Identifying change on the CERAD Word List with standardized regression‐based scores

Abstract

AbstractBackgroundList learning and memory tests are frequently used to evaluate older adults for Mild Cognitive Impairment and Alzheimer’s disease. One such test, the Word List subtest of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD), has been widely‐used in clinical trials track such disorders. However, data is lacking on the amount of change that is needed to accurately identify decline.MethodParticipants included 897 older adults (age in years: M = 81.1, SD = 8.1, range 55‐100; 67% women; education in years: M = 14.8, SD = 2.8, range = 7‐24) who were cognitively intact at two annual visits in a longitudinal observational study from the Uniform Data Set of the National Alzheimer’s Coordinating Center. Each participant completed the Word List subtest at two visits, which were approximately 1 year apart (interval in years: M = 1.3, SD = 1.0). Multiple linear regression predicted Time 2 scores from Time 1 scores and simple demographic variables (age, sex, education). The predicted Time 2 scores were compared to observed Time 2 scores to preliminarily validate this change data.ResultFor three primary scores from the Word List subtest (Acquisition Total Correct, Delay Recall Total Correct, Recognition Total Correct), the Time 2 scores were significantly predicted from the other variables, with R2s ranging from 0.16 – 0.41 (p‐values<0.001). When these predicted Time 2 scores were compared to observed Time 2 scores, there were no significant differences (p‐values≧0.19), indicating preliminary support for these prediction scores.ConclusionThe current prediction equations appear to allow clinicians and researchers to accurately track change on the Word List subtest of the CERAD by comparing their patients/participants to the current sample of robustly intact older adults. Consistent with past work in this area, Time 1 cognitive scores were the best predictors of Time 2 cognitive scores, although demographic variables added to that prediction. Future work should consider if additional variables can improve the prediction equations. Future work should also validate these equations in independent samples to examine their clinical utility.