative case-control studies are placed above the line, indicating a potential role for the design of study in the observed efficacy.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    The study used censoring to account for neutralisation titres below the limits of detection of the assay.

    “We used censoring to account for neutralisation titres below the limits of detection of the assay (appendix pp 7–14).”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Boosting data were included only if they could be compared with data from naive vaccinated individuals from the same laboratory using the same assay.

    “data were taken from studies of boosting of previously infected or vaccinated individuals and were included if the boosting data could be compared with data from naive vaccinated individuals from the same laboratory using the same assay (appendix pp 1–14).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Neutralising antibody titres decrease over the first 8 months of infection, with a half-life of around 3-4 months.

    “A previous study24 showed a decrease in neutralising antibody titres over the first 8 months of infection, with a half-life of around 3–4 months”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [24] Dan, 2021, Immunological memory to SARS-CoV-2 assessed for up to 8 months after infection, Science source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Vaccination of previously infected individuals, even with existing vaccines targeting ancestral virus, will provide robust protection against the current variants of concern, considerably prolonging the duration of efficacy of existing vaccines against these variants.

    “Together these data and modelling suggest that vaccination of previously infected individuals, even with existing vaccines targeting ancestral virus, will provide robust protection against the current variants of concern, considerably prolonging the duration of efficacy of existing vaccines against these variants.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: Neutralisation decays with a half-life of 108 days.
    • The authors state this as an assumption: Variant neutralisation decreases as estimated (appendix p 23).
    • The authors state this as an assumption: Decay after boosting is the same as after initial infection or primary vaccination.

  • May combine two claims.

    Interpretationreasoning

    Previous infection is thought to provide about 90% protection against the ancestral virus early after infection, but by 6 months protection is predicted to decrease to 41% against the delta variant and 24% against the beta variant.

    “For example, previous infection is thought to provide about 90% protection against the ancestral virus early after infection, but, by 6 months, protection is predicted to decrease to 41% against the delta variant and 24% against the beta variant.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Previous infection provides about 90% protection against the ancestral virus early after infection.

  • May combine two claims.

    Interpretationreasoning

    Boosting at 6 months with an mRNA vaccine is expected to increase normalised neutralising antibody titres to 12-fold that of early convalescent titres, leading to protection of 95% for the delta variant and 87% for the beta variant soon after boosting.

    “However, boosting at 6 months with an mRNA vaccine is expected to increase normalised neutralising antibody titres to 12-fold that of early convalescent titres, leading to protection of 95% for the delta variant and 87% for the beta variant soon after boosting.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: Neutralisation decays with a half-life of 108 days.
    • The authors state this as an assumption: Variant neutralisation decreases as estimated (appendix p 23).
    • The authors state this as an assumption: Decay after boosting is the same as after initial infection or primary vaccination.

  • May combine two claims.

    Interpretationreasoning

    Boosting is predicted to provide 68% protection from symptomatic infection (lower bound of 95% CI 45) and 94% protection from severe infection (lower bound of 95% CI 72) even against the beta variant of concern 6 months after boosting.

    “Although the decay of neutralisation titres after boosting is similar to that after primary infection,25 boosting is still predicted to provide 68% protection from symptomatic infection (lower bound of 95% CI 45) and 94% protection from severe infection (lower bound of 95% CI 72) even against the beta variant of concern 6 months after boosting (figure 4).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [25] Goel, 2021, mRNA vaccination induces durable immune memory to SARS-CoV-2 and variants of concern, Science source

      Not checked in the free run.

    • The authors state this as an assumption: Neutralisation decays with a half-life of 108 days.
    • The authors state this as an assumption: Variant neutralisation decreases as estimated (appendix p 23).
    • The authors state this as an assumption: Decay after boosting is the same as after initial infection or primary vaccination.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    Immune imprinting or cross-reactivity to previous immunogens might need to be considered in strategies aimed at boosting neutralising responses to new variants.

    “Hence, immune imprinting or cross-reactivity to previous immunogens might need to be considered in strategies aimed at boosting neutralising responses to new variants.4, 40”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [4] Liu, 2021, Reduced neutralization of SARS-CoV-2 B.1.617 by vaccine and convalescent serum, Cell source

      Not checked in the free run.

    • cites [40] Wheatley, 2021, Immune imprinting and SARS-CoV-2 vaccine design, Trends Immunol source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    The main limitations of this study centre around difficulties in aggregating neutralisation data from multiple laboratories, most of which use different assays.

