A randomized controlled trial of Scanning Eye trAining as a Rehabilitation Choice for Hemianopia after stroke (SEARCH)

Background: Hemianopia is common after stroke. We aimed to evaluate clinical effectiveness of visual scanning training (VST) versus sham training, for homonymous hemianopia. Design: Randomized controlled, parallel, double-blind, two-arm trial. Methods: Prospective, multicentre randomized controlled trial (RCT) with 34 UK stroke units. Participants: Adult stroke survivors with confirmed stable homonymous hemianopia. Inclusion criteria: Clinically diagnosed stroke, 18+ years, stable hemianopia, >4 weeks and <26 weeks post-stroke onset, able to engage in training, informed/proxy consent. Interventions: Arm A (VST) or arm B (sham training) for minimum 30 minutes, 7 days per week over 6 weeks. Follow-up to 26 weeks. Objective: Evaluate clinical effectiveness of VST versus sham training, for homonymous hemianopia. Outcomes: Primary outcome measurement was change in the National Eye Institute Visual Function Questionnaire 25 (NEI VFQ-25) score from baseline to 26 weeks. Secondary outcome measurements were change in the Nottingham Extended Activities of Daily Living (NEADL), EuroQoL (EQ-5D-5L), Brain Injury-Related Visual Impairment—Impact Questionnaire (BIVI-IQ), visual field measurement (Esterman program), visual scanning performance, and adverse events from baseline to 26 weeks. Randomization: Web-based randomization system stratified by partial/complete hemianopia. Blinding: Participants and primary outcome assessor blinded to treatment allocation. Results: In total, 161 participants were randomized; 80 to the VST group and 81 to the sham group. One participant was randomized in error and two withdrew consent to use data so were not included. Of 158 participants, 78 were in the VST group and 80 in the sham group. No participants were unblinded. All participants began their training allocation. During the first 6 weeks of training, 56 (72%) and 58 (73%) undertook training every day or most days in the VST and sham groups, respectively. There were 37 withdrawals from the trial: 18 in the VST group and 19 in the sham group. Both groups were comparable in terms of baseline characteristics. Primary analysis of covariance (ANCOVA) analysis was carried out on 104 participants with VFQ-25 data at both baseline and 26 weeks; sensitivity analysis was undertaken for 120 participants. Estimated mean difference at 26 weeks, adjusting for baseline score and hemianopia type was −4.04 (95% confidence interval (CI): –9.45 to 1.36; p = 0.141) for primary analysis and −2.33 (95% CI: −7.42 to 2.75; p = 0.365) for sensitivity analysis. There were no significant differences between groups for primary and secondary outcome measure comparisons from baseline to 26 weeks. Adverse events, reported for 20 participants, included eye strain, headache, and blurred vision. Conclusion: Both groups improved in all primary and secondary outcomes. There were no differences between the groups for any outcome measure. However, it was possible that there was a placebo effect from additional information resources and clinician input during the first 6 weeks of treatment and for the sham training providing a treatment effect. These aspects warrant further research.