An economic evaluation alongside a randomised controlled trial of Scanning Eye trAining as a Rehabilitation Choice compared to sham training for Hemianopia after stroke (the SEARCH trial)

ObjectiveTo explore the cost-effectiveness of visual scanning training for hemianopia compared to sham training.DesignRandomised controlled trial.SettingAcute stroke unit and outpatient eye clinics.ParticipantsAdult stroke survivors with homonymous hemianopia.InterventionVisual scanning training.ComparatorSham eye pursuit movements.Main measuresCost-utility and cost-effectiveness analyses were conducted from UK National Health Service (NHS) and wider societal perspectives. The primary analysis was a cost-utility analysis using EQ-5D-5L utility values. Cost-effectiveness analyses using changes in National Eye Institute-Visual Function Questionnaire-25 item composite and Brain Injury-associated Visual Impairment-Impact Questionnaire scores formed the secondary analyses.ResultsBoth intervention and sham training packages cost £9.58 (printing costs, clinician time to explain self-administration). Mean total health resource use costs at 26-weeks were higher in the intervention (£1406) than in the sham group (£1339). At 26-weeks, the intervention group gained 0.001 (95%CI:-0.018, 0.021) Quality-Adjusted Life Years (QALY) while the sham group evidenced a QALY loss of -0.002 (95%CI:-0.017, 0.012); not statistically significant (p = 0.71). The Incremental Cost-Effectiveness Ratio of visual scanning training compared to sham was £19,790/QALY, below the National Institute for Health and Care Excellence (NICE) threshold of £20,000-£30,000/QALY. Sensitivity analysis (5000 replications) suggested potential cost-effectiveness of visual scanning training, which had a 67% probability of being cost-effective at £20,000/QALY; this was not significant.ConclusionsThis is the first economic evaluation comparing visual scanning training to sham training for treating post-stroke hemianopia. While both interventions had similar costs, the costs were slightly under NICE thresholds. The intervention evidenced marginal but non-significant QALY gain over sham.