Dementia Is Often Misclassified by Brief Cognitive Assessments

Brief cognitive assessments often produce a misclassification of dementia, according to a new study.

This inaccuracy in classifying patients with dementia can be largely attributed to test-specific biases, the researchers said.

IF YOU LIKE THIS, READ MORE...

Dementia Is Associated With Higher Rate of Headaches, Movement Disorders

Pearls of Wisdom: Do Statins Cause Dementia?

The study included 824 older adults from the population-based US Aging, Demographics, and Memory Study who had been diagnosed with dementia (DSM-III-R and DSM-IV criteria).

The researchers sought to identify predictors of misclassification in 3 brief cognitive assessments the participants were administered: Mini-Mental State Examination (MMSE), Memory Impairment Screen (MIS), and animal naming.

Predictors of false-negative, false-positive, and overall misclassification were analyzed for each assessment. The researchers analyzed 22 candidate predictors, including sociodemographics, dementia risk factors, and potential sources of test bias.

In all, 35.7% of participants were misclassified by at least 1 assessment, and 1.7% of participants were misclassified by all 3 assessments.

The only consistent predictor of overall misclassification across all the assessments was absence of informant-rated poor memory. Otherwise, each assessment had a different pattern of misclassification predictors.

For example, within the MMSE, years of education predicted higher false-negatives and lower false-positives. For the animal naming test, nursing home residency predicted lower false-negatives and higher false-positives.

Age, nursing home residency, and non-Caucasian ethnicity were the most consistent predictors for false-positives.

—Colleen Murphy

Reference:

Ranson JM, Kuźma E, Hamilton W, Muniz-Terrera G, Langa KM, Llewellyn DJ. Predictors of dementia misclassification when using brief cognitive assessments [published online November 28, 2018]. Neurology. https://doi.org/10.1212/CPJ.0000000000000566.