CoChat
Sign Up
Cited by 105Smoking Behavior and CessationGlobal Cancer Incidence and ScreeningChronic Obstructive Pulmonary Disease (COPD) Research

Estimated Worldwide Mortality Attributed to Secondhand Tobacco Smoke Exposure, 1990-2016

Open access
Yousuf, Hamza, Hofstra, Martijn, Tijssen, Jan, Leenen, Brian, Lindemans, Jan Willem, van Rossum, Albert, Narula, Jagat, Hofstra, Leonard·2020·JAMA Network Open
DOI 10.1001/jamanetworkopen.2020.1177PMC PMC7078760
Open this paper in CoChat — free accountOpen PDF

Abstract

Importance: The World Health Organization estimates that the 1 billion individuals who smoke worldwide contribute to the 880 000 secondhand smoke (SHS)-related deaths among individuals who do not smoke each year. A better understanding of the scale of harm of SHS to those who do not smoke could increase awareness of the consequences of smoking and help to design measures to protect individuals who do not smoke, especially children. Objective: To calculate the number of individuals who smoke associated with the death of 1 individual who died of SHS exposure both on a global scale and in various World Bank regions. Design, Setting, and Participants: In this cross-sectional epidemiologic assessment, data from Our World in Data were used to tabulate the number of individuals who smoke in each country and number of premature deaths related to SHS in that country from 1990 to 2016. The mean number of cigarettes consumed in all countries was also included in analyses. Data were collected for the following World Bank regions: North America, Latin America and the Caribbean, Europe and Central Asia, the Middle East and North Africa, sub-Saharan Africa, South Asia, and East Asia and the Pacific from 1990 and 2016. Statistical analysis was conducted in July 2019. Exposure: Secondhand smoke. Main Outcomes and Measures: The pack-year index, calculated as the number of pack-years associated with the death of 1 individual who does not smoke but was exposed to SHS, and the SHS index, calculated as the number of individuals who smoked for 24 years (ie, the mean duration of smoking) associated with the death of 1 individual who does not smoke. Results: Globally, the SHS index changed favorably, from 31.3 (95% CI, 30.6-32.0) individuals who smoked associated with the death of 1 individual who did not smoke in 1990 to 52.3 (95% CI, 51.2-53.5) individuals who smoked in 2016. There was a wide regional variation in the 2016 secondhand smoke index, from 42.6 (95% CI, 41.6-43.5) individuals who smoked in the Middle East and North Africa to 85.7 (95% CI, 83.8-87.7) individuals who smoked in North America. Worldwide, the pack-year index also changed favorably from 751.9 (95% CI, 736.3-770.7) pack-years associated with 1 death in 1990 to 1255.9 (95% CI, 1227.2-1284.4) pack-years in 2016. Conclusions and Relevance: In this study, the substantial disparity among regions in both the SHS index and pack-year index reflected large differences in the scale of the harm of SHS on those who do not smoke. This information may help local policy makers implement measures to better protect those who do not smoke and increase public engagement. Although the number of pack-years and the number of individuals who smoke associated with the death of 1 individual who did not smoke favorably changed over the study period, as of 2016, 52.3 individuals who smoked were associated with the death of 1 individual who did not smoke.

AI snapshot

Written from the paper’s full text.

TL;DR

This cross-sectional epidemiologic analysis used Our World in Data to estimate how many individuals who smoke are associated with the death of 1 individual who does not smoke but is exposed to secondhand smoke (SHS). Globally, the secondhand smoke index (SHSI) increased from 31.3 individuals who smoked in 1990 to 52.3 in 2016, while the pack-year index (PYI) rose from 751.9 to 1255.9 pack-years.

Key findings

  • The global SHSI increased from 31.3 (95% CI, 30.6-32.0) individuals who smoked in 1990 to 52.3 (95% CI, 51.2-53.5) in 2016, based on a mean smoking duration of 24 years.
  • The global PYI increased from 751.9 (95% CI, 736.3-770.7) pack-years in 1990 to 1255.9 (95% CI, 1227.2-1284.4) pack-years in 2016.
  • The absolute number of individuals who smoked rose from 853,619,292 in 1990 to 987,985,450 in 2016, while estimated SHS deaths decreased from 946,041 in 1990 to approximately 848,702 in 2006 and then increased to 883,930 in 2016.
  • Regional SHSI varied substantially: North America increased from 72.5 (95% CI, 71.0-74.3) in 1990 to 85.7 (95% CI, 83.8-87.7) in 2016; South Asia rose from 38.3 (95% CI, 37.5-39.3) to 70.6 (95% CI, 69.0-72.2); Sub-Saharan Africa rose from 35.4 (95% CI, 34.6-36.2) to 56.6 (95% CI, 55.3-57.9); Middle East and North Africa was 42.6 (95% CI, 41.6-43.5) in 2016.

