Beverages

Four beverages dominate the longevity conversation: water, coffee, tea, and alcohol. The evidence on each has shifted notably in the last decade. Water is no longer "drink eight glasses" folklore — chronic mild dehydration tracks with measurably faster biological aging, and per- and polyfluoroalkyl substance (PFAS) and microplastic contamination of municipal and bottled supplies has made filtration a real longevity decision. Coffee has accumulated a remarkably consistent positive signal across major outcomes — mortality, type 2 diabetes, liver disease, Parkinson's disease and several cancers — and one randomised trial has now undercut the standard advice to avoid coffee in atrial fibrillation. Tea shows smaller, less rigorously studied benefits in the same general direction. Alcohol's historical "J-curve" cardio-protection has largely collapsed under Mendelian randomisation — analyses that use inherited gene variants as a natural experiment — and the World Health Organization's 2023 position that no level of drinking is safe is increasingly the consensus, though one large recent review still dissents on mortality.

Two framings are worth carrying through. The first: what you don't drink matters as much as what you do. Sugar-sweetened beverages sit on the negative side of the ledger — see Sugar substitutes and Foods to limit — and so does alcohol past the first drop. The second is easy to miss because it cuts across three of the four: temperature. Hot drinks raise oesophageal cancer risk — measurably from about 60 °C, with the International Agency for Research on Cancer drawing its classification line at 65 °C — which makes letting a fresh pour stand a few minutes the most actionable habit change on this page.

What the evidence actually supports

Strong:

• Alcohol is a Group 1 carcinogen, and the dose-response starts at the first drop. A landmark meta-analysis covering 572 studies and 486,538 cancer cases found clean monotonic curves with no protective dip at the bottom — even light drinking, around one drink a day, carries measurably raised breast, oral and oesophageal cancer risk.^[1] For scale: alcohol accounts for about 4% of cancers worldwide, roughly 741,000 cases in 2020, of which about 103,000 arose in people drinking under 20 g a day — under two drinks — so roughly one alcohol-attributable cancer in seven occurs at intakes most people would not call heavy drinking.^[2] The US Surgeon General's January 2025 advisory named alcohol the third leading preventable cause of cancer in the US, after tobacco and obesity.^[3] See Alcohol.
• Alcohol raises blood pressure and stroke risk with no protective threshold. The China Kadoorie Biobank enrolled 512,715 adults and genotyped 161,498 men; among those men, any genetically predicted increase in drinking raised blood pressure and stroke incidence linearly — although the same analysis found no significant effect on heart attack, the endpoint the J-curve was built on.^[4]
• Drinking more water prevents kidney stones — the most robust hydration finding in the literature, and the only one resting on a large long-term trial. Randomising 199 first-time stone formers to a high-water target halved five-year recurrence, 12% versus 27%, and delayed the stones that did recur by about a year.^[5] See Water.
• Drink to thirst, not ahead of it. Overdrinking during endurance exercise dilutes blood sodium dangerously, and because it is caused by fluid volume rather than salt loss, sports drinks do not prevent it. The international consensus on the condition — linked to at least a dozen deaths — is that thirst prevents both dehydration and over-hydration, and that "drink before you're thirsty" is wrong for most people.^[6]

Moderate:

