Comparing the reliability of physical activity questionnaires in community-dwelling adults with stroke.

Background: Physical activity (PA) is important for people with stroke, but the reliability of PA questionnaires used in this population has been relatively unexplored.

Objective: To compare the internal consistency, test-retest, and absolute reliability of 3 commonly used PA questionnaires in adults with stroke.

Methods: Participants reported their PA levels twice, 2-3 days apart, using the Physical Activity Scale for Individuals with Physical Disabilities (PASIPD), International Physical Activity Questionnaire (IPAQ), and Global Physical Activity Questionnaire (GPAQ). Intraclass correlation coefficients (ICC_2,1) were calculated for test-retest reliability, Cronbach's alpha (α) for internal consistency, and standard error of measurement (SEM) and minimal detectable change (MDC₉₅) for absolute reliability.

Results: Twenty-eight people (64.4 years, 50% female, 5.2 years post-stroke) participated. Internal consistency was acceptable for total scores on the IPAQ ($\alpha$ = 0.79) and GPAQ ($\alpha$ = 0.74), but only domain-level scores for the GPAQ ($\alpha$ = 0.71-0.88). In the full sample, test-retest reliability was good for the PASIPD (ICC_2,1 = 0.83) but poor for the IPAQ and GPAQ (ICC_2,1 <0.50). After excluding participants self-reporting true changes in PA, all questionnaires had good test-retest reliability (ICC_2,1 = 0.77-0.88). SEM and MDC₉₅ were lowest for the PASIPD (188.8 and 523.3 MET-minutes/week, respectively).

Conclusions: In adults with stroke, the IPAQ and GPAQ had adequate total-questionnaire internal consistency, and the GPAQ had acceptable domain-level internal consistency. When true change in PA did not occur, test-retest reliability was good for all questionnaires. We suggest clinicians and rehabilitation scientists can use any of the three questionnaires, but may consider the GPAQ due to acceptable internal consistency and test-retest reliability.