Coffee and Health: The Evidence

What different types of evidence can tell us

An observational study records what people do and what happens to their health without assigning their behaviour. A prospective cohort study follows participants forwards in time, comparing outcomes among people with different habits.

These studies are useful for investigating diseases that develop over many years. Their weakness is confounding: coffee drinkers and non-drinkers may differ in smoking, diet, income or other characteristics that also influence health. Statistical adjustments help, but cannot guarantee that those differences have been removed.

Reverse causation is another difficulty. Someone may stop drinking coffee because they are already unwell, making the non-drinking group appear less healthy even though avoiding coffee did not cause the illness.

A randomised controlled trial assigns participants to an intervention or comparison group. Randomisation reduces confounding and gives stronger evidence about cause and effect, but a short trial measuring sleep or cholesterol cannot establish whether coffee prevents disease decades later.

A systematic review collects studies using stated methods; a meta-analysis combines their results statistically. An umbrella review assesses multiple reviews. These can reveal patterns, but pooling studies does not remove weaknesses in the underlying research. Our coffee and health hub provides routes into individual topics.

What the major reviews found

Poole and colleagues’ 2017 BMJ umbrella review included 201 meta-analyses of observational research, alongside 17 meta-analyses of trials. Coffee consumption was more often associated with benefit than harm. For several outcomes, including mortality and cardiovascular disease, the most favourable associations occurred around 3 to 4 cups daily compared with none.

That finding is not a prescription. Study definitions of a cup differed, and cup counts cannot be translated reliably into a personal caffeine allowance. The review also rated much of the evidence as low or very low quality. A correction published in 2018 did not change its findings.

Grosso and colleagues’ 2017 umbrella review in Annual Review of Nutrition also found favourable associations across several outcomes, including type 2 diabetes, Parkinson’s disease and cardiovascular disease. It distinguished these observational findings from trial evidence, including increases in blood lipids with coffee and blood pressure with caffeine.

The practical interpretation is narrower than many headlines suggest: coffee can fit within a healthy diet, but the research does not establish it as preventive treatment.

Evidence summary

The strength descriptions below are editorial summaries, not a new formal evidence grading.

Likely benefits, without promises

The strongest recurring positive signals concern lower rates of type 2 diabetes, some liver conditions and mortality among habitual coffee drinkers. These are reasons for continued research, not grounds for claiming that drinking coffee will prevent a particular illness.

Some favourable associations also appear with decaffeinated coffee. This suggests that caffeine alone cannot explain every finding, although it does not establish which other constituents are responsible.

Coffee contains biologically active compounds, and proposed explanations include effects on inflammation and metabolism. A plausible mechanism is supporting evidence, not proof: showing an effect in a laboratory is different from demonstrating fewer illnesses in people.

The distinction matters when deciding what to do. An association between coffee and lower disease risk does not mean that adding coffee to your routine will produce the same result. For a closer look at the positive findings and their limitations, see coffee health benefits.

Risks: distinguish immediate effects from long-term outcomes

Sleep disruption, anxiety, jitteriness and palpitations are practical concerns even when someone remains below a population caffeine benchmark. The Food Standards Agency warns that sensitive people and those with underlying conditions may experience adverse effects at lower intakes.

The 2023 CRAVE randomised trial illustrates why “healthy” is too simple a label. It followed 100 adults for 14 days, alternating instructions to drink caffeinated coffee or avoid caffeine. Coffee did not significantly increase the primary outcome, premature atrial contractions, which are early beats arising in the heart’s upper chambers.

Secondary findings suggested more physical activity, more premature ventricular contractions, which arise in the lower chambers, and less sleep. Those secondary results were hypothesis-generating, not proof of long-term benefit or harm. People with certain heart conditions were excluded, so the findings cannot simply be applied to them.

A favourable long-term association therefore does not cancel an immediate unwanted effect. If coffee repeatedly leaves you anxious or sleeping poorly, a reassuring disease headline is not a reason to ignore that response.

