ObjectiveTo evaluate the added value of sensor-based tests over conventional tests in explaining the variance in independence in activities of daily living (ADL) and independent walking in patients during inpatient stroke rehabilitation.DesignCross-sectional designSettingInpatient stroke rehabilitationParticipantsData were collected from 115 patients with stroke admitted to inpatient rehabilitation.InterventionNo intervention has been applied.Main measuresConventional measures and sensor-based tests in which postural sway and gait variables were measured using inertial measurement units. Outcome measures were ADL independence (determined by Barthel Index [BI]) and independent walking (determined by Functional Ambulation Categories [FAC]).ResultsWith univariable linear regression analyses showed that the Berg Balance Scale (BBS) explained most variance in ADL independence (R² = .527) and independent walking (R² = .727). In hierarchical multivariable regression analyses, symmetry during walking without a walking aid contributed most (ΔR²=18.6%) in explaining variance in ADL independency, resulting in a model explaining 15.7% (p = .029) of the variance in the BI. Tempo during walking with a walking aid contributed most (ΔR² = 20.1%) in explaining variance in independent walking, resulting in a model explaining 23.3% (p = .002) of the variance in the FAC. Adding sensor-based variables to models with the BBS did not significantly improve variance explanation.ConclusionsThe added value of variables measured with an inertial measurement unit in explaining ADL independence and walking ability after stroke is limited. These findings contribute to understanding the use of inertial measurement units in stroke rehabilitation, but caution is needed when applying them to predict physical recovery.
Keywords: ADL; IMU; Stroke; association; prediction; sensor-based; walking.