    “The main limitations of this study centre around difficulties in aggregating neutralisation data from multiple laboratories, most of which use different assays.”
    Quote in the paperSubstantial claim
21 of 21

  • The authors state this as an assumption.

    The decay rate of neutralisation before and after boosting is the same.

    “The decay rate of neutralisation before and after boosting was assumed to be the same.25”

  • The authors state this as an assumption.

    Boosting increases the normalised geometric mean neutralisation titre to the mean reported in the boosting studies.

    “Boosting was assumed to increase the normalised geometric mean neutralisation titre to the mean reported in the boosting studies (appendix p 18).”

  • The authors state this as an assumption.

    Neutralisation alone drives protection against severe disease.

    “It should be noted that our model assumes neutralisation alone drives protection against severe disease”

  • The authors state this as an assumption.

    Neutralisation of variants is reduced by the fold changes estimated in this study.

    “and assumed that neutralisation of variants was reduced by the fold changes estimated in this study”

  • The authors state this as an assumption.

    Neutralisation decays with a half-life of 108 days.

    “It is assumed that neutralisation decays with a half-life of 108 days”

  • The authors state this as an assumption.

    Variant neutralisation decreases as estimated (appendix p 23).

    “and variant neutralisation decreases as estimated (appendix p 23).”

  • The authors state this as an assumption.

    Decay after boosting is the same as after initial infection or primary vaccination.

    “and assumes decay after boosting is the same as after initial infection or primary vaccination.”

  • The paper relies on this without support.

    Previous infection provides about 90% protection against the ancestral virus early after infection.

    “For example, previous infection is thought to provide about 90% protection against the ancestral virus early after infection”
8 of 8

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Pan, 2021, Immunogenicity and safety of a third dose, and immune persistence of CoronaVac vaccine in healthy adults aged 18–59 years: interim results from a double-blind, randomized, placebo-controlled phase 2 clinical trial, medRxiv

Cited by

Sorted by citations

Authors
Cromer, D., Steain, M., Reynaldi, A., Schlub, T. E., Wheatley, A. K., Juno, J. A., Kent, S. J., Triccas, J. A., Khoury, D. S., & Davenport, M. P.
Affiliations
  • UNSW Sydney (AU)
  • The University of Sydney (AU)
  • The University of Melbourne (AU)
Corresponding author
Khoury, David S
ORCID
Venue
The Lancet Microbe
Publisher
Elsevier BV
Published
2022-01
Volume
3
Issue
1
Pages
e52-e61
ISSN
2666-5247 (ISSN-L)
DOI
10.1016/s2666-5247(21)00267-6
PMC
PMC8592563
OpenAlex
W3212447088
Semantic Scholar
d246f823f9d54743bcf57ab5387c63f965bbe0a4
Cited by
590
References
63
Citations by year
2026: 37 · 2025: 61 · 2024: 88 · 2023: 158 · 2022: 220 · 2021: 26
FWCI
14.45
Type
journal article
Language
en
Open access
Licence
Locations
  • The Lancet Microbe (publishedVersion, cc-by)
  • medRxiv (acceptedVersion)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • UNSWorks (University of New South Wales, Sydney, Australia) (submittedVersion, cc-by-nc-nd)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Minerva Access (University of Melbourne) (submittedVersion, cc-by)
  • The Sydney eScholarship Repository (The University of Sydney) (submittedVersion)
  • UNSWorks (University of New South Wales, Sydney, Australia) (submittedVersion, cc-by)
Funders
  • National Health and Medical Research Council (GNT1149990)
  • State Government of Victoria
Topics
  • Health Sciences › Medicine › Infectious Diseases › SARS-CoV-2 and COVID-19 Research
  • Physical Sciences › Mathematics › Modeling and Simulation › COVID-19 epidemiological studies
  • Health Sciences › Medicine › Infectious Diseases › COVID-19 Clinical Research Studies
Keywords
SARS-CoV-2 variants of concern, neutralizing antibodies, vaccine efficacy, waning immunity, booster vaccination, vaccine escape, correlates of protection, COVID-19 vaccines
MeSH terms
COVID-19, SARS-CoV-2, COVID-19 Vaccines, Antibodies, Viral, Humans, Antibodies, Neutralizing

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review