Did a 4-Week Diet Reverse Biological Age? What the Sydney Trial Actually Measured

TL;DR: In a 2026 University of Sydney trial, 104 adults aged 65 to 75 ate one of four fully provided diets for 4 weeks. Three of the diets moved a biomarker score called KDM biological age by roughly 3 to 4 points compared with a diet close to what people already ate, but only some of those differences were statistically significant. The score is built from routine blood and body measures, not DNA methylation, and the authors themselves say the shift may reflect short-term physiology rather than slower aging.

Table of Contents

• The headline claim and where it came from
• What the trial actually did
• What “biological age” meant in this study
• What the authors say the result can and cannot mean
• What you can reasonably take from it
• What evidence would be needed for a real claim
• FAQ
• References

The Headline Claim and Where It Came From

In September 2026 a story moved through science news feeds with a striking headline: scientists had reversed biological age in older adults in just four weeks, with nothing more than a change of diet. It is the kind of claim that gets shared, screenshotted and turned into a meal plan within days.

The study behind it is real and worth reading. It was published in Aging Cell in April 2026 by Caitlin Andrews, Alistair Senior and colleagues at the University of Sydney’s Charles Perkins Centre, under the much more modest title “Short-Term Dietary Intervention Alters Physiological Profiles Relevant to Ageing.” That title is closer to what the data show. The gap between the two framings is the whole story of this article.

So let’s take the paper on its own terms. What did the researchers do, what did they measure, and what does a lower score on their measure actually tell you?

What the Trial Actually Did

The paper is a secondary analysis of an earlier study called the Nutrition for Healthy Living (NHL) trial, registered in the Australian New Zealand Clinical Trials Registry as ACTRN12616001606471 and described by Rosilene Ribeiro and colleagues in Nutrition and Healthy Aging in 2019. The NHL trial was a single-blinded, randomised feeding trial with a 2 x 2 factorial design. “Secondary analysis” matters here: the biological age question was asked of data collected in a trial that was set up to study protein sources and healthy ageing more broadly.

Who took part

The analysis includes 104 Australian adults aged 65 to 75, 34 men and 70 women, with complete biomarker data. To enter the trial they had to have a BMI between 20 and 35, be non-smokers and non-vegetarians, and be free of type 2 diabetes, cancer, kidney or liver disease and food allergies. In other words, these were relatively healthy older Australians, not people with metabolic disease.

The four diets

Every diet supplied 14% of energy from protein. What changed was where the protein came from and how the rest of the energy was split between fat and carbohydrate:

• OHF (omnivorous, higher fat): about half of protein from animal sources, 37 to 41% of energy from fat, 41 to 43% from carbohydrate. This was the diet closest to what participants already ate, and it is the reference group in the analysis.
• OHC (omnivorous, higher carbohydrate): same protein split, 28 to 29% fat, 53% carbohydrate.
• VHF (semi-vegetarian, higher fat): about 70% of protein from plants, 37 to 41% fat, 41 to 43% carbohydrate.
• VHC (semi-vegetarian, higher carbohydrate): about 70% of protein from plants, 28 to 29% fat, 53% carbohydrate.

All meals were delivered to participants’ homes for four weeks. People could eat as much as they wanted, and more food was supplied than they needed. Weighed food records showed high compliance with the target diets. That level of control is a genuine strength. Most diet studies rely on people remembering what they ate.

One detail is easy to miss and matters a lot: the authors describe the carbohydrates in these diets as complex, whole and minimally processed. “Higher carbohydrate” here does not mean more white bread and sugar.

What “Biological Age” Meant in This Study

When most people hear “biological age” in 2026 they think of an epigenetic clock, a test that reads chemical marks on DNA. No epigenetic clock was measured in this trial. The researchers used the Klemera-Doubal Method (KDM), which estimates a biological age from a set of ordinary clinical measurements.

KDM works by learning, in a large reference population, how each biomarker tends to change with chronological age, then combining a person’s values into one age estimate. Here the reference population was the US National Health and Nutrition Examination Survey (NHANES IV), using non-pregnant adults aged 30 to 75, with separate models for men and women. The result is expressed as deltaAge: KDM age minus chronological age. A negative deltaAge means your measurements look like those of a younger person in the reference data.

