Coffee, Tea, and Your Brain: What 43 Years of Data Tells Us
Source: Zhang et al., JAMA, February 2026
Link: https://jamanetwork.com/journals/jama/fullarticle/2826569
Executive Summary (Your Dinner Table Pitch)
Drinking 2-3 cups of caffeinated coffee or 1-2 cups of tea daily was associated with an 18% lower risk of dementia in a massive 43-year study of over 131,000 health professionals—but decaf showed no benefit, suggesting caffeine itself drives the protective effect. The sweet spot appears to be moderate consumption; drinking more didn't provide additional benefits, and the associations held up regardless of genetic risk factors like APOE4 status. While this observational study can't prove causation, the consistency across multiple cognitive measures (diagnosed dementia, self-reported decline, and objective testing) and the biological plausibility make this one of the strongest pieces of evidence yet that your morning coffee habit might be doing your brain a favor.
Authors & Affiliations
Lead/Corresponding Author:
- Dong D. Wang, MD, ScD (Brigham and Women's Hospital, Harvard Medical School)
Key Institutions:
- Harvard T.H. Chan School of Public Health (Department of Nutrition, Department of Epidemiology)
- Brigham and Women's Hospital (Channing Division of Network Medicine)
- Broad Institute
First Authors (co-equal contribution):
- Yu Zhang, MBBS
- Yuxi Liu, PhD
Senior Authors:
- Frank B. Hu, MD, PhD (major figure in nutritional epidemiology)
- Meir J. Stampfer, MD, DrPH
- Walter C. Willett, MD, DrPH (legendary nutritional epidemiologist)
Conflicts of Interest
Disclosed Conflicts:
- Dr. Hu: Received research support from Analysis Group (consulting firm that works with pharmaceutical/food companies)
- All others: No conflicts reported
Assessment:
- Minimal concerning conflicts for a study of this scope
- No direct coffee/tea industry funding reported
- NIH-funded study (grants listed: UM1 CA186107, U01 HL145386, and others)
The Data: What They Actually Did
Study Design:
- Prospective cohort study using Nurses' Health Study (NHS, 86,606 women starting 1980) and Health Professionals Follow-up Study (HPFS, 45,215 men starting 1986)
- Follow-up: Up to 43 years (median 36.8 years) — this is genuinely impressive
- Dietary assessment: Validated food frequency questionnaires every 2-4 years (not just one snapshot)
- 11,033 dementia cases identified during follow-up
What They Measured:
- Primary outcome: Incident dementia (from death certificates and physician diagnoses)
- Secondary outcomes:
- Subjective cognitive decline (self-reported, 6-7 item questionnaire)
- Objective cognitive testing (phone-based tests, NHS cohort only)
Key Findings:
- Caffeinated coffee: Highest quartile (median 4.5 cups/day for women, 2.5 for men) vs. lowest had 18% lower dementia risk (HR 0.82, 95% CI 0.76-0.89)
- Tea: Similar protective effect (14% lower risk in highest vs. lowest tertile)
- Decaf coffee: No association with dementia risk
- Dose-response: Nonlinear relationship—benefits plateaued around 2-3 cups coffee or 1-2 cups tea daily; more wasn't better
- Caffeine intake: ~300mg/day associated with lowest risk (that's about 2-3 cups of coffee)
Strengths: Why This Study Matters
- Duration is exceptional: 43 years of follow-up is rare in nutritional epidemiology; most studies are 10-15 years, which may miss late-onset dementia
- Repeated dietary measures: Updating coffee/tea intake every 2-4 years captures changes in habits rather than relying on a single baseline assessment—huge advantage over most studies
- Multiple cognitive outcomes assessed: Diagnosed dementia, self-reported decline, AND objective phone-based testing creates a comprehensive picture across the dementia continuum
- Large sample with excellent follow-up: Over 131,000 participants with >98% mortality ascertainment is methodologically stellar
- Validation using biomarkers: They confirmed their dementia ascertainment was valid by showing APOE4 carriers and those with elevated p-tau217 had higher dementia risk (essentially validating their outcome measurement)
- Robust adjustments: Controlled for education, socioeconomic status, comorbidities, lifestyle factors, diet quality, and even mutual adjustment between different caffeinated beverages
- Caffeinated vs. decaffeinated distinction: Separating these categories is critical and often overlooked—shows it's likely the caffeine, not just "coffee lifestyle"
- Genetic stratification showed independence: The benefit persisted regardless of APOE4 status or Alzheimer's polygenic risk score, suggesting this isn't just for low-risk people
- Dose-response curve makes biological sense: The plateau effect (benefits max out around 2-3 cups) aligns with known pharmacokinetics of caffeine metabolism
Weaknesses: The Caveats to Remember
- Observational study = correlation, not causation: People who drink coffee may differ systematically from non-drinkers in ways not fully captured (healthier overall, different social patterns, etc.)
- Reverse causation concern: Early cognitive decline might cause people to change coffee habits or reduce intake, though lag analyses attempted to address this
- FFQ limitations on tea specificity: Couldn't distinguish green vs. black tea or caffeinated vs. decaffeinated tea—these subtypes have different polyphenol profiles
- Homogeneous population limits generalizability: Almost entirely white health professionals (nurses, doctors, dentists); unclear if findings apply to other racial/ethnic groups or socioeconomic backgrounds
- Dementia ascertainment may miss cases: Relying on death certificates and self-reported diagnoses likely undercounts milder cases; no systematic neurological screening
- Couldn't separate Alzheimer's from other dementias: Would be valuable to know if the association is specific to AD vs. vascular dementia vs. Lewy body disease
- Objective cognitive testing only in women: HPFS cohort didn't undergo phone testing, so can't independently replicate those findings in men
- Modest effect sizes on objective tests: The 0.11-point improvement in TICS score is statistically significant but clinically small (~0.6 years of aging); hard to know if this matters for an individual
- Potential unmeasured confounding: Despite extensive adjustments, could never fully account for things like cognitive stimulation, social engagement, or other neuroactive medications taken sporadically
- Publication bias consideration: Positive findings are more likely to be published; we don't know how many null coffee-cognition studies exist in file drawers
The Bottom Line for Your Readers
This is one of the best-designed studies we have on coffee and brain health, but it's still observational. The fact that decaf showed no benefit points toward caffeine as the active ingredient, which has plausible mechanisms (adenosine receptor antagonism, reduced neuroinflammation, improved insulin sensitivity). The nonlinear dose-response—where 2-3 cups is optimal and more doesn't help—fits with caffeine pharmacology and makes the finding more credible.
If you drink coffee: This is reassuring data, but it doesn't mean you should force yourself to drink more to reach some magic number.
If you don't drink coffee: This study doesn't prove you should start; there are other ways to reduce dementia risk (exercise, Mediterranean diet, cognitive engagement).
The real story: Moderate caffeine consumption appears compatible with, and possibly supportive of, long-term cognitive health—but as always in nutrition science, your overall lifestyle pattern matters far more than any single food or beverage.