JMIR Publications

ABSTRACT

Background:

Various chronic conditions are risk factors for decreased cognitive performance, making assessment of cognition in people with chronic illness important. Formal mobile cognitive assessments measure cognitive performance with higher temporal density and greater ecological validity compared to traditional laboratory-based testing, but can be burdensome for participants. Given that responding to a survey is considered a cognitively demanding task, response times that are passively collected as a byproduct of ecological momentary assessments (EMA) may be a means through which cognitive performance in people’s natural environment can be captured, while limiting participant burden. We examine if EMA response times can serve as approximations of cognitive processing speed.

Objective:

To investigate if response times from EMA surveys can serve as approximate indicators of between-person differences and momentary within-person variability in cognitive processing speed.

Methods:

Data from a two-week EMA study investigating relationships between glucose, emotion, and functioning in adults with type 1 diabetes (T1D) were analyzed. A validated mobile cognitive processing speed test (Symbol Search) was administered with EMA surveys 5-6 times/day via smartphones. Multilevel modeling was used to examine the reliability of EMA response times and their convergent validity with the Symbol Search task. Other tests of convergent validity of EMA response times included examining associations with age, depression, fatigue, and time of day.

Results:

In between-person analyses, overall, evidence was found supporting the reliability and convergent validity of EMA response times as a measure of average processing speed from even a single EMA item administered repeatedly over a two-week period. Between-person correlations between the Symbol Search task and EMA response times ranged from 0.43 to 0.58 (P<.001). EMA response times had significant between-person associations with age as expected, but not depression or average fatigue. In within-person analyses, response times from 16 slider items and all 22 EMA items (including the 16 slider items) had acceptable (>0.70) within-person reliability. Within-person reliability was much lower for sets of items with heterogeneous response options (e.g., multiple-choice questions with different numbers of response options) and for individual items. After correcting for unreliability in multilevel models, EMA response times from most combinations of items showed moderate within-person correlations with the Symbol Search task (ranged from 0.29 to 0.58, P<.001), and demonstrated theoretically expected relationships with momentary fatigue and time of day.

Conclusions:

Response times from EMA items may be a minimally burdensome method of approximating average levels and momentary fluctuations in processing speed, both of which may be influential on daily functioning, particularly for populations with chronic conditions.