Applying a hierarchical linear model to investigate intra-individual variability in response time as a predictor for neurocognitive impairment

Abstract

The Alzheimer’s Association Research Roundtable has called for research to focus on the development of diagnostics that identify dementia and Alzheimer’s’ disease (AD) before overt symptoms arise and the disease accelerates. Intra-individual variability in response time (IIV-RT) has been identified as a potential predictor for cognitive impairment in its earliest stages, making it a promising candidate for sensitive diagnostic measures. This study applies a longitudinal hierarchical linear model to accurately accommodate the nested structure of the present data set. Participants, consisting of young-age (YA; 21-35 years) and old-age (OA; 60-75 years) cohorts, underwent a cognitive battery followed by encoding and retrieval phases of an associative memory task. It was hypothesized that accuracy at retrieval and retrieval trial number would significantly influence IIV-RT, while age group membership and cognitive test scores that reflect neurocognitive health would moderate the effects of these trial-level variables. It was also expected that OA participants would exhibit a higher IIV-RT than YA participants. Performance-wise and trial-wise split-cohort analyses were also conducted to further investigate retrieval performance and early versus late trial effects, respectively. Statistical tests showed that both retrieval accuracy and retrieval trial number significantly explained variance in response time. Cognitive test scores and age group membership significantly moderated the effects of retrieval trial number, but not accuracy, on response time. Collectively, OA participants had a higher IIV-RT than YA participants. In OA participants, IIV-RT was higher in late trials, whereas in YA participants it was higher in early trials. Within each age cohort, those who retrieved with lesser accuracy had higher IIV-RT than those who retrieved with greater accuracy. The present study poses evidence that accuracy at retrieval influences IIV-RT, that a given response time is in part determined by its location within a series of trials, and that cognitive health and age moderate these effects. Importantly, this study also emphasizes that IIV-RT is related to performance in a memory retrieval task, regardless of age group membership. These results further support the notion that IIV-RT may serve as a useful cognitive marker for the identification of cognitive deficits, and its potential for incorporation in more sensitive dementia diagnostic techniques.