Retrospective studies in the 1970s showed that there was often a 30 to 50% error rate simply in deciding whether a patient did or did not have dementia, let alone what kind of dementia the patient might have. For example, in an institution as sophisticated as the Maudsley Hospital, the major psychiatric hospital in Great Britain, it had been found that the diagnosis of the presenile form of Alzheimer’s disease was in error by 30%, most of the patients misdiagnosed as having been depressed.

LEARNING TO DIFFERENTIATE DEMENTIA FROM DEPRESSION

The misdiagnosis of depression as dementia was first described by Kiloh in 1961 as “pseudodementia”. During the 1970s differentiation of the dementia syndrome based on studies from major depression were undertaken by a number of psychiatrists, and, in 1980, the American Psychiatric Association, in the third edition of its Diagnostic and Statistical Manual (DSM-III), set forth clear-cut operational criteria for the two disorders. These criteria limited the diagnosis of dementia to individuals who were alert and awake and who had experienced a decline in functional abilities secondary to cognitive impairment, with evidence of such cognitive impairment in at least two areas of cognition. This powerful definition separated not only dementia from depression but also dementia from amnestic syndromes (in which only memory is involved) and delirium (which occurs with acute medical conditions and in which there is a fluctuating state of awareness). In contrast to the very low accuracy rate of diagnosis of dementia syndrome in the 1970s, by the mid- 1980s, clinicians using DSM-III criteria were able to diagnose the dementia syndrome with greater than 95% accuracy.

THE DEVELOPMENT OF MENTAL STATUS AND FUNCTIONAL MEASURES OF DEMENTIA SEVERITY

The decline in cognitive and functional abilities in dementia can be usefully measured as a first approximation by use of mental status and functional instruments. An additional major contribution of the 1968 Blessed, Tomlinson, Roth study described above was their validation of formal scales for testing cognitive status and function against autopsy diagnosis. Their initial information–memory–concentration (IMC) test is still a widely used mental status test, and a brief version is now widely used in nursing homes and as part of the test battery used by the multicenter group the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). This test was developed after decades of investigation by psychiatrists interested in identifying test items that differentiated functional from “organic” mental syndromes; the final version was dependent on work by Roth and associates and Shapiro and associates. Subsequently, the Mini-Mental State Examination (MMSE) was described by Folstein and the MMSE has enjoyed wider usage than the IMC test. The two tests have many items in common; they differ in that the MMSE tests a broader range of function with items that test reading, writing, and manipulation of objects, whereas the IMC is entirely verbal but includes items of greater difficulty than those in the MMSE, items that are useful in testing subjects in the earliest stages of Alzheimer’s disease. The items include the recall of a five-part name and address and recitation of the months of the year backward. Mental status tests — and many variations of the above tests have been proposed and field tested — are relatively easy to administer and dominate both clinical and epidemiological studies. Such tests, however, are subject to a variety of problems including the effect of education and culture on the response.

Perhaps a more universal measure of the presence of dementia and dementia severity are questionnaires that quantitate impairment of so-called instrumental activities of daily living, such as the ability to handle money or shop independently, and, in late stages of dementia, instruments that estimate impairment of activities of daily living, such as dressing, eating, or toileting. Both types of activities of daily living were included in the dementia score that had been validated by the Blessed, Tomlinson, and Roth study. Alternate forms of these tests, particularly the scales of Lawton and Brody, Katz, and Pfefferare more often used today. All of these functional questionnaires depend on the availability of a reliable informant, but when used together with a mental status test are excellent indicators of the presence of a dementing illness and hence are especially useful in epidemiological studies.

DIAGNOSTIC CRITERIA FOR ALZHEIMER’S DISEASE

The success of the DSM-III criteria for the diagnosis of dementia led to the formation of a task force by the National Institute of Neurological and Communicative Disorders and Strokes (NINCDS) and the ADRDA to establish criteria for the diagnosis of Alzheimer’s disease. The task force agreed that the definitive diagnosis of Alzheimer’s disease depends on neuropathological confirmation. However, their criteria for “probable” Alzheimer’s disease has turned out to have an accuracy of 85 to 95% based on subsequent clinical–pathological analyses.

