Estimating the true extent of cognitive decline in the old old

OBJECTIVE: To measure cognitive change using a brief measure over a period of 9 years and to adjust for attrition in the sample. DESIGN: The Cambridge City over 75 Cohort (CC75C), a complete sample of the 75 years and older age group from five group general practices in the city of Cambridge with a systematic one-third of a further practice, all followed on four occasions. SETTING: Cambridge city, UK, the respondents' place of residence. PARTICIPANTS: A total of 2106 subjects were included at study entry. MEASUREMENTS: A brief interview, administered by a trained interviewer, containing a short cognitive scale and the Mini-Mental State Examination (MMSE) at baseline, 2.4 years, 6 years, and 9 years. RESULTS: Decline in MMSE scores occurred across the population and was greater in the oldest age groups. Attrition at later stages of the follow-up was associated with greater decline at earlier stages. Adjusting the results for loss to the sample leads to considerably higher estimates of decline, with the older age groups declining faster from lower levels. CONCLUSIONS: To date, cognitive decline in the very old has been considerably underestimated by longitudinal studies. If studies of population samples are to reflect the health and social needs of this frail group accurately, adjustments for the effect of attrition must be included before true decline can be estimated.