Abstract
Purpose:
Validate functional ability measures estimated from timed independent activities of daily living (TIADL) tasks and compare to those estimated from patient self-report.
Methods:
Participants were home-health patients being served for non-vision related conditions, but all had low vision. 40 patients completed the study and all measures were made at baseline in the patients’ homes. Participants were timed while performing 17 standardized IADL tasks in their customary manner. The tasks were: 2 telephone tasks (dialing 911 and a 10 digit phone number), 8 related to cash transactions (i.e. paying and counting change), 3 search tasks (i.e. finding items on a shelf), 3 short reading tasks (i.e. package information), and a sorting task (identifying/sorting medications). Participants were encouraged to place highest priority on performing tasks accurately.<br /> Performance time was recorded for each task and assigned to one of 5 ordinal categories (quintiles across all tasks/ participants). Rasch analysis of the ordinal time categories was used to estimate an item measure for each task and a person measure for each participant.<br /> In addition to the TIADL tasks, participants completed the Activity Inventory (AI) by telephone interview. Visual ability was estimated from difficulty ratings of AI reading, visual motor and visual information tasks, all of which would normally be performed inside the home (equivalent to IADLs). The AI item measures and response category thresholds were anchored to calibrated values previously estimated from the responses of 3200 low vision patients.
Results:
Estimated item measures in the TIADL tasks are linear with average log time for the item across all participants. Item measure reliability is 0.94; person measure reliability is 0.89. Both item measures and person measures proved to be valid and unidimensional, as defined by Rasch model expectations. Principal component analysis of response residuals shows that the estimated measures explain 62% of the observed variance and that the unexplained variance is random. Person measures estimated from TIADLs agree with person measures estimated from AI difficulty ratings for the same participants (Pearson r=0.52).
Conclusions:
Valid interval-scaled functional ability measures can be estimated from TIADL tests. At baseline these performance-based functional ability measures agree with those estimated from patient self-report.