CoChat
Sign Up
Cited by 2Amyotrophic Lateral Sclerosis ResearchNeurogenetic and Muscular Disorders ResearchVoice and Speech Disorders

Head trauma and environment progression of amyotrophic lateral sclerosis: long-term data from the National ALS Registry

Open access
Martinez-Nunez, Alfonso Enrique, Guo, Jingchuan, Dorsey, Earl Ray, Ruffing, Kyle W, Wymer, James, Okun, Michael S·2026·BMJ Neurology Open
DOI 10.1136/bmjno-2026-001574PMC PMC13140978
Open this paper in CoChat — free account

Abstract

Background: Environmental exposures have been linked to increased risk of amyotrophic lateral sclerosis (ALS); however, their impact on disease progression remains unclear. This study examined whether prior environmental and occupational exposures influenced functional decline in patients with an established ALS diagnosis. Methods: We conducted a retrospective cohort analysis using the National ALS Registry from 2010 to 2024. Participants with complete exposure histories were included. Disease progression was measured with the ALS Functional Rating Scale-Revised (ALSFRS-R) at baseline and every 3 months. Mixed-effects linear regression models assessed associations between exposures and ALSFRS-R decline, adjusting for age, sex and time since diagnosis. Results: The cohort included 8618 participants with ALS. The median time from diagnosis to enrolment was 2 years (IQR= 1.1–2.9), with a median of 1 year of follow-up (IQR=1–4). Exposure to herbicides (β=−0.57. IC 95% =−0.86 to −0.28, p<0.001), metal dust and fumes (β=−0.28, IC 95% =−0.51 to −0.04, p=0.020) and oil paint (β=−0.27, IC 95% =−0.48 to −0.06, p=0.011) prior to diagnosis were each associated with accelerated decline. Head injury was associated with an overall lower ALSFRS-R score (β=−1.74, IC 95% =−2.21 to −1.27, <0.001), based on our non-linear mixed effects model. Conclusion: Environmental and occupational exposures, particularly herbicides, metal dust/fumes and oil-based paints, were associated with faster ALS progression, and head injury was associated with overall worse function.

AI snapshot

Written from the paper’s full text.

TL;DR

In a retrospective National ALS Registry cohort of 8618 people with ALS, prior herbicide, metal dust/fume, and oil-based paint exposures were associated with faster ALSFRS-R functional decline, while head injury was associated with lower overall function but not faster progression. The authors conclude exposure history may add prognostic information and suggest cumulative exposure burden may modify ALS trajectory.

Key findings

  • Among 8618 participants, herbicide exposure was associated with an additional yearly ALSFRS-R decline of 0.57 points (β=−0.57, 95% CI −0.86 to −0.28, p<0.001), metal dust/fumes with 0.28 points (β=−0.28, 95% CI −0.51 to −0.04, p=0.020), and oil-based paint with 0.27 points (β=−0.27, 95% CI −0.48 to −0.06, p=0.011).
  • Head injury was associated with a lower overall ALSFRS-R score (β=−1.74, 95% CI −2.21 to −1.27, p<0.001) but did not significantly interact with time since diagnosis (p=0.802), indicating lower baseline function rather than accelerated progression.
  • The null cubic model showed an initial yearly ALSFRS-R decline of 2.80 points (time since diagnosis coefficient −2.80, 95% CI −3.32 to −2.29, p<0.001), with deceleration in later disease stages.
  • Fungicides, solvents, and insecticides did not show a significant effect on ALSFRS-R after correction for time since diagnosis.

Methods

  • Retrospective cohort analysis using the National ALS Registry from 2010 to 2024, including 8618 participants with ALS and complete exposure histories; ALSFRS-R was measured at baseline and every 3 months.
  • Exposures included herbicides, fungicides, insecticides, rodenticides, gasoline, metal dust/fumes, unleaded paint, leaded paint, oil paint, and head injury, ascertained by questionnaire.
  • Mixed-effects regression models assessed ALSFRS-R decline, adjusting for age, sex, and time since diagnosis; a cubic time model best fit by AIC/BIC, with random intercepts, multiple imputation for missing predictors, ℓ1-penalisation, and Benjamini-Hochberg correction. Sensitivity analyses restricted follow-up to ≤1, ≤2, and ≤3 years.