The team built two versions of the score for the trial:

• KDM-S (7 markers): C-reactive protein, albumin, total cholesterol, systolic blood pressure, alkaline phosphatase, blood glucose and urinary creatinine.
• KDM-M (15 markers): the KDM-S markers plus HDL cholesterol, diastolic blood pressure, insulin, the AST/ALT liver enzyme ratio, total bilirubin, triglycerides, waist circumference and BMI.

Look at that list and the core issue becomes obvious. Cholesterol, glucose, insulin, triglycerides, blood pressure, CRP and waist size are exactly the kinds of numbers that respond quickly when you change what people eat. A score built from them will move when the diet moves. That does not make the score useless. It does mean a fast change in the score is not, by itself, evidence that the underlying rate of aging changed.

The authors did check that their scores track something real. In NHANES IV, KDM-S correlated with chronological age at r = 0.92 and KDM-M at r = 0.82. In men, a higher KDM-M deltaAge was linked to higher mortality (hazard ratio 1.44 per standard deviation, 95% CI 1.21 to 1.71). In women in that same test data set, none of the KDM measures significantly predicted mortality. So the scores carry information about risk in populations, unevenly, but that is a different question from whether a four-week shift in one person means anything for their future.

We covered how much these scores can wobble for reasons that have nothing to do with aging in our comparison of biological age tests. The same caution applies here.

The numbers

At baseline, participants already scored younger than their chronological age on average, meaning negative deltaAge. After four weeks, the OHF group, eating close to their usual diet, showed no meaningful change. The other three groups showed lower deltaAge. The key analysis compared each diet with OHF after four weeks, adjusting for each person’s starting score. The figures below are regression coefficients in years of deltaAge relative to OHF, with standard errors (SE) and p-values:

Diet vs OHF	KDM-S (7 markers)	KDM-M (15 markers)
OHC (omnivorous, higher carb)	-3.51 (SE 1.62), p = 0.032	-4.15 (SE 1.69), p = 0.016
VHF (semi-vegetarian, higher fat)	-3.52 (SE 1.62), p = 0.032	-3.22 (SE 1.68), p = 0.057, not significant
VHC (semi-vegetarian, higher carb)	-3.14 (SE 1.62), p = 0.056, not significant	-2.93 (SE 1.68), p = 0.083, not significant

Three things stand out. First, only the omnivorous higher-carbohydrate diet was significantly different from the reference on both versions of the score. Second, half of the six comparisons did not reach the study’s own threshold for significance (p < 0.05). Third, the standard errors are large relative to the estimates, which is what you expect with 104 people split across four diets. The paper does not report confidence intervals for these coefficients, and it does not report body weight change in the main text, so we won’t guess at either. Adding sex to the models did not change the conclusions.

Notice what these numbers are. They are differences between groups in a composite score after a month, compared with a diet close to normal eating. They are not “years of life gained,” and they are not a measurement of anyone becoming younger.

What the Authors Say the Result Can and Cannot Mean

To their credit, the Sydney team is far more careful than the headlines. Their own reading of the data comes down to five points:

• Acute response, not necessarily slower aging. Many of the biomarkers in KDM are tightly regulated by the body and shift with diet, activity, stress or medication. The authors say the improvements may reflect short-term physiological responses to recent eating rather than a change in the trajectory of aging, and that the study was not designed to tell those apart.
• No persistence data. Nobody was followed to see whether the lower scores lasted after the diets ended, or whether they predicted any health outcome.
• Several results were not significant. The semi-vegetarian diets pointed in the same direction, but most of their comparisons missed the significance threshold. The authors argue for larger trials rather than claiming an effect.
• Mechanisms were not measured directly. The authors suggest the balance of fat and fibre may matter at least as much as protein source, but this trial did not test that directly.
• Do not extrapolate to refined carbohydrates. The higher-carbohydrate diets used complex, minimally processed carbohydrates, and the authors explicitly warn that the findings should not be extended to diets high in refined or simple carbohydrates.

They also make a useful comparison. Biological age estimates are known to shift temporarily during transient states such as pregnancy and then return. A short-term change in a biological age score can be real as a change in physiology without telling you anything about long-term aging.

Put plainly: This trial showed that a blood-and-body-measure score called KDM biological age moved within 4 weeks of a diet change. It did not show that aging itself was reversed or slowed.

We saw a similar pattern when we looked at the claim that hyperbaric oxygen therapy reverses aging: a real change in a biomarker or a function, reported as if it were proof of reversed aging. The distinction is the same here.

