Reward During Arm Training Improves Impairment and Activity After Stroke: A Randomized Controlled Trial

Abstract

Background: Learning and learning-related neuroplasticity in motor cortex are potential mechanisms mediating recovery of movement abilities after stroke. These mechanisms depend on dopaminergic projections from midbrain that may encode reward information. Likewise, therapist experience confirms the role of feedback/reward for training efficacy after stroke.

Objective: To test the hypothesis that rehabilitative training can be enhanced by adding performance feedback and monetary rewards.

Methods: This multicentric, assessor-blinded, randomized controlled trial used the ArmeoSenso virtual reality rehabilitation system to train 37 first-ever subacute stroke patients in arm-reaching to moving targets. The rewarded group (n = 19) trained with performance feedback (gameplay) and contingent monetary reward. The control group (n = 18) used the same system without monetary reward and with graphically minimized performance feedback. Primary outcome was the change in the two-dimensional reaching space until the end of the intervention period. Secondary clinical assessments were performed at baseline, after 3 weeks of training (15 1-hour sessions), and at 3 month follow-up. Duration and intensity of the interventions as well as concomitant therapy were comparable between groups.

Results: The two-dimensional reaching space showed an overall improvement but no difference between groups. The rewarded group, however, showed significantly greater improvements from baseline in secondary outcomes assessing arm activity (Box and Block Test at post-training: 6.03±2.95, P = .046 and 3 months: 9.66±3.11, P = .003; Wolf Motor Function Test [Score] at 3 months: .63±.22, P = .007) and arm impairment (Fugl-Meyer Upper Extremity at 3 months: 8.22±3.11, P = .011).

Conclusions: Although neutral in its primary outcome, the trial signals a potential facilitating effect of reward on training-mediated improvement of arm paresis.

Trial registration: ClinicalTrials.gov (ID: NCT02257125).

Keywords: feedback; rehabilitation; reward; stroke; upper extremity; virtual reality.

Conflict of interest statement

Competing interests: Prof. Luft reports personal fees from Amgen, personal fees from Moleac, and personal fees from Bayer, outside the submitted work. The other authors have declared that no conflict of interest exists.

Figures

Figure 1.

Participant flow through the study. Consolidated Standards of Reporting Trials (CONSORT) flow chart.

Figure 2.

ArmeoSenso-Reward: Device and interventions. (A) Healthy subject using the ArmeoSenso training system. (B) Arm workspace assessment: Gray cubic voxels arranged in the transverse plane reflecting 10 cm × 10 cm active workspace relative to the patient’s trunk. (C) Rewarded training using the METEORS game: The hand of the virtual arm was used to catch the falling meteors before they crash onto the planet. If caught, the meteor exploded (visual and auditory feedback), and a score appeared (visual feedback). The earlier the meteor was caught, the higher was the produced score. If missed, the planet got damaged (note the impact crater (visual and auditory feedback)). Monetary rewards were given for each completed level. Patients could win up to 1 Swiss Franc (CHF), if they succeeded, but .1 CHF was deducted for every missed meteor. As a new level could be started approximately every 3 minutes, a maximum of 20 CHF (approx. 20 US-Dollars) could be won per training session in case of an uninterrupted winning streak. Summary statistics and monetary rewards were displayed visually after successfully completing a level. (D) Control game. The virtual hand was a green decagon that could be used to touch the pill-shaped, single-colored targets dropping in from the top of the screen, which then disappeared with a delay of 1 s without producing visual or sound effects and without producing a score. For more details, see Widmer et al.

Figure 3.

ArmeoSenso integrated assessments and motivation. Results from the ArmeoSenso Workspace (primary outcome) and Pointing Task Assessment, as well as from the motivation questionnaire. These outcomes were assessed each day during the training period. Data is presented as mean and confidence interval.

Figure 4.

Secondary clinical outcomes. Fugl-Meyer Assessment–Upper Extremity (FMA-UE), Wolf Motor Function Test (WMFT) and Box and Block Test showing significant between-group differences in change from baseline. Data is presented as mean and confidence interval.