Related Experiment Video
Updated: Aug 7, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Measurement Error, Reliability, and Minimum Detectable Change in the Mini-Mental State Examination, Montreal
Joanne Feeney1,2, George M Savva2,3, Claire O'Regan2
1Centre for Public Health, Queens University Belfast, Royal Victoria Hospital, Belfast, UK.
Background:
Knowing the reliability of cognitive tests, particularly those commonly used in clinical practice, is important in order to interpret the clinical significance of a change in performance or a low score on a single test.
Objective:
To report the intra-class correlation (ICC), standard error of measurement (SEM) and minimum detectable change (MDC) for the Mini-Mental State Examination (MMSE), Montreal Cognitive Assessment (MoCA), and Color Trails Test (CTT) among community dwelling older adults.
Methods:
130 participants aged 55 and older without severe cognitive impairment underwent two cognitive assessments between two and four months apart. Half the group changed rater between assessments and half changed time of day.
Results:
Mean (standard deviation) MMSE was 28.1 (2.1) at baseline and 28.4 (2.1) at repeat. Mean (SD) MoCA increased from 24.8 (3.6) to 25.2 (3.6). There was a rater effect on CTT, but not on the MMSE or MoCA. The SEM of the MMSE was 1.0, leading to an MDC (based on a 95% confidence interval) of 3 points. The SEM of the MoCA was 1.5, implying an MDC95 of 4 points. MoCA (ICC = 0.81) was more reliable than MMSE (ICC = 0.75), but all tests examined showed substantial within-patient variation.
Conclusion:
An individual's score would have to change by greater than or equal to 3 points on the MMSE and 4 points on the MoCA for the rater to be confident that the change was not due to measurement error. This has important implications for epidemiologists and clinicians in dementia screening and diagnosis.
