Years of Exercise, Blood Pressure Drugs Failed to Slow Cognitive Decline in Seniors at Dementia Risk

They were, by any reasonable measure, doing everything right. The 513 older adults who enrolled in the Risk Reduction for Alzheimer’s Disease trial (rrAD, to the researchers running it) weren’t sedentary people shrugging at their health. They were people with something concrete at stake: a parent lost to dementia, perhaps, or that creeping private awareness that their own memory wasn’t quite what it used to be.

Over two years, some of them exercised at moderate to vigorous intensity for nearly three hours a week. Others took statins and blood pressure medication aggressive enough to push their systolic readings below 130 millimetres of mercury. A third group did both. They were, in effect, running a personal experiment in dementia prevention.

The results, published in JAMA Neurology this week, found no significant cognitive benefit from any of it.

Rong Zhang at the Texas Health Presbyterian Hospital Dallas led the trial alongside colleagues at four US sites. The core question was whether combining aerobic exercise with intensive cardiovascular risk reduction would do more for brain health than either intervention alone, a reasonable hypothesis, since both approaches work through different mechanisms and there were good reasons to think their effects might compound. Exercise boosts cerebral blood flow and promotes neuroplasticity; aggressive treatment of hypertension and high cholesterol reduces the vascular damage that quietly undermines cognition over decades. On paper, the combination looked like a sensible one-two punch.

Except it didn’t work. Not the exercise group. Not the medication group. Not the combined group either.

The team measured cognitive performance using two established batteries: the Preclinical Alzheimer Cognitive Composite, which leans toward memory function, and the NIH Toolbox Cognition Battery, weighted toward executive function. Both showed improvement over the 24-month trial period, but all four groups improved by roughly the same amount, including the people who received only usual care. The difference between the exercise group and the non-exercise group on the primary outcome was a tenth of a point on a composite z-score. Statistically indistinguishable from zero.

The interventions did work physiologically. Exercise participants maintained their cardiorespiratory fitness over two years while the control group’s fitness quietly declined, a finding consistent with what decades of exercise research would predict. The medication group achieved meaningful reductions in both systolic blood pressure (roughly 5 millimetres of mercury lower than the no-treatment group) and LDL cholesterol (about 17 milligrams per decilitre lower). The biology was responding. The cognition wasn’t, at least not within the study’s timeframe.

This is genuinely difficult to interpret, partly because the field is muddier than the popular press tends to acknowledge. Observational studies consistently show that people who exercise more and keep their blood pressure controlled have lower dementia rates. But randomised trials, which can actually test causality rather than just association, have been less convincing. A 2023 umbrella review found that the apparent benefits of exercise on cognition largely evaporated after correcting for publication bias. The SPRINT MIND trial, which tested intensive blood pressure lowering in a large population, found no statistically significant improvement in cognitive test scores either; though it did reduce the incidence of mild cognitive impairment and dementia when those outcomes were combined.

That last distinction matters. There’s a difference between improving cognitive function in people who are still cognitively normal and preventing a slide into impairment. The rrAD trial was designed to measure the former. It’s possible, perhaps even likely, that the protective effects of these interventions operate on a longer timescale than two years, or manifest as a reduced rate of eventual decline rather than near-term measurable gains. The fact that the American Heart Association now recommends keeping systolic blood pressure below 130 mm Hg specifically to help prevent cognitive decline, the threshold both treatment and control arms effectively reached, complicates the picture further. Zhang and colleagues note this explicitly: by the time the trial ran, the usual-care group was already receiving fairly aggressive cardiovascular management from their own doctors.

There are other possible explanations for the null result. The improvements seen across all groups may partly reflect practice effects (people simply getting better at the cognitive tests through repeated exposure) and an expectancy effect from participating in a clinical trial at all. The sample size, originally planned at 640 participants and trimmed to 510 for logistical reasons, may not have been large enough to detect modest true effects. And crucially, participants weren’t stratified by amyloid status or other biological markers of Alzheimer’s risk. It’s quite possible that the intervention worked well in some subgroup, and the trial couldn’t see it.

What does appear to work, somewhat frustratingly, is doing more. The FINGER trial in Finland and, more recently, the large US POINTER trial both showed that genuinely multicomponent interventions combining exercise, dietary change, cognitive training, and vascular risk monitoring can improve cognitive function in at-risk older adults. Exercise alone, or medication alone, or even the two together, may not be enough.

Zhang’s team call for longer trials with richer participant profiling, including brain imaging and blood-based biomarkers, to identify who would actually benefit from which combination of interventions. That work is coming; the infrastructure for it is being built across multiple research programmes right now, and the question of dementia prevention has never had more resources pointed at it. The rrAD trial’s contribution is, in its way, an important one. Ruling out what doesn’t work, under these conditions, in this population, over this timeframe, narrows the field.

The 513 people who exercised for two years and still showed no cognitive edge over those who didn’t: they aren’t a cautionary tale. They’re a data point. And the data are still being assembled.

DOI / Source: 10.1001/jamaneurol.2026.0359