CoChat
Sign Up
Cited by 144Coffee research and impactsAlcohol Consumption and Health EffectsNutritional Studies and Diet

Does coffee drinking increase the risk of coronary heart disease? Results from a meta‐analysis

Open access
Kawachi, I, Colditz, G A, Stone, C B·1994·Heart
DOI 10.1136/hrt.72.3.269PMC PMC1025516
Open this paper in CoChat — free accountOpen PDF

Abstract

BACKGROUND: The association between coffee drinking and risk of coronary heart disease remains controversial despite many epidemiological studies. A meta-analysis was carried out on these studies to resolve some of the uncertainties. Particular attention was paid to details of study design. METHODS: Eight case-control studies and 15 cohort studies were analysed. Weighted, fixed effects linear regression of log relative risks (or odds ratios) was used to pool the study results. The pooling procedures were performed separately by study design, sex, coronary heart disease end points, smoking habit, and period of study. RESULTS: The pooled case-control odds ratio (for the effect of drinking five cups of coffee/day v none) was 1.63 (95% confidence interval (95% CI) 1.50 to 1.78). The pooled cohort study relative risk (five cups/day v none) was 1.05 (95% CI 0.99 to 1.12). The discrepancy between the pooled case-control and cohort study results could not be attributed to differences in the end points chosen, period of study, or to confounding by smoking status or sex. CONCLUSIONS: The cohort study data suggest very little excess risk of coronary heart disease among habitual coffee drinkers. The case-control data do not rule out an increased risk of heart disease among a subgroup of people who acutely increase their coffee intake. Further epidemiological studies are needed to assess the risk of drinking boiled or decaffeinated coffee.

AI snapshot

Written from the paper’s full text.

TL;DR

This meta-analysis pooled 8 case-control and 15 cohort studies and found that case-control data suggested a positive association between coffee drinking and coronary heart disease, while cohort data suggested little excess risk. The discrepancy persisted across end points, sex, and smoking-restricted analyses.

Key findings

  • The pooled case-control odds ratio for drinking five cups/day versus none was 1.63 (95% CI 1.50 to 1.78); the pooled cohort relative risk was 1.05 (95% CI 0.99 to 1.12).
  • Among non-smokers, the pooled case-control odds ratio was 1.85 (95% CI 1.42 to 2.42), while the pooled cohort relative risk was 1.04 (95% CI 0.71 to 1.52).
  • Cohort analyses by end point gave pooled relative risks of 0.97 (95% CI 0.94 to 1.01) for CHD death, 1.16 (95% CI 1.02 to 1.30) for myocardial infarction, and 1.25 (95% CI 1.08 to 1.46) for total CHD.
  • The pooled odds ratio from three case-control studies of decaffeinated coffee was 1.42 (95% CI 1.01 to 1.99), not substantially different from the pooled odds ratio for caffeinated coffee (1.65, 95% CI 1.52 to 1.81).

Methods

  • Meta-analysis of 8 case-control and 15 cohort studies identified by computer-aided literature search and bibliographic searches of review articles and previous meta-analyses; early case-control studies with insufficient data and prevalence-only studies were excluded.
  • Weighted, fixed-effects linear regression of log relative risks or odds ratios was used, with inverse variance weights; coffee exposure categories were assigned midpoints, and sensitivity analyses used the lowest number of cups in open-ended categories.
  • Pooling was performed separately by study design, sex, CHD end points, smoking habit, and period of study, with additional analyses restricted to non-smokers and to decaffeinated coffee.

Limitations

  • Stated: cohort studies had extremely high heterogeneity (χ² = 85.4, df = 14, P < 0.0001), so the SEM for the pooled coefficients is almost certainly an underestimate; prospective designs may miss acute effects because of lag time between exposure assessment and outcome.
  • Stated: recall bias in case-control studies needs empirical quantification, and no epidemiological studies to date assessed boiled versus filtered coffee; further studies are needed on boiled and decaffeinated coffee.
  • Evident from the text: case-control estimates may be affected by control selection bias (including controls admitted for chronic conditions), recall bias, and residual confounding by smoking; the report also notes support in part by the National Coffee Association.

