Purchase this article with an account.
Micaela R Gobeille, Kyoko Fujiwara, Robert W Massof; Rasch Analysis and Validation of a Timed Instrumental Activities of Daily Living Instrument in Low Vision Patients. Invest. Ophthalmol. Vis. Sci. 2021;62(8):3586.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Assess internal and external validity of a Timed Instrumental Activities of Daily Living (TIADL) instrument to evaluate visual ability based on task performance time and error rate for patients with low vision when the instrument is administered in patients’ homes.
Data were collected within patients’ homes from patients with low vision receiving home care from an occupational therapist (n=144). Patients completed a series of ADL tasks as quickly and accurately as possible. Tasks included: 2 telephone tasks, 8 cash transaction tasks (i.e. making change), 3 search tasks (i.e. finding items on a shelf), 3 short reading tasks (i.e. reading package information), and a medication management task. Completion time and number of errors were recorded for each task. Data were then quantized and underwent Rash analysis to estimate the latent visual ability variable based on observed time and accuracy. The Activity Inventory (AI) was also administered and data underwent Rasch analysis to estimate self-reported visual ability.
At baseline, average TIADL task performance time was 26 seconds (range 1-441 seconds), and 72% of the 2561 tasks performed were completed without error.Pseudo-R2 was large for TIADL item measures estimated by time (0.996) and error (0.954), as well as for person measures estimated by time (0.929). Pseudo-R2 for TIADL person measure estimated by error was small (0.16). The distribution of TIADL person measure infit mean squares approximated the expected χ2/df distribution for both time and error. Item infit z-scores fell within 2 standard deviations of 0 except for one item corresponding to medication management. TIADL item measures estimated from time versus those from error were well correlated (R=0.73), while person measures estimated from time versus those from error showed a more moderate correlation (R=0.38). Visual ability based on TIADL time was moderately correlated (R=0.34) with visual ability based on the AI.When rehabilitation outcome measures were compared between TIADL time and the AI, change scores were positive (0.65 logit for AI, 0.54 logit for TIADL time) and Cohen’s effect sizes were moderate (d=0.78 for AI, d=0.53 for TIADL time). Weak agreement was seen between AI and TIADL time change scores (R=0.16).
Valid outcome measures can be estimated from TIADL performance time when observed in the patient’s home environment.
This is a 2021 ARVO Annual Meeting abstract.
This PDF is available to Subscribers Only