Your Daily Coffee Is Reshaping Your Gut — And Your Brain Feels the Difference
A 2026 Nature Communications study followed 62 healthy adults through three phases — drinking, quitting, and resuming coffee — to map what changes inside the gut and brain along the way.
Most people drink coffee for energy or taste. But a study published on 21 April 2026 in Nature Communications shows that the cup you pour every morning is also associated with measurable changes in the bacterial communities living in your gut — and those shifts are linked to differences in how impulsive you feel, how well you remember things, and how your body handles stress. The research, led by scientists at University College Cork, is one of the first controlled human studies to map coffee’s influence on the microbiota–gut–brain axis — the bidirectional communication pathway between the gut microbiome and the brain — while simultaneously tracking what happens when coffee is removed and reintroduced.
Unlike previous research observing coffee drinkers versus non-drinkers at a single point in time, this prospective trial ran participants through three distinct phases: a baseline comparison of 31 moderate coffee drinkers (CDs) against 31 non-coffee drinkers (NCDs), a 14-day coffee-free washout for the CDs, and a 21-day reintroduction of either caffeinated or decaffeinated coffee. The design allowed researchers to separate what caffeine does from what the rest of coffee’s compound mix does — and the differences turned out to be substantial.
How the Study Was Structured
Three phases, two groups, 62 participants — tracked through withdrawal and reintroduction.
What Changed — and Where
Select a category to see what the study found at each level of the microbiota–gut–brain axis.
Higher in Coffee Drinkers
- Cryptobacterium curtum
- Eggerthella sp. CAG:209
- Eggerthella sp. CAG:51_9
- Firmicutes CAG:94
Higher in Non-Coffee Drinkers
- Veillonella parvula
- Veillonella sp. ACP1
- Haemophilus parainfluenzae
Key Findings on Diversity & Reintroduction
- Alpha-diversity was significantly different between NCD and CD groups at baseline
- Within CDs, alpha-diversity did not change significantly during washout or after reintroduction — coffee appears to affect specific strains rather than overall diversity
- On day 21 of reintroduction, a significant increase in abundance of Veillonella sp. ACP1 was observed in both caffeinated and decaffeinated groups
- All seven strains different between NCD and CD were significantly affected upon reintroduction
Elevated in Coffee Drinkers (Fecal)
- Caffeine
- Theophylline (coffee alkaloid)
- 1,7-Dimethylxanthine (caffeine metabolite)
- Hippuric acid (phenolic metabolite)
Reduced in Coffee Drinkers (Fecal)
- Indole-3-propionic acid (IPA) — tryptophan-derived, microbial metabolite
- Indole-3-carboxyaldehyde (ICA) — gut barrier integrity, IL-10 & AHR signalling
- γ-Aminobutyric acid (GABA) — neurotransmitter
After Abstinence & Reintroduction
- After washout: caffeine, theophylline, 1,7-dimethylxanthine, and hippuric acid all significantly reduced; ICA significantly increased
- After reintroduction: caffeine, theophylline, hippuric acid, and 1,7-dimethylxanthine did not fully return to baseline measurement levels
- Caffeine metabolites returned to CD values only upon caffeinated — not decaffeinated — reintroduction
Coffee Drinkers vs. Non-Drinkers at Baseline
- Higher impulsivity (UPPS-P scale) and “Sensation Seeking” sub-scores
- Higher emotional reactivity (ERS) — sensitivity, intensity, and persistency
- No significant difference in memory (ModRey) or learning (PAL) at baseline
After 14-Day Coffee Abstinence
- Impulsivity (UPPS-P total and Negative Urgency) decreased significantly
- ERS scores for sensitivity, intensity, persistency, and total fell
- PASAT performance improved — paper notes “possibly due to task repetition”
Caffeinated Coffee (21 Days)
- Reduced anxiety (STAI-Trait) and psychological distress (HSCL)
- Improved PASAT — not solely due to task repetition per authors
- Reduced impulsivity and negative urgency
Decaffeinated Coffee (21 Days)
- Improved episodic memory (ModRey) — chance of learning effect noted
- Fewer PAL learning errors — non-caffeine components may play a role
- Better sleep quality (PSQI); increased physical activity (IPAQ)
Coffee Drinkers at Baseline
- Lower CRP levels in peripheral blood
- Higher IL-10 (anti-inflammatory cytokine)
- Less IL-6 secreted upon LPS (TLR4) stimulation than non-drinkers
After 2-Week Abstinence
- CRP levels increased
- TNFα levels rose
- IL-6 secretion increased upon LPS stimulation
Cortisol (HPA Axis)
- No significant difference in salivary cortisol between CD and NCD during SECPT stress test at baseline
- Cortisol awakening response (CAR) similar between groups at baseline; no change after washout
- Caffeinated group had lower salivary cortisol upon awakening post-intervention
- Decaffeinated group: CAR remained unchanged throughout
The microbiome findings add depth to what was already known about coffee and health. A metagenomics study of over 1,000 participants had previously identified coffee as the most strongly diet-correlated food item for gut microbiome composition in a dose-dependent manner, validated across a second cohort. The 2026 UCC study tracks those associations through withdrawal and reintroduction in real time — and extends the picture into neuroactive chemical pathways. For readers familiar with coffee’s broader health associations, this study adds mechanistic detail at the level of gut bacteria, their metabolites, and the neuroactive compounds they produce.
