Chronic low-grade inflammation is the connective tissue of modern health writing. It is invoked to explain heart disease, diabetes, depression, ageing and almost everything else, and every food gets sorted into inflammatory or anti-inflammatory on that basis. Coffee has been sorted into both. Articles calling coffee a powerful anti-inflammatory and articles warning that it drives inflammation are citing real, published studies. This article is about why they disagree, because the reason is more useful than picking a side.
What inflammation markers actually measure
Inflammation in this sense is not a swollen ankle. It is a persistent, low-level activation of the immune system that can be measured in blood. The markers researchers use most are C-reactive protein, made by the liver in response to inflammatory signalling, and interleukin-6 and tumour necrosis factor alpha, which are signalling molecules released by immune and fat cells. Adiponectin, a hormone made by fat tissue, usually moves the other way: higher levels are associated with better metabolic health and lower inflammation.
Two things are worth knowing before reading any of this literature. These markers fluctuate, responding to infection, exercise, poor sleep, stress and body weight, so a single reading says little. And they are proxies. A study showing coffee shifts a marker has not shown coffee changes anyone's risk of disease.
The study that says coffee is inflammatory
The most-cited evidence on that side is the ATTICA study, reported by Zampelas and colleagues[1]. They surveyed 1,514 men and 1,528 women in Greece, took fasting blood samples and assessed diet with a validated questionnaire. Compared with non-drinkers, men consuming more than 200 ml of coffee a day had 50% higher interleukin-6, 30% higher C-reactive protein, 12% higher serum amyloid-A, 28% higher tumour necrosis factor alpha and 3% higher white blood cell counts. In women the same comparisons gave 54%, 38%, 28%, 28% and 4%. The findings held after adjustment for age, sex, smoking, body mass index, physical activity and other covariates. The authors concluded that moderate to high coffee consumption is related to an increased inflammation process.
Those are not small percentages, and this is not a small or sloppy study. It is, however, cross-sectional: everything was measured at one moment, so it cannot tell whether coffee raised the markers, or people with higher markers drink more coffee, or something else drives both. Greek coffee is also typically prepared unfiltered, which matters, as the cholesterol research covered in our article on cafestol shows that brewing method changes what ends up in the cup.
The study that says the opposite
Against it sits a much larger biomarker analysis by Hang and colleagues[2], covering 15,551 women in the Nurses' Health Study and 7,397 men in the Health Professionals Follow-up Study, all free of diabetes, cardiovascular disease and cancer when blood was drawn. Compared with non-drinkers, participants drinking 4 or more cups a day had 16.6% lower C-reactive protein, 8.1% lower interleukin-6, 5.8% lower soluble tumour necrosis factor receptor 2, 9.3% higher total adiponectin and 17.2% higher high-molecular-weight adiponectin. They also had lower C-peptide, leptin and oestrogen measures. The results were largely similar for caffeinated and decaffeinated coffee.
So one cross-sectional study in Greece found substantially higher inflammatory markers in coffee drinkers, and another, much larger cross-sectional analysis in the United States found substantially lower ones. Both adjusted for the obvious confounders. Both are in respectable journals.
Why two careful studies disagree
Several differences could produce this, and it is worth being honest that no one has definitively identified which. Populations differ in what else they eat and in how coffee fits their lives. Brewing differs, and unfiltered coffee carries compounds filtered coffee does not. Cohorts of health professionals are unusually health-conscious, and in such a group the person drinking four cups a day may differ from the abstainer in ways that differ from the general population. And in cross-sectional data, reverse causation is always live: people who feel unwell may drink less coffee, which by itself would make coffee drinkers look healthier.
The honest summary is that cross-sectional biomarker studies cannot settle this. For that you need trials.
What happens when you randomise people to drink coffee
Wedick and colleagues[3] randomised 45 healthy overweight adults, all regular coffee drinkers, to 5 cups a day of instant caffeinated coffee, decaffeinated coffee, or water, for 8 weeks. Compared with no coffee, caffeinated coffee increased adiponectin by 1.4 micrograms per millilitre and increased interleukin-6 by 60%. Decaffeinated coffee decreased fetuin-A, a liver-derived protein, by 20%. There were no significant differences in glucose tolerance, insulin sensitivity or insulin secretion.
That single trial contains the contradiction in miniature. Adiponectin moved in the direction associated with better metabolic health. Interleukin-6 moved in the direction associated with worse. Any article quoting one of those two results and not the other is not misquoting the study, it is choosing from it.
What the trial evidence says when it is pooled
Paiva and colleagues[4] systematically reviewed 15 clinical trials, 8 of coffee and 7 of caffeine. Adiponectin increased in four of seven coffee trials. None of the five studies testing coffee found changes in C-reactive protein, though one of three caffeine trials found it decreased. Interleukin-6 was increased by caffeinated coffee in one of four coffee trials, and by caffeine in three of five. Caffeine increased interleukin-10, an anti-inflammatory signal, in two of three trials. Their conclusion was carefully worded: the data suggest a predominantly anti-inflammatory action of coffee but not of caffeine, and caffeine's mixed pro- and anti-inflammatory responses point to complex effects.
