Is case-ascertainment of Alzheimer's disease in field surveys practicable?

Research on Alzheimer's disease (AD) has recently burgeoned because of its public health importance and scientific opportunities. While biological investigations are energetically pursued and now carry considerable promise, clinical and epidemiological research may also have a contribution to make (Khachaturian, 1985). Such work may help in identifying subtypes of AD, in establishing its natural history and in finding clues to its aetiology. But here a major obstacle arises: the cases studied should be of Alzheimer's disease, not other dementias. This requires highly reliable and valid diagnosis during life. Only five years ago, Masters et al. (1981, p. 165) expressed pessimism about the problem, asserting that' Until criteria are established for diagnosis, it seems almost pointless to attempt an epidemiological analysis . . . There does not even appear to be any reliable information on the accuracy of clinical diagnosis in a large enough series of patients with AD. Therefore, no estimate is available on the degree of case ascertainment that might be expected from population surveys'. For clinical research, the situation is now much improved: there are research diagnostic criteria for AD; and information on the accuracy of clinical diagnosis has been obtained. But there is no experience of trying to use these criteria in field surveys; and nothing is known about the accuracy of the diagnoses obtained under survey conditions. Ascertainment of cases of Alzheimer's disease necessarily must proceed in two stages: the diagnosis of dementia; then the diagnosis of AD itself, with the exclusion of other causes of dementia. Here we examine what is presently known about reliability and validity at both these stages; and how practicable it is to achieve accurate case-ascertainment in field studies.