Methods

  • Cross-sectional epidemiologic assessment using Our World in Data, which combines data from sources including the WHO Global Health Observatory, International Mortality and Smoking Statistics, and the Institute of Health Metrics and Evaluation Global Burden of Disease database.
  • Calculated the SHSI (number of individuals who smoke associated with the death of 1 individual who does not smoke) and the PYI (pack-years associated with the death of 1 individual who does not smoke), using 0-, 2-, 10-, and 20-year lead-time delays; linear interpolation was used for SHS death data available in 5-year clusters.
  • Descriptive analyses were performed in R; no statistical significance threshold was set. Ethical approval was provided by the medical ethics committee of VU University Medical Center, and informed consent was neither required nor possible because preexisting open-access data sets were used.

Limitations

  • The paper states that data collected from diverse world regions vary in quality and measure, and that the data sets supplied only the number of individuals who smoked in a given year and the number who died of SHS exposure, requiring correction for the validity of the association.
  • The paper notes uncertainty about the delay between SHS exposure and disease occurrence, although applying 0-, 2-, 10-, and 20-year delays did not substantially change the PYI or SHSI; it also states that lung cancer plays only a minor role among SHS-related causes of death.
  • Evident from the text: the analysis relies on modeled, population-level estimates from OWID/GBD and is ecological and cross-sectional, so it cannot establish individual-level causation; the SHSI and PYI are derived aggregate indices rather than direct measurements of exposure or harm.

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

Integrity signals

Registry

Crossref · 10.1001/jamanetworkopen.2020.1177

Cites

11 of 11 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

2 of 20 cited sources checked · 0 supported · 0 flagged · accuracy not yet measured

  • Premisecitation

    In many countries, there is still a 40% level of SHS exposure among children.

    “Studies from Malaysia and Bangladesh20 have demonstrated that, in many countries, there is still a 40% level of exposure among children.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [20] Mbulo L, Palipudi KM, Andes L,; GATS Collaborative Group Secondhand smoke exposure at home among one billion children in 21 countries: findings from the Global Adult Tobacco Survey (GATS). Tob Control. 2016; 25(e2): e95- e100. doi: 10.1136/tobaccocontrol-2015-052693 26869598 PMC5488799 source

      The Results report that 48.7% of children under 15 across the 21 countries studied were exposed to SHS at home, supporting the claim that exposure remains widespread but not its stated 40% figure. The Introduction cites a separate estimate of 40% worldwide exposure in any environment; this paper does not present itself as a study from Malaysia and Bangladesh. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
  • Findingown data

    Even after the use of variable time intervals, the PYI and SHSI did not substantially change.

    “Even after the use of variable time intervals, the PYI and SHSI did not substantially change.”
    Quote in the paperSubstantial claim
  • Findingcitation

    Mean daily cigarette consumption varied significantly across time, regions, and countries.

    “Mean daily cigarette consumption varied significantly across time, regions, and countries.”
    Quote in the paperNot a substantial claim
  • Findingown data

    Globally, the secondhand smoke index (SHSI), the number of individuals who smoked associated with the death of 1 individual who did not smoke, increased from 31.3 in 1990 to 52.3 in 2016.

    “This cross-sectional epidemiologic assessment used Our World in Data to calculate how many individuals who smoked for a mean of 24 years were associated with the death of 1 individual who died of exposure to secondhand smoke; globally, this changed from 31.3 individuals who smoked in 1990 to 52.3 individuals who smoked in 2016.”
    • Also stated in: § The SHSI · plain
      “Worldwide, a gradual change in the number of individuals who smoked associated with the death of 1 individual who did not smoke was observed, decreasing from 31.3 (95% CI, 30.6-32.0) individuals who smoked in 1990 to 52.3 (95% CI, 51.2-53.5) individuals who smoked in 2016 (Figure 2B).”
    • Also stated in: § Conclusions · plain
      “In addition, we calculated the number of individuals who smoked for 24 years associated with the death of 1 individual who does not smoke; this increased from 31.3 in 1990 to 52.3 in 2016.”
    Quote in the paperSubstantial claim
  • Findingown data

    In regions where smoking bans have been widely adopted, such as North America, the SHSI was more favorable than in areas with minimal or no protection from SHS, such as the Middle East and North Africa.

