← Back

INTERSTROKE

Tea and coffee consumption and risk of acute stroke: The INTERSTROKE Study

Share Share Copied! Compare

Year of Publication: 2024

Authors: Andrew Smyth, Graeme J Hankey, Peter Langhorne, ..., Martin O’Donnell

Journal: International Journal of Stroke

Citation: Int J Stroke. 2024;19(9):1053–1063. doi:10.1177/17474930241264685

Link: https://doi.org/10.1177/17474930241264685

PDF: https://journals.sagepub.com/doi/epub/10...474930241264685

Clinical Question

Does the consumption of tea or coffee affect the risk of first stroke?

Study Overview

Objective

To examine the associations between tea and coffee consumption and risk of acute stroke using a large, multicenter, matched case-control study.

Study Summary

High coffee intake (>4/day) was associated with higher stroke odds (OR 1.37), lower coffee consumption had no effect on stroke.
Tea intake linked to reduced stroke odds

Intervention

Observational exposure comparison: none, 1–2, 3–4, or >4 cups/day for coffee and each tea type (all tea, black tea, green tea, other tea).

Patients per Arm

Tea only: 12,666; Coffee only: 4024; Both: 5021; None: 5239

Bottom Line

High coffee intake (>4 cups/day) was associated with increased odds of all stroke and ischemic stroke, while tea consumption (including black, green, and other tea types) was associated with reduced odds of stroke.

        Major Points
        INTERSTROKE is a large, international case–control study of 26,950 participants (13,462 cases, 13,488 controls) from 32 countries across 142 centers — the largest study of tea/coffee and stroke risk to date.
High coffee consumption (>4 cups/day) was associated with increased odds of all stroke (OR 1.37, 95% CI 1.06–1.77) and ischemic stroke (OR 1.32, 95% CI 1.00–1.74), contrasting with prospective cohort studies that mostly show coffee as protective — likely reflecting case-control design limitations.
Low-to-moderate coffee intake (1–4 cups/day) showed no significant association with stroke risk, suggesting a threshold effect rather than linear dose-response.
Tea consumption (any type, ≥4 cups/day) was associated with reduced odds of all stroke (OR 0.81) and ischemic stroke, with the strongest effect at 3–4 cups/day — consistent with polyphenol/catechin-mediated endothelial protection.
Black tea intake (3–4 cups/day) showed a striking protective effect against ICH (OR 0.41, 95% CI 0.24–0.70), though based on relatively few ICH events and requiring confirmation.
Green tea (≥4 cups/day) was associated with reduced odds of all stroke (OR 0.70, 95% CI 0.54–0.90), consistent with prior Asian cohort data from the JPHC and Shanghai Women's Health studies.
Marked regional heterogeneity: tea was protective in China and South America but paradoxically increased stroke risk in South Asia — likely reflecting confounding by preparation method, additives (sugar, milk), and co-consumed foods.
Milk addition modified the tea-stroke association: tea without milk was protective, while tea with milk showed attenuated benefit, possibly through casein binding of catechins.
The study is a subanalysis of the landmark INTERSTROKE study (O'Donnell et al., Lancet 2010/2016) which identified 10 modifiable risk factors accounting for 90% of stroke risk globally.
Hypertension modified the coffee-stroke association: the harmful effect of high coffee consumption was more pronounced in hypertensive individuals, consistent with the acute pressor effect of caffeine.

      

Design

Study Type: International matched case–control study

Randomization:

Enrollment Period: March 2007 – July 2015

Centers: 142

Countries: Canada, Australia, Germany, Denmark, Sweden, United Kingdom, Ireland, Croatia, Poland, Turkey, Iran, United Arab Emirates, Russia, Saudi Arabia, China, Argentina, Brazil, Chile, Colombia, Ecuador, Peru, Thailand, Philippines, Malaysia, India, Pakistan, South Africa, Mozambique, Uganda, Sudan, Nigeria

Sample Size: 26950

Analysis: Multivariable conditional logistic regression with adjustment for demographic, clinical, and dietary variables, and beverage interactions

Inclusion Criteria

Adults with first acute stroke (ischemic or hemorrhagic) presenting within 5 days of symptom onset and admitted within 3 days of hospital presentation
Controls without acute stroke, matched to cases for age (<5 years difference, or <10 years if aged >90 years), sex, and community/hospital
Neuroimaging (CT or MRI) performed in 99.9% of cases
Informed consent provided by patient or by a proxy respondent when appropriate
Dietary intake data (tea and coffee consumption) collected via structured questionnaire

Exclusion Criteria

Cases with recurrent stroke (only first-ever stroke events were included)
Controls with hospital referral or diagnosis related to stroke (excluded to reduce selection bias)

Baseline Characteristics

Overall cohort (n=26,950):

Age (mean, SD): 61.7 (13.4)
Female (%): 40.4% (10,894)
Education <8 years: 48.3%
BMI (mean, SD): 25.7 (4.8)
WHR (mean, SD): 0.93 (0.08)
Hypertension: 61.4%
Diabetes: 25.0%
Cardiac risk factors: 9.5%
Myocardial infarction: 3.5%
Atrial fibrillation: 3.2%

