Diabetes and cognitive impairment: a community-based study of elderly subjects.

To ascertain whether diabetes in elderly people is associated with cognitive impairment, we offered all residents of Melton Mowbray aged 75, 80 and 85 years both a modified glucose tolerance test (1985 WHO criteria) and Folstein mental state examination (MMSE, 23/24 cut-off). Analysis of the results, stratified by age, revealed that subjects with known diabetes were more likely than normal subjects to have low MMSE [odds ratio 3.30 (95% CI 1.3 to 8.5)], whilst newly found diabetic subjects were less likely to have low MMSE [odds ratio 5(-14) (95% CI 9(-25) to 0.003)]. The difference between known and newly found diabetic subjects might relate to duration of diabetes. The increased frequency of cognitive impairment in known diabetic subjects may be pertinent to the safe use of hypoglycaemic agents.