File
Andersson S.1,2, Danielsson A.1,2, Ohlsson F.3, Wipenmyr J.3, Alt Murphy M.1,2
1University of Gothenburg, Inst. of Neuroscience and Physiology, Rehabilitation Medicine, Gothenburg, Sweden, 2Sahlgrenska University Hospital, Occupational Therapy and Physiotherapy, Gothenburg, Sweden, 3Acreo Swedish ICT AB, Gothenburg, Sweden
Background: Individuals after stroke have difficulties to reach a sufficient physical activity level within and outside the therapy time. The number of studies using accelerometers in clinical settings is increasing, but clinical feasibility and implications for accelerometers are still unclear.
Purpose: To quantify activity levels and possible differences in physical activity in weekdays and weekends and investigate clinical feasibility of accelerometers in the subacute stage after stroke.
Methods: The study included 21 persons with stroke undergoing inpatient rehabilitation. Data were collected using five triaxial accelerometers (trunk, wrists, and ankles) during two 48 hours sessions on weekdays and weekend, respectively. Acceleration raw data were filtered, and activity level expressed as the Signal Magnitude Area (SMA) averaged over 2 min epochs from day-time periods of 2 consecutive days (in total 24hrs). Measurements including acceleration data less than 22 hrs for one session were not included in the analyses. Mean activity level from each sensor and symmetry indices, expressed as ratios, for upper and lower extremities were calculated. Motor impairment was assessed using the Fugl-Meyer Assessment (FMA) and comfort was assessed using a 5-point Likert scale.
Results: Full accelerometer data (≥ 22 hours) were collected in 82% of all measurement periods. In 8% data was partly missing (1-21 hours of registered data) and in 10% completely missing. Malfunctioning of sensor(s) and/or software and patient-related failures as inadequate support or cognitive problems were identified as main reasons for data loss. Data from18 individuals were included in the final analysis (61% male, mean age: 54.6 yrs, mean time since stroke: 54 days, mean FMA score: 55/100). Overall, the mean acceleration of all sensors was low (0.56-2.2 m/s2). Both the affected (95%CI: 0.08 to 0.25, p=0.001) and non-affected arm (95%CI: 0.03 to 0.41, p=0.024) were used less during weekends. The affected leg (95%CI:-0.09 to 0.21) and non-affected leg (95%CI:-0.02 to 0.28) also showed tendency of less activity over weekend. No difference was found for the trunk sensor. The ratio between paretic and non-paretic arm/leg indicated that the affected side was less active but there were no differences between weekdays and weekends. Of 17 participants, one strongly agreed, six agreed, eight were undecided, two disagreed and none strongly disagreed that the sensors were comfortable to wear and use.
Conclusion(s): Relatively low-cost accelerometers allowing raw-data handling, as used in this study, can with customized data handling provide meaningful information about activity levels during real-life measurement. Accelerometer data from extremities and the ratio between affected and non-affected extremity were the most informative measures. In clinical settings, individual modifications for application and technical support are required for generating meaningful information.
Implications: Multiple sensors allowing raw-data handling have a potential but further development is needed to become a user-friendly, simple and easily interpretable tool useful for monitoring patients´ activity levels in clinical settings.
Funding acknowledgements: Local Research and Development Board for Gothenburg and Södra Bohuslän, Swedish National Stroke Foundation, Swedish Foundation for Strategic Research (WearItmed).
Topic: Neurology: stroke
Ethics approval: Regional Ethical Review Board in Gothenburg, Sweden (507-15).
All authors, affiliations and abstracts have been published as submitted.
