VALIDATION OF TWO COMMERCIALLY AVAILABLE ACTIVITY MONITORS IN HOSPITALIZED GERIATRIC REHABILITATION PATIENTS

M. van Dijk¹, C. Van Driessche², T. Van Meerbeek², H. Van Cleynenbreugel³, S. Verschueren², W. Bijnens⁴, J. Tournoy⁵, G. Verheyden², J. Flamaing⁵

¹UZ Leuven, Campus Pellenberg, Physical Medicine and Rehabilitation, Pellenberg, Belgium, ²KU Leuven, Department of Rehabilitation Sciences, Leuven, Belgium, ³UZ Leuven, Physical Medicine and Rehabilitation, Leuven, Belgium, ⁴Maastricht University, IDEE, Maastricht, Netherlands, ⁵KU Leuven, Department of Public Health and Primary Care, Leuven, Belgium

Background: Activity monitors are frequently used to count steps and evaluate the intensity of their activities in the community. This information is highly relevant to evaluate functional performance and daily activity of older inpatients. In older patients using a walking aid and with walking speed >0.8m/s, activity monitors have been investigated and some have been found sufficiently accurate. However, patients in a geriatric rehabilitation ward tend to walk much slower and sometimes even require physical assistance. At the same time, new activity monitors with more and different types of sensors are constantly emerging. Can we use commercially available activity monitors in this specific population?

Purpose: The aim of this study was to validate the accuracy of two types of trackers in a predefined test setting: the general wrist-worn Fitbit sense and the MOX which is attached to the upper leg.

Methods: Twenty older participants (age 70+) using a walking aid were recruited in 2021 on the geriatric rehabilitation unit at UZ Leuven. There were two groups: those who walked independently (mean walking speed of 0.63 m/s), and those who required one person to support them physically (mean walking speed of 0.35 m/s). They completed a set of tasks while wearing the two devices, namely lying, sitting, standing, and walking, each lasting four minutes. The sessions were video recorded and analyzed on a second-by-second basis, and on step count, by three raters. We examined the Fitbit's accuracy in counting steps and activity intensity. For the MOX, we check for correctly registering sedentary activities (lying and sitting), standing, and dynamic activities (walking) sec by sec. By comparing the seconds of a specific activity or the number of steps observed by the raters on the video with the seconds recorded by the MOX or steps by the Fitbit, the percentage error was calculated.

Results: The Fitbit delivered poor results, with step count percentage error rates ranging from 71% to 100%. The program did not distinguish between sedentary and dynamic activities in this group, even though it supplied the intensity of activities based on heart rate zones. The Fitbit does not provide lying, sitting, or standing detection. For sedentary actions (lying or sitting), the MOX was error-free. There were some minor errors distinguishing between standing and walking, and the recording of standing and walking started consistently 4 seconds late. For standing and walking, the average percentage errors were 2.2% and 2.7%, respectively. There was no significant difference in the accuracy if physical support was used or not.

Conclusions: We conclude that for hospitalized geriatric rehabilitation patients, the MOX on the upper leg performs noticeably better compared to the Fitbit. This is a result of two factors:
1) a better positioning on the body, making it simpler to detect orientation and swing movement even when using a walking aid; and
2) an algorithm that has been modified to detect slower walking.

Implications: The accuracy of activity monitors cannot be generalized to different populations. The MOX can accurately track sedentary, standing and walking activities in the slow walking hospitalized geriatric rehabilitation population.

Funding acknowledgements: No funding was received.

Keywords:
activity monitors
older people
in-hospital rehabilitation

Topics:
Older people
Disability & rehabilitation
Health promotion & wellbeing/healthy ageing/physical activity

Did this work require ethics approval? Yes

Institution: UZ Leuven

Committee: Ethische Commissie Onderzoek UZ/KU Leuven

Ethics number: S64932