[Creation and normalisation of a verbal episodic memory task in elderly adults: "GERIA-12"]

Early damage to episodic memory encoding and consolidation processes has been demonstrated in dementia of the Alzheimer type. However, in the domain of verbal episodic memory assessment, there are few diagnostic tools adapted to the old and oldest old as far as ease of administration and accuracy of normative data are concerned. Classic tasks are either too effortful (like the free recall/cued recall of 16 items), not sensitive enough (like the 5 words test), or insufficiently accurate for people above 70 years old in terms of normative data.

The aim of this study was to develop a reduced task (in terms of number of items and number of trials) assessing verbal episodic memory in people aged between 70 and 89 years old.

The task (GERIA-12) used the same procedure as the RL/RI-16 task but the list comprised only 12 words and there were only 2 learning trials. In order to assess consolidation processes, we included 2 delayed recall trials, one after 20 minutes and the other after 24 hours. We also calculated indexes adapted from the Item-Specific Deficit Approach developed by Wright et al., which has the advantage of providing measures specific to encoding, consolidation and retrieval processes. Standardization was done with data from 220 people aged between 70 and 89 years old and belonging to 3 education levels.

We obtained a significant effect of age and education level: scores decrease with age and increase with education. Norms have thus been calculated taking those two variables into consideration. Concerning the standardization, Barona method has been used for free recall scores while percentiles have been used for all other scores (total recall, free recall, encoding, consolidation and retrieval indexes). Normative data are also provided for intrusions and perseverations.

This new task allows encoding, consolidation and retrieval processes assessment in older people and has the following advantages: the procedure is more suitable (ease and time of administration), there are accurate normative data for old and oldest old people, and there are normative data for two delayed recalls (at 20 minutes and at 24 hours).