A lower score on a cognitive test is usually interpreted as information about how well a person is thinking at that moment. New longitudinal research suggests that, for people in their 60s, it may also contain information about vascular health years before a stroke occurs.
Researchers following 4,912 older adults in southern Sweden found that participants aged 60 to 69 with the lowest overall cognitive performance had an 86% higher risk of a first stroke than those with the highest performance. Lower memory and information-processing speed were also associated with substantially higher risks. The same pattern did not remain statistically clear in participants who entered the study in their 70s or 80s after health and lifestyle factors were taken into account.
The findings, published in the Journal of the American Heart Association, do not mean that poor memory or slower thinking causes stroke. Instead, they raise a more subtle possibility: cognitive performance in the early years of older adulthood may reveal underlying vascular or brain changes that conventional risk assessment does not fully capture.
Following nearly 5,000 adults for more than a decade
The study drew on the Good Aging in Skåne project, a population-based longitudinal study in Sweden. The analysis included 4,912 adults aged 60 years and older who had no history of dementia, stroke or transient ischaemic attack when they entered the study. Forty-six per cent of participants were men.
Baseline and follow-up data were collected between 2001 and 2023. Participants completed standardised cognitive assessments covering memory, processing speed, verbal fluency and executive function. The researchers also created an overall cognitive score by combining available test results. Within each age group, performance was classified as low, middle or high so that people were compared with peers of a similar age.
Stroke outcomes were identified through the Swedish National Patient Register, which captures specialised inpatient and outpatient care. Over an average follow-up of 12.3 years, 572 participants experienced a first stroke. That represents 11.6% of the study population.
The long follow-up is important. Rather than asking whether people who had already experienced a stroke performed poorly on cognitive tests, the study established cognitive performance first and then observed who subsequently experienced a first recorded stroke.
The strongest associations appeared among people in their 60s
Age substantially changed the pattern. Among participants aged 60 to 69, those in the low overall cognitive-performance group had an 86% higher subsequent stroke risk than those in the high-performance group after the researchers accounted for relevant health and lifestyle factors.
The domain-specific results pointed in the same direction. Low memory performance was associated with a 68% higher risk of stroke, while low information-processing speed was associated with a 66% higher risk.
These are relative differences in risk, not statements that an individual with a low score has an 86% chance of having a stroke. Across the full cohort, 572 of 4,912 participants experienced a stroke during follow-up. An individual’s absolute risk still depends on age, blood pressure, smoking, diabetes, cardiovascular disease and many other factors.
Crucially, the cognitive associations were not simply repeated at every age. Among adults aged 70 to 79 and those aged 80 or older, the same pattern was not evident after adjustment for other health and lifestyle variables.
One detail helps put the estimates into perspective: the investigators used Cox regression, which compares the rate at which first strokes occurred during follow-up after accounting for other measured characteristics. The 1.86 hazard ratio for overall cognition among adults aged 60 to 69 therefore describes an association between groups over time, rather than a prediction that a particular patient will experience a stroke.
Across the complete study population, the fully adjusted association for lower perceptual speed was more modest, with a hazard ratio of 1.40 and a 95% confidence interval from 1.13 to 1.72. In the 60-to-69 subgroup, the corresponding confidence intervals were 1.27 to 2.69 for global cognition, 1.21 to 2.33 for episodic memory and 1.19 to 2.32 for perceptual speed. These intervals convey statistical uncertainty around the estimated associations; they are not individual risk ranges.
Why age may change what a cognitive score tells us
The age difference is one of the study’s most informative findings. Cognitive performance becomes increasingly heterogeneous with advancing age. People in their 70s and 80s are more likely to have multiple chronic conditions, use several medications and experience age-related neurological changes, all of which can influence test performance independently of future stroke risk.
In a person in their 60s, unusually low performance relative to same-age peers may therefore stand out more clearly as a marker of underlying brain or vascular health. One possibility raised by the researchers is that some participants already had clinically silent brain injury. Such injury can affect cognition without producing the obvious symptoms that would lead to a stroke diagnosis.
