Taking clinical judgment out of the equation : a call for the standardization of MCI diagnostic criteria and construction of a model to predict conversion to dementia.

Abstract

Although the diagnostic criteria for mild cognitive impairment have evolved considerably since their inception, they remain varied and able to be interpreted and implemented in different ways depending on the judgment of the clinician. Because of this issue, a wide range of incidence, prevalence, and conversion rates are found in the research literature. Using data collected from 400 patients with amnestic mild cognitive impairment, limitations inherent in current mild cognitive impairment diagnostic criteria were addressed. First, using Alzheimer’s Disease Neuroimaging Initiative diagnostic criteria, an equation was constructed to predict conversion from mild cognitive impairment to dementia by analyzing the predictive ability of variables representative of a number of categories (i.e., demographic, psychiatric, functional, biomarker, imaging, and cognitive). This model accounted for over 60% of variance in conversion and exhibited an area under the curve of 0.93. Then, separate models were constructed using different applications of the current diagnostic criteria for mild cognitive impairment. As expected, criteria utilizing a one standard deviation clinical cutoff on a measure of delayed verbal recall in combination with the allowance for some functional change (i.e., scores =5 on the Functional Activities Questionnaire) exhibited the greatest utility of any combination of diagnostic criteria. Taken together, these results indicate that statistical equations can be constructed to predict conversion from mild cognitive impairment to dementia and be tailored for widespread clinical use. Moreover, these results show that the standardization of current diagnostic criteria for mild cognitive impairment is clearly needed.