Reliability of intensity-based physical activity measurement using an activity monitor in people with subacute stroke in the hospital setting: a cross-sectional study.

Background Activity monitors are known to provide quantitative evaluation of physical activity in people with stroke. However, little evidence is available on the test-retest reliability and suitable measuring conditions for evaluating physical activity with an activity monitor in people with subacute stroke at a hospital setting. Objectives To evaluate reliability of physical activity measurement using an activity monitor in different measuring conditions in hospitalized people with subacute stroke according to gait ability. Methods Sixty-nine participants with subacute stroke were categorized based on their gait ability: group 1 (n = 22), could walk independently at sufficient gait speed (≥0.9 m/s); group 2 (n = 11), could walk independently with insufficient gait speed (<0.9 m/s); group 3 (n = 36), required assistance in walking. Physical activity (sedentary behavior, light-intensity physical activity, moderate-to-vigorous physical activity) was measured using an activity monitor (OMRON, HJA350-IT) for seven consecutive days. All physical activity variables were calculated in five conditions defined according to measurement periods (3, 5, or 7 days with or without weekends). Results In groups 1 and 2, intraclass correlation coefficients were high across all measuring conditions in all physical activity variables (>0.7). In group 3, intraclass correlation coefficients were high in all measuring conditions (>0.7) except for three weekdays measuring condition (0.563-0.922). Conclusions In hospitalized people with subacute stroke, the central tendency of intensity-based physical activity could be assessed with good reliability using an activity monitor. For reliable measurement, more measurement days are needed in those with dependent walking compared to those with independent walking.