The bottom line: A study of 6,150 U.S. adults found that higher reported caffeine intake and several caffeine metabolites were associated with younger biological-age profiles, less frailty and lower risk of death. The most consistent signal involved the amount of paraxanthine relative to caffeine in a spot urine sample. This was observational research. It cannot show that caffeine slowed aging, reversed biological age or extended life.
Caffeine research usually begins with a familiar question: how much coffee, tea or soda did people say they consumed? A new study went further. Researchers paired dietary reports with urine measurements of caffeine and 14 compounds produced as the body processes it.
That distinction matters because two people who report the same caffeine intake can have very different levels of caffeine and its metabolites. Genetics, smoking, medications, liver function, recent food and drink, and the time since the last dose can all affect what appears in a urine sample.
The study, published in Nutrition & Metabolism, linked those exposure measures with several ways of estimating biological aging, a 36-item frailty index and deaths recorded over a median follow-up of 8.9 years. The overall pattern was favorable, but it was also complex. The results are better viewed as a clue about caffeine metabolism than as evidence that drinking more caffeine produces healthier aging.
Evidence at a Glance
- Study type: Observational analysis of U.S. National Health and Nutrition Examination Survey data, with cross-sectional aging and frailty analyses plus prospective mortality follow-up.
- Participants: 6,150 adults ages 20 to 80, with a median age of 46.
- Exposure measures: Dietary caffeine estimated from one or two 24-hour recalls, caffeine and 14 related compounds measured in a single spot urine sample, and several metabolite ratios.
- Outcomes: Clinical and DNA methylation-based biological-age acceleration, a 36-item frailty index and all-cause mortality.
- Follow-up: Median 8.9 years, during which 1,016 participants died.
- Evidence level: Early human association evidence. Useful for generating hypotheses, not for proving a treatment effect.
Why It Matters
Caffeine is one of the most widely consumed psychoactive compounds, yet intake alone gives an incomplete picture of exposure. The liver converts most caffeine into paraxanthine, with smaller amounts becoming theobromine and theophylline. Much of this work is carried out by an enzyme called CYP1A2.
The speed and pattern of that conversion vary substantially among people. As a result, a cup count cannot capture everything happening after caffeine enters the body. Measuring downstream metabolites may reveal information about recent exposure and metabolism that a dietary questionnaire misses.
This study is notable because it examined reported intake, individual metabolites and ratios between metabolites in the same population, then compared them across biological age, frailty and mortality. That broader view is more informative than treating every cup of coffee as an identical biological dose.
How the Study Worked
The researchers analyzed NHANES data from 1999 to 2002 and 2009 to 2014. NHANES uses a multistage sampling design intended to represent the U.S. population. Participants reported food and drink intake through 24-hour recalls. Urinary caffeine and related compounds were measured with high-performance liquid chromatography and tandem mass spectrometry, a laboratory method that separates and identifies chemicals at very low concentrations.
The team assessed biological-age acceleration in several ways. Two measures, Klemera-Doubal biological age and PhenoAge, combine routine clinical markers such as blood chemistry into an estimate of how old a person’s physiology appears relative to their chronological age. Positive acceleration means the estimate is older than expected; negative acceleration means younger than expected.
DNA methylation clocks were also available, but only for 619 participants. These clocks examine chemical marks attached to DNA. They are research biomarkers, not direct measurements of how long someone will live.
Frailty was calculated from 36 health deficits, including reported conditions and biochemical measures. Mortality came from records linked to the National Death Index through the end of 2019.
What the Researchers Found
Higher reported caffeine intake, several urinary metabolites and higher metabolite ratios generally tracked with less biological-age acceleration, lower frailty and lower mortality. Many of the individual compounds showing favorable associations were xanthines, a chemical family that includes caffeine, paraxanthine, theobromine and theophylline.
The pattern was not uniform. Caffeine itself in urine was not significantly associated with frailty or mortality. Some associations for reported caffeine intake and individual metabolites also weakened after the researchers adjusted for multiple statistical tests or for caffeine-containing foods.
