Why Repeat Memory Tests Can Flatter Aging Brains: A New Analysis of the ACHIEVE Hearing Trial

Researchers behind one of the largest hearing studies in the United States report that simply taking the same cognitive tests again can lift scores by roughly the equivalent of one year of age-related decline.

Studies that follow older adults over several years usually measure thinking and memory with the same battery of tests, repeated at regular visits. It is a sensible design, but it carries a quiet complication: people get better at tests they have seen before. That improvement has nothing to do with healthier brains and everything to do with familiarity.

A new analysis published in JAMA Network Open puts numbers on that problem using data from the ACHIEVE study, a landmark randomized trial that enrolled older adults with untreated hearing loss. The question the authors asked is one that matters for the entire field of aging research: how much do these practice gains distort what we think we know about cognitive decline, and do they change the apparent results of interventions?

Background: Why the Researchers Looked at This

The ACHIEVE trial, short for Aging and Cognitive Health Evaluation in Elders, is one of the most closely watched studies in hearing research. Conducted from 2017 to 2022 at four community sites in North Carolina, Mississippi, Minnesota, and Maryland, it enrolled community-dwelling older adults with untreated hearing loss and followed their thinking and memory for three years. Its results have shaped how scientists discuss the relationship between hearing and cognitive health in later life.

Any trial like this depends on repeated cognitive testing. Here the complication known as the retest effect, sometimes called the practice effect, enters the picture. When a person sits the same memory or problem-solving tasks a second or third time, their score tends to rise because the format is familiar, the strategies are rehearsed, and the anxiety is lower. In a study of aging, that artificial lift can mask genuine decline.

The authors of this new analysis wanted to measure the size of that lift in ACHIEVE, check whether it differed across groups of participants, and answer a practical question for the field: if you correct for retest effects, do the estimated benefits of an intervention change?

How the Study Was Done

The team analyzed 977 older adults from the ACHIEVE trial, with an average age of 76.3 years, 54 percent of them women. Together these participants contributed 2,571 in-person cognitive assessments across the three-year follow-up. About 57 percent had mild hearing loss, and 76 percent were recruited fresh from the community rather than from an earlier study.

Cognition was summarized as a global factor score drawn from a full neurocognitive battery, along with separate scores for language, executive function, and memory. To isolate the practice effect, the researchers used statistical models that included a simple flag: zero for a participant's first visit, one for any follow-up visit. The difference attached to that flag, after adjusting for the same covariates used in the main ACHIEVE analysis, represents the score change associated with having taken the tests before.

The authors then compared cognitive trajectories and intervention-versus-control estimates with and without this retest adjustment, and tested whether the retest effect varied by hearing loss severity, by trial arm, or by how participants were recruited.

What the Researchers Found

Retest effects were real and sizable. Global cognition scores rose by 0.11 standard units simply from repeat testing, with a 95 percent confidence interval of 0.07 to 0.15. Executive function showed the largest bump at 0.15 units, and memory rose by 0.12 units. In practical terms, the authors estimate this lift roughly offsets one full year of model-estimated age-related cognitive decline.

Language tests behaved differently. There, the retest difference was minimal and statistically indistinguishable from zero, suggesting vocabulary and naming tasks are less coachable through familiarity than memory puzzles or speeded executive tasks.

Importantly, the size of the practice effect did not meaningfully differ by hearing loss severity, by whether a participant was in the intervention or control group, or by recruitment source. The boost showed up across the board.

The most reassuring result came last. When the researchers re-ran the trial's comparisons with the retest adjustment in place, the estimated cognitive trajectories shifted, but the estimated effects of the hearing intervention itself did not meaningfully change. In other words, practice effects were inflating everyone's scores about equally, so the between-group comparisons held up.

What It Means for People with Hearing Loss

For anyone following the science on hearing and brain health, this analysis is quietly good news. A common worry about long-running studies is that hidden measurement quirks could be driving headline results. Here, a rigorous check found that one of the best-known quirks, the practice effect, was substantial but did not distort the trial's core comparisons between groups.

The findings also carry a caution for interpreting any single set of test scores. An older adult who scores a little better at a second memory assessment has not necessarily gained ground; some of that gain may be familiarity with the test itself. Researchers, and clinicians who track cognition over time, get a clearer picture when they account for this.

Finally, the study strengthens the foundation under future research on hearing and cognition. The authors recommend that longitudinal studies routinely adjust for retest effects, which should make the next generation of findings about hearing care and brain health easier to trust.

Careful Evidence on Hearing and Aging Supports Everyday Hearing Care

Trials like ACHIEVE exist because hearing in later life is now understood as a mainstream health topic, not a niche concern. The same period that produced this research also produced FDA-OTC hearing aids, a category created to make hearing devices easier to obtain for adults with mild to moderate hearing loss. People with severe or profound loss still benefit most from clinical fittings.

Panda Quantum sits in that category. It is a 16-channel receiver-in-canal device with adaptive noise reduction aimed at clearer speech in noisy environments, Bluetooth for calls, TV, and music, 20 hours of use per charge with a case that holds three additional full charges, a 5-year warranty, and a 45-day trial that starts when the device arrives.

Quantum also offers an optional companion app with an optional in-ear hearing test that can personalize gain and frequency response to a user's hearing profile, a form of app-based hearing personalization. The device works fully out of the box without the app or the test. Details are at pandahearing.com.

Limitations of This Research

This was a secondary analysis rather than a study designed from the start to measure practice effects, and it modeled retesting as a simple first-visit versus later-visit flag, which cannot capture how familiarity builds across a third or fourth assessment. Participants were community-dwelling adults averaging 76 years of age with untreated hearing loss at enrollment, so the size of the retest effect may differ in younger, older, or clinical populations.

Readers should also note that one coauthor lists an affiliation with Amplifon SpA, a hearing device company, which is disclosed in the article's author information.

Where This Leaves Us

Repeat testing flatters aging brains by about a year's worth of decline, and careful studies need to account for that. The encouraging part is that when the ACHIEVE researchers did exactly this, their conclusions about hearing intervention stood firm, a small but meaningful vote of confidence in the evidence base connecting hearing care and healthy aging.

Wang Y, Pike JR, Deal JA, Lin FR, Reed NS, Garcia-Morales EE, Huang AR, Jabs DA, Gross AL. Differences in Cognitive Aging Assessments Associated With Retest Effects. JAMA Network Open. 2026;9(8):e2628782. Retrieved from PubMed. https://doi.org/10.1001/jamanetworkopen.2026.28782