Related Experiment Video
Updated: Sep 29, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Six-item screener to identify cognitive impairment among potential subjects for clinical research
Christopher M Callahan1, Frederick W Unverzagt, Siu L Hui
1Indiana University Center for Aging Research, Indianapolis 46202, USA. ccallahan@regenstrief.org
Objective:
To design a brief cognitive screener with acceptable sensitivity and specificity for identifying subjects with cognitive impairment.
Design:
Cohort one is assembled from a community-based survey coupled with a second-stage diagnostic evaluation using formal diagnostic criteria for dementia. Cohort two is assembled from referrals to a specialty clinic for dementing disorders that completed the same diagnostic evaluation.
Setting:
Urban neighborhoods in Indianapolis, Indiana and the Indiana Alzheimer Disease Center.
Patients:
Cohort one consists of 344 community-dwelling black persons identified from a random sample of 2212 black persons aged 65 and older residing in Indianapolis; cohort two consists of 651 subject referrals to the Alzheimer Disease Center.
Measurements:
Formal diagnostic clinical assessments for dementia including scores on the Mini-mental state examination (MMSE), a six-item screener derived from the MMSE, the Blessed Dementia Rating Scale (BDRS), and the Word List Recall. Based on clinical evaluations, subjects were categorized as no cognitive impairment, cognitive impairment-not demented, or demented.
Results:
The mean age of the community-based sample was 74.4 years, 59.4% of the sample were women, and the mean years of education was 10.1. The prevalence of dementia in this sample was 4.3% and the prevalence of cognitive impairment was 24.6%. Using a cut-off of three or more errors, the sensitivity and specificity of the six-item screener for a diagnosis of dementia was 88.7 and 88.0, respectively. In the same sample, the corresponding sensitivity and specificity for the MMSE using a cut-off score of 23 was 95.2 and 86.7. The performance of the two scales was comparable across the two populations studied and using either cognitive impairment or dementia as the gold standard. An increasing number of errors on the six-item screener is highly correlated with poorer scores on longer measures of cognitive impairment.
Conclusions:
The six-item screener is a brief and reliable instrument for identifying subjects with cognitive impairment and its diagnostic properties are comparable to the full MMSE. It can be administered by telephone or face-to-face interview and is easily scored by a simple summation of errors.