The paraxanthine-to-caffeine ratio was the standout. In the adjusted mortality model, a one-unit increase in the natural logarithm of this ratio was associated with about a 22 percent lower hazard of death. The same ratio was associated with lower frailty and more favorable values on the two clinical biological-age measures. Its mortality association remained statistically significant after false-discovery-rate correction, a procedure used to reduce the chance of treating random results as real when many comparisons are tested.
Even here, association strength should not be confused with a treatment effect. The ratio was measured once, its value depends on recent caffeine timing and other factors, and the analysis cannot determine which direction the relationship runs.
The Paraxanthine Signal
Paraxanthine is the main product formed when the body breaks down caffeine. Under tightly controlled conditions, the ratio of paraxanthine to caffeine in blood or saliva can provide information about CYP1A2 activity. This study did not use a controlled caffeine dose or timed sampling. It used a single spot urine sample collected after normal, uncontrolled dietary exposure.
That makes the ratio intriguing but difficult to interpret. It may partly reflect caffeine conversion, but it can also be influenced by how recently someone consumed caffeine, urine concentration, kidney handling, medications, smoking, inflammation and other factors.
The researchers therefore described it as an exploratory spot-urine association signal, not a validated measure of a person’s caffeine-metabolizing speed. A different urine ratio with stronger validation for estimating CYP1A2 activity did not show significant associations with aging outcomes. That finding argues against a simple conclusion that faster caffeine metabolism is necessarily healthier.
Biological Age May Be Part of the Pattern, but Mediation Is Not Proof
The researchers also asked whether biological-age acceleration might statistically account for part of the association between caffeine-related measures, frailty and mortality. The two physiology-based age measures produced several mediation signals, especially for the paraxanthine-to-caffeine ratio.
Mediation analysis separates an observed association into statistical pathways. In an observational dataset with measurements taken at one point in time, it cannot establish that caffeine changed biological aging, which then changed frailty or survival. The authors treated these results as exploratory, and the mediation tests were not corrected for multiple comparisons.
Limitations and Caveats
The design cannot establish cause and effect. Biological age and frailty were analyzed cross-sectionally. Poorer health could change caffeine habits or metabolism, rather than caffeine exposure changing aging.
Exposure was measured over a short window. Dietary estimates came from one or two 24-hour recalls, and metabolites came from one spot urine sample. Neither necessarily represents a person’s long-term caffeine pattern.
Coffee is more than caffeine. Coffee was the main caffeine source and contains chlorogenic acids, trigonelline and many other compounds. When the researchers adjusted for caffeine-containing foods, several associations for caffeine intake and individual metabolites became weaker or disappeared.
Residual confounding remains possible. The models accounted for age, sex, body mass index, smoking, alcohol, education, income and other factors, with additional sensitivity analyses for influences on CYP1A2. Unmeasured or imperfectly measured differences could still explain some of the pattern.
The epigenetic subgroup was small. DNA methylation results were available for only 619 participants, limiting statistical power and confidence in those clock findings.
Multiple testing raises false-positive risk. The study examined many exposures, outcomes, subgroups and possible pathways. The authors applied false-discovery-rate correction to primary analyses, but not to all exploratory tests.
The paper is an accepted manuscript. It has passed peer review and carries a permanent DOI, but Springer notes that this early version will still undergo copyediting and may contain errors before the final version of record appears.
What This Means
The study strengthens the case for looking beyond self-reported caffeine intake. How the body processes caffeine may carry useful information that questionnaires alone cannot capture. The paraxanthine-to-caffeine ratio deserves replication in independent cohorts using repeated samples, standardized timing and better genetic and metabolic characterization.
It does not justify increasing caffeine intake, using caffeine to target biological age or treating a urine metabolite ratio as a longevity test. Caffeine can disrupt sleep, worsen anxiety and cause palpitations in some people. Individual responses, medications, pregnancy and cardiovascular conditions also matter.
The most defensible conclusion is narrower: in this nationally representative observational sample, several caffeine-related measures, especially a spot-urine paraxanthine-to-caffeine ratio, tracked with healthier aging markers and lower mortality. Whether that signal reflects caffeine metabolism, other compounds in coffee, underlying health, behavior or some combination remains unresolved.