This interpretation would also help explain why cognition can precede a recognised stroke without necessarily causing it. The cognitive test may be detecting consequences of the same vascular processes that later contribute to a clinically apparent event.
That distinction matters clinically. A risk marker can be useful even when it is not itself a causal factor. Blood pressure, cholesterol, smoking status and diabetes remain established targets for stroke prevention. Cognitive performance could potentially add another layer of information that helps identify people who warrant closer assessment of those modifiable risks.
Cognitive testing is not yet a stand-alone stroke prediction tool
The findings should not be interpreted as evidence that routine cognitive testing can currently predict who will have a stroke. The study demonstrates an association at the group level. It did not establish a clinical threshold that separates individuals who will and will not experience an event, nor did it show that adding cognitive scores to existing stroke-risk models improves prediction enough to change treatment.
There is also an important issue of direction. Vascular disease can affect the brain before a recognised stroke, but cognition is shaped by education, socioeconomic circumstances, sleep, depression, medication, sensory impairment and numerous other influences. A low test score is therefore not specific to cerebrovascular disease.
What the study does provide is evidence that cognition and vascular health should not always be treated as separate domains. In the 60-to-69 group, several different measures of cognitive performance carried information about later stroke incidence even though participants had no previous diagnosed stroke, transient ischaemic attack or dementia.
The results strengthen the case for earlier brain-health prevention
Stroke prevention often becomes more salient as people move deeper into older age. These findings suggest that the preceding decade may be particularly informative. If lower cognition in the 60s partly reflects accumulated vascular burden or silent injury, waiting for obvious neurological symptoms would miss an earlier opportunity to investigate risk.
The practical implication is not that people should attempt to self-test their memory and infer their stroke risk. Rather, unexpected or concerning cognitive change can be considered alongside established cardiovascular information in clinical care. The study also supports the broader principle that protecting cardiovascular health and protecting brain health are closely connected.
Other Research Today coverage explores how sleep loss affects thought control and how external reminders relate to memory traces. These are separate research questions, not direct evidence about stroke prediction.
For researchers, the next question is whether cognitive measures add meaningful predictive value beyond conventional risk factors. That would require prospective validation, comparison with established risk models and evidence that acting on the additional information improves outcomes.
Important limitations
The study was observational, so it cannot establish that lower cognitive performance causes stroke. Residual confounding is possible even after statistical adjustment, and unmeasured vascular or neurological changes could contribute to both cognition and subsequent stroke.
The cohort was also based in Sweden. Healthcare access, socioeconomic conditions, population health, education and lifestyle differ across countries, meaning the size of the associations may not transfer directly to other populations. Replication in more geographically and socially diverse cohorts would strengthen the evidence.
Finally, the absence of the same adjusted pattern in older groups should not be read as evidence that cognition or brain health ceases to matter after age 70. Instead, it suggests that the relationship between cognitive test performance and future stroke becomes harder to isolate as competing health conditions and age-related changes accumulate.
A potential early signal rather than a diagnosis
The central finding is therefore less about predicting an individual stroke than about when subtle signs of vascular vulnerability may become visible. In this Swedish cohort, cognitive differences among people in their 60s preceded clinically recognised strokes by years, with the lowest overall performers experiencing an 86% higher relative risk than the highest performers.
That makes cognitive performance a potentially valuable piece of the prevention puzzle, but not a diagnosis and not a replacement for established cardiovascular assessment. The challenge now is to determine whether incorporating cognition into risk assessment can identify preventable strokes earlier and, most importantly, whether doing so changes outcomes.
Source Information
Study: Research examining cognitive performance and subsequent first-time stroke risk in older adults from the Good Aging in Skåne cohort.
Journal: Journal of the American Heart Association
Published: 30 September 2026
Sample: 4,912 Swedish adults aged 60 years and older without previous dementia, stroke or transient ischaemic attack; 572 first strokes occurred during an average 12.3 years of follow-up.
