The topic of amblyopia is overflowing with promises to "retrain the brain". This lesson is the only reason Module 4 stands apart from "how to patch" at all. Each line goes into one basket. Mixing baskets in a prescription is worse than not treating: the family spends the time of the critical window on a method without effect.
Basket 1. Proven in randomised trials (and listed in the PPP as an appropriate option)
- Optical correction as treatment, not as preparation: ATS5 and related studies, PPP highlighted findings.
- Occlusion at doses that withstood comparison: 2 h for moderate, 6 h no worse than full-time for severe (ATS2B, ATS2A), in the 3–<7-year population.
- Atropine comparable with occlusion in moderate preschool amblyopia (ATS1; Cochrane, Li 2019). Weekend comparable with daily in moderate amblyopia (ATS4).
- Bangerter filters — a reasonable option in moderate amblyopia (ATS13), with the caveat about the formally unmet non-inferiority criterion.
- Treatment of older children: ATS3 — offer an attempt; the realistic response is smaller, and at 13–17 years adding occlusion/atropine to optics did not win on the proportion of responders.
- Elimination of deprivation in time (cataract/ptosis surgery as a condition for treating amblyopia, not as a replacement for occlusion).
- A specific desk-based dichoptic therapeutic versus 2 h of occlusion — not "any binocular game". Wygnanski-Jaffe et al., Ophthalmology 2023;130:274–285: 103 children aged 4–<9 years with anisometropic, small-angle strabismic or combined amblyopia; a desk-based computer, red-blue anaglyph glasses and an eye tracker versus a prescribed 2 hours of patching, 16 weeks. Gain 2.8 versus 2.3 lines (0.28 ± 0.13 vs 0.23 ± 0.14 logMAR); 90% CI of the difference −0.008 to 0.076, non-inferiority criterion met; adherence 91% versus 83%. The February 2024 update of the Amblyopia PPP adds this reference on P153 and, on P156, the panel's wording: therapeutic non-inferior to patching 2 hours per day, GRADE I+, Good, Discretionary. This is proof for this device in this population against this comparator — not an indulgence for the whole class of "neuro-game instead of patching".
The highlighted findings of the original 2022/2023 journal text already listed digital therapeutics among appropriate options after a search up to May 2022. The February 2024 revision did not strike out this class and did not declare any iPad a replacement for the patch: it added one RCT with a "2 hours of occlusion" comparator. Luminopia was still compared with glasses, not with the patch (basket 2 as regards replacing patching). ATS18 still did not prove non-inferiority of a specific iPad game against the patch (not basket 1 either).
Basket 2. Promising and under investigation
This basket holds methods with biological plausibility and incomplete or heterogeneous clinical evidence. They can be discussed, included in a study, sometimes prescribed within the regulator's indications — they must not be sold as a replacement for basket 1 "because they are more modern".
Dichoptic / binocular digital methods are not one box. The general idea: reduce the contrast of the fellow eye and distribute parts of the image between the eyes so that the scene can only be assembled jointly, weakening suppression. Beyond that, the comparator and the device decide the basket. The desk-based system from the 2024 update (above, basket 1, I+, Good, Discretionary, vs 2 h of patching) ≠ the ATS18 iPad game ≠ Luminopia vs glasses. From one failed NI trial or from one RCT against glasses you cannot derive a general "binocular vs patch — basket 2 only".
- PEDIG ATS18, Holmes et al., JAMA Ophthalmol 2016: 385 children aged 5–<13 years, a binocular iPad game 1 h/day versus 2 h of patching, 16 weeks. Gain 1.05 versus 1.35 lines; the upper bound of the one-sided 95% CI, 0.53, exceeded the non-inferiority limit of 0.5. Only 22% in the game group completed >75% of what was prescribed (median 46%). Primary analysis inconclusive; post hoc — this particular iPad game is not as good as 2 hours of patching. In adolescents aged 13–<17 (Manh et al., Am J Ophthalmol 2018, n=100) the game was not better than the patch and possibly worse; 13% completed >75% of the dose.
