Objectives: Older people commonly present with memory loss although on assessment are not found to have a full dementia complex. Previous studies have suggested however that people with subjective and objective cognitive loss are at higher risk of dementia. We aimed to determine from the literature the rate of conversion from mild cognitive impairment to dementia.

Methods: Systematic review of MedLine, PsychLit and EmBase.

Results: We identified 19 longitudinal studies published between 1991 and 2001 that addressed conversion of mild cognitive impairment to dementia. Overall the rate of conversion was 10% but with large differences between studies. The single biggest variable accounting for between study heterogeneity was source of subjects, with self-selected clinic attenders having the highest conversion rate. The most important factor accounting for heterogeneity within studies was cognitive testing, with poor performance predicting conversion with a high degree of accuracy.

Conclusions: These data strongly support the notion that subjective and objective evidence of cognitive decline is not normal and predicts conversion to dementia. The more stringent the measures of both variables the better the prediction of conversion. Mild cognitive impairment, appropriately diagnosed, is a good measure with which to select subjects for disease modification studies.