Saltar al contenido principal
Pan y Hierro Nutrition and exercise with sources

Blog

Coffee: what the evidence says, and what it does not, about drinking it or quitting

An umbrella review of 201 meta-analyses on coffee and health, the ISSN position stand on performance, and why almost all of it is observational.

A 2017 review synthesizes 201 meta-analyses of observational studies on coffee and health, and that evidence does not demonstrate causality [1].

What that review summarizes

Published in the BMJ in 2017, it reviewed 201 meta-analyses of observational studies, with 67 different health outcomes, plus 17 meta-analyses of interventional studies with nine outcomes [1].

Its main results:

What that review summarizes: Outcome · What was seen
Outcome What was seen
All-cause mortality Between 12 % and 21 % less
Cardiovascular mortality Around 19 % less
Cardiovascular disease Around 15 % less
Incident cancer (high vs. low consumption) Around 18 % less

The first three figures correspond to consumption of three to four cups a day versus none [1]. The association is not linear: the greatest benefit sits in that range, and beyond it the benefit shrinks without clear signs of harm appearing [1].

The finding that forces a caveat

The review is unusually honest about why the results look so favorable. It points out that the harmful associations were largely cancelled out when adjusted properly for tobacco [1]. A good part of the myth that links coffee to serious disease was explained by studies that did not adjust properly for tobacco.

The exception the review itself highlights is pregnancy. There the harmful associations hold and are not cancelled out by adjusting for tobacco [1]. Those are associations of high versus low or no consumption, not of three to four cups. The risk of low birth weight rose by around 30 %, the risk of prematurity by between 12 % and 22 %, depending on the trimester, and pregnancy loss by around 45 % [1].

A minority finding with its own nuance also appears: an association between coffee and fracture risk in women, while in men the association points the other way [1]. The authors advise caution for women at increased risk of fracture [1].

The methodological limit, stated by the authors

The conclusion of the paper is clear on both counts. First, that coffee consumption “seems generally safe within usual levels of intake” and that, in the summarized figures, it is more likely to be healthy than harmful [1]. Second, and this is the important part: robust randomized trials are needed to know whether those associations are causal [1].

A meta-analysis of observational studies can say that people who drink three cups live longer than people who do not. It cannot say that drinking three cups makes you live longer. That is the whole difference.

Coffee and exercise: a different question

The ISSN Position Stand on coffee and sports performance is a different document, and it is worth not mixing it with the previous one [2]. The ISSN defines it in its own title as a document on coffee and sports performance: it is a position stand, the position of a scientific society, not a meta-analysis. It does not pool estimates into a single combined result.

What it holds:

  • Coffee and its components improve performance across a range of activities running from reaction time, through brief power exercises, to the aerobic range, in most —but not all— of the studies [2].
  • The optimal doses and timing sit at around two to four cups (about 473 to 946 ml) of standard coffee, equivalent to 3-6 mg of caffeine per kilo of body weight, 60 minutes before exercise [2][3].
  • Coffee also provides other compounds, such as chlorogenic acids, at around 100-400 mg per cup [2].
  • In that dose range it is considered generally safe and beneficial for healthy people who exercise [2].

ISSN itself admits two limits: that the literature specific to coffee as an ergogenic aid is scarce compared with the literature on caffeine, and that operationally defining “coffee” across studies is difficult [2]. It also notes that the effects depend on dose, timing, habituation, genetics, gut microbiome, sex and training status [2].

And it puts practical limits on that dose. Anyone sensitive to caffeine or with sleep disorders may have to lower the dose, especially in afternoon or evening sessions; ISSN recommends limiting caffeinated coffee to the morning or the afternoon, considering decaf at the end of the day and, in general, considering the low end of the range (two cups) [2]. The high end is also a significant volume of liquid, and people over 100 kg would need even more [2]. Anyone looking for the benefit should talk to a healthcare professional and start at the bottom [2].

Worth saying without softening it: “generally safe and beneficial” refers to healthy people who exercise, in that dose range. It is not a claim of universal safety, and pregnancy falls outside it [2].

The diterpenes

The 2017 review pays attention to the brewing method, and it is a detail with consequences [1]: there is fairly consistent evidence of small increases in total cholesterol, LDL cholesterol and triglycerides in trials, attributed to the diterpenes. The important thing is that the diterpene content varies a lot with preparation: instant and filtered coffee contain negligible amounts, while espresso, boiled coffee and cafetière coffee contain more [1].

And it adds a fact that balances the message: those changes in the lipid profile reverse with abstinence [1].

What you take away

  1. A review of 201 meta-analyses associates consumption of three to four cups a day with lower mortality, and high consumption with low consumption with lower cancer risk [1].
  2. But almost all of that evidence is observational, and the authors ask for randomized trials before causality can be discussed [1].
  3. For exercise, ISSN —the position of a society, not a meta-analysis— puts the useful dose at 2-4 cups, 60 minutes before, and considers it safe and beneficial in that range for healthy people who exercise, adjusting the dose if there is sensitivity or sleep problems [2].
  4. The brewing method matters: filtered and instant coffee have fewer diterpenes than espresso or cafetière coffee [1].
  5. In pregnancy, the association with low birth weight, prematurity and pregnancy loss is not corrected by adjusting for tobacco [1].

If you are pregnant, the available evidence justifies talking to your healthcare professional before keeping up your consumption.

More in: the supplements page, with the figures and the sources of the Nutrition section.

Related articles

Sources for this article (3) 3 references
  1. Poole R, Kennedy OJ, Roderick P, Fallowfield JA, Hayes PC, Parkes J (2017) Coffee consumption and health: umbrella review of meta-analyses of multiple health outcomes. BMJ 2017;359:j5024. See source (opens in a new tab) DOI: 10.1136/bmj.j5024 Revisión paraguas (umbrella review) de metaanálisis sobre café y salud Licence: Artículo de acceso abierto (texto completo en PubMed Central, CC BY 4.0). Se cita con enlace. Consulted:
  2. Lowery LM, Anderson DE, Scanlon KF, Stack A, Guest NS, Pearlson R, et al. (2023) International Society of Sports Nutrition position stand: coffee and sports performance. Journal of the International Society of Sports Nutrition 20:100. See source (opens in a new tab) DOI: 10.1080/15502783.2023.2237952 Position Stand de la ISSN (café, cafeína y rendimiento; revisión narrativa, no metaanálisis) Licence: Artículo de acceso abierto (texto completo en PubMed Central, CC BY 4.0). Se cita con enlace. Consulted:
  3. Guest NS, VanDusseldorp TA, Nelson MT, Grgic J, Schoenfeld BJ, Jenkins NDM, Arent SM, Antonio J, Stout JR, Trexler ET, Smith-Ryan AE, Goldstein ER, Kalman DS, Campbell BI (2021) International society of sports nutrition position stand: caffeine and exercise performance. Journal of the International Society of Sports Nutrition. See source (opens in a new tab) DOI: 10.1186/s12970-020-00383-4 Posición de la ISSN sobre cafeína: 3-6 mg/kg de masa corporal, con la dosis mínima todavía incierta Licence: Artículo de revista con derechos de autor. Se cita con enlace al editor; no se reproduce texto salvo cita breve verificable. Consulted:

How to cite

Pan y Hierro (2026). Coffee: what the evidence says, and what it does not, about drinking it or quitting. https://panyhierro.com/en/blog/coffee-what-the-evidence-says/. Updated October 12, 2026.