Background and Purpose: People with stroke have low levels of activity, are sedentary, and have poor sleep, which have a negative impact on outcomes. Compositional data analysis (CoDA) provides a method to analyze movement behaviors that preserves the relative nature of the different movement behaviors within a day. The purpose of this study was to use CoDA analysis to explore the association between the composition of movement behaviors and physical function and to compare it to traditional analyses that consider movement behaviors as independent variables. Methods: Participants (n = 68) wore activity monitors that measured time in active, sedentary, and sleep behaviors at 60 days post stroke. CoDA using the isometric log-ratios of the composition of the movement behaviors in a linear regression was used to explore the association between movement behaviors and the Stroke Impact Scale 16. Traditional analysis considered individual movement behaviors separately. Results: The compositional mean of the movement behaviors found that participants spent 7.3% of the day in active, 58.1% in sedentary, and 34.6% in sleep behaviors. The composition of the movement behaviors was significantly associated with Stroke Impact Scale 16 (adjusted R 2 = 0.5007, P < 0.05). This was greater than when the individual movement behaviors were considered separately (adjusted R 2 = 0.3926, P < 0.05). Discussion and Conclusions: The composition of movement behaviors (active, sedentary, and sleep) was associated with physical function. Using CoDA with movement behavior data in people with stroke may improve the interpretability of the data by accounting for the dependency between compositional components. Clinicians may want to consider a comprehensive intervention to support behavioral change that seeks to increase physical activity, enhance sleep, and reduce sedentary time.
Fulk et al. (Tue,) studied this question.