Objectives: To develop an informant-based semistructured interview to determine the onset and duration of symptoms of Alzheimer's disease, and to use this instrument with informants to characterize a cohort of mildly impaired patients with Alzheimer's disease.
Design: In study 1, interrater and interinformant reliability was examined for the date of onset and the order of appearance for specific symptoms that were elicited by the semistructured onset interview. In study 2, the instrument was used to characterize disease onset in a cohort of patients with Alzheimer's disease who were participating in a large multicenter study.
Subjects: Informants of patients with Alzheimer's disease.
Results: In study 1, interrater reliability for duration of illness was excellent (intraclass correlation coefficient = .99, P < .001), and interinformant reliability was good (intraclass correlation coefficient = .86, P < .001). Agreement for the presence of a given symptom was highest for those that were most commonly reported (eg, memory and performance difficulty). In study 2, 89% of the cohort had memory problems, and 63.9% had performance difficulties as the first or second symptom. Depression and language problems were less commonly reported. Psychosis and behavioral disturbances were rarely reported as the first problem.
Conclusion: This instrument provides a reliable procedure for standardizing the estimation of duration of illness based on retrospective report.