Accelerometry as an objective measure of upper-extremity activity

Most studies evaluating the effectiveness of treatments targeting shoulder pathologies use subjective outcome measures such as self-administered questionnaires. To date, there are no validated tools that objectively measure shoulder-specific functional activity. The purpose of this study was to validate wearable accelerometers as an objective proxy for shoulder activity. Ten healthy volunteers wore accelerometers placed at both wrists, the dominant upper arm and the chest while performing standardised shoulder and non-shoulder activities. Recorded tridimensional acceleration was computed into activity counts for epochs of 10 s. Receiver operating characteristics (ROC) curves were built to determine the optimal configuration to classify shoulder-type activities. For single accelerometer placement, the area under the ROC curve (AUC) was optimal for the 10-s epoch (AUC = 0.779) using the wrist placement, with a sensitivity of 94.1% and specificity of 67.5%. The combined upper arm and chest placement had an AUC of 0.985 (94.8% sensitivity, 94.8% specificity). Dual-accelerometer placement (upper arm and chest) is the optimal configuration to classify shoulder activity. However, a sole wrist-based accelerometer can be used as an objective proxy for shoulder activity in long-term unsupervised monitoring with excellent sensitivity and acceptable specificity.