There are many screening instruments available for the detection of dementia. However, they were not tested in very old age and often not in the primary care setting for which they are developed.

We condensed a screening battery of all elements of common tests (MMSE, TFDD, DEMTECT, Clock Test…). Nine primary care practices took part in the study. The practice assistants were trained to approach all patients over 75y visiting the practice. Only patients with already diagnosed dementia, severe hearing or vision disability or communication deficits were excluded. After informed consent they applied the screening. Within the next days a second independent neuropsychological examination was arranged in the memory clinic of the University of Goettingen. This included established neuropsychological tests (CERAD-NP, WMS-R, TMT.) and further scales were applied (NPI, CDR..). All patients with CDR=0.5 were approached 1.5y later for a follow-up by phone.

N=90 patients (25M, 65F) took part in the initial evaluation. From N=54 with CDR=0.5 only N=14 could be reached for follow-up. None of them returned to CDR=0, most progressed up to CDR=4. There were some correlations to education and dementia severity. All available screenings showed a good effect size of >0.70 (MMSE, DEMTECT, TFDD, RDST).

Elements for an optimal screening of the oldest old are discussed.