Preparation, sugar and milk change the question

Coffee preparation matters independently of caffeine. A meta-analysis of randomised trials found that unfiltered, rather than filtered, coffee increased total and LDL cholesterol. LDL is the cholesterol fraction commonly described as “bad” cholesterol.

Coffee oils contain diterpenes, including cafestol and kahweol, which help explain this effect. Paper filtering substantially reduces their passage into the drink. A cafetière’s metal mesh is not equivalent to a paper filter. If you have raised cholesterol and regularly drink unfiltered coffee, discuss preparation with your clinician rather than assuming every brewing method is interchangeable.

Additions deserve separate consideration. Research about coffee cannot automatically justify a drink containing substantial added sugar, syrup or cream. NHS guidance recommends limiting sugary drinks and choosing lower-fat milk or unsweetened plant-based drinks where appropriate.

Milk itself is not a reason to label coffee unhealthy. Consider the whole drink, its portion size and your wider diet. There is no need to force yourself to drink black coffee in pursuit of an unproven preventive benefit.

NHS and EFSA guidance: apply milligrams, not cup counts

EFSA’s 2015 assessment found no safety concerns for healthy non-pregnant adults consuming:

• Up to 400 mg caffeine daily, spread throughout the day.
• Single doses up to 200 mg, approximately 3 mg per kg body weight.

It also found that 100 mg close to bedtime may affect sleep in some adults. These are population safety benchmarks, not recommended intakes or guarantees of symptom-free use.

Count tea, chocolate, energy drinks, supplements and relevant medicines as well as coffee. The FSA specifically advises adding supplement caffeine to other dietary sources.

Caffeine content varies with the drink and serving. NHS examples include about 100 mg in a mug of instant coffee, 140 mg in a mug of filter coffee and 75 mg in a mug of tea. These are estimates, not values for every serving. Do not apply the filter figure to espresso, pods or an unfamiliar workplace drink. Our caffeine and health guide examines caffeine separately from coffee.

Who should be more careful?

During pregnancy, follow the NHS advice of no more than 200 mg caffeine daily from all sources. Its guidance links excessive intake with a higher chance of miscarriage and stillbirth. This lower allowance takes priority over the general adult benchmark.

During breastfeeding, the sources checked differ. NHS Best Start in Life advises cutting caffeine out if possible, or trying not to exceed 300 mg daily; EFSA’s assessment identifies intake up to 200 mg daily as raising no safety concerns for breastfed infants. These are differently framed recommendations, not identical thresholds. A cautious approach is to stay within 200 mg and discuss concerns, particularly infant restlessness, with a health visitor or GP.

People with troublesome symptoms or medical conditions should seek individual advice. Sleep problems, anxiety, heart symptoms and poorly controlled blood pressure deserve attention rather than reassurance based solely on an intake total.

People taking medicines should check instructions with a pharmacist. For example, NHS guidance says to take levothyroxine on an empty stomach 30 to 60 minutes before food or a caffeinated drink. Do not change prescribed treatment to accommodate coffee.

The adult caffeine benchmark should not be treated as a child’s allowance. This article provides general information, not diagnosis or personal medical advice; guidance from your GP, pharmacist or midwife takes priority.

How to read coffee and health headlines

Before changing your habits, ask:

1. Was this observational research or a trial? “Associated with” should not become “prevents”.
2. What actually changed? A cholesterol measurement is not the same outcome as a heart attack.
3. Who was studied? Findings in healthy volunteers may not apply to pregnancy or an existing condition.
4. What was the exposure? Coffee, caffeine tablets, decaf and sweetened drinks are not interchangeable.
5. How large and certain was the result? Look for absolute risks, uncertainty and agreement across studies, not just a striking relative percentage.

Treat a new study as a contribution to the evidence, not an instruction to overhaul your routine. Coffee and health decisions should combine research with your actual drink, sleep, symptoms and medical circumstances.