Prior large-scale epidemiological work — including a cross-sectional study of 468,629 individuals without clinical cardiovascular disease — linked light-to-moderate coffee consumption with lower all-cause mortality, cardiovascular mortality, and stroke rates. One meta-analysis of cohort studies examining cognitive decline, cited within the paper, reported an association between coffee consumption and a 27% lower incidence of Alzheimer’s disease. Multiple meta-analyses have also linked coffee with a lower risk of depression. The 2026 trial does not establish causation for any of these outcomes, but it identifies biological pathways — specific gut bacteria, their metabolites, and neuroactive compounds they produce — through which such associations may operate.
“These findings reveal previously unrecognised effects of coffee on the microbiota–gut–brain axis, suggesting that microbiome profiles could potentially predict coffee consumption patterns and highlighting a close association between coffee intake and gut microbial composition.”
— Boscaini et al., Nature Communications, Volume 17, Article 3439, 21 April 2026 (verbatim from abstract)
Caffeinated vs. Decaffeinated — Not the Same Thing
Caffeine and coffee’s non-caffeine compounds had distinct — and in some cases, opposite — effects across the trial.
- Reduced anxiety (STAI-Trait)
- Lower psychological distress (HSCL)
- Improved PASAT attention performance
- Better stress coping (PASA secondary appraisal)
- Reduced impulsivity and negative urgency
- Lower cortisol awakening response (CAR)
- Improved episodic memory (ModRey)
- Fewer associative learning errors (PAL)
- Better sleep quality (PSQI)
- Increased physical activity (IPAQ)
- Higher positive affect (PANAS)
- Reduced emotional reactivity intensity (ERS)
- IL-10 and IL-6 plasma levels decreased
- LPS-stimulated IL-6 halved post-intervention
- Caffeine metabolites (1,7-dimethylxanthine) restored to CD baseline values
- Cortisol awakening response reduced
- CRP and TNFα levels increased post-reintroduction
- LPS-stimulated IL-6 also halved (same as caffeinated group)
- Phenolic metabolites rose equally in feces and urine
- Veillonella parvula significantly increased (not seen in caffeinated)
One of the study’s more detailed findings involves the link between GABA, IPA, and the microbiome. GABA — the brain’s primary inhibitory neurotransmitter — was measurably lower in the fecal samples of coffee drinkers. The authors note that further work is needed to explore this relationship, given both chemicals’ ability to influence anxiety pathways. Lower IPA in coffee drinkers’ fecal samples is noted in the context of separate literature on IPA and cognition — these are associations observed in a specific study population and do not establish cause.
The immune findings are equally layered. At baseline, coffee drinkers had lower CRP and higher IL-10 — a pattern consistent with a lower inflammatory state. When they stopped coffee for two weeks, CRP and TNFα levels rose. The authors describe this as suggesting “a protective role of coffee or caffeine against inflammation,” while acknowledging conflicting findings in existing literature on coffee and CRP. These changes were partially reversed upon resuming coffee, with differences between the caffeinated and decaffeinated groups pointing to distinct roles for caffeine and polyphenols in modulating cytokine responses. This aligns with broader evidence on how dietary habits interact with systemic inflammation markers.
The study’s integrated analysis — connecting nine key metabolites to both specific microbial species and measurable cognitive outcomes — provides a detailed picture of how a daily dietary habit maps onto human biology at multiple levels. Two clinical trials underpinning this work are registered at ClinicalTrials.gov (NCT05927038) and NCT05927103. The full metabolomics dataset is deposited at the EBI MetaboLights repository (MTBLS13494).
The authors acknowledge limitations including a modest sample size, the absence of direct stool transit time measurements, and the possibility that the study was not fully powered to detect small-to-medium effect sizes beyond the primary microbiome outcome. For further reading on how daily habits intersect with health outcomes, see related coverage on daytime napping and mortality risk and screen time, cognition, and mental health.
The study covered coffee’s effects on emotional reactivity, immune responses, gut microbial composition, urinary and fecal metabolite profiles, cognition, mood, cortisol physiology, and gastrointestinal function across three structured phases. Differences between caffeinated and decaffeinated outcomes across all these domains were described and documented.