• Coffee at 2–4 cups per day is associated with the lowest all-cause mortality across multiple large cohorts — consistently, but observationally, which is what keeps it out of the Strong tier. A meta-analysis pooling 40 cohort studies across roughly 3.85 million adults found the nadir at 3.5 cups per day, with about 15% lower all-cause mortality at the optimum.^[7] A UK Biobank cohort of 171,616 adults found 1.5–3.5 cups per day associated with 29–31% lower all-cause mortality.^[8] Both caffeinated and decaffeinated forms show benefit, with one exception noted below. See Coffee.
• Chronic mild dehydration tracks with faster biological aging. A 2023 analysis of the Atherosclerosis Risk in Communities (ARIC) cohort, following 15,752 US adults for 25 years, found that serum sodium chronically above 142 mmol/L was associated with up to 50% higher odds of being biologically older than chronological age and roughly 39% higher chronic-disease risk.^[9] The entire 137–146 mmol/L range is labelled "normal" by lab reference — the risk gradient sits inside that window. It stays at Moderate rather than Strong because serum sodium is a proxy for habitual intake rather than a readout of it, and the one trial that tested whether drinking more actually helps an already-diseased kidney was null.
• Coffee and type 2 diabetes: each cup per day is associated with about 9% lower incident type 2 diabetes risk for caffeinated coffee and about 6% for decaf, pooled across 28 cohorts and over 1.1 million adults.^[10]
• Coffee and liver disease: each additional 2 cups per day is associated with about 35% lower risk of hepatocellular carcinoma, the main form of primary liver cancer.^[11] The signal extends to cirrhosis and to fibrosis in fatty liver disease.
• Coffee and Parkinson's disease: the most consistent inverse association in the coffee literature, and the one most often argued to be causal — though the genetic evidence that would settle it has not been done for Parkinson's specifically — roughly 30–40% lower risk at 3–4 cups per day, with protection peaking near 3 cups in a dose-response meta-analysis of about 900,000 people.^[12]
• Coffee and atrial fibrillation: the DECAF randomised trial, in 200 coffee-drinking adults whose heart rhythm had just been reset electrically to treat atrial fibrillation, found that drinking one or more cups per day of caffeinated coffee reduced recurrence versus complete caffeine abstinence — 47% versus 64%, roughly a 39% lower hazard.^[13] It challenges the long-standing avoidance instruction, but it is a single open-label trial of 200 patients, so the honest claim is narrow: telling an established coffee drinker with atrial fibrillation to quit does not appear to help.
• Tea and mortality: in the UK Biobank, about 498,000 adults mostly drinking black tea, 2 or more cups a day was associated with 9–13% lower all-cause mortality — the largest reduction at 2–3 cups — and the benefit held regardless of milk, sugar, or caffeine-metabolism genotype.^[14] See Tea.
• Tea and dementia: a 2026 JAMA analysis pooling the Nurses' Health Study and Health Professionals Follow-Up Study found 1–2 cups of tea per day associated with about 14% lower dementia risk; the larger 18% reduction in the same study was for 2–3 cups of caffeinated coffee, and decaffeinated coffee showed no association at all.^[15] The tea figure comes from the authors' own summary without a reported confidence interval, so treat the exact percentage as approximate.
• Tea and bone density: a 2025 meta-analysis covering roughly 560,000 adults found tea drinkers had about 25% lower odds of osteoporosis; polyphenol effects on the bone-resorbing cells probably outweigh the small caffeine-induced calcium loss.^[16]
• Alcohol's J-curve has collapsed — but the argument is not over. Bias-corrected pooling of 107 cohort studies, 4.8 million people, reproduced the classic J-curve in naïve estimates and then lost it once former drinkers were separated from lifelong abstainers: low-volume drinkers showed no mortality advantage, and risk turned significantly upward from about 45 g a day in men and 25 g in women.^[17] The surviving dissent is a 2025 US National Academies review, which pooled eight studies and found moderate drinkers had about 16% lower all-cause mortality than never-drinkers — graded moderate certainty, entirely observational, and resting on the never-drinker comparison group that is itself the problem.^[18] This is why the rating here is Moderate for the absence of benefit while the harm signals above are Strong.
• Hydration for other clinical endpoints: a 2024 JAMA Network Open systematic review of 18 randomised trials found only two areas reached statistical significance — kidney stones and weight loss. The rest is thinner than the popular version: recurrent urinary-tract infections rest on one clearly positive trial against two null ones, migraine prevention on two trials of which the larger changed quality of life but not headache days, and glycaemic control and orthostatic symptoms on single small trials. Eight of the 18 trials were negative overall.^[19]

Weak / preliminary:

• Hydrogen water has plausible biology and small trials in metabolic syndrome; the only "functional water" with real evidence behind it, and still not conclusive.
• Successful Aging Index–green tea associations (ATTICA/MEDIS 2019, about 3,300 Greek adults over 50) are real but cross-sectional and confounded — green tea drinkers were also more physically active and less hypertensive at baseline.^[20]

Caution:

• Very hot drinks. This is the one caution that applies to coffee and tea alike, and it is about temperature rather than the drink. In a cohort of about 50,000 adults with tea temperature measured objectively rather than self-reported, drinking at 60 °C or hotter carried 41% higher risk of oesophageal squamous-cell cancer, rising to roughly 90% higher at high daily volumes.^[21] The International Agency for Research on Cancer draws its line a little higher, classifying any beverage consumed above 65 °C as probably carcinogenic.^[22] Typical Western drinking temperatures land around 56–60 °C — at the boundary rather than safely below it — so 60 °C is the number worth acting on.
• Heavy added sugar in coffee likely cancels the favourable signal, and certainly adds substantial calories. The nuance matters: in the UK Biobank cohort above, a modest spoonful — around 4 g per cup — tracked with benefit similar to unsweetened coffee. It is the sugary café drink, at 200–500 kcal, that puts coffee on the wrong side of the ledger, not the teaspoon at home.
• Concentrated green-tea-extract supplements carry rare but real liver-injury reports serious enough to require transplant; brewed tea does not.
• Microplastics and PFAS in drinking water. Improved detection now puts bottled water on the order of 240,000 plastic particles per litre — around 90% of them nanoplastics small enough to cross the gut lining — far above older microplastic-only counts, though a published critique of the control blanks warrants caution on the absolute number.^[23] PFAS exposure links to immunosuppression, thyroid disruption, abnormal blood lipids and reproductive harm,^[24] and the evidence hardened in late 2023 when the WHO's cancer agency reclassified perfluorooctanoic acid as a Group 1 human carcinogen, its highest certainty tier — a hazard classification, which grades what the chemical can do rather than the risk at everyday drinking-water levels.^[25]
• Tea, coffee and non-heme iron. Both inhibit absorption of the plant-derived iron that makes up most of the diet, tea substantially more than coffee. This matters for menstruating women, plant-based eaters, and anyone on iron therapy — and the timing fix is directional rather than symmetric, which is where most advice goes wrong. Drink before the meal, not after.^[26]
• Alcohol in pregnancy. No safe level is established, harm is dose-dependent, and individual susceptibility varies too widely to identify a threshold. Every major authority recommends complete avoidance, ideally from the point of trying to conceive.

Water: the underrated lever

For decades, hydration science was binary: you were either fine or clinically dehydrated, and anything in between was thought irrelevant. The 2023 ARIC analysis broke that view by showing that serum sodium at the high end of the "normal" range — driven mostly by chronic low water intake — predicts accelerated biological aging, chronic disease, and earlier death.^[27] The middle of the reference range is where lowest risk sits; above 142 mmol/L, still inside "normal", the odds of being biologically older than chronological age run up to 50% higher, and mortality about 21% higher above 144 mmol/L. An Israeli cohort of 407,187 adults healthy at baseline points the same way — sodium of 140–142 mmol/L carried about 13% higher risk of later hypertension, rising to about 29% above 143 mmol/L, with heart-failure risk about 20% higher in that top band — though it shares its senior authors with the ARIC analysis, so it corroborates rather than independently replicates.^[28] The two disagree about where the line falls: ARIC's optimum runs up to 142 mmol/L while the Israeli risk starts climbing from 140, so the low 140s are best read as a grey zone rather than a safe plateau.

How you drink may also matter. In a small post-exercise rehydration trial, bolus drinking — chugging a large volume in a short window — captured only about 55% of the fluid, against around 75% for metered sipping of the same total volume.^[29] That is eight participants in a heat chamber, so treat the specific numbers loosely; the mechanism is the more convincing part. A sudden flood dilutes the blood quickly, the hypothalamus responds by switching off the hormone that tells the kidneys to conserve water, and the kidneys clear most of the water before it reaches the cells. Metered intake keeps the regulation steady.

The modern wrinkle is what's in the water. Microplastics and nanoplastics are now ubiquitous in tap and bottled supplies, and the particles do not pass inertly through: nanoplastics appear able to cross the gut lining, and have been reported in human blood, lung and placenta, where the concern is localised chronic inflammation and the endocrine-disrupting plasticisers the particles carry with them. PFAS — virtually indestructible carbon-fluorine compounds — accumulate in groundwater and human serum, with documented links to immunosuppression and reproductive harm.^[30] Reverse osmosis with a remineralisation stage is the protocol of choice for most modern municipal supplies; mineral-stripping reverse osmosis without remineralisation actively works against the magnesium signal in drinking water.

Water covers the serum-sodium dose-response in full, the bolus-versus-metered physiology, the PFAS and microplastic literature including the arsenic, nitrate and chlorination-byproduct story, filter technology comparison, the honest read on "functional waters" (hydrogen, alkaline, structured, marine plasma), the dehydration burden in older adults, and the magnesium-in-drinking-water stroke cohort.