Limitations

  • The authors state reliance on self-reported exposure history introduces recall bias, head injury assessment was broad, and later follow-up beyond 3 years was sparse, so long-term cubic model estimates may reflect extrapolation rather than directly observed decline.
  • The authors state time since diagnosis was used instead of symptom onset, diagnostic delay could bias results, and residual confounding is possible from unmeasured genetic testing, disease-modifying medications, rurality/access to care, and other factors; the mean enrolment ALSFRS-R of 26 indicates later-stage disease, limiting generalisability to early ALS.
  • The authors state the observational design precludes causal inference and reverse causation cannot be ruled out; they also note a male-to-female ratio near 1.5, possibly related to veteran recruitment.

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

Integrity signals

Registry

Crossref · 10.1136/bmjno-2026-001574

Cites

32 of 32 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

3 of 5 cited sources checked · 1 supported · 0 flagged · accuracy not yet measured

  • Premisecitation

    Environmental exposures including herbicides and heavy metals have been associated with increased risk of developing ALS.

    “Growing epidemiological evidence suggests that exposure to certain environmental contaminants, including herbicides and heavy metals, has been associated with a higher incidence of ALS.2”
    • Also stated in: § WHAT IS ALREADY KNOWN ON THIS TOPIC · plain
      “Environmental exposures including herbicides, heavy metals and head trauma have been associated with increased risk of developing amyotrophic lateral sclerosis (ALS), but their impact on disease progression after diagnosis has remained unclear.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [2] source

      The Introduction and Discussion report epidemiological associations between pesticide exposure, including herbicides, and ALS risk. Our copy does not report an association between heavy-metal exposure and ALS; searches for heavy metals and specific metals did not find one. No page numbers were provided for these passages. (LLM, conf=0.87)

      Experimental — accuracy not yet measured
  • Premisecitation

    Diagnostic delay for ALS is shorter compared with other neurodegenerative syndromes.

    “However, diagnostic delay for ALS is shorter compared with other neurodegenerative syndromes.31”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [31] Richards, 2020, Time to diagnosis and factors affecting diagnostic delay in amyotrophic lateral sclerosis, J Neurol Sci source

      Section 5 reports that most studies found an ALS diagnostic delay of 10–16 months. Section 3 discusses delayed dementia diagnosis, but gives no comparable duration or finding that ALS is diagnosed sooner than other neurodegenerative syndromes. The claimed comparison is therefore unsupported. (LLM, conf=0.78)

      Experimental — accuracy not yet measured
  • Findingown data

    Male gender did not significantly influence the rate of progression over time (Time×Male interaction p=0.860).

    “Time×Male gender | 0.01 | (−0.14 to 0.17) | 0.860”
    • Also stated in: § Mixed effects models of disease progression · hedged
      “Male gender was associated with higher baseline ALSFRS-R scores (coefficient: 1.39, p<0.001) but did not significantly influence the rate of progression over time (Time×Male interaction: p=0.860), indicating that sex differences affected initial disease severity rather than trajectory of decline.”
    Quote in the paperNot a substantial claim
  • Findingown data

    Metal dust and fumes exposure prior to diagnosis was associated with accelerated ALSFRS-R decline (β=−0.28, 95% CI −0.51 to −0.04, p=0.020).

    “Exposure to herbicides (β=−0.57. IC95%=−0.86 to −0.28, p<0.001), metal dust and fumes (β=−0.28, IC95%=−0.51 to −0.04, p=0.020) and oil paint (β=−0.27, IC95%=−0.48 to −0.06, p=0.011) prior to diagnosis were each associated with accelerated decline.”
    • Also stated in: § Mixed effects models of disease progression · plain
      “Time×Metal dust/fumes | −0.28 | (−0.51 to −0.04) | 0.020”
    • Also stated in: § Mixed effects models of disease progression · plain
      “we found that herbicide, metal dust, fumes and oil paint exposures were significantly associated with accelerated functional decline as measured by ALSFRS-R scores over time (all interaction terms p<0.05, see table 2).”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Self-reported exposure and head injury histories accurately classify the exposures of interest.
  • Findingown data

    Male gender was associated with higher baseline ALSFRS-R scores (coefficient 1.39, p<0.001).

    “Male gender | 1.39 | (0.99 to 1.79) | <0.001”
    • Also stated in: § Mixed effects models of disease progression · plain
      “Male gender was associated with higher baseline ALSFRS-R scores (coefficient: 1.39, p<0.001) but did not significantly influence the rate of progression over time (Time×Male interaction: p=0.860), indicating that sex differences affected initial disease severity rather than trajectory of decline.”
    Quote in the paperNot a substantial claim
  • Classified as probably not a substantial claim. 20
  • May combine two claims. 13
  • The quote only partly states this claim. 1

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

    Premisecitation

    Head trauma has been associated with increased risk of developing ALS.

