Since its 1975 publication in the Journal of Psychiatric Research, the Mini-Mental State Exam (MMSE) has become a staple of dementia screening. The simple test, which requires no special equipment, provides a cursory assessment for cognitive impairment, testing for things like orientation to time and place and the ability to follow complex instructions. The MMSE is widely popular, both in the U.S and abroad, having been translated into ten languages since its initial publication in English.
Over time, other screening tests have been added to the clinician’s toolbox to evaluate the possibility of dementia under a variety of circumstances. They provide valuable insight for diagnosing and following the course of dementia.
However, these tests’ utility and popularity don’t mean they are without issues. It has been suggested,and in some cases confirmed, that factors such as patient age, education level, sensory impairment, and race may affect the results of these assessments.
A recent evaluation of misclassification due to bias was performed and published in Neurology: Clinical Practice in November. Researchers reviewed the MMSE, the Memory Impairment Screen (MIS), and Animal Naming (AN) to look for predictors of misclassification. They assessed misclassification in 824 older adults using diagnostic criteria in DSM-III-R and DSM-IV as the reference standard. Of the subjects, 35 percent had confirmed dementia, and 65 percent did not.
Misclassification on at least one assessment tool occurred in 35 percent of patients, while misclassification on all three occurred in only 1.7 percent of patients. Test-specific results were as follows:
- More years of education predict more false-negatives while fewer years of education predict more false-positives by the MMSE.
- Nursing home resident had fewer false negatives and more false-positives by AN.
Other findings were consistent across the three assessments:
- “Absence of informant-rated poor memory” most consistently predicted false-negatives.
- “Age, nursing home residency and non-Caucasian ethnicity” most consistently predicted false-positives.
Absence of informant-rated poor memory was the only consistent predictor of overall misclassification across all assessments. This emphasizes the contribution a patient’s partner can provide by offering their observations about memory impairment. A patient may state that they have no memory loss, but it can be difficult to tell if this an accurate self-assessment, denial, or simply an inaccurate assessment.
In addition, the testing should be chosen with an eye to its specific biases. AN testing might not be the best choice for a nursing home patient, for example. And regardless of the test used, awareness of the role common biases play should always be considered along with the results.