An accelerometry and observational study to quantify upper limb use after stroke during inpatient rehabilitation

Objectives: The objectives of the study are to characterize paretic upper limb (UL) use in people with different levels of impairment 4\ua0weeks poststroke and to compare accelerometry and direct observational approaches. Methods: Twelve stroke inpatients (five mild, three moderate, and four severe UL impairment) were recruited from a rehabilitation hospital. UL use was measured using accelerometry (24\ua0hr) and direct observation (12\ua0hr of behavioural mapping). Accelerometry variables included duration of use, use ratio, magnitude ratio, bilateral magnitude, and variation ratio. Direct observation recorded the duration of use and type of UL movement (e.g., functional vs. non-functional). Results: From accelerometry data, people with mild, moderate, and severe UL impairments used their paretic UL 59%, 45%, and 22% of a 24\ua0hr-day, respectively. People with severe UL impairment had the lowest paretic UL use duration (median 1.49\ua0hr/day), magnitude ratio, and variation ratio compared with people with mild and moderate UL impairment. From 12\ua0hr of observational data, people with mild impairment were using their UL for 37.8% of the observed time, whereas the people with moderate and severe impairment were using their UL 15.8% and 4.9%, respectively. UL movements for the mild cohort were mainly functional, whereas UL movements of the moderate and severe cohorts were mainly non-functional. UL movements were predominantly active for the mild and moderate cohorts but passive for the severe cohort. Duration of paretic UL use from accelerometry and observation data were highly correlated (ICC\ua0>\ua00.8), but the absolute percentage error between methods ranged from 34.2% to 42.7%. Conclusions: Paretic UL use within the first 4\ua0weeks poststroke differs across levels of impairment in this exploratory study. Accelerometry and observation findings of paretic UL use were correlated and may be needed in different situations as they capture different information