- Cochrane, Tailor 2022 (CD011347.pub3), binocular versus standard occlusion/fogging in children aged 3–8 years: at the time of the review one RCT, comparable gain at 16 weeks (about −0.21 versus −0.24 logMAR), moderate certainty, no data at 52 weeks, limited sample. The authors' conclusion: firm conclusions about safety and durability cannot yet be drawn; further studies with recognised methods of measuring acuity and stereopsis are needed.
- Luminopia One: Xiao et al., Ophthalmology 2022; 105 children aged 4–7 years, 21 US centres, dichoptic viewing of selected video 1 h/day 6 days a week plus glasses versus glasses alone. At 12 weeks +1.8 lines (95% CI 1.4–2.3) versus +0.8 (0.4–1.3), difference 1.0 line; the study was stopped early per protocol. There were no serious adverse events. FDA De Novo DEN210005, 20 October 2021: indication — improvement of acuity in patients aged 4–7 years with amblyopia associated with anisometropia and/or mild strabismus, prescribed by a specialist, for home use. This is an RCT against glasses, not against the patch. A direct comparison with occlusion (PEDIG ATS24) as a published primary result was not found. So Luminopia is not "proven better than patching". It is "proven better than glasses alone in its age and diagnostic niche, approved by the US regulator, comparison with the gold standard still pending". It must not be confused with Wygnanski-Jaffe 2023: a different device, a different comparator, a different age range (4–7 vs 4–<9), and only the second study was explicitly placed by the panel in 2024 alongside 2 hours of occlusion.
- Small laboratory RCTs (for example, Kelly et al., 28 children, 2 weeks, the game beat the patch over a short period) remain hypothesis-generating. They cannot be put into basket 1 against ATS18.
Simultaneous start of glasses + patch versus sequential. EuPatch 2024 — in favour of earlier occlusion by its own definition of success; ATS22 — awaiting publication. Basket 2, not a clinic dogma.
Refractive surgery in selected children, which the PPP mentions through references: not a first-line standard.
Basket 3. Not supported — and repeating the promises is worse than saying "no"
- Levodopa as augmentation of occlusion in residual amblyopia. PEDIG ATS17 (Repka et al., Ophthalmology 2015): 139 children aged 7–12 years, 20/50–20/400 after occlusion, levodopa 0.76 mg/kg + carbidopa 0.17 mg/kg 3 times a day for 16 weeks plus 2 h of patching versus placebo plus patching. Gain 5.2 versus 3.8 ETDRS letters; adjusted difference +1.4 letters, two-sided 95% CI −0.4 to +3.3, one-sided P=0.06. No clinically or statistically meaningful effect. There were no serious adverse events. This is not a case of "a higher dose is needed": the upper bound of the CI is 3.3 letters. Earlier small and uncontrolled series promising to "reopen the critical period" do not make up basket 1.
- An "intensive final push" of combined occlusion and atropine for mild residual amblyopia that has stopped improving: a small PEDIG RCT (children aged 3–10 years) showed no advantage of 6 h of patching plus daily atropine over gradual weaning. It has not been proven that "adding everything at once" squeezes out the remainder.
- Systemic nootropics, citicoline, dietary supplements "for the visual cortex" — within the foundations of this course (PPP, PEDIG ATS, Cochrane on occlusion/atropine/binocular treatment) there is no confirmed effect as a standard. An honest gap, not a hint of "try it; if it doesn't help, at least it does no harm": the window is finite.
- Device "stimulation", biofeedback courses, clinics' promises of "we treat all forms in 10 sessions without patching" — are not PEDIG RCTs and are not included in the Cochrane reviews on which the course rests. Prescribing them instead of optics and proven penalisation replaces treatment with spectacle.
- Acupuncture, "trainer glasses" without a dichoptic protocol, colour therapy — not a basis for management.
- Transferring Levi 2005 adult perceptual learning into a mandatory protocol for childhood amblyopia is extrapolation.
How to talk to a family that has brought a brochure. Do not humiliate them. Show the basket, the population and the comparator. "This game was tested against glasses in children aged 4–7 with anisometropia or mild strabismus" is not the same as "it replaces patching for any schoolchild". "A desk-based system with an eye tracker was not inferior to 2 hours of patching in one RCT in children aged 4–<9, and the panel gave it Discretionary" is not the same as "any binocular game = basket 1". "Levodopa added nothing to patching in a large RCT" is not the same as "science hasn't got there yet".