    “We observed that in regions where smoking bans have been widely adopted, such as North America, the SHSI was more favorable than in areas with minimal or no protection from SHS, such as the Middle East and North Africa.”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim. 14
  • May combine two claims. 8

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

    Premisecitation

    The two measures most associated with reductions in tobacco use are increasing taxes and discussing SHS exposure, especially among children.

    “The evaluation of antitobacco legislation and campaigns has shown that the 2 measures that are most associated with reductions in tobacco use are increasing taxes and discussing SHS exposure, especially among children.17,18”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites [17] Wilkinson AL, Scollo MM, Wakefield MA, Spittal MJ, Chaloupka FJ, Durkin SJ Smoking prevalence following tobacco tax increases in Australia between 2001 and 2017: an interrupted time-series analysis. Lancet Public Health. 2019; 4(12): e618- e627. doi: 10.1016/S2468-2667(19)30203-8 31759897 source

      The Results and Discussion report reductions in Australian smoking prevalence associated with tobacco tax increases. The paper does not establish that taxes and discussion of secondhand-smoke exposure are the two measures most associated with reduced tobacco use; the passages retrieved for secondhand-smoke exposure, including exposure among children, do not address that measure. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites [18] Schar EH, Gutierrez KK; World Health Organization Smoking cessation media campaigns from around the world. Accessed February 5, 2020. https://apps.who.int/iris/handle/10665/108529

      Not checked in the free run.

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

    Findingown data

    The mean number of cigarettes consumed by an individual who smokes daily decreased from 25 to 18 cigarettes per day between 1990 and 2016.

    “The mean number of cigarettes consumed by an individual who smokes daily decreased from 25 to 18 cigarettes per day between 1990 and 2016 (Figure 1C).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The SHSI showed substantial changes across World Bank regions during the study period; for example, South Asia increased from 38.3 in 1990 to 70.6 in 2016.

    “The data showed substantial changes in the SHSI during the study period; the index was low in 1990 in South Asia (38.3 [95% CI, 37.5-39.3] individuals who smoked) and significantly improved in 2016 (70.6 [95% CI, 69.0-72.2] individuals who smoked).”
    Quote in the paperNot a substantial claim

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

    Findingown data

    The absolute number of people who smoked worldwide increased from 853,619,292 in 1990 to 987,985,450 in 2016.

    “In 1990, 853 619 292 individuals smoked, and in 2016, 987 985 450 individuals smoked.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The estimated absolute number of individuals who died of SHS exposure decreased from 946,041 in 1990 to approximately 848,702 in 2006, then gradually increased to 883,930 in 2016.

    “The estimated absolute number of individuals who died of SHS exposure had decreased from 946 041 in 1990 to approximately 848 702 in 2006, but it started to gradually increase thereafter to 883 930 in 2016 (Figure 1B).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The absolute number of people who smoked between 1990 and 2016 increased, mostly owing to the increase in smoking in low- to middle-income countries, especially China and India.

    “The absolute number of people who smoked between 1990 and 2016 has increased, mostly owing to the increase in smoking in low- to middle-income countries, especially China and India (Figure 1A).”
    Quote in the paperNot a substantial claim

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

    Study design and conductnone given

    Ethical approval for the investigation was provided by the medical ethics committee of the VU University Medical Center.

    “Ethical approval for our investigation was provided by the medical ethics committee of the VU University Medical Center.”
    Quote in the paperNot a substantial claim

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

    Study design and conductnone given

    The report follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

    “Our report follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    It has been reported that 90% of individuals who smoke daily begin before the age of 25 years and that the mean duration of daily smoking is 24 years.

    “It has been reported that 90% of individuals who smoke daily begin before the age of 25 years and that the mean duration of daily smoking is 24 years.14”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [14] Centraal Bureau Statistiek Gezondheidsenquête vanaf 2014. Accessed February 11, 2020. https://www.cbs.nl/nl-nl/onze-diensten/methoden/onderzoeksomschrijvingen/korte-onderzoeksbeschrijvingen/gezondheidsenquete-vanaf-2014

      Not checked in the free run.

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

    Premisecitation

    The Global Burden of Disease report has used a definition from a meta-analysis of 19 studies, largely based on comparative risk assessment approaches in which the proportion of disease-specific burden associated with SHS exposure was calculated.