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

Integrity signals

Registry

Crossref · 10.1136/hrt.72.3.269

Cites

59 of 59 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 42 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    The pooled relative risk for myocardial infarction among cohort studies was 1.16 (95% CI 1.02 to 1.30).

    “The pooled relative risks were 0 97 (95% CI 0'94-1-01) for CHD death, 1-16 (95% CI 1-02 to 1-30) for myocardial infarction, and 1-25 (95% CI 1.08 to 1 46) for total CHD.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: A log linear relation holds between coffee intake and the log relative risk or odds ratio, such that the estimated relative risk for five cups/day is the antilog of five times the coefficient for one cup versus none.
  • Findingown data

    When the cohort data were stratified according to type of end point, there was no evidence to suggest that the effect of coffee drinking was specific to a particular category of CHD.

    “When we stratified the cohort data according to type of end point (table 2) there was no evidence to suggest that the effect of coffee drinking was specific to a particular category of CHD.”
    Quote in the paperSubstantial claim
  • Findingown data

    The pooled odds ratio for case-control studies with myocardial infarction as the end point was 1.66 (95% CI 1.52 to 1.81).

    “The pooled odds ratio for case- control studies with myocardial infarction as the end point was 1i66 (95% CI 1i52 to 1.81);”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: A log linear relation holds between coffee intake and the log relative risk or odds ratio, such that the estimated relative risk for five cups/day is the antilog of five times the coefficient for one cup versus none.
  • Findingown data

    Among cohort studies restricted to non-smokers, there was no suggestion of increased risk (pooled relative risk = 1.04; 95% CI 0.71 to 1.52).

    “Among cohort studies there was no sugges- tion of increased risk (pooled relative risk = 1-04; 95% CI 0-71 to 1-52).”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: A log linear relation holds between coffee intake and the log relative risk or odds ratio, such that the estimated relative risk for five cups/day is the antilog of five times the coefficient for one cup versus none.
  • Findingnone given

    The case-control studies consistently pointed to an excess risk of coronary heart disease.

    “By contrast, the case-control studies consistently pointed to an excess risk.”
    Quote in the paperSubstantial claim
  • The quote only partly states this claim. 2
  • May combine two claims. 15
  • Classified as probably not a substantial claim. 3

  • The quote only partly states this claim.

    Findingown data

    The discrepancy between the results of case-control and cohort studies persisted in analyses by sex.

    “The discrepancy between the results of case-control and cohort studies persisted.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: A log linear relation holds between coffee intake and the log relative risk or odds ratio, such that the estimated relative risk for five cups/day is the antilog of five times the coefficient for one cup versus none.

  • May combine two claims.

    Findingcitation

    Both prospective data by Grobbee et al and a pooled analysis of three case-control studies that included data on decaffeinated coffee suggest a marginally increased risk of coronary heart disease with decaffeinated coffee consumption.

    “suggest a marginally increased risk of CHD with decaffeinated cof- fee consumption.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [41] Murray SS, Bjelke E, Gibson RW, et al. Coffee consumption and mortality from ischemic heart disease and other causes: results from the Lutheran Brotherhood study. AmJEpidemiol 1981;113:661-7.

      Not checked in the free run.

  • May combine two claims.

    Findingcitation

    In 1975, 75% of Americans drank percolated coffee and only 20% filtered, whereas by 1985, 25% drank percolated coffee and 66% filtered.

    “In 1975, 75% of Americans drank percolated coffee and only 20% filtered, whereas by 1985, 25% drank percolated cof- fee and 66% filtered (International Coffee Organization, 1989-quoted by Bak54).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Bak54

      Not checked in the free run.

  • May combine two claims.

    Findingcitation

    No studies to date have shown a positive association between tea intake and coronary heart disease, even though tea contains caffeine (64 mg/cup of tea, compared with 136 mg/cup for coffee).