Neither tea nor coffee (n=5239):

Age (mean, SD): 61.4 (13.2)
Female (%): 50.0% (2621)
Education <8 years: 61.0%
BMI (mean, SD): 25.0 (4.1)
WHR (mean, SD): 0.91 (0.07)
Hypertension: 58.8%
Diabetes: 16.8%
Cardiac risk factors: 5.4%
Myocardial infarction: 1.5%
Atrial fibrillation: 2.1%

Both tea and coffee (n=5021):

Age (mean, SD): 63.4 (13.6)
Female (%): 41.0% (2057)
Education <8 years: 26.1%
BMI (mean, SD): 27.1 (4.9)
WHR (mean, SD): 0.94 (0.08)
Hypertension: 64.0%
Diabetes: 26.1%
Cardiac risk factors: 15.8%
Myocardial infarction: 6.7%
Atrial fibrillation: 6.1%

Arms

Field	Control	1–2 cups/day	3–4 cups/day	>4 cups/day
Intervention	0 cups/day of the beverage type	1–2 cups/day of the beverage type	3–4 cups/day of the beverage type	>4 cups/day of the beverage type
Duration	Habitual intake (self-reported at single time point)	Habitual intake (self-reported)	Habitual intake (self-reported)	Habitual intake (self-reported)

Outcomes

Outcome	Type	Control	Intervention	HR / OR / RR
Association of tea and coffee consumption with odds of all stroke and ischemic stroke (adjusted ORs from conditional logistic regression)	Primary	None consumption of the beverage (reference)	Coffee >4 cups/day: OR 1.37 (all stroke), OR 1.32 (ischemic stroke), OR 1.66 (ICH, NS)
All tea (>4 cups/day) — all stroke (adjusted)	Secondary	No tea (reference)	OR 0.81	0.81 (95% CI 0.69–0.94)
All tea (>4 cups/day) — ischemic stroke (adjusted)	Secondary	No tea (reference)	OR 0.81	0.81 (95% CI 0.68–0.97)
Black tea (3–4 cups/day) — ICH (adjusted)	Secondary	No black tea (reference)	OR 0.41	0.41 (95% CI 0.24–0.70)
Black tea (3–4 cups/day) — all stroke (adjusted)	Secondary	No black tea (reference)	OR 0.71	0.71 (95% CI 0.58–0.86)
Green tea (>4 cups/day) — all stroke (adjusted)	Secondary	No green tea (reference)	OR 0.70	0.70 (95% CI 0.54–0.90)
Green tea (>4 cups/day) — ischemic stroke (adjusted)	Secondary	No green tea (reference)	OR 0.69	0.69 (95% CI 0.52–0.91)
Other tea (>4 cups/day) — all stroke (adjusted)	Secondary	No other tea (reference)	OR 0.79	0.79 (95% CI 0.64–0.98)
Case-control study	Adverse		Observational case–control study of dietary exposures; no intervention-related adverse event data reported.

Criticisms

Case-control design is inherently susceptible to recall bias — stroke patients may recall dietary habits differently from controls, and hospitalized cases may have altered intake patterns preceding the event.
Self-reported intake measured at a single time point may not reflect long-term cumulative exposure or recent changes in consumption patterns before stroke.
No data on coffee preparation method (filtered vs unfiltered/boiled), roast type, or decaffeination — diterpenes in unfiltered coffee have different cardiovascular effects than filtered.
Limited power for ICH subgroup analyses — the striking black tea OR 0.41 for ICH was based on few events and had wide confidence intervals (0.24–0.70).
Potential for residual confounding despite multivariable adjustment — tea and coffee consumption correlates with socioeconomic status, diet quality, exercise, and healthcare access differently across regions.
Cup volume was assumed at 250 mL across all 32 countries; the paper did not report regional cup-size or preparation-method comparisons, but such variation is a plausible unmeasured source of exposure misclassification.
Contradicts large prospective cohort studies (Nurses' Health Study, Health Professionals Follow-up Study) showing coffee as protective — case-control design and reverse causation may explain this discrepancy.
Regional heterogeneity in tea-stroke associations (protective in China, harmful in South Asia) likely reflects unmeasured confounding by preparation method and additives rather than a true biological interaction.
No biomarker validation of self-reported intake — caffeine metabolites or polyphenol biomarkers would strengthen causal inference but were not measured.

Funding

The INTERSTROKE study was funded by the Canadian Institutes of Health Research, Heart and Stroke Foundation (Canada), Canadian Stroke Network, Swedish Research Council, Swedish Heart and Lung Foundation, Health & Medical Care Committee of the Västra Götaland Regional Executive Board (Sweden), unrestricted grants from AstraZeneca, Boehringer Ingelheim (Canada), Pfizer (Canada), and MSD, Chest, Heart and Stroke Scotland, the Stroke Association, UK Stroke Research Network, German Research Council (DFG), German Ministry of Education and Research (BMBF), European Union, National Institutes of Health, Bertelsmann Foundation, and Heinz-Nixdorf Foundation.

Based on: INTERSTROKE (International Journal of Stroke, 2024)