Related Experiment Video
Updated: May 2, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Improving diagnostic accuracy of the Montreal Cognitive Assessment to identify post-stroke cognitive impairment
Laura Gallucci1,2, Christoph Sperber1, Andreas U Monsch3
1Department of Neurology, University Hospital, Inselspital, University of Bern, Freiburgstr. 16, 3010, Bern, Switzerland.
Abstract:
Given advantages in reperfusion therapy leading to mild stroke, less apparent cognitive deficits can be overseen in a routine neurological examination. Despite the widespread use of the Montreal Cognitive Assessment (MoCA), age- and education-specific cutoffs for the detection of post-stroke cognitive impairment (PSCI) are not established, hampering its valid application in stroke. We aimed to establish age- and education-specific MoCA cutoffs to better discriminate patients with and without acute PSCI. Patients with acute ischemic stroke underwent the MoCA and a detailed neuropsychological assessment. PSCI was defined as a performance < - 1.5 SD in ≥ 2 cognitive domains. As secondary data analysis, the discriminant abilities of the MoCAraw-score (not adding + 1 as correction for ≤ 12 years of education, YoE) cutoffs were automatically derived based on Youden Index and evaluated by receiver operating characteristic analyses across age- (< 55, 55-70, > 70 years old) and education-specific (≤ 12 and > 12 YoE) groups. 351 stroke patients (67.4 ± 14.1 years old; 13.1 ± 2.8 YoE) underwent the neuropsychological assessment 2.7 ± 2.0 days post-stroke. The original MoCA cutoff < 26 falsely classified 26.2% of examined patients, with poor sensitivity in younger adults (34.8% in patients < 55 years > 12 YoE) and poor specificity in older adults (55.0%, in > 70 years ≤ 12 YoE). By maximizing both sensitivity and specificity, the optimal MoCAraw cutoffs were: (i) < 28 in patients aged < 55 with > 12 YoE (sensitivity = 69.6%, specificity = 77.8%); (ii) < 22 and < 25 in patients > 70 years with ≤ 12 and > 12 YoE (sensitivity = 61.6%, specificity = 90.0%; sensitivity = 63.3%, specificity = 84.0%, respectively). In other groups the optimal MoCAraw cutoff was < 26. Age and education level should be considered when interpreting MoCA-scores. Though new age- and education-specific cutoffs demonstrated higher discriminant ability for PSCI, their performance in young stroke and adults with higher education level was low due to ceiling effects and MoCA subtests structure, and cautious interpretation in these patients is warranted.Trial registration: ClinicalTrials.gov Identifier: NCT05653141.
Insights
New age- and education-specific cutoffs for the Montreal Cognitive Assessment (MoCA) improve detection of post-stroke cognitive impairment (PSCI). Standard MoCA scores can misclassify stroke patients, highlighting the need for tailored interpretation based on demographics.
Area of Science:
- Neurology
- Cognitive Science
- Medical Diagnostics
Background:
- Mild stroke cognitive deficits are often overlooked.
- The Montreal Cognitive Assessment (MoCA) is widely used but lacks age- and education-specific cutoffs for post-stroke cognitive impairment (PSCI).
- This deficiency hinders accurate PSCI detection in stroke patients.
Purpose of the Study:
- To establish age- and education-specific MoCA cutoffs for improved PSCI detection.
- To enhance the diagnostic accuracy of the MoCA in diverse stroke patient populations.
Main Methods:
- 351 acute ischemic stroke patients underwent MoCA and detailed neuropsychological assessment.
- PSCI was defined as performance below -1.5 SD in at least two cognitive domains.
- Receiver operating characteristic analyses identified optimal MoCA raw-score cutoffs across age and education groups.
Main Results:
- The standard MoCA cutoff (<26) misclassified 26.2% of patients.
- Optimal MoCA raw cutoffs varied significantly by age and education level (e.g., <28 for <55 years/>12 years of education; <22 or <25 for >70 years).
- New cutoffs showed improved discriminant ability but had limitations in young stroke patients and those with higher education due to ceiling effects.
Conclusions:
- Age and education level are critical factors for interpreting MoCA scores in stroke patients.
- Tailored MoCA cutoffs enhance PSCI detection accuracy.
- Cautious interpretation is advised for specific subgroups, particularly young stroke patients and those with higher education, due to potential ceiling effects.
