How many words can you remember? Comparing delayed recall tests in The Mexican Cognitive Aging Ancillary Study (Mex-Cog)

Abstract Background The assessment of episodic memory is a key component of cognitive screening and is usually assessed using a verbal delayed recall task. The number of words and trials often varies. This study will compare the diagnostic properties of recall tests using data from the Mexican Cognitive Aging Ancillary Study (Mex-Cog). Methods Data were obtained from of the Mex-Cog wave 2 (2021), a sub-analysis of the Mexican Health and Ageing Study focusing on cognition including a representative sample of community dwelling adults aged ≥ 58 years. Objective measures included: 3-word recall (scored 0–3), 10-word recall (scored 0–10), short story recall (scored 0–12) and long story recall (scored 0–50). Diagnostic accuracy for cognitive impairment was measured using Area Under the Curve (AUC) from Receiver Operating Characteristic curves. Results In all, 3,208 participants with a mean age of 72 years (range 58-99) were included of which 58% were female. The average number of words recalled was 2, standard deviation (SD) ± 1 for 3-word recall and 3 SD (± 2) for 10-word recall. For 3-word recall, 66% recalled ≤ 2 words while 42% recalled ≤ 2 words on the 10-word recall test. The 3-word and 10-word recall tests had poor agreement (Cohen’s Kappa < 0.2). The 10-word recall test had better diagnostic accuracy than 3-word recall for the lowest quartile of long story recall logical memory (AUC: 0.75 vs. 0.61, p < 0.001). Conclusion There was poor agreement between scores for 3-word and 10-word recall tests; and 10-word recall displayed better diagnostic accuracy for logical memory. The 3-item and 10-item word recall were not interchangeable subtests. Longer word lists to assess episodic memory (delayed recall) are preferable when conducting cognitive testing, acknowledging the well-established time-accuracy trade off when using cognitive screening instruments.