Abstract
OBJECTIVE: The primary objective was to develop a Swedish short version of the Montreal Cognitive Assessment (s-MoCA-SWE) for use with patients with stroke. Secondary objectives were to identify an optimal cut-off value for the s-MoCA-SWE to screen for cognitive impairment and to compare its sensitivity with that of previously developed short forms of the Montreal Cognitive Assessment. DESIGN: Cross-sectional study. SUBJECTS/PATIENTS: Patients admitted to stroke and rehabilitation units in hospitals across Sweden. METHODS: Cognition was screened using the Montreal Cognitive Assessment. Working versions of the s-MoCA-SWE were developed using supervised and unsupervised algorithms. RESULTS: Data from 3,276 patients were analysed (40% female, mean age 71.5 years, 56% minor stroke at admission). The suggested s-MoCA-SWE comprised delayed recall, visuospatial/executive function, serial 7, fluency, and abstraction. The aggregated scores ranged from 0 to 16. A threshold for impaired cognition ≤ 12 had a sensitivity of 97.41 (95% confidence interval, 96.64-98.03) and positive predictive value of 90.30 (95% confidence interval 89.23-91.27). The s-MoCA-SWE had a higher absolute sensitivity than that of other short forms. CONCLUSION: The s-MoCA-SWE (threshold ≤ 12) can detect post-stroke cognitive issues. The high sensitivity makes it a potentially useful "rule-out" tool that may eliminate severe cognitive impairment in people with stoke.