Cognitive testing and the hazards of cut-offs

Abstract

The article reviews some basic statistical concepts used in medicine, including the mean, standard deviation, sensitivity and specificity. Using this background the authors describe how these can be applied to cognitive tests, taking the Montreal Cognitive Assessment (MoCA) as an example. Two different approaches to using the MoCA in diagnosing dementia are considered: one using a fixed cut-off score, the other taking account of normative data about the effects of age and educational level on MoCA scores. It is recommended that clinicians assessing cognitive function should not rely on a fixed cut-off score, but where possible compare the patient's result with those of people of comparable age and educational background, although normative data of this kind are not always available.