    “No studies to date have shown a positive association between tea intake and CHD even though tea contains caffeine (64 mg/cup of tea, compared with 136 mg/cup for coffee).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites l30 3639

      Not checked in the free run.

  • May combine two claims.

    Findingcitation

    Regular ingestion of caffeine does not seem to lead to an increase in blood pressure, although infrequent use among people who have not developed tolerance does seem to cause a small, but clinically unimportant, rise in blood pressure.

    “Nor does regular ingestion of caffeine seem to lead to an increase in blood pressure (although infre- quent use among people who have not devel- oped tolerance does seem to cause a small, but clinically unimportant rise in blood pres- sure).354”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [354]

      Not checked in the free run.

  • May combine two claims.

    Findingcitation

    In the Norwegian study by Tverdal et al, relative risks of coronary heart disease death (comparing more than nine vs less than one cup/day) were 2.2 (95% CI 1.1 to 4.5) for men and 5.1 (95% CI 0.4 to 60.3) for women.

    “The Norwegian study by Tverdal et al found relative risks of CHD death (comparing more than nine v less than one cup/day) of 2-2 (95% CI 1 1 to 4 5) for men, and 5-1 (95% CI 0 4 to 60 3) for women.42”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [42] Tverdal A, Stensvold I, Solvoll K, et al. Coffee consumption and death from coronary heart disease in middleaged Norwegian men and women. BMJ 1990;300: 566-9. source

      Not checked in the free run.

  • May combine two claims.

    Findingcitation

    In a meta-analysis of 21 cross sectional studies, Bak found that filtered coffee was associated with an average increase in total cholesterol of 0.008 mmol/l per daily cup consumed, compared with an increase of 0.038 mmol/l for boiled (unfiltered) coffee.

    “In a meta-analysis of 21 cross sectional studies, Bak found that filtered coffee was associated with an average increase in total cholesterol of 0-008 mmol/ldaily cup consumed, compared with an increase of 0-038 mmol/l for boiled (unfiltered) coffee.54”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [54] Bak AAA. Coffee and cardiovascular risk: an epidemiological study. Rotterdam: Erasmus University, 1990. (PhD thesis.)

      Not checked in the free run.

  • The quote only partly states this claim.

    Findingown data

    Under sensitivity analysis with conservative assignment of cups/day to open-ended categories, the pooled case-control odds ratio for five cups/day vs none was 1.66 (95% CI 1.52 to 1.81).

    “OR * 1-66 (1-52 to 1-81)”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    These endpoint-specific results suggest a possible slight increase in risk of coronary heart disease, although the magnitude of the effect is such that confounding, for example by smoking, could not be ruled out.

    “These results suggest a possible slight increase in risk of coronary heart disease, although the magni- tude of the effect is such that confoundingfor example, by smoking-could not be ruled out.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: A log linear relation holds between coffee intake and the log relative risk or odds ratio, such that the estimated relative risk for five cups/day is the antilog of five times the coefficient for one cup versus none.

  • May combine two claims.

    Interpretationown data

    Failure to control for smoking cannot entirely account for the discrepancy between the pooled case-control and cohort study results, as analyses restricted to non-smokers continued to show excess risk among case-control studies (OR = 1.85) but no increased risk among cohort studies (RR = 1.04).

    “Alternatively, failure to control for smoking cannot entirely account for the discrepancy between the pooled case-control and cohort study results, as the analyses restricted to data on non-smokers continued to show an excess risk among case-control studies (pooled odds ratio = 1i85, 95% CI 1-42 to 2.42) but no increased risk among cohort studies (pooled risk ratio = 1-04, 95% CI 0-71 to 1-52).”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: A log linear relation holds between coffee intake and the log relative risk or odds ratio, such that the estimated relative risk for five cups/day is the antilog of five times the coefficient for one cup versus none.

  • May combine two claims.

    Interpretationreasoning

    This meta-analysis supersedes previous studies because it includes case-control studies, pools data from a more comprehensive set of studies, and stratifies the analyses by disease end point, smoking, and sex.

