Background: Cognitive impairment is common post-stroke. There is a need to understand patterns of early cognitive recovery post-stroke to guide both clinical and research practice.

Aims: To map the trajectory of cognitive recovery during the first week to 90-days post-stroke using serial computerised assessment.

Methods: An observational cohort study recruited consecutive stroke patients admitted to a stroke unit within 48 hours of onset. Cognitive function was assessed using the computerised Cambridge Neuropsychological Test Automated Battery (CANTAB) daily for seven days, then 14, 30 and 90-days post-stroke. The CANTAB measured visual episodic memory and learning, information processing speed, visuo-spatial working memory, complex sustained attention, and mental flexibility. Repeated measures MANOVA/ANOVA with Least Squares Difference post-hoc analysis were performed to ascertain significant change over time.

Results: Forty-eight participants with primarily mild, ischaemic stroke completed all time-points for analysis. For all subtests, significant improvement in performance was captured at multiple assessment points. In the first week post-stroke, a distinct period of early, rapid, global cognitive improvement was found, that largely stabilized between 8 and 14 days post mild stroke.

Conclusions: Serial, computerised cognitive assessment can effectively map post-stroke cognitive recovery and revealed an early phase of rapid global improvement that is evidence for an acute post-stroke delirium. Post-stroke delirium appears to be far more common than is appreciated and can be measured. Resolution of post-stroke delirium in the second week following mild stroke indicates more extensive neuropsychological testing can be undertaken earlier than previously thought.