| Sumario: | Background: The lack of information on the accuracy of physical activity (PA) intensity estimation using PA monitors in nonambulatory individuals with stroke has hindered the advancement of epidemiological research in this population. Objective: This study aimed to validate the accuracy of accelerometer-based PA monitors for estimating PA intensity in nonambulatory patients with subacute stroke. Methods: During physical therapy sessions involving 50 nonambulatory participants with subacute stroke, sedentary behavior (SB), light intensity physical activity (LPA), and moderate-to-vigorous-intensity physical activity (MVPA) were assessed using PA monitors placed on two wearing sides (affected/unaffected) and recorded using two epoch lengths (10/60 s), along with video recordings. The time-series concordance rates for the PA intensity classification by the PA monitors were calculated using video observations as the benchmark. The outcomes were analyzed using a hierarchical linear model, with the wearing side, epoch length, intensity, and their interactions as independent variables to assess the measurement accuracy. Results: MVPA was excluded due to limited observations. Concordance rates were approximately 70% for SB and 80% for LPA. A significant product term between epoch length and intensity indicated that a 10-second epoch improved concordance for SB, whereas a 60-second epoch enhanced concordance for LPA. Conclusion: PA intensity estimation using monitors in nonambulatory patients with subacute stroke was generally accurate. A 10-second epoch length was optimal for measuring SB, while a 60-second epoch length was preferable for measuring LPA. These findings provide scientific evidence on the optimal measurement conditions for conducting epidemiological studies on PA in patients with nonambulatory stroke.
|
|---|