Overview
Cognitive problems are common after stroke and can affect language, memory, attention, executive function, and the ability to process numbers. These changes can interfere with daily life, return to work, medication management, and rehabilitation. Yet many studies have only measured cognition briefly and have relied on broad screening tools, making it difficult to understand how thinking abilities change over time after stroke.
This study used the Oxford Cognitive Screen, a stroke-friendly cognitive assessment designed to detect domain-specific difficulties that may be missed by general tests. The researchers followed patients from the acute stage after stroke, reassessed them at 6 months, and again at 2 years or later to map long-term cognitive recovery and decline.
Why this study matters
After a stroke, some people improve quickly, some recover slowly, and others continue to have persistent cognitive problems. Clinicians often need to know which patients are likely to recover and which may need longer-term support. Standard mental status tests can miss subtle but important deficits, especially in attention, executive function, and number processing. By tracking multiple cognitive domains over several years, this study helps clarify which functions tend to recover, which remain impaired, and which patterns of recovery are most common.
How the study was done
Participants were recruited from a regional acute stroke unit at John Radcliffe Hospital in Oxford, United Kingdom, between 2012 and 2019. Cognitive testing took place at three time points: during the acute hospital phase, at 6 months, and at 2 years or more after stroke.
The Oxford Cognitive Screen was used at each visit. Rather than giving only one overall score, this tool looks at several separate cognitive domains, including:
- Language
- Memory
- Attention
- Executive function
- Number processing
The investigators measured overall impairment by calculating the proportion of subtasks that were impaired. They also used statistical models to identify predictors of long-term impairment and to group patients into distinct recovery patterns, known as latent trajectories. Analyses were adjusted for the severity of acute impairment and for time.
Who was included
A total of 866 patients were assessed in the acute stage. Of these, 105 were followed up at 2 years or beyond, and 98 had complete Oxford Cognitive Screen data available for analysis. The median follow-up time was 4.1 years. The average age was 69 years, and 41% of participants were women.
As with many long-term stroke studies, only a subset of the original group returned for extended follow-up. This is an important limitation because people who do not return may differ in health, disability, or cognitive status from those who do.
Main findings: overall cognitive recovery
Overall cognitive impairment improved substantially by 6 months after stroke and continued to improve, though more gradually, over the longer term. In statistical terms, the severity of impairment fell significantly at 6 months and again at later follow-up.
The most important predictor of long-term cognition was how impaired the patient was at the acute stage. In other words, people with more severe early cognitive problems were much more likely to have difficulties years later. By contrast, demographic factors such as age and sex, as well as vascular risk factors, explained only a small amount of the variation in long-term cognitive outcome.
This finding is clinically important: early cognitive status appears to be a stronger guide for prognosis than many traditional background characteristics.
Distinct recovery patterns
The trajectory analysis identified four broad patterns of overall cognitive change:
- No or mild acute impairment with stability: 47.6%
- Moderate but improving impairment: 32.3%
- Large improvement over time: 15.2%
- Decline: 4.8%
These groups show that recovery after stroke is not uniform. Nearly half of the patients had little or no major cognitive impairment from the start and remained relatively stable. About one-third had moderate impairment that improved. A smaller group showed marked improvement, while a very small minority declined over time.
The existence of a decline group is especially important because it suggests that some patients may worsen later, possibly because of recurrent vascular injury, underlying neurodegenerative disease, frailty, or limited cognitive reserve. Long-term follow-up is therefore valuable even for patients who seem to recover well early on.
Domain-specific recovery: not all cognitive skills improve equally
When the researchers looked at each cognitive domain separately, they found that recovery was uneven.
Memory showed the greatest improvement, followed by language. Attention improved too, but less dramatically. Executive function improved the least and was more likely to remain impaired. Number processing was also part of the domain-specific analysis, though the abstract emphasizes the strongest findings for memory, language, attention, and executive skills.
Reported odds ratios suggested large gains in some domains:
- Memory: odds ratio 16.40
- Language: odds ratio 8.17
- Attention: odds ratio 5.41
- Executive function: odds ratio 4.14
These results indicate that some functions, especially memory and language, are more likely to recover substantially over time, whereas executive difficulties can persist and may require targeted rehabilitation.
Persistent and delayed recovery patterns
The domain-specific trajectory models revealed additional subgroups beyond the overall patterns. In particular, some patients showed persistent impairment or delayed recovery, especially in executive function and attention.
This matters because these domains are often crucial for real-world independence. Executive function includes planning, organizing, flexible thinking, and self-monitoring. Attention affects concentration, task switching, and the ability to manage complex environments. Even when a patient performs reasonably well on everyday conversation or basic memory tasks, ongoing problems in these areas can still limit safe driving, medication adherence, financial management, and return to work.
What the findings mean for stroke care
The study supports several practical conclusions for clinicians and rehabilitation teams:
- Cognitive recovery is greatest within the first 6 months after stroke, but improvement can continue beyond that.
- Early cognitive impairment is the strongest predictor of long-term outcome.
- Executive dysfunction may persist for years and should not be overlooked.
- Different cognitive domains recover at different rates, so a single global screen may miss important problems.
- Long-term cognitive monitoring should be individualized rather than based only on age, sex, or vascular risk factors.
For patients and families, this means that early improvement does not always mean full recovery, and a normal-looking recovery in one area may hide continuing difficulties in another. Follow-up assessments can help identify ongoing needs and guide referrals to neuropsychology, occupational therapy, speech and language therapy, and cognitive rehabilitation.
Strengths of the study
This study has several important strengths. It used a stroke-specific cognitive tool rather than a generic screen, making it more sensitive to the types of deficits commonly seen after stroke. It also followed patients over a long period, allowing the researchers to describe both short-term and long-term change.
Another strength is the focus on separate cognitive domains. Stroke does not affect cognition in a single, uniform way, so analyzing language, memory, attention, executive function, and number processing separately gives a more useful clinical picture.
Limitations to keep in mind
The main limitation is the relatively small number of patients who returned for long-term follow-up compared with the original acute cohort. This raises the possibility of selection bias. Patients with worse outcomes may have been less likely to attend follow-up, or conversely, patients with more problems may have been more motivated to return. Either way, the long-term findings should be interpreted with caution.
In addition, observational trajectory studies can identify patterns but cannot prove why those patterns occur. Factors such as recurrent stroke, depression, fatigue, sleep problems, and pre-existing cognitive decline may all influence long-term cognition but were not necessarily fully captured in the summary.
Clinical takeaway
The central message is that cognitive recovery after stroke is real, often substantial, and most pronounced early on, but it is not the same across all domains. Memory and language tend to recover better than executive function, which often remains a challenge. The severity of cognitive impairment at the acute stage gives the best clue to long-term outcome.
For stroke services, this supports routine and repeated cognitive assessment, ideally with tools designed for stroke survivors. For patients, it reinforces the importance of follow-up, rehabilitation, and ongoing monitoring, even years after the initial event.
Reference
Milosevich E, Kusec A, Pendlebury ST, Demeyere N. Longitudinal Trajectories of Global and Domain-Specific Cognition After Stroke Using the Oxford Cognitive Screen. Stroke. 2026-05-19. PMID: 42153299.