Objective: To estimate the predictive value of four IADLs on 3- and 5-year incident dementia.
Design: Prospective cohort study.
Setting: A community survey in 37 randomly selected parishes.
Subjects: A random sample of French community dwellers aged 65 and older included in the PAQUID study followed-up at 3 (1582 subjects) and 5 years (1283 subjects).
Main outcome measures: Incidence of dementia at 3 and 5 years, diagnosed by two-step screening: (1) MMS and DSM-III-R; (2) diagnosis of dementia confirmed by a neurologist.
Results: A score summing up the number of dependencies at baseline on four IADLs is a predictor of 3-year but not of 5-year incident dementia. Increase in the level of dependence between baseline and 3-year follow-up is associated with an increased risk of incident dementia at 5-year follow-up.
Conclusion: These four IADLs can help to identify older subjects at high risk of dementia who should then undergo neuropsychological testing.