An eye-tracking-based dichoptic home treatment for amblyopia: a multicenter randomized clinical trial

To compare the effectiveness and safety of a novel binocular eye-tracking-based-home-treatment (CureSight) to patching.Prospective, multi-center, randomized, masked, controlled non-inferiority pivotal trial.A total of 103 children aged 4 to ≤9 years with anisometropic, small-angle strabismic, or mixed-mechanism amblyopia were enrolled at six clinical sites, randomized 1:1 to either CureSight treatment or patching.Binocular treatment group used the CureSight for 90 min/day, 5 days/week for 16 weeks (120 hours). The treatment combined anaglyph glasses and an eye tracker to induce dominant eye real-time blur around the fovea in dichoptic streamed video content. Patching group received 2-hour patching 7 days/week (224 hours). The pre-specified non-inferiority margin was 1 logMAR line.The primary outcome was the improvement in the amblyopic eye distance visual acuity (VA) from baseline at 16 weeks, modeled with a repeated measures ANCOVA. Secondary outcomes included stereoacuity, binocular VA, and treatment adherence rates, analyzed by a one-sample Wilcoxon-test within each group and a two-sample Wilcoxon-test comparing groups. Safety outcomes included the frequency and severity of study-related adverse events.Binocular treatment group VA improvement at 16 weeks was found to be not inferior to patching group improvement (0.28 logMAR (±0.13, p0.0001) and 0.23 logMAR (±0.14, p0.0001) in binocular treatment group and patching group (90% CI of difference [-0.008, 0.076]), respectively)), since the lower confidence bound of -0.008 falls within the non-inferiority margin of -0.1 logMAR. Stereoacuity improvement of 0.40 log-arcseconds (p0.0001) and improved binocular VA (0.13 logMAR, p0.0001) was observed in binocular treatment group, with similar improvements in patching group in stereoacuity (0.40 log-arcseconds, p0.0001) and binocular VA (0.09 logMAR, p0.0001), with no significant difference between improvements in the two groups for both stereoacuity (difference 0 95% CI[-0.27, -0.27]; p=0.76) and binocular VA (difference 0.041 95% CI[-0.002, 0.085]; p=0.07). A significantly higher adherence was observed in treatment compared with patching group (91% vs. 83%, difference 8% 95% CI[-4.0%-21%]; p=0.011). No serious adverse events were found.Binocular treatment was well tolerated, noninferior to patching in amblyopic children aged 4 to ≤9 years. High adherence may provide an alternative treatment option for amblyopia.