Alternate forms of logical memory and verbal fluency tasks for repeated testing in early cognitive changes

Abstract

Background: Repeat cognitive testing is an essential diagnostic strategy to measure changes in cognition over time when following people with memory problems. Alternate forms may avert practice effects that can mimic improvements in cognition. We evaluated alternate forms of verbal fluency and logical memory (paragraph recall) tasks to evaluate their equivalence for clinical use.

Methods: Participants with mild cognitive impairment (MCI) and dementia were recruited from five outpatient memory clinics and one nursing home. Participants with normal cognition (NC) were recruited from family members or friends. Verbal fluency categories of animals, cities & towns, fruits & vegetables and first names were used. Scores were recorded for 0–30 seconds, 31–60 seconds and errors. For the logical memory task, participants were read one of three different paragraphs and then were asked to recall the story. Immediate recall and delayed recall scores were recorded. The Standardized Mini-mental State Examination, the AB Cognitive Screen and the 15-point Geriatric Depression Scale were administered as part of the assessment. Analyses were performed using means, frequency distributions, t-tests, receiver-operating characteristic curves and effect sizes.

Results: There were 46 NC participants, 45 with MCI and 55 with dementia. For verbal fluency, the mean number of animals, cities & towns, names or fruits & vegetables named in 60 seconds did not differ significantly within each cognitive group. First names was an easier category than the others: NC named 16.9–22.3 items, MCI named 11.6–14.4 items and dementia named 8.1–11.4 items. The mean number of items immediately recalled in logical memory was not significantly different for the three paragraphs. The verbal fluency task (in 60 seconds) and logical memory immediate recall were highly sensitive and specific to differences between NC and MCI (areas under the curves 0.87 and 0.76, respectively).

Conclusions: Alternate forms allow serial testing without learning bias. Verbal fluency and logical memory tasks are sensitive to early cognitive changes.