This high degree of accuracy was attained by the widespread adoption of an intensive clinical evaluation that required a careful clinical history, mental status testing, and neuropsychological and neuropsychiatric examination, all office procedures. In addition, this evaluation required blood tests to rule out various unusual but important metabolic disorders that might produce cognitive impairment, and an imaging procedure (such as a CT scan) to rule out the rare dementias secondary to hydrocephalus or midline tumors that do not present with focal signs. In addition, the imaging procedure assists in the diagnosis of vascular dementia by confirming the presence of cerebral infarcts. However, neither biopsies nor more complicated procedures such as positron emission tomography (PET) were needed to achieve these accuracy rates. This is perhaps remarkable among the major diseases of our time. Definitive diagnosis was recognized to require autopsy confirmation. The many families who actively participated in giving autopsy permission for their loved ones with presumed Alzheimer’s disease were rewarded in most cases with a definitive diagnosis and, in some cases, a diagnosis of important but less common neurodegenerative diseases such as Pick’s disease.

A REMAINING PROBLEM: DIFFERENTIATING ALZHEIMER’S DISEASE AND CEREBROVASCULAR DEMENTIA IN SUBJECTS WHO HAVE HAD A STROKE OR OTHER EVIDENCE OF SEVERE CEREBROVASCULAR DISEASE

The nosology of dementia is complicated by the existence of a progressive dementia in some individuals with strokes, a relationship that was recognized during the 19th century [Alzheimer published a paper on atheroma and dementia in 1898 (described by Bick1)]. Strokes, like Alzheimer’s disease, are age-dependent disorders which increase exponentially during the same period of life. Hence, there is a significant number of brains in the very elderly that contain evidence both of Alzheimer changes and of cerebrovascular disease. How can the clinician dissociate what role, if any, vascular or Alzheimer changes play in the development of dementia in an elderly patient? Historically, great confusion was caused by the assumption of some earlier investigators that progressive dementia in the elderly had a single cause, and they could not decide between these two pathologies. The diagnosis of vascular dementia achieved a vogue in the 1940s and 1950s when Dr. Walter Alvarez, a highly respected physician from the Mayo Clinic, ignoring the extant neuropathological literature, speculated that most cases of progressive cognitive symptoms in the elderly were due to multiple small strokes. He was a persuasive writer, and families of elderly dements were told that the condition was due to rigid arteries. It should be noted that, at the middle of the 20th century, there was still not adequate treatment for hypertension, and major stroke was four times more common than today. One can therefore assume that vascular dementia was undoubtedly more common than today, but in our own opinion most of the cases of progressive dementia even in the 1960s were likely due to Alzheimer’s disease.

It was the prospective clinical–pathological study of Blessed and colleagues described above that led to an initial, but not fully satisfying, resolution of this problem. Tomlinson and colleagues described in detail the neuropathological changes in the brains of the 17 patients in this study who were thought to have vascular dementia. All had multiple cerebral infarcts — areas of dead brain tissue secondary to strokes — with a total volume greater than 50 ml and in most cases with a volume in excess of 100 ml of infarcted brain. It was the infarcted brain tissue, not the presence of atheromatous plaques in blood vessels, that caused the dementia in these cases.

The concept that most cases of cerebrovascular dementia are due to multiple large infarcts became widely accepted. However, with the successful prevention of stroke by control of hypertension, the type of patient with large multiple infarcts described by Tomlinson in their original clinical pathological description is seldom seen today. Indeed, large cerebral infarcts occur with only one fourth the number that occurred in the 1960s. Nevertheless, as Tatemichi has recently shown, there are patients who have a single major infarct who develop progressive dementia without additional infarcts and others with multiple small infarcts or lacunes who develop dementia secondary to these vascular events. The problem has been further complicated by the overreading of hyperintense spots on the T2 image of the MRI which have been interpreted by some radiologists as indicative of ischemic disease but which are also seen in normal elderly and in Alzheimer’s disease.

A further diagnostic complication is that the amyloid angiopathy of Alzheimer’s disease can, if severe, lead to cerebral infarcts, particularly in patients who also are hypertensive. Thus, we are at an impasse today. We do not understand the pathogenesis of vascular dementia. As a consequence we continue to make diagnostic errors. It is now evident that scoring systems based on the original concept of a true multi-infarct dementia — the Hachinski ischemic score or the Rosen modification thereof — lead to only about a 50% accuracy of diagnosis of vascular dementia based on post-mortem findings. Recent attempts to develop better criteria for the diagnosis of vascular dementia are now being evaluated. The question of whether diagnosis can improve prior to an improvement in our basic understanding of this type of dementia is yet to be determined.