1040 — Performance of Cognitive Screening Tests in Primary Care
Holsinger TC (Durham VAMC), Williams JW
(Durham VAMC), Coffman C
(Durham VAMC), Stechuchek KM
(Durham VAMC), Burke J
(Durham VAMC), Plassman BP
(Duke University Medical Center)
To determine the performance of four cognitive screening tests for detecting dementia and cognitive impairment, not dementia (CIND).
Primary care patients age 65 or older with no diagnosis for dementia were randomly selected. Subjects were administered screening tests followed by neurocognitive battery and complete medical record review. Subjects named an informant to provide collateral information for diagnosis. A multidisciplinary consensus conference blinded to screening test results assigned research diagnoses using standardized criteria. Sensitivity, specificity, and positive likelihood ratio (LR+) were used to compare results of the screening tests with the research diagnosis.
630 subjects were diagnosed; 3.3% met criteria for dementia, 39.2% met criteria for CIND, and 57.5% had normal cognition.
For detecting dementia, the modified mini mental state exam (3MS), the longest screening test, had the highest sensitivity of 0.86, a specificity of 0.79 and LR+ of 4.06. The Mini-Cog had a sensitivity of 0.76, a specificity of 0.73, and LR+ of 2.79. The memory impairment screen (MIS) had a sensitivity of 0.43, a specificity of 0.93, and LR+ of 5.90. The two-item functional screening test had a sensitivity and specificity of 0.38 and 0.87 when administered to the subject, and 0.79 and 0.83 when given to the informant.
For detecting dementia or CIND, the sensitivity for each screening test was low. For the 3MS, the sensitivity was 0.39 with a specificity of 0.89 and LR+ of 3.45. For the Mini-Cog™, the sensitivity was 0.39, the specificity 0.78 and LR+ 1.76. For the MIS, the sensitivity and specificity were 0.17 and 0.98, LR+ of 7.59. Sensitivities for the two-item screening test were 0.24 when given to the subject and 0.35 when given to the informant. Specificity for the two-item screen was 0.93 for subject completion and 0.92 for informant completion.
These brief cognitive screening tests are, at best, moderately efficacious at detecting dementia level symptoms, but are not practical for use in detecting CIND.
Patients and primary care providers have indicated a preference for cognitive evaluation via screening tests. These brief screening tests are not adequate for detecting those patients needing evaluation for CIND and show only modest efficacy for detecting the mildly severe dementia seen in this population.