Larry Leach, PhD, Cpsych
Psychologist, Department of Psychology,
Baycrest Centre for Geriatric Care,
Adjunct Professor, Department of Psychology,
University of Toronto,
Toronto, ON.

The accurate and reliable assessment of dementia is essential not only for the purpose of diagnosis, but also because recent advances offer promise for treatment. With the advent of new treatments, early diagnosis becomes even more important because those individuals in the early stages of a disease such as Alzheimer are likely to benefit the most from treatment. Dementia is not a disease but rather a constellation of symptoms that reflect brain dysfunction. The most common clinical criteria of dementia (the DSM-IV) requires the presence of dysfunction in memory and one or more other mental functions such as language, reasoning, perception, attention or spatial awareness. In addition, the mental dysfunction must interfere with the performance of daily activities. There are many neurological diseases that can lead to dementia, Alzheimer disease being the most common. Recent statistics published by the Canadian Consensus Conference on Dementia indicate that dementia is present in approximately 8% of Canadians. The prevalence of dementia changes dramatically with age, increasing from 2.4% of Canadians aged 65-74 to 34.5% over the age of 85.

In order to determine if an individual has some form of dementia they must undergo special tests of mental functions referred to as mental status exams or neuropsychological assessments. Such tests much meet three general criteria: they must be reliable; they must be sensitive to dementia; and they must be specific to the dementia. Reliability ensures that results can be repeated within acceptable limits of error. Sensitivity requires that the test must be able to identify individuals who are known to have dementia. Specificity requires that the test must be able to identify individuals who are known not to have dementia. Over the years, various mental status examinations have also been developed to help identify the presence of cognitive impairments that are sufficiently severe to represent a dementia. The reliability, sensitivity and specificity of tests are often documented but unfortunately, one characteristic--the predictive ability--has not always been thoroughly documented or appreciated. The predictive ability involves two complementary characteristics, positive and negative prediction. In contrast to sensitivity and specificity, which reflect how well a test identifies individuals who are known to have (or not to have) dementia, positive and negative predictions reflect the likelihood that a person has the disease given a certain test result. Positive prediction is the probability that a positive test result indicates the presence of dementia. Negative prediction is the probability that a negative test result represents the absence of dementia. Positive or negative prediction is what a clinician should be most concerned with. After all, if a clinician knew that the person has a dementia then there would be no problem. There is one aspect about predictive ability that is not fully appreciated: predictive ability varies with the prevalence of dementia. As disease prevalence increases, positive prediction increases but negative prediction decreases. Conversely, as disease prevalence decreases negative prediction increases while positive prediction decreases. This lack of appreciation of the change in predictive ability of a test with disease prevalence can lead unwary clinicians into a false sense of security with respect to the outcomes of test results.

One of the most commonly used tests of mental status is the Mini-Mental Status Examination (MMSE). This test is the most frequently used diagnostic tool to determine presence of cognitive impairment for the diagnosis of dementia. Although the sensitivity and specificity of the MMSE appear acceptable, evaluation of the positive and negative prediction across different levels of prevalence reveals that the MMSE is less than stellar. For example, a positive test result of an MMSE (based on a cut-off score of 24/30) obtained in a situation where the prevalence of dementia is expected to be 2.4% is only 12% accurate. In other words, 88% of the positive test results would likely come from normal individuals! Furthermore, the sensitivity and specificity of the MMSE has also been shown to vary with age and education of people, but age corrections are seldom applied in many clinical settings. The Mattis Dementia Rating Scale (DRS) has been shown to have excellent sensitivity and specificity; in addition, it has excellent positive and negative predictive ability across a wide range of prevalence of dementia. The MMSE has a large following due to the short amount of time it takes to administer (approximately 10 minutes), but the greater predictive ability of the DRS makes it well worth the 25 minutes it may take to administer.

Although a short test such as the DRS is sensitive to cognitive impairment, it has the shortcoming of failing to identify specific patterns of dysfunction relating to various types of dementia. For this purpose, neuropsychological tests are far superior. A detailed neuropsychological assessment can clearly and reliably define those cognitive domains that are deficient and intact. Furthermore, the neuropsychological tests have been shown to correlate with regional brain dysfunction. Although patterns of dysfunction relate to brain dysfunction they do not always identify cause. Numerous case studies have demonstrated that individuals, who have been diagnosed as having one form of dementia on the basis of patterns of impairment, have been shown to have another cause of dementia at autopsy. This represents a limitation to the present diagnostic techniques based primarily on clinical signs and symptoms. Nevertheless, an appreciation of the extent as well as type of cognitive impairment can prove to be valuable in the treatment of individuals with dementia. Despite this limitation, the pattern of deficits due to dementia are clearly distinguishable from those cognitive disruptions associated with depression.