A Danish Study of 1,199 Adults Says Gut Bacteria Predict Inflammation Better Than Age. Here’s What “37 Years Earlier” Actually Measured

The most quotable line in gut science this month is “37 years earlier,” and it comes from a paper in Nature Communications that, as of this writing, has no press release, no author interviews and no coverage outside one research-news paraphrase. Thirteen researchers at the Technical University of Denmark, Aalborg University and the University of Copenhagen took 1,199 Danish adults aged 20 to 72, measured 30 inflammation-signalling molecules in their blood and 19 physiological markers, sequenced their gut bacteria, and asked a plain question: which predicts a person’s inflammatory tone better, their age or their microbes? The microbes won, for 97 percent of the molecules. And one bacterial pattern, present in about one adult in six, looked in its twenty-year-olds the way the other patterns looked at nearly sixty.

That is a striking result and an easy one to over-read, so we read the full text and the 39-page supplement rather than the abstract. What follows is what was measured, what the headline numbers compare, what the authors say the finding cannot show, and why none of it is something a mail-in stool kit can tell you about yourself.

The studyWho was measured, and how

The people come from DanFunD, a general-population cohort of 9,656 adults from greater Copenhagen examined between 2011 and 2015. Of 7,493 adults in the relevant wave, 4,101 had a 30-cytokine blood panel, 2,460 provided a stool sample, and 1,199 had both, which is the core group. Bacteria were profiled by 16S ribosomal RNA sequencing, which identifies genera rather than individual species or genes; that is a coarser lens than the shotgun sequencing used in some larger studies, and worth keeping in mind. The blood panel included the molecules anyone who has read about “inflammaging” would expect, among them interleukin-6, tumour necrosis factor alpha, interferon gamma and C-reactive protein, alongside cholesterol fractions, triglycerides, blood pressure, glucose control and cardiorespiratory fitness.

The first analysis is the one behind the title. Age, on its own, was associated with 40 percent of the cytokines and most of the physiological measures, as expected: low-grade inflammation rises with the years, a pattern named “inflamm-aging” in 2000. But when the researchers partitioned the variance, gut microbiota composition explained more of it than age did for 29 of the 30 cytokines and 16 of the 19 physiological variables. That is the “outperforms chronological age” of the title, and it was reproduced, for the measures the two datasets shared, in MetaCardis, a separate European cohort.

The four patternsEnterotypes, and the one that stood out

To turn thousands of bacterial counts into something a statistician can compare, the team sorted the 1,199 microbiomes into clusters by a standard unsupervised method, and the data settled on four. Three are familiar from the 2011 paper that coined the word enterotype: a Bacteroides-dominant group, a Prevotella-dominant group, and a high-diversity group the authors call Richness. The fourth is the one that matters here. Bacteroides 2 is Bacteroides-dominant too, but with the lowest diversity of the four, roughly half the Richness group’s on the standard index, and an enrichment of bacteria that recur in inflammation research: Escherichia/Shigella, Ruminococcus gnavus, Lactobacillus. This configuration was first described in 2017 as a low-cell-count pattern over-represented in Crohn’s disease, and a 2020 Nature paper found it becomes more common as body-mass index rises and less common in people taking statins. Seventeen percent of the Danish sample fell into it.

Two cautions from the authors themselves belong next to the table. Enterotypes, they write, “represent dominant community patterns rather than discrete biological states”; a well-known 2014 critique made the same point, that the distribution “is continuous and can vary widely within an individual.” Nobody has a B2 gut the way they have a blood type. They have a microbiome that, on the day of sampling, sat closer to that end of a gradient than to the others.

The 37 yearsWhat the number is, and what it is not

Here is the sentence the headline comes from, exactly: “Individuals with the B1, P, and R enterotypes reached cytokine levels comparable to those of 20-year-old B2 individuals only after an additional ~37 years.” After adjusting for age, B2 participants ran higher on five of the sixteen cytokines that normally climb with age, namely interferon gamma, interleukin-6, tumour necrosis factor alpha, C-reactive protein and VEGF, and on four physiological measures, HDL and LDL cholesterol, triglycerides and VO₂max, the fitness marker. Plotted against age, the B2 line for those molecules started where the other lines finished. That is what “an inflammaging-like phenotype up to 37 years earlier” means: not that these people are biologically 57 at 20, but that on a specific handful of blood markers, their twenties resembled other people’s late fifties.

Three details in the full text soften the picture without dissolving it. The values were “within standard clinical reference ranges” throughout, so this is a shift inside normal, not an abnormal result. The offset was largest in young adults and the lines converged by around 65, when everybody’s inflammatory tone has risen; the B2 pattern seems to bring the curve forward rather than lift it permanently. And the comparison is cross-sectional: different people at different ages, sampled once, not one person followed for 37 years.

The disease findingA hazard ratio of 1.21, read carefully

The second half of the paper is the reason it will be cited. Denmark records every hospital-associated diagnosis in a national register, so the researchers could follow 2,259 people with a stool sample forward from 2013–15 to the end of August 2022, an average of eight years and a maximum of nine, and count the first appearance of any new diagnosis in each chapter of the ICD-10 code book. Against the other three enterotypes pooled, with age as the clock and adjusting for sex and smoking, the B2 group’s hazard of a new diagnosis was 1.21, with a 95 percent confidence interval of 1.07 to 1.36. It was raised in seven of thirteen chapters, most for infections (1.53), respiratory (1.46) and digestive (1.44) diseases, then nervous, musculoskeletal, circulatory and eye conditions. The Richness group ran the other way, at 0.79.

Plainly: a relative hazard about a fifth higher, across a broad range of conditions, in a pattern that a sixth of adults carried. The paper reports no absolute risks, so neither do we; a fifth higher hazard of a common outcome and of a rare one are very different things, and this study pools them. The mediation analysis is the part most likely to be skipped and most worth reading. The inflammatory cytokines that gave the paper its title accounted for only 4.1 percent of B2’s overall excess risk, rising to about a quarter for respiratory and infectious disease. Whatever links the low-diversity pattern to later diagnoses, most of it is not running through the 30 molecules measured here. The authors are direct about the limits: “causal relationships cannot be inferred”; the B2 pattern already co-occurred with more disease at baseline; older participants who volunteered a stool sample may have been healthier than average; diet was not fully accounted for.

What to do with itNothing you can buy, one thing you can do

The obvious next thought is to find out which pattern you are, and this is where a careful reader stops. No consumer microbiome test reports an enterotype defined the way this paper defines it, and when one stool sample was sent to seven labs, it came back with seven different answers about what was in it. A composition that shifts along a gradient, sampled once, with 16S sequencing, is not a diagnostic; the paper does not present it as one, and the companies that would like to sell you the idea have not earned it either.

What the paper does support is old advice with a new reason. The protective pattern here is the high-diversity one, and in the largest citizen-science microbiome dataset, diversity tracked with how many different plants people ate in a week, which is a lever that Americans have been pulling less even as probiotic sales climb. Working out where your own intake sits is a more useful response to this study than a stool kit. The B2 pattern, meanwhile, is a research signal: a description of one end of a spectrum that inflammation researchers will now follow forward, ideally in the same people over time, which this study could not do. It is a good paper, and its headline is true. Its fine print is where the meaning is. This article is educational and is not medical advice.