    “Environmental exposures including herbicides, heavy metals and head trauma have been associated with increased risk of developing amyotrophic lateral sclerosis (ALS), but their impact on disease progression after diagnosis has remained unclear.”
    • Also stated in: § Discussion · plain
      “Head injuries were associated with earlier diagnosis of ALS.15”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [15] Raymond, 2025, Head Injury and Amyotrophic Lateral Sclerosis: Population-Based Study from the National ALS Registry, Brain Sci source

  • May combine two claims.

    Findingown data

    At enrolment, the cohort reported exposure to head injury (60.7%), insecticides (33.1%), oil paint (30.6%), metal dust and fumes (29.2%), fungicides (27.9%), herbicides (26.4%) and solvents (24.9%).

    “At enrolment, our cohort reported exposure to head injury (60.7%), insecticides (33.1%), oil paint (30.6%), metal dust and fumes (29.2%), fungicides (27.9%), herbicides (26.4%) and solvents (24.9%).”
    Quote in the paperNot a substantial claim

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

    Findingown data

    The median follow-up was 1 year (IQR 1-4).

    “The median time from diagnosis to enrolment was 2 years (IQR= 1.1–2.9), with a median of 1 year of follow-up (IQR=1–4).”
    • Also stated in: § Study population · plain
      “The mean follow-up time was 1.90 years (SD=1.36, median=1, IQR=1–4).”
    Quote in the paperNot a substantial claim

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

    Findingown data

    The subject male to female ratio is close to 1.5, showing a male ratio in the higher limit reported in other studies.

    “Third, our subject male to female ratio is close to 1.5, which shows a male ratio in the higher limit reported in other studies.”
    Quote in the paperNot a substantial claim

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

    Findingown data

    The median number of surveys answered per subject was 6 (range 2 to 16, IQR 4-8).

    “The median number of surveys answered per subject was 6 (range=2 to 16, IQR=4–8).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    A cubic model for time since diagnosis provided the best fit to the data, with the lowest AIC and BIC values compared with linear, quadratic and logarithmic alternatives.

    “A cubic model for time since diagnosis provided the best fit to the data (figure 2), with the lowest AIC and BIC values compared with linear, quadratic and logarithmic alternatives (see online supplemental table 1).”
    • Also stated in: § Statistical analysis · plain
      “We compared different mixed-effects models: linear, exponential, log and cubic. We used the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC).”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Findingown data

    The cohort included 8618 participants with ALS.

    “The cohort included 8618 participants with ALS.”
    • Also stated in: § Study population · plain
      “A total of 8618 patients with ALS were included in this analysis with surveys ranging from 2010 to 2024.”
    • Also stated in: § Mixed effects models of disease progression · plain
      “A total of 8618 unique participants were included in the models.”
    Quote in the paperSubstantial claim

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

    Findingown data

    The study population was predominantly male (59.65%), consistent with the known male predominance in ALS.

    “The study population was predominantly male (59.65%), which is consistent with the known male predominance in ALS.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Head injury did not significantly interact with time since diagnosis (p=0.802), indicating no association with progression rate.

    “Time×Head injury history | 0.02 | (−0.15 to 0.19) | 0.802”
    • Also stated in: § Mixed effects models of disease progression · hedged
      “Notably, history of head injury showed a significant main effect on baseline ALSFRS-R scores (coefficient: −1.74, p<0.001) but did not significantly interact with time (p=0.802), indicating it was associated with lower initial functional status rather than accelerated progression.”
    • Also stated in: § Discussion · plain
      “The history of head injury was associated with an overall more severe disease; however, the progression rate was not different when compared with those without history of head trauma.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Self-reported exposure and head injury histories accurately classify the exposures of interest.

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

    Findingown data

    The mean age at ALS diagnosis was 59.50 years (SD=10.62).

    “The mean age at ALS diagnosis was 59.50 years (SD=10.62)”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The mean ALSFRS-R score at enrolment was 26, indicating participants were already substantially disabled at recruitment.

