Objectives: To compare the accuracy of commonly used tasks with that of the Visual Association Test (VAT), a conceptually different test involving cued recall of pictorial stimuli, in the recognition of dementia within primary care.
Design: A cross-sectional diagnostic study of concurrent validity.
Setting: Twenty-nine German primary care practices.
Participants: Four hundred twenty-three individuals in primary care participating in a longitudinal cohort study.
Measurements: Participants underwent a comprehensive neuropsychological interview. The validated clinical dementia diagnosis was used as reference standard. Index tests comprised the VAT, Mini-Cog, clock drawing, verbal fluency, episodic memory, and subjective complaints. Validity parameters were calculated; possible confounders of test performance (age, sex, education, comorbidity, depression, language) were evaluated.
Results: Twenty-one participants (5%) had dementia according to the reference standard. The VAT distinguished dementia from nondementia with a sensitivity of 95.2% (95% confidence interval (CI) = 86.1-100.0), a specificity of 96.0% (95% CI = 94.1-97.9), a positive predictive value (PPV) of 55.6% (95% CI = 39.3-71.8), and a negative predictive value (NPV) of 99.7% (95% CI = 99.2-100.0). The next-most-accurate tasks were the Mini-Cog and immediate and delayed recall. Their sensitivity and NPV are similar to those of the VAT, but their PPV and specificity were significantly lower than those of the VAT. Age and depression affected all test scores.
Conclusion: The VAT (cued recall of pictorial material) is superior to other tasks for the recognition of dementia in terms of higher specificity and PPV. Age-specific cutoff scores may improve the validity of all tests.
© 2011, Copyright the Authors Journal compilation © 2011, The American Geriatrics Society.