Has familial aggregation in Alzheimer's disease been overestimated?

Studies of the familial aggregation of Alzheimer's disease have primarily u sed samples ascertained from tertiary care clinics which may not be represe ntative of many AD patients, for example those residing at geriatric nursin g homes. Survival analysis was used to investigate whether estimates of fam ilial aggregation of AD based on a clinic-based AD proband (C-AD) sample (p robands: N = 544; first degree relatives; N = 4267) differ from one ascerta ined at a nursing home (NH-AD; probands: N = 225; first degree relatives: N = 1772). The cumulative survival from AD was significantly worse in relati ves of the C-AD probands and the overall relative risk (RR) of AD in this g roup was greater than twice that of relatives of the NH-AD probands. Howeve r, age at onset in C-AD probands was significantly earlier than in the NH-A D group and in both groups this factor was negatively associated with famil ial aggregation. When, for this reason, the proband samples were matched on e-to-one by age at onset, dropping those probands with no match, the two cu rves were close to identical and the RR for the C-AD group of relatives was 1.0. The results suggest that estimates of familial risk of AD based on C- AD samples are not applicable and overestimate the extent of increased risk for relatives of more prevalent, later onset AD probands. However, the ove restimate can be explained by the typically earlier age at onset in C-AD sa mples as opposed to a sampling bias related to the proband's family history status per se. The relationship between onset age and familial aggregation suggests that no single estimate of the age-dependent risk (survival curve ) is uniformly appropriate for relatives of AD probands. Copyright (C) 2000 John Wiley & Sons, Ltd.