    “Our meta-analysis supersedes previous stud- ies12 14 because it includes case-control studies (omitted in the study of Myers and Basinski12); pools data from a more compre- hensive set of studies (15 cohort studies compared with 11 analysed in Myers and Basinski,12 and five by Greenland14); and stratifies the analyses by disease end point, smoking, and sex.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [12] Myers MG, Basinski A. Coffee and coronary heart disease. Arch Intern Med 1992;152:1767-72. source

      Not checked in the free run.

    • cites [14] Greenland S. Quantitative methods in review of epidemiologic literature. Epidemiol Rev 1987;9: 1-30. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Interpretationreasoning

    In theory, a prospective study design is more capable of making reliable assessments of both coffee intake and smoking habits.

    “In theory, a prospective study design is more capable of making reli- able assessments of both coffee intake and smoking habits.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    In a meta-analysis of 11 cohort studies, Myers and Basinski reported relative risks of 1.01 among drinkers of one to four cups/day, 1.01 for four to six cups/day, and 1.09 for six or more cups/day, with drinkers of less than one cup/day as the reference category.

    “In their meta-analysis of 11 cohort studies Myers and Basinski reported relative risks of 1 01 among drinkers of one to four cups/day, 1-01 for four to six cups/day, and 1-09 for six or more cups/day (with drinkers of less than one cup/day as the refer- ence category).'2”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [12] Myers MG, Basinski A. Coffee and coronary heart disease. Arch Intern Med 1992;152:1767-72. source

      Not checked in the free run.

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

    Interpretationcitation

    The increased risk with decaffeinated coffee may be attributable to an adverse effect of decaffeinated coffee on serum lipids.

    “which may be attributable to an adverse effect of decaffeinated coffee on serum lipids.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    The secular trends in method of coffee preparation may be consistent with the generally positive associations between coffee intake and heart disease found in the United States case-control studies, all but one of which were completed during the 1970s.

    “The secular trends in method of coffee preparation may be consistent with the gener- ally positive associations between coffee intake and heart disease found in the United States case-control studies, all but one of which were completed during the 1970s-”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Grobbee et al found no association between coffee intake and risk of heart disease, even though the mean lag time between exposure assessment and outcome was less than two years.

    “Grobbee et al however, found no association between coffee intake and risk of heart disease, even though the mean lag time between exposure assessment and outcome was less than two years.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [11] Grobbee DE, Rimm EB, Giovannucci E, et al. Coffee, caffeine, and cardiovascular disease in men. N Engi Med 1990;323: 1026-32. source

      Not checked in the free run.

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

    Interpretationnone given

    It is possible that cohort studies may have missed an acute effect of coffee drinking on coronary heart disease risk, although the evidence for this hypothesis is sparse and inconsistent.

    “It is possible that cohort studies may have missed an acute effect of coffee drinking on CHD risk, although the evidence for this hypothesis is sparse and inconsistent.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationcitation

    The higher risk estimates from individual case-control studies may have resulted from a combination of bias in control selection, recall bias, or confounding by cigarette smoking.

    “The higher risk estimates from individual case- control studies may have resulted from a combination of bias in control selection30 35; recall bias; or be confounded by cigarette smoking3033 35;”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [30] Jick H, Miettinen OS, Neff RK, et at. Coffee and myocardial infarction. N EnglJ Med 1973;289 63-7.

      Not checked in the free run.

    • cites [35] Boston collaborative drug surveillance program. Coffee drinking and acute myocardial infarction. Lancet 1972: fi: 1278-81.

      Not checked in the free run.

    • cites [33] Mann JI, Thorogood M. Coffee-drinking and myocardial infarction. Lancet 1975;ii:1215. source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Percolation involves boiling coffee, which has been shown to have adverse effects on serum lipids.

    “Percolation involves boiling coffee, which has been shown to have adverse effects on serum lipids.54”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [54] Bak AAA. Coffee and cardiovascular risk: an epidemiological study. Rotterdam: Erasmus University, 1990. (PhD thesis.)

