Proprioceptive deficits affect around 50% of stroke survivors. Most commonly used proprioception assessments have poor psychometric properties and are not adapted for people with multiple impairments. To address these limitations, we developed an assessment protocol that enables the quantification of elbow position sense without using the opposite arm, involving active movement of the evaluated limb or relying on working memory. The objectives of this study were to quantify elbow position sense of stroke survivors and to compare it with a control group.

Elbow position sense of the affected arm was quantified in subacute stroke patients and compared to age-matched healthy persons. Elbow position sense was evaluated using a KINARM exoskeleton and a virtual reality display. The exoskeleton passively moved the participant's arm from an initial to a target position (no visual cues of arm position). A virtual arm representation was then projected on a screen placed over the participant's arm. The participant had to indicate verbally its relative position (more flexed or more extended) compared to the virtual representation. A 75% detection threshold was extracted from a sigmoid curve fit representing the relationship between the angular difference and the percentage of successful trials. Independent T-tests were used to compare the results of both groups.

Eleven people with a stroke (67±9 yrs old; 6 males; 72±30 days post-lesion) and 19 age-matched controls (63±6 yrs old; 8 males) were recruited. A significant difference in the mean elbow position sense detection thresholds between the stroke group (12.8±5.2 degrees) and the control group (7.2±3.5 degrees) was noted (P=0.001).

This robotic assessment enables the quantification of elbow position sense in patients with stroke. Difference in detection thresholds observed between both groups demonstrated that stroke survivors might have impaired proprioception. More participants are required to confirm these results.