The validity of the family history method for identifying Alzheimer disease

To examine the validity of the family history method for identifying Alzheimer disease (AD) by comparing family history and neuropathological diagnoses.

Seventy-seven former residents of the Jewish Home and Hospital for the Aged, New York, NY, with neuropathological evaluations on record were blindly assessed for the presence of dementia and, if present, the type of dementia through family informants by telephone interviews. The Alzheimer's Disease Risk Questionnaire was used to collect demographic information and screen for possible dementia. If dementia was suspected, the Dementia Questionnaire was administered to assess the course and type of dementia, i.e., primary progressive dementia (PPD, likely AD), multiple infarct dementia, mixed dementia (i.e., PPD and multiple infarct dementia), and other dementias based on the modified Diagnostic and Statistical Manual of Mental Disorders, Third Edition, criteria.

Sixty (77.9%) of 77 elderly subjects were classified as having dementia and 17 (22.1%) were without dementia by family history evaluation. Of the 60 elderly subjects with dementia, 57 (95%) were found at autopsy to have had neuropathological changes related to dementia. The sensitivity of the family history diagnosis for dementia with related neuropathological change was 0.84 (57 of 68) and the specificity was 0.67 (6 of 9). Using family history information to differentiate the type of dementia, the sensitivity for definite or probable AD (with or without another condition) was 0.69 (36 of 51) and the specificity was 0.73 (19 of 26). The majority (9 of 15) of patients testing false negative for PPD had a history of stroke associated with onset of memory changes, excluding a diagnosis of PPD.

Identifying dementia, in general, and AD, in particular, has an acceptable sensitivity and specificity. As is true for direct clinical diagnosis, the major issue associated with misclassifying AD in a family history assessment is the masking effects of a coexisting non-AD dementia or dementia-related disorders, such as stroke. Including mixed cases, ie, PPD and multiple infarct dementia in estimates of the familial risk for AD can reduce the extent of underestimation of PPD.