What You Can Reasonably Take From It

If you are between 40 and 75 and wondering whether to change how you eat because of this study, here is a sober reading of what it supports.

What it supports

• Diet changes your metabolic and inflammatory markers quickly. In a tightly controlled setting, healthy people aged 65 to 75 saw a composite of cholesterol, glucose, insulin, blood pressure, inflammation and related measures shift within a month.
• The pattern that did best was not exotic. The only diet with a consistent, significant difference was omnivorous, with about half its protein from animal sources, and a little more of its energy from complex, minimally processed carbohydrates and less from fat. No supplements, no fasting protocol, no special foods.
• The direction fits the wider evidence. The authors note that semi-vegetarian and higher complex-carbohydrate patterns are linked to better metabolic health in other studies. This trial adds a short, controlled data point in the same direction.

What it does not support

• Any claim that a diet makes you years younger. The trial measured a score, for a month, in a small group.
• Cutting protein. Every diet held protein at 14% of energy. The trial says nothing about eating less protein, and older adults’ protein needs are a separate question, covered in why older adults need more protein than you think.
• More refined carbohydrates. The authors specifically rule this out.
• Buying a biological age test to track a 4-week diet. If a score built from routine labs moves within weeks of a diet change, a single before-and-after test mostly tells you your recent labs changed. Your standard blood panel, reviewed with your doctor, gives you the same information without the “age” label.

If you are on medication for blood pressure, cholesterol or blood sugar, talk to your doctor before making a large change to your diet, since the very markers in this score are the ones those medicines target.

What Evidence Would Be Needed Before Anyone Claims a Diet Reverses Aging

A claim that a diet reverses or slows aging would need much more than a four-week change in a composite score. At a minimum:

1. Longer trials. Months to years, not weeks, so that short-term swings in blood markers settle.
2. Follow-up after the diet ends. Does the change persist, or does the score return to baseline within weeks?
3. Hard outcomes. Fewer heart attacks, less diabetes, less frailty, better function, longer survival. A score is only as useful as its link to these.
4. Several validated measures that agree. For example, a biomarker score like KDM alongside validated epigenetic clocks and functional tests, all moving in the same direction.
5. Larger samples and pre-registered outcomes. Enough people per group that a 3 to 4 point difference is clearly distinguishable from noise, with the biological age outcome declared in advance.

The Sydney authors call for exactly this kind of work. Until it exists, the honest summary is that diet quickly changes the numbers that biological age scores are built from, and whether that changes how fast you age is still an open question.

FAQ

Did the diet reverse biological age?

No. Three diets lowered a biomarker score called KDM deltaAge compared with a near-usual diet after four weeks, and only some of those differences were statistically significant. The authors say this may reflect short-term physiology, not slower aging.

Was an epigenetic clock used?

No. The study used the Klemera-Doubal Method, which combines routine measures such as cholesterol, glucose, blood pressure, CRP and albumin. No DNA methylation was measured.

Which diet did best?

The omnivorous higher-carbohydrate diet (14% protein, 28 to 29% fat, 53% complex carbohydrate) was the only one significantly different from the reference on both versions of the score.

Should I go vegetarian to lower my biological age?

This trial does not show that. The semi-vegetarian diets moved the score in the same direction, but most of their comparisons were not statistically significant, and none of the results speak to long-term aging.

How long was the study?

Four weeks, with all meals provided. There was no follow-up after the diets ended.

If you want the full system behind how I decide which studies change what I eat and do, I wrote it all down in The MVHK Protocol.

References

1. Andrews CJ, Ribeiro RV, Gosby A, Le Couteur DG, Raubenheimer D, Tan J, Simpson SJ, Senior AM. Short-Term Dietary Intervention Alters Physiological Profiles Relevant to Ageing. Aging Cell. 2026;25(5):e70507. PMID: 42037549. PMCID: PMC13112187.
2. Ribeiro RV, Simpson SJ, Le Couteur DG, Raubenheimer D, Eberhard J, Ruiz K, Macia L, Todorova Y, Gosby AK. The nutrition for healthy living study: A randomised clinical trial assessing the effect of protein sources on healthy ageing. Nutrition and Healthy Aging. 2019;5(1):43-51. DOI: 10.3233/NHA-180055. Trial registration: ACTRN12616001606471.