A second review took a harder line. Daneschvar and colleagues[5] screened 1,631 records and analysed 17 randomised controlled trials of coffee and cardiovascular biomarkers. They noted that boiled coffee raised total and LDL cholesterol and apolipoprotein B while filtered coffee did not, and that one study showed a significant increase in interleukin-6 among caffeinated coffee drinkers. Their conclusion was that they could not confidently conclude that an anti-inflammatory effect of coffee is a major contributing factor to the lower all-cause mortality reported in observational studies.
That is an important sentence. Coffee's association with lower mortality is one of the more robust findings in this field, covered in our article on coffee and lifespan. The inflammation explanation for it is popular, intuitive, and not well supported by trials.
Where the anti-inflammatory story does hold up
One place the picture is more consistent is metabolic signalling rather than classical inflammation. Osama and colleagues[6] pooled 69 cross-sectional and cohort studies. Coffee consumption was inversely associated with type 2 diabetes risk, at a pooled relative risk of 0.73 comparing highest with lowest intake. Looking at possible mediators, coffee was associated with higher adiponectin and lower leptin, while interleukin-6 showed no change. They concluded adiponectin and leptin appear to be potential mediators of coffee's effect on diabetes, and interleukin-6 does not.
That is a more specific and better-supported claim than coffee being anti-inflammatory in general, and it lines up with the trial data, where adiponectin is the marker that moves most consistently.
People with inflammatory bowel disease
A specific group worth mentioning, because they search this question often, are people with inflammatory bowel disease. Neamți and colleagues[7] studied 148 patients with Crohn's disease or ulcerative colitis. Sixty percent drank coffee regularly, with no significant difference between the two conditions. About 46% reported no effect on their symptoms, and among those who did report their symptoms worsening, 48% carried on drinking coffee anyway. Faecal calprotectin, a marker of gut inflammation, was significantly lower in coffee drinkers than non-drinkers, particularly for natural coffee and particularly in ulcerative colitis. Other markers showed no differences.
This is a small cross-sectional study and the authors did not claim coffee treats anything. It is a useful illustration that perceived symptoms and measured inflammation can diverge, which is a recurring theme whenever people are asked whether a food makes them feel worse.
Where the anti-inflammatory claim comes from in the first place
The popular version of this story rarely starts with blood markers. It starts with chlorogenic acids, the main polyphenols in coffee, which have antioxidant activity in laboratory assays. Coffee is one of the largest dietary sources of them for many people, and it is a short and appealing step from an antioxidant compound in a test tube to an anti-inflammatory drink in a mug.
How much is actually there varies enormously. Farah and colleagues[8] measured chlorogenic acids and their lactones in regular and water-decaffeinated arabica coffees, and roasting degree is the dominant variable: the compounds break down as roasting proceeds, so a light roast and a dark roast are not delivering the same dose. Angelino and colleagues[9] measured phenolic compounds across widely sold capsule coffees and found substantial differences between products on the shelf. This is the weak link in the chain from compound to health claim. Before you can say coffee reduces inflammation because of its polyphenols, you have to know how much survived the roast, how much made it into the cup, how much was absorbed, and whether any of that changed a marker in a person. The trial evidence above is where that chain is tested, and it does not hold together cleanly.
Does what you put in it change the answer?
Almost every study here measured coffee, not coffee drinks. That distinction matters more than it sounds. Henn and colleagues[10], working across three large US cohorts, examined changes in coffee intake, added sugar and long-term weight gain, and the added sugar is the part relevant here: body fat is itself an active producer of inflammatory signalling, and adiponectin, the marker that moves most consistently with coffee, is made by fat tissue.
So a sweetened iced coffee drunk daily and a black long black are being treated as the same exposure in most of this research, when their effect on the thing that drives low-grade inflammation in the first place, body weight, may differ. None of the studies above were designed to separate that, which is one more reason to hold the conclusions loosely.
Reading this from the Philippines
None of these studies were run in Southeast Asia, and the coffee in them was filtered, instant or boiled European-style rather than the brews common in Filipino homes and cafés. The ATTICA finding in particular came from a population drinking mostly unfiltered coffee, which carries compounds a paper filter removes. Given that brewing method changed the cholesterol results in the cafestol research, it is reasonable to suspect it matters here too, and no study has tested that directly. Anyone extrapolating these marker percentages to a Marikina café latte or a cup of kapeng barako is going well beyond what was measured.
The limits of this evidence
The large studies here are cross-sectional and cannot establish direction. The trials are small, short, mostly 4 to 8 weeks, and mostly in healthy or overweight adults rather than people with inflammatory conditions. Coffee was defined differently everywhere, from instant to boiled to unspecified, and brewing method demonstrably changes the chemistry. Markers fluctuate, and none of these studies followed people long enough to see whether a marker shift changed any outcome. The evidence rating on this article is contested for a reason.
The practical answer
There is no clean answer, and anyone giving you one is not reading the same papers. Coffee is not established as inflammatory or as anti-inflammatory. The most consistent signal across trials and pooled analyses is that coffee raises adiponectin, which is favourable, while its effect on interleukin-6 is inconsistent and sometimes clearly upward, and its effect on C-reactive protein in trials is usually nothing. The large observational finding that coffee drinkers live slightly longer is real, but the inflammation mechanism popularly used to explain it is not supported by the trial evidence. If you are drinking coffee because you enjoy it, none of this is a reason to stop. If you are drinking it as an anti-inflammatory intervention, the evidence does not support that. See also our article on coffee and type 2 diabetes and our article on coffee and the gut microbiome. Nothing here is medical advice.