    “In addition, the Global Burden of Disease report has used the definition from a meta-analysis of 19 studies,2 largely based on comparative risk assessment approaches in which the proportion of disease-specific burden associated with SHS exposure was calculated.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [2] Öberg M, Jaakkola MS, Woodward A, Peruga A, Prüss-Ustün A Worldwide burden of disease from exposure to second-hand smoke: a retrospective analysis of data from 192 countries. Lancet. 2011; 377(9760):-. doi: 10.1016/S0140-6736(10)61388-8 21112082 source

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

    Study design and conductown data

    Because only descriptive statistics were calculated, no threshold of statistical significance was set.

    “Because only descriptive statistics were calculated, no threshold of statistical significance was set.”
    Quote in the paperNot a substantial claim

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

    Premisecitation

    The OWID database comprises a combination and interpretation of numerous existing high-quality data sets, including the World Health Organization Global Health Observatory Data Repository, the International Mortality and Smoking Statistics, and the Institute of Health Metrics and Evaluation Global Burden of Disease database.

    “The OWID database comprises a combination and interpretation of numerous existing high-quality data sets, including the World Health Organization Global Health Observatory Data Repository,10 the International Mortality and Smoking Statistics,11 and the Institute of Health Metrics and Evaluation Global Burden of Disease database.12”
    Quote in the paperNot a substantial claim0 of 4 sources checked
    • cites [3] Roser M, Ritchie H Smoking. Accessed February 5, 2020. https://ourworldindata.org/smoking

      Not checked in the free run.

    • cites [10] World Health Organization Mortality and global health estimates. Accessed February 5, 2020. http://apps.who.int/gho/data/node.main.686?lang=en

      Not checked in the free run.

    • cites [11] P N Lee Statistics and Computing Ltd IMASS: International Mortality and Smoking Statistics. Accessed February 5, 2020. http://www.pnlee.co.uk/imass.htm

      Not checked in the free run.

    • cites [12] Global Health Data Exchange Global Burden of Disease Study (GBD 2017) burden by risk 1990-2017. Accessed February 5, 2020. http://ghdx.healthdata.org/record/ihme-data/gbd-2017-burden-risk-1990-2017

      Not checked in the free run.

  • May combine two claims.

    Interpretationcitation

    SHS affects those who do not smoke, including fetuses and children, within a short time span.

    “These data suggest that SHS affects those who do not smoke, including fetuses and children, within a short time span.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [2] Öberg M, Jaakkola MS, Woodward A, Peruga A, Prüss-Ustün A Worldwide burden of disease from exposure to second-hand smoke: a retrospective analysis of data from 192 countries. Lancet. 2011; 377(9760):-. doi: 10.1016/S0140-6736(10)61388-8 21112082 source

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

    Study design and conductown data

    Because the data on individuals who died of SHS exposure were only available in clusters of 5 years, linear interpolation was conducted to determine intermediate numbers.

    “Because the data on individuals who died of SHS exposure were only available in clusters of 5 years, linear interpolation was conducted to determine the intermediate numbers of individuals in this group.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationown data

    Larger proportions of the population in countries such as China and India have started smoking, and the number of individuals who actively smoke is likely to increase further.

    “Unfortunately, larger proportions of the population in countries such as China and India have started smoking, and the number of individuals who actively smoke is likely to increase further.”
    Quote in the paperNot a substantial claim

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

    Premisecitation

    The approach of confirming long-term associations based on risk factors can be calculated as the Framingham Risk Score for 10 years or lifetime risk.

    “The approach of confirming long-term associations based on risk factors can be calculated as the Framingham Risk Score for 10 years or lifetime risk.13”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [13] Lloyd-Jones DM, Wilson PWF, Larson MG, Framingham risk score and prediction of lifetime risk for coronary heart disease. Am J Cardiol. 2004; 94(1): 20- 24. doi: 10.1016/j.amjcard.2004.03.023 15219502 source

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

    Study design and conductnone given

    The data collected from diverse world regions vary in quality and measure.

    “The data collected from diverse world regions vary in quality and measure.”
    Quote in the paperNot a substantial claim

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

    Study design and conductown data

    Most data were available from 1990 until 2016, but some data sets only contained data points until 2012, so data were extrapolated to 2016.

