Background: Dementia often remains undiagnosed until it has reached moderate or severe stages, thereby preventing patients and their families from obtaining optimal care. Tools that are easy to use in primary care might facilitate earlier detection of dementia.
Aim: Develop and validate a very brief test for the detection of dementia.
Methods: In the derivation study, we recorded educational level, Mini Mental State Examination (MMSE) scores and subscores and results of a simplified clock-drawing test (sCDT) for consecutive patients attending a single memory clinic over a two-year period,. Dementia was diagnosed according to DSM-IV criteria. The independent variables related to dementia were determined by a multivariable logistic model (MLM) and used to develop a decision tree to predict this diagnosis. In the validation study, the decision tree was applied to consecutive patients of six memory clinics for whom status about dementia was previously determined with DSM-IV criteria. The decision tree, MLM, and MMSE were applied to detect dementia in these patients. The sensitivity and specificity of each diagnostic tool were estimated and compared.
Result: Of 242 patients in the derivation study, the following independent variables were correlated with dementia: sex, sCDT, and two MMSE subscores - the 3-word recall test and spatial orientation. We used Bayesian statistics to develop a brief 2-step decision analysis tree (2-3 min.), which we named Codex (cognitive disorders examination). The validation study applied Codex to 323 patients. Sensitivity was 93% and specificity 85%. The corresponding values were 88% and 87% for the MLM, 94% and 67% or 91% and 70% for the MMSE, depending on the MMSE cutoff score. The sensitivity of Codex was significantly higher than that of MLM, and its specificity was significantly greater than that of MMSE.
Conclusion: Codex is a simple, brief, and reliable test for detecting dementia and requires three minutes or less to administer. Its simplicity and brevity make it appropriate for and easy to use in primary care.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1016/j.lpm.2007.03.016 | DOI Listing |
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!