    “Additionally, the mean ALSFRS-R score at enrolment was 26, indicating that participants were already substantially disabled at recruitment.”
    • Also stated in: § Study population · plain
      “Total (mean±SD) | 26.21±9.31 points”
    Quote in the paperSubstantial claim

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

    Findingown data

    The median time from ALS diagnosis to enrolment was 2 years (IQR 1.1-2.9).

    “The median time from diagnosis to enrolment was 2 years (IQR= 1.1–2.9), with a median of 1 year of follow-up (IQR=1–4).”
    • Also stated in: § Study population · plain
      “having been diagnosed an average of 1.84 years (SD=1.70, median=2, IQR=1.1 to 2.9) prior to study enrolment.”
    Quote in the paperNot a substantial claim

  • The quote only partly states this claim.

    Study design and conductown data

    The direction and general pattern of environmental effects remained stable across model specifications and time windows.

    “While the magnitude and precise significance levels varied across model specifications, the direction and general pattern of these environmental effects remained stable, supporting the robustness of these key findings.”
    • Also stated in: § Mixed effects models of disease progression · plain
      “The direction and general pattern of all other exposure effects remained stable across time windows, supporting the robustness of the primary findings.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Self-reported exposure and head injury histories accurately classify the exposures of interest.

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

    Interpretationreasoning

    Geographic factors such as rurality may confound the findings.

    “Additionally, geographic factors such as rurality may confound our findings, as patients in non-urban settings may have had greater environmental exposures alongside reduced access to specialised ALS care, potentially influencing observed progression rates.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductreasoning

    Education level was not included in the regression models because it served primarily as a socioeconomic proxy with no established direct biological effect on ALS progression rate, and its correlation with occupational exposure history risked introducing collinearity.

    “Although education level was collected and is reported as a descriptive characteristic, it was not included in the regression models as it served primarily as a socioeconomic proxy with no established direct biological effect on ALS progression rate, and its correlation with occupational exposure history risked introducing collinearity into the model.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    The models were not adjusted for genetic status, medication use, site of onset, respiratory function or socioeconomic factors.

    “These models were not adjusted for genetic status, medication use, site of onset, respiratory function or socioeconomic factors.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    The study handled missing data in predictors by using multiple imputations by chained equations.

    “We handled missing data in the predictors by using multiple imputations by chained equations.11”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [11] source

  • Classified as probably not a substantial claim.

    Study design and conductown data

    The time variable reflects time since ALS diagnosis because symptom onset data are not available.

    “The time variable reflects the time since ALS diagnosis, since the data from symptom onset are not available.”
    Quote in the paperSubstantial claim

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

    Study design and conductown data

    The study used Benjamini-Hochberg false discovery rate correction to control for multiple comparisons.

    “We used the Benjamini-Hochberg false discovery rate post hoc correction to control for the multiple comparisons of separate interaction terms. All the p values presented have been already corrected using this method.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    The study used l1-penalisation to prevent estimate bias in the case of unbalanced classes.

    “To prevent estimate bias in the case of unbalanced classes (eg, if a certain exposure is relatively rare), we used ℓ1-penalisation.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    Patients with more aggressive disease might be diagnosed faster, creating systematic bias.

    “Patients with more aggressive disease (potentially those with exposures) might be diagnosed faster, creating systematic bias.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning

    Symptom onset data and disease survival were not available from the CDC, so the study conservatively used the overall rate of progression to reduce bias from lack of exact symptom onset.

    “Symptom onset data and disease survival were not available from the CDC and because of this, we conservatively used the overall rate of progression in order to reduce the bias from lack of exact symptom onset.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning

    The model's behaviour at later time points is informed by fewer patient observations, and long-term trajectory estimates should be interpreted with caution.

    “As such, the model’s behaviour at later time points is informed by fewer patient observations, and long-term trajectory estimates should therefore be interpreted with caution as they may reflect mathematical extrapolation rather than directly observed decline.”
    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 conducted sensitivity analyses restricting the dataset to observations within ≤1 year, ≤2 years and ≤3 years of ALS diagnosis.

    “To assess the robustness of our findings across different follow-up durations, we conducted sensitivity analyses by restricting the dataset to observations within ≤1 year, ≤2 years and ≤3 years of ALS diagnosis.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductcitation

    The ALSFRS-R is a validated clinical rating scale that accurately tracks functional decline in patients with ALS.

    “This is a validated clinical rating scale that has shown to accurately track functional decline in patients with ALS, and it is the most used marker of disease progression used in clinical trials.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [10] source

  • May combine two claims.

