Biological Age Tests: Expensive Noise or Useful Signal?

The epigenetic clock is one of the more remarkable findings in ageing biology. It is also the basis for a consumer product category that is running well ahead of what the science currently supports for individual use.

Both statements are true, and keeping them apart is the whole exercise.

The underlying science is good

DNA methylation — chemical marks on cytosines, mostly at CpG sites — changes in patterned ways with age. In 2013 Steve Horvath showed that methylation at a few hundred sites could predict chronological age across many tissue types with a median error of a few years. That was a genuinely surprising result.

More useful were the clocks that followed. Rather than predicting chronological age, second-generation clocks such as PhenoAge and GrimAge were trained to predict mortality and morbidity. They perform better at the thing anyone actually cares about: people whose predicted age exceeds their real age have, on average, higher rates of disease and death.

At population scale, this is real and useful.

Where the consumer product breaks

Test-retest reliability. This is the central problem. Work by Higgins-Chen and colleagues in 2022 documented substantial technical noise in first-generation clocks — the same DNA sample, split and run twice, can return biological ages differing by several years. Their proposed fix, principal-component-based clocks, materially improves reliability, and most consumer tests do not use it.

Population validity does not transfer to individuals. A clock that stratifies risk across ten thousand people can be badly wrong about any one of them. The confidence interval around a single reported value is wide, and it is almost never shown on the report.

Almost no intervention has been validated against it. The TRIIM trial is regularly cited as showing epigenetic age reversal. It had nine participants, no control group, and used a drug combination including growth hormone and DHEA. It is hypothesis-generating and nothing more. The 2023 CALERIE analysis of caloric restriction is stronger — randomised, larger — and found a small effect on one clock, not on others.

Tissue specificity. You are sampling blood or saliva. What that says about your heart, brain or kidneys is not established.

What to do instead, if the goal is knowing where you stand

The unglamorous panel that clinicians already use has decades of outcome data behind it and costs less:

• Blood pressure
• ApoB or a full lipid panel
• HbA1c and fasting insulin
• hs-CRP
• Renal and hepatic function
• Grip strength and VO2 max, both of which you can improve and both of which carry robust mortality associations

Every item there has a known relationship to outcomes, a known reference range, and a known response to intervention. None of that is true of your saliva-derived biological age.

The fair verdict

Epigenetic clocks are an important research tool and will likely become a clinically useful one. Today, the consumer version delivers a number with unquoted error bars, unvalidated responsiveness, and considerable capacity to prompt expensive decisions.

If you want one out of curiosity, enjoy it. Do not build a protocol around the result, and be sceptical of anyone selling both the test and the thing that improves it.

Theo Lindqvist covers the wellness industry. Not medical advice.