Coffee: the most consistently positive beverage on the longevity ledger

Coffee accumulates a remarkably clean signal across major endpoints. An umbrella review of 201 meta-analyses put the mortality nadir at 3 cups per day;^[31] the 40-cohort, 3.85-million-adult meta-analysis put it at 3.5;^[32] a UK Biobank cohort of 449,000 found the greatest mortality reduction at 2–3 cups across ground (27% lower), instant (11%), and decaffeinated (14%) forms.^[33] That decaffeinated coffee shows benefit suggests non-caffeine constituents — chlorogenic acids and other polyphenols — are doing meaningful work, not just the caffeine. Dementia is the one documented exception: in the 2026 JAMA analysis, only caffeinated coffee carried the 18% lower risk, and decaf showed nothing, which points to caffeine itself being necessary for that particular outcome.^[34]

Beyond mortality, the disease-specific signals are coherent. A pooled cardiovascular cohort covering 1.28 million adults found a J-shape with lowest risk at 3–5 cups per day.^[35] Each cup per day is associated with about 9% lower incident type 2 diabetes for caffeinated coffee and about 6% for decaf,^[36] each additional 2 cups per day tracks with about 35% lower hepatocellular carcinoma risk,^[37] and Parkinson's disease runs 30–40% lower at 3–4 cups.^[38]

The heart-rhythm question is the one corner of this topic with actual randomised trials, and the two of them point in usefully different directions. DECAF undercut the long-standing instruction to avoid coffee in atrial fibrillation: among 200 established coffee drinkers whose heart rhythm had just been reset electrically, one or more cups per day of caffeinated coffee cut recurrence from 64% to 47% — about a 39% lower rate (hazard ratio 0.61, 95% confidence interval 0.42–0.89; that bracketed range is where the true effect most plausibly lies, and here it sits entirely below 1.0, so the benefit is unlikely to be chance alone).^[39] CRAVE is the counterweight: 100 adults randomised to drink or avoid coffee on alternating days while wearing continuous monitors showed essentially unchanged extra beats from the heart's upper chambers — the ones that tend to precede atrial fibrillation — but extra beats from the lower chambers rose by about half, and participants slept about 36 minutes less on coffee days.^[40] Read together: coffee does not appear to drive atrial arrhythmia, but it does provoke ventricular ectopy, and it does cost sleep.

The European Food Safety Authority and US Food and Drug Administration converge on up to 400 mg of caffeine per day as safe for healthy adults, with a 200 mg ceiling in pregnancy. Two consumer worries can be set down: ochratoxin A exposure from coffee runs at roughly 2% of the tolerable threshold even at 4 cups a day, and the acrylamide formed in roasting is an animal-derived concern that human data across 14 cancer sites has not confirmed.

Coffee covers the dose-response across endpoints, both arrhythmia trials in full, the iron interaction and why the timing rule runs in only one direction, the filtered-versus-unfiltered question (cafestol and kahweol raise LDL — low-density lipoprotein — cholesterol slightly), caffeine safety by population including slow metabolisers and pregnancy, hot versus cold brew, Mendelian-randomisation caveats, the reflux evidence, and why decaffeinated and caffeinated track together for every outcome except dementia.

Tea: a smaller, parallel signal

Tea sits in the same general direction as coffee but with smaller effect sizes and a thinner evidence base. In the UK Biobank, 2 or more cups a day was associated with 9–13% lower all-cause mortality, and the benefit survived adjustment for milk, sugar and caffeine-metabolism genotype — which is part of why the polyphenols rather than the caffeine are the usual explanation.^[41] The most striking recent result is the 2026 JAMA analysis pooling the Nurses' Health Study and the Health Professionals Follow-Up Study, which found 1–2 cups of tea per day associated with about 14% lower dementia risk — smaller than the 18% the same study reported for 2–3 cups of caffeinated coffee.^[42] Black tea benefits plateau around 2 cups per day; green tea shows closer to a linear response up to 3–4 cups. The genetic evidence is the honest caveat: Mendelian-randomisation studies have been largely null for cardiovascular outcomes, so some of the observational signal may be confounding.