    Study design and conductreasoning

    The observational nature of the study precludes causal inference, and reverse causation cannot be entirely ruled out.

    “Finally, the observational nature of this study precludes us from making causal inference, and thus reverse causation cannot be entirely ruled out.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Environmental and occupational exposures, particularly herbicides, metal dust/fumes and oil-based paints, were associated with faster ALS progression, and head injury was associated with overall worse function.

    “Environmental and occupational exposures, particularly herbicides, metal dust/fumes and oil-based paints, were associated with faster ALS progression, and head injury was associated with overall worse function.”
    • Also stated in: § Conclusion · plain
      “This study provides evidence that prior to ALS diagnosis, environmental and occupational exposures to metal dust/fumes, herbicides and oil-based paints were all associated with a rapid functional decline in persons with ALS, while head trauma was associated with a more severe disease course.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Self-reported exposure and head injury histories accurately classify the exposures of interest.

  • May combine two claims.

    Interpretationown data

    The effects of exposures are independent associations that together suggest a cumulative exposure burden.

    “These effects are independent associations that together suggest a cumulative exposure burden.”
    • Also stated in: § Discussion · hedged
      “The effects were independent associations that together suggest a cumulative exposure burden following a diagnosis of ALS experienced a more aggressive disease course when compared with those with single exposures or with no exposure history.”
    • Also stated in: § Conclusion · hedged
      “The independent nature of these effects suggests that cumulative exposure burden is a potentially important modifier of later disease trajectory.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Self-reported exposure and head injury histories accurately classify the exposures of interest.

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

    Study design and conductnone given

    The study was approved by the University of Florida's Internal Review Board under approval number IRB202500452.

    “This study was approved by the University of Florida’s Internal Review Board under the approval number IRB202500452.”
    Quote in the paperNot a substantial claim

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

    Study design and conductown data

    The dataset does not contain information on the intensity, timing or cumulative dose of each exposure.

    “This data set does not contain information on the intensity, timing or cumulative dose of each exposure.”
    Quote in the paperNot a substantial claim
30 of 30

  • The paper relies on this without support.

    Self-reported exposure and head injury histories accurately classify the exposures of interest.

    “Information on toxic exposures was obtained through questionnaires that evaluated for lifetime exposure to different types of toxic substances prior to disease onset and history of head trauma during childhood.”
    “First, the reliance on self-reported exposure history introduces the possibility of recall bias.”
1 of 1

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Consistentt(8616) = 9.515reported p < .001 · recomputed p < .001

    “t(8616)=9.515, p < .001”

  • Consistentt(8616) = 5.854reported p < .001 · recomputed p < .001

    “t(8616)=5.854, p < .001”

  • Consistentt(8616) = 2.466reported p = .014 · recomputed p = .014

    “t(8616)=2.466, p = .014”

  • Consistentt(8616) = 2.326reported p = .02 · recomputed p = .020

    “t(8616)=2.326, p = .020”

4 of 4

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Self-check results appear here once an independent check has confirmed them.

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Martinez-Nunez, A. E., Guo, J., Dorsey, E. R., Ruffing, K. W., Wymer, J., & Okun, M. S.
Affiliations
  • University of Florida (US)
  • University of Rochester Medicine (US)
ORCID
Venue
BMJ Neurology Open
Publisher
BMJ
Published
2026-01
Volume
8
Issue
1
Pages
e001574
ISSN
2632-6140 (ISSN-L)
DOI
10.1136/bmjno-2026-001574
PMC
PMC13140978
OpenAlex
W7158246173
Semantic Scholar
d2a72d691c4662412e4bce0fc389370a459634a6
Cited by
1
References
31
Citations by year
2026: 1
FWCI
4.10
Type
journal article
Language
en
Open access
Licence
Locations
  • BMJ Neurology Open (publishedVersion, cc-by-nc)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • PubMed Central (submittedVersion, cc-by-nc)
Topics
  • Health Sciences › Medicine › Neurology › Amyotrophic Lateral Sclerosis Research
  • Health Sciences › Medicine › Genetics › Neurogenetic and Muscular Disorders Research
  • Health Sciences › Medicine › Physiology › Voice and Speech Disorders
Keywords
amyotrophic lateral sclerosis, traumatic brain injury, environmental exposure, occupational exposure, herbicides, oil-based paints, disease progression, functional decline, ALSFRS-R, National ALS Registry

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 2026-10-18.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-18.

Cited sources are checked by gpt-6-sol.

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