A simple method of classifying impairment could be to choose a specific score on a test as a cut-off for indicating impairment. For example, anyone with a score below 27 on MMSE is classified as ?impaired? and everyone scoring above that point is considered ?normal?. Although simple, this method is not an accurate way of classifying cognitive deficits, especially in elderly persons. This is because in many cognitive domains, performance tends to decline with age and differs according to a person?s educational level and sex. The figure (Figure 6) shows how average performance on MMSE differs according to the age and amount of formal education that a subject has received, according to data from a longitudinal population-based study.

Let us consider a scenario on which we use a cut off score of 26 on MMSE as a measure of ?impairment.? Subject A is a 75-year-old woman with 10 years of education. Her MMSE score is 24. Thus, she would be classified as impaired. Subject B is a 95-year-old uneducated woman. Her MMSE score is 26. Thus, she would be classified as impaired. Subject C is a 75-year-old man with 12 years of formal education. His MMSE score 27. Thus, he would be classified as normal.

In this case, it is not unexpected that subject A?s cognitive performance is low and that she has a cognitive deficit. However, the situation is less clear for subject B. Is it really abnormal for a 95 year old woman with a no education to score 26 on MMSE? Probably not. Using the current method, subject B would be classified as impaired, which would probably be a ?False positive? result. Similarly, MMSE score 27 for a 75 year old man with high education might be considered as unusually low. Thus, subject C would be classified as having ?normal cognitive functioning? despite the fact that he should really be classified as having a cognitive deficit (False negative).