The mechanisms are partly polyphenol-driven, and partly still speculative. Green-tea polyphenols are extensively metabolised by gut microbes and appear to reshape microbial communities in return, with a gut bacterium linked to metabolic health, Akkermansia muciniphila, the one most often named — but that link rests on animal models and observational microbiome data rather than tea-dosing trials, so treat it as plausible biology rather than an established human effect. The bone signal is firmer: a 2025 meta-analysis covering roughly 560,000 adults found tea drinkers had about 25% lower odds of osteoporosis,^[43] and a 2025 analysis of the Study of Osteoporotic Fractures found higher hip bone mineral density in tea-drinking women aged 65 and over.^[44]

Two practical cautions. Tea inhibits non-heme iron absorption substantially more than coffee does — and separating tea from an iron-rich meal by an hour roughly halves the interaction rather than abolishing it, cutting inhibition from about 37% to 18%.^[45] The effect is local to the gut, not a systemic depletion, but it matters for menstruating women, plant-based eaters, and anyone on iron therapy. And concentrated green-tea-extract supplements have caused rare cases of serious liver injury; brewed tea has not.

Tea covers green versus black, the L-theanine question, the very-hot-tea and oesophageal cancer evidence, the iron interaction in detail, the bone-density evidence, the mixed cancer literature, green-tea-extract hepatotoxicity, and an honest read on the "Successful Aging Index" signal from cross-sectional Mediterranean cohorts.

Alcohol: the J-curve collapse

For 30 years, observational cohort data showed a U- or J-shaped relationship between alcohol intake and cardiovascular mortality — light-to-moderate drinkers appeared healthier than abstainers. That curve has collapsed under three methodological pressures: better handling of "sick quitters" (former drinkers, many of whom stopped because they were already ill, landing in the abstainer reference group), the rise of Mendelian-randomisation studies using genetic variants as instruments, and an expanding cancer signal that the cardiovascular story had been masking.

The genetic data is the most consequential. The China Kadoorie Biobank enrolled 512,715 adults and genotyped 161,498 men; among those men, any genetically predicted increase in drinking raised blood pressure and stroke incidence linearly, with no protective threshold — although the same analysis found no significant effect on heart attack, the endpoint the J-curve was built on.^[46] A 94-variant analysis found about 27% higher stroke risk and roughly triple the risk of peripheral artery disease among people whose genes predispose them to drink substantially more.^[47] A UK Biobank analysis found each additional daily drink causally associated with a 2.65 mmHg rise in systolic blood pressure, roughly double the risk of bleeding stroke, and about 26% higher atrial fibrillation risk.^[48] A Million Veteran Program analysis replicated the U-shape observationally and then found no causal benefit at moderate intake — the signature of confounding.^[49]

What has not happened is a clean end to the argument, and the honest hub-level summary says so. The 2025 US National Academies review remains the strongest surviving pro-moderate signal: pooling eight studies, it found moderate drinkers had about 16% lower all-cause mortality than never-drinkers, graded moderate certainty.^[50] It is entirely observational, none of its 34 included articles was a randomised trial, and the comparison rests on the never-drinker group that carries the sick-quitter problem in the first place — but it is not nothing, and the one endpoint where the genetic evidence genuinely fails to show harm is heart attack.

The cancer side is where the evidence is unambiguous. The Bagnardi dose-response meta-analysis covering 572 studies and 486,538 cases found clean monotonic curves, with breast, oral and oesophageal cancer risk already measurably raised at about one drink a day, and colorectal and liver cancer rising as intake climbs.^[51] Acetaldehyde, the first metabolic product of ethanol, is the proximate carcinogen; the East Asian variant that slows acetaldehyde clearance and produces facial flushing gets about as close to a controlled experiment as this field allows, since carriers who drink develop far more mouth, throat and oesophageal cancer at the same intake.^[52] The WHO's January 2023 position is that no level of alcohol consumption is safe for health.^[53] Even moderate drinking is associated with reduced grey-matter volume in cohort imaging — a reason not to feel reassured rather than a diagnosis, since what a few per cent of volume means for how you think and function is not established.^[54]

Guidance has not moved in one direction, though. In January 2026 the US 2025–2030 Dietary Guidelines removed the previous numeric limits and the distinction between men and women, replacing them with unquantified advice to drink less — a change the American Association for the Study of Liver Diseases objected to as the removal of evidence-based guidance, and which The Lancet Gastroenterology & Hepatology called a step backwards.^[55] The scientific case for lower limits strengthened over the same period the largest national guidance dropped its numbers. That is a policy divergence, not an evidentiary one.

