The relationship between age and the prevalence of senile dementia: a meta-analysis of recent data.

A linear regression model derived from a meta-analysis of 13 epidemiological studies of senile dementia conducted since 1980, and employing internationally-known case-finding procedures, suggests a much lower general rate of dementia prevalence than has been previously estimated. An exponential increase with age is observed, with senile dementia prevalence diagnosed by Diagnostic and Statistical Manual (DSM-III) criteria doubling every 6 years and senile dementia of the Alzheimer's type (SDAT) every 4.2 years. Studies providing data for the oldest ages indicating a drop in the rate of increase after the age of 80 suggest that senile dementia may be age-related rather than ageing-related. Estimates derived from this model may provide a reasonably accurate means of estimating dementia prevalence in the general population. The limitations of this method for the purposes of prediction and studies of risk factors are discussed in relation to the hypothesized heterogeneity of senile dementia and possible cohort effects.