A novel tool accurately predicts dementia risk within a decade following acute ischemic stroke (AIS) or transient ischemic attack (TIA).

“Quantifying and predicting dementia risk after stroke is a major step towards tackling it just as risk scores have transformed heart disease prevention,” study investigator, assistant professor, Department of Medicine, Division of neurology, McMaster University, Hamilton, Ontario, Canada, told Medscape Medical News.

He said this new tool could potentially be used to identify and select patients for clinical trials evaluating emerging dementia treatments.

The findings were presented on February 5 at International Stroke Conference (ISC) 2026.

Reliable Risk Prediction

To develop the tool, investigators analyzed data from nearly 45,000 adults without dementia in the Ontario Stroke Registry between 2002 and 2013, including 25,996 patients with AIS, 3399 with intracerebral hemorrhage (ICH), and 14,149 with TIA. The mean age was 70 years, and 53% of participants were men.

Patients were followed through linked provincial administrative databases for dementia, recurrent stroke, or death over a mean of 8.4 years, with follow-up extending up to 20 years. Dementia developed in approximately 30% of patients with AIS, 29% with ICH, and 27% with TIA.

Among individuals with TIA, the strongest predictors of dementia included older age, preexisting dependence, diabetes, depression, cognitive symptoms at presentation, and greater disability at hospital discharge, as measured by the modified Rankin Scale.

For those with AIS, key risk factors included older age, female sex, diabetes, prior stroke or TIA, depression, ICH (vs ischemic stroke), visual field deficits, cognitive symptoms during hospitalization, and higher discharge disability.

Researchers used the strongest predictors to calculate dementia risk scores and stratified patients into five risk categories, from lowest to highest. Separate scores were generated for TIA and stroke, with models estimating 1-, 5-, and 10-year dementia risk. The models were externally validated using the Ontario Stroke Audit, which conducts periodic audits of randomly selected patients across all hospitals in Ontario.

‘Excellent Calibration’

Investigators found good discrimination for all models. In the derivation cohort, the area under the curve (AUC) for 1-, 5-, and 10-year dementia risk was 0.81, 0.78, and 0.76, respectively, for TIA, and 0.78, 0.73, and 0.71 for stroke. 

Model performance declined only minimally in the validation cohort. Corresponding AUC values for ICH were 0.79 at 1 year, 0.75 at 5 years, and 0.73 at 10 years.

Joundi noted the predicted probabilities closely matched observed dementia risk, demonstrating what he described as “excellent calibration.” Based on these estimates, dementia risk varied widely, from < 5% in the lowest risk quintile to nearly 50% in the highest.

“In the highest risk quintile, almost half of people are diagnosed with dementia over 10 years,” said Joundi.

However, he emphasized that the estimates reflect probabilities, not certainties. “There is no certainty in prediction for any individual,” he said. 

Knowing the risk level may facilitate enrolment of participants into observational cohort studies, biomarker studies, or most importantly, interventional studies and clinical trials, said Joundi. “Thedementia risk prediction tool would be used to stratify patients into different levels of risk for research studies and clinical trials of dementia prevention.”

A Research Tool…for Now

A key strength of the risk score is its flexibility, with separate prediction models for TIA, ischemic stroke, and ICH, as well as for different time horizons after the cerebrovascular event. “That makes it adaptable to the specific research goal or question,” Joundi said.

For now, the tool is meant to be used in research settings, not for directing care for individual patients, said Joundi.

He added that all patients with TIA or stroke should receive aggressive vascular risk reduction, which may also reduce the risk for subsequent dementia. Key interventions include optimal control of hypertension and diabetes, smoking cessation, and regular physical activity. 

Joundi said external validation outside Ontario would strengthen confidence that the tool’s accuracy is maintained across different geographic settings.

A key limitation of the study, he said, is the lack of data on dementia subtype. Investigators also did not have access to neuroimaging, although Joundi noted that the score demonstrated good discrimination and excellent calibration without requiring detailed MRI information.

A Shift in Poststroke Care

Commenting for Medscape Medical News, vascular neurologist Jonathan Solomonow, MD, Cerebrovascular Center, Neurological Institute, Cleveland Clinic in Cleveland, said this new tool reflects a shift in after-stroke care.

“The focus has traditionally centered solely on preventing recurrent strokes, and often has not directly addressed the cognitive, behavioral, and psychological complications patients experience,” Solomonow said.

Having a practical tool to flag high-risk individuals could eventually allow physicians “to proactively address cognitive decline, which is often a more devastating threat to a survivor’s quality of life than a recurrent stroke,” said Solomonow.

He noted that risk factors such as hypertension and diabetes wear down blood vessels over time and contribute to both stroke and cognitive decline.

Early identification of high-risk patients enables timely, aggressive intervention to prevent avoidable decline and support long-term functional independence, he added.

In a press release, Deborah A. Levine, MD, a professor of internal medicine and neurology at the University of Michigan, Ann Arbor, Michigan, and a volunteer expert with the American Stroke Association who was not involved in the study, said the research was well done and offers a useful tool that could speed research and help deliver new treatments to stroke survivors sooner.

This study was funded by Brain Canada, Canadian Stroke Consortium, and Heart & Stroke Foundation of Canada Catalyst Grant.

Joundi reported he was supported by a New Investigator Award from the Heart & Stroke Foundation of Canada and the Anne Marie Brune/PHRI Chair in Vascular Dementia.