Does the same biological age test give the same result when taken twice?
The same test yields reasonably consistent results when repeated, but different tests do not measure the same thing and are largely interchangeable. Only compare results when the test and the measurement platform are identical.
Taking the same test at two different points in time generally produces comparable results, but this holds more strongly for some tests than for others. The DunedinPACE DNA methylation test scores well on repeatability: during the development of the test, measurement sites with low reliability were already filtered out, which benefits consistency. The MRI-based brain age test also performed well within the same scanner, with a reliability coefficient of 0.90 to 0.99. When switching to a different scanner, the grey matter measurement remained reliable (0.83 to 0.96), but the white matter measurement fell short (0.51 to 0.77).
There are, however, two serious pitfalls. First, the platform on which the measurement runs matters. When existing epigenetic clocks are used on a newer generation of measurement equipment, the results can differ substantially, because some of the measurement sites on which the clock was built are simply no longer available. In such cases, the outcomes of the same clock can vary considerably.
Second, different biological age tests run on the same person do not always agree. A comparison of eleven methods, including telomere length, epigenetic clocks and composite biomarker scores, in nearly a thousand people revealed low mutual agreement. The tests therefore do not all measure the same aspect of ageing. This was also evident in people with schizophrenia: one clock indicated accelerated ageing, while two other clocks showed no difference.
In practical terms, this means: if you have taken the same test twice on the same platform, the results are reasonably comparable. If you compare two different tests with each other, or use an older clock on a new measurement platform, the results are far from consistent.
This answer draws on five studies, including comparisons of multiple measurement methods in hundreds of people. The findings on repeatability within a single test are reasonably consistent, but the evidence on the interchangeability of tests is more limited and partly contradictory. No studies were included that randomly assigned participants or followed them over the long term.