Alcohol covers the J-curve collapse in detail, the genetic data and where it stops short, the National Academies dissent, the flushing natural experiment, the cancer dose-response by site with absolute burden figures, the liver thresholds that reclassify a fatty-liver diagnosis, the brain-imaging and telomere signals at "moderate" intake, sleep effects and the deep-sleep tracker trap, the 2026 US guidance reversal, and Canada's 2023 low-risk threshold of two drinks per week or fewer.

Practical beverage guidance

1. Coffee, 2–4 cups per day if you already drink it — the effect size doesn't justify starting a habit you don't enjoy. Filtered or unfiltered both qualify; paper filtration removes the coffee oils that raise LDL cholesterol slightly, which matters if yours is already elevated.
2. Let hot drinks cool below about 60 °C — comfortably warm rather than scalding. This applies to coffee and tea equally and is the cheapest risk reduction on this page.
3. Keep the caffeine early. A single afternoon cup is defensible for most people, but a heavy total load wants a 10–12 hour buffer before bed. A crossover trial found that 400 mg — a full day's allowance in one hit — altered sleep architecture taken within 12 hours of bedtime and fragmented sleep within 8, while 100 mg had no measurable effect even 4 hours before bed.^[56] The reason to distrust your own judgement here is an earlier trial, in which a 400 mg dose 6 hours before bed cut total sleep by more than an hour and the subjects did not notice it in their own ratings.^[57]
4. Skip the sugary café drink, not the teaspoon. A modest spoonful of sugar tracked with benefit similar to black coffee; a 200–500 kcal sweetened drink does not.
5. For alcohol, the minimum-risk dose is at or near zero. Cancer risk begins from the first drop; don't start drinking for "heart benefits" — the surviving observational signal is for mortality, is moderate-certainty, and does not reproduce genetically. If you currently drink, less is better, and Canada's 2023 guidance puts the lower-risk threshold at two drinks per week or fewer. Finish at least 3 hours before bed if sleep matters, and don't drink at all in pregnancy.
6. Drink water steadily, not in boluses — roughly 2–3 L per day of total fluids for most healthy adults, but let thirst and pale-yellow urine be the actual guide rather than a target volume. Filter for PFAS and microplastics; reverse osmosis with remineralisation handles most modern municipal supplies.
7. Tea (green or black), 2–3 cups per day — where most cohort signals converge, though the dementia signal sits at 1–2. Drink it before iron-rich meals or supplements rather than after if iron status is a concern, and, as with coffee, don't take it up purely for the health benefits.
8. Sugar-sweetened beverages are off the list. They aren't on the same axis — they're on the negative side of the ledger, alongside alcohol.

What's overrated

• "Wine is healthy because of resveratrol." Wine contains 0.3–1 mg of resveratrol per glass; therapeutic doses in research are 100–500 mg. The observational cardiovascular signal is overwhelmingly explained by lifestyle confounding, and no special protective effect of wine over other alcohol survives rigorous methods.
• "Bulletproof coffee" / butter coffee. Calorie-dense, no demonstrated benefit over plain coffee; the saturated-fat content is meaningful.
• Alkaline water, "structured" water, marine-plasma water. No clinical evidence of benefit beyond plain mineralised water; "structured" water is physicochemical pseudoscience. Hydrogen water has plausible biology but is still preliminary.
• Daily electrolyte powders for sedentary adults. They didn't prevent blood-sodium disturbances even in desert ultramarathoners; the conventional rule of thumb puts any benefit past roughly an hour of sustained exertion or heavy heat, and many powders add substantial sugar and unnecessary sodium.
• The "J-curve" for moderate drinking. The cardiovascular protection it promised does not survive Mendelian randomisation or sick-quitter adjustment. What survives is an observational mortality signal of moderate certainty that no genetic analysis reproduces — enough to keep the argument alive, not enough to make drinking a health behaviour.
• Blanket coffee avoidance for atrial fibrillation. The reflex instruction to give up coffee is not supported by the one randomised test of it: in DECAF, established coffee drinkers who kept drinking after cardioversion had fewer recurrences than those told to abstain.^[58] That is a single open-label 200-patient trial, and CRAVE found more ventricular ectopy on coffee days, so this belongs here as "the blanket rule is overstated," not as "coffee is proven antiarrhythmic." Individuals who reliably notice palpitations after coffee should still trust that.
• Vitamin-fortified "wellness" waters. Vitamin doses are typically negligible; the sugar and sweetener loads usually are not.

---

For the full ordered action list drawn from every pillar, see the healthspan long list.