      Not checked in the free run.

19 of 19

  • The authors state this as an assumption.

    A log linear relation holds between coffee intake and the log relative risk or odds ratio, such that the estimated relative risk for five cups/day is the antilog of five times the coefficient for one cup versus none.

    “A log linear assumption was made, in which the estimated relative risk for five cups/day is the antilog of five times the coefficient for one cup (v none) shown on appendices to tables 1 and 2.”
1 of 1

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?

  • Previous meta-analyses have consistently reported positive associations between habitual consumption of caffeine-containing foods and beverages, mostly coffee, and the risk of cardiovascular disease from case-control studies, but not from prospective cohort studies.

    “Previous meta-analyses have consistently reported positive associations between habitual consumption of caffeine-containing foods and beverages (mostly coffee) and the risk of CVD from case-control studies, but not from prospective cohort studies (Greenland, 1993;Kawachi et al., 1994;Sofi et al., 2007).”

    From Scientific Opinion on the safety of caffeine (2015) · cites this paper as Kawachi et al., 1994

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.

  • Sensitivity-analysis footnote conflicts with Table 1 comparison

    Minor

    Table 1 describes its pooled estimates as comparing five cups per day with none, but its footnote describes the starred sensitivity estimates as comparing one cup per day with none.

1 of 1

Cites

Flagged first, then by citations

5 of the 64 references couldn’t be matched to a known paper.

Research works printed but not linked: 6

  1. Coffee and cholesterol. Lancet 1985;i:1283-4.
  2. Yano K, Reed DM, MacLean CJ. Letter to the editor. N Engl_ Med 1987;316:946.
  3. Boston collaborative drug surveillance program. Coffee drinking and acute myocardial infarction. Lancet 1972: fi: 1278-81.
  4. Heyden S, Tyroler HA, Heiss G, et al. Coffee consumption and total mortality. Arch Intern Med 1978-138:1472-5.
  5. Jacobsen BK, Bjelke E, Kvale G, Heuch I. Coffee drinking, mortality, and cancer incidence: results from a Norwegian prospective study. Nad Cancer Inst 1986; 76:823-31.
  6. Superko HR, Bortz W, Albers JJ, Wood PD. Lipoprotein and apolipoprotein changes during a controlled trial of caffeinated and decaffeinated coffee drinking in men. Circulation 1989;80 (suppl II):II-86.

Cited by

Sorted by citations

Related

Authors
Kawachi, I., Colditz, G. A., & Stone, C. B.
Affiliations
Harvard University (US)
ORCID
Venue
Heart
Publisher
BMJ
Published
1994-09-01
Volume
72
Issue
3
Pages
269-275
ISSN
1355-6037 (ISSN-L)
DOI
10.1136/hrt.72.3.269
PMC
PMC1025516
OpenAlex
W2165557546
Semantic Scholar
2e2ee12eea427412ed987a0f2a277234eddb2566
Cited by
144
References
64
Citations by year
2022: 4 · 2021: 1 · 2020: 2 · 2019: 2 · 2018: 1 · 2017: 3 · 2016: 4 · 2015: 3 · 2014: 6 · 2013: 9 · 2012: 10
FWCI
0.72
Type
journal article
Language
en
Open access
Locations
  • Heart (publishedVersion)
  • PubMed Central (publishedVersion)
  • PubMed (publishedVersion)
Topics
  • Health Sciences › Medicine › Pharmacology › Coffee research and impacts
  • Health Sciences › Medicine › Pathology and Forensic Medicine › Alcohol Consumption and Health Effects
  • Health Sciences › Medicine › Public Health, Environmental and Occupational Health › Nutritional Studies and Diet
Keywords
coffee consumption, coronary artery disease, cardiovascular risk, case-control study, cohort studies, study design bias, smoking confounding, decaffeinated coffee, boiled coffee
MeSH terms
Coffee, Coronary Disease, Female, Humans, Male, Risk Factors, Cohort Studies, Odds Ratio, Case-Control Studies

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