    “Most data were available from 1990 until 2016. Some data sets only contained data points until 2012, and we extrapolated the data to 2016 (eAppendix in the Supplement).”
    Quote in the paperNot a substantial claim

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

    Study design and conductcitation

    It is justifiable to allow a 2-year delay in the remaining calculations.

    “Given that studies on the consequences of smoking bans show improvement for up to 2 years, it seems justifiable to allow a 2-year delay in the remaining calculations.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [2] Öberg M, Jaakkola MS, Woodward A, Peruga A, Prüss-Ustün A Worldwide burden of disease from exposure to second-hand smoke: a retrospective analysis of data from 192 countries. Lancet. 2011; 377(9760):-. doi: 10.1016/S0140-6736(10)61388-8 21112082 source
19 of 19

  • The paper relies on this without support.

    The scale of the consequences of smoking on those who do not smoke has not been developed.

    “Although it is widely believed that SHS is associated with extensive harm and the number of individuals who smoke by country and globally has been reported alongside the number of individuals who have died of SHS exposure, the scale of the consequences of smoking on those who do not smoke has not been developed.”
1 of 1

Self-check

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

Each contradiction between two passages listed here was confirmed by an independent check.

  • 1990 cigarette-consumption estimates disagree

    Minor

    One Results section reports 25 cigarettes per day in 1990, while another reports a mean of 24.0 for that year. Rounding 24.0 to a whole cigarette gives 24, not 25.

  • SHSI trend is described as decreasing despite rising values

    Minor

    The Results call the change from 31.3 to 52.3 a decrease, while the Discussion calls the same change an increase.

  • Figure 2 caption excludes the cited 1990 result

    Minor

    The caption dates Figure 2 from 1996, but the Results explicitly attribute a 1990 PYI value to Figure 2A.

  • The stated number-needed-to-harm division is reversed

    Minor

    Deaths divided by smokers gives deaths per smoker, whereas the defined SHSI is smokers per death. The stated division is the reciprocal of the index it is meant to calculate.

4 of 4

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Yousuf, H., Hofstra, M., Tijssen, J., Leenen, B., Lindemans, J. W., van Rossum, A., Narula, J., & Hofstra, L.
Affiliations
  • Amsterdam University Medical Centers (NL)
  • Erasmus University Rotterdam (NL)
  • Duke University (US)
  • Department of Cardiology, VU University Medical Center, Amsterdam University Medical Center, Amsterdam, the Netherlands
  • Mount Sinai Hospital (US)
Corresponding author
  • Narula, Jagat
  • Hofstra, Leonard
ORCID
Venue
JAMA Network Open
Publisher
American Medical Association (AMA)
Published
2020-03-17
Volume
3
Issue
3
Pages
e201177
ISSN
2574-3805 (ISSN-L)
DOI
10.1001/jamanetworkopen.2020.1177
PMC
PMC7078760
OpenAlex
W3011036940
Semantic Scholar
ff65c16bb09938ac09e9ab36088c3b227b3aa797
Cited by
105
References
11
Citations by year
2026: 7 · 2025: 11 · 2024: 11 · 2023: 21 · 2022: 30 · 2021: 21 · 2020: 2 · 2018: 1
FWCI
11.15
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JAMA Network Open (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Pure Amsterdam UMC (publishedVersion, cc-by)
  • Amsterdam UMC (VU Amsterdam) - Institutional Repository (publishedVersion, other-oa)
  • EUR Research Repository (Erasmus University Rotterdam) (publishedVersion, other-oa)
  • Data Archiving and Networked Services (DANS) (submittedVersion, other-oa)
  • EUR Research Repository (Erasmus University Rotterdam) (submittedVersion, other-oa)
  • RePub (Erasmus University Rotterdam) (submittedVersion, other-oa)
Topics
  • Health Sciences › Medicine › Physiology › Smoking Behavior and Cessation
  • Health Sciences › Medicine › Oncology › Global Cancer Incidence and Screening
  • Health Sciences › Medicine › Pulmonary and Respiratory Medicine › Chronic Obstructive Pulmonary Disease (COPD) Research
Keywords
passive smoking, pack-years, smoking duration, tobacco control, global burden of disease, cigarette consumption, tobacco control policy evaluation, tobacco taxation
MeSH terms
Adolescent, Adult, Aged, Child, Child, Preschool, Cross-Sectional Studies, Environmental Exposure, Female, Humans, Male, Middle Aged, Mortality, Smoking, Tobacco Smoke Pollution, Global Health, Young Adult, Mortality, Premature, Tobacco Products

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 2026-10-29.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-29.

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