. 2020 Nov 24;15(1):3–34. doi: 10.1016/j.optom.2020.08.002

Table 2.

Most relevant results, conclusions and limitations of studies included in the review.

Study	Results	Limitations / recomendations exposed by the authors	Compliance
Hess et al., 2014¹²	• Binocular perception improved in 13 of 14 cases • VA: 1.1 lines^* • Stereo: 0.61 log units^* • The anaglyph and lenticular platforms were equally effective	• Unspecified	• Considerable variability, 9 of 14 (85.7%) achieving close to the expected levels or above expected levels
Mansouri et al., 2014¹³	• VA: 3.4 lines^ • Follow-up (6 months): Improvement VA is maintained^ • Increased significantly as a function of the number of sessions completed (r² = 0.27^*)	• It would be necessary to determine the number sessions to maximize the result and the minimum number of sessions needed to treatment	• Although no compliance data is provided, pre-post results of the 22 participants are presented, so it is understood that all of them completed the training. Mean 146 sessions over a period of 4–6 weeks
Xi et al., 2014¹⁴	• VA: 0.9 lines^* • Stereo: 200.3″ to 81.6″ ^ • Disparity threshold: 776.7″ to 490.4″ ^ • 2 of the 3 subjects reevaluated maintained the effects in VA and stereo at 5 months reevaluation	• Unspecified	• Although no compliance data is provided, pre-post results of the 11 participants are presented, so it is understood that all of them completed the training
Zhang et al., 2014¹⁵	• VA: 1.55 lines^* • Stereo: 53%^* • CS: specially on high spatial frequencies near the cutoff frequency^**	• Unspecified	• All participants completed the first stage. Only a subset of them (63.15%) completed the second stage. 58.33% completed training at an oblique orientation
Birch et al., 2015¹⁶	• VA: 0.2 lines^ns (Sham iPad); 0.9 lines^*** (Binocular iPad) • Stereo acuity: (ns)	• Not a randomized clinical trial, cohort study • Patching at a different time was allowed, which can be a confussion factor	• Binocular iPad game play time of 16 h was reported to be ≥50% by 62% of the participants
Khan et al., 2015¹⁷	• VA: 6 lines^***	• Small sample size • Short follow-up time	• Although no compliance data is provided, pre-post results of the 61 participants are presented, so it is understood that all of them completed the training
Li SL et al., 2015¹⁸	• VA: 2.0 lines^** • Suppression^ns • Stereo^ns	• Small sample size • Anecdotal nature of the data, it is not possible to draw firm conclusions	• All children completed the study
Li J et al., 2015¹⁹	• CS: improved across all spatial frequencies tested for both groups • No significant correlation between the change in CS and changes in VA or suppression in neither group	• Data were combined from two different studies that used different experimental designs • Both groups of participants were not directly matched in terms of age and amblyopic eye VA • CS measured at a relatively high mean luminance	• Although no compliance data is provided, pre-post results of the 30 participants are presented, so it is understood that all of them completed the training
Verdamurthy et al., 2015a²⁰	• VA: 1.4 lines^** • Stereo (1/arcsec): 0.007* • CS: 3.07 cpd^** • IOR increased by a factor of ≈1.6, indicating a reduction in suppression^* • The reduction in suppression was not significantly correlated with the improvement in the visual function • Stereopsis improved in 9 subjects (39%), of 200 in 20 arcseg (a 10 factor)	• The participants were “A” and “S” amblyopic patients, it would be necessary to conduct a randomized clinical trial in a large population of amblyopia to determine the most effective method according to the type of amblyopia	• Compliance data unespeficied
Verdamurthy et al., 2015b²¹	• VA: 1.4 lines^* (game group); 0.7 lines^* (movie group) • VA (difference between groups): 0.7 lines^* • Stereo acuity: Overall significant change^*, but no difference between groups^ns • CS: Overall significant change^, but no difference between groups^ns • No effect for timeXgroup in reading speed^ns • Most improvements were largely retained following a 2-month no-contact period	• Dichoptic VG intervention required that subjects receive extensive training (40 h) in the laboratory, which resulted in a large drop-out rate (38%)	• Drop-out rate 38% vs 28% in VG group and movie group, respectively
Chen et al., 2016²²	• VA (PL group): 1.64 lines^* • CS (PL group): Significantly improved^and it varies with frequency^** • Stereo (PL group): Increases significantly^*	• Test-retest reliability was assessed in a patching group instead of a non-intervention group because of the regulation of clinical practice • There were not well matched in terms of age and interventional periods between both groups • Patching could be considered as a confounding factor (too short-term monocular deprivation)	• Compliance data unespeficied
Dadeya & Dangda, 2016²³	• VA: 2.9 lines^* (group A); 4.2 lines^* (group B) • VA (difference between groups): 1.3 lines^** • Stereo acuity: (Group A): 5 subjects had 200 arcseg; (group B): 7 subjects had 100 arcseg	• Small sample size • Short duration of exposure • Initial dropouts have been excluded from analysis • Lack of objectively recorded compliance at home for both groups	• Compliance to the VG was fully ensured as the examiner was present throughout the session and monitored each child
Herbison et al., 2016²⁴	• VA: ≈ 0.7 lines in all three groups at 6 and 10 weeks^*** • VA (difference between groups): No difference between I-BiT DVD (1 line) and non-I-BiT games (0.3 lines) compared with I-BiT games (0.6 lines) in terms of gain in vision^ns • Stereo: No significant changes in any of the three groups^ns	• Short duration of treatment, and duration and frequency of the sessions • High proportion of patients with strabismic (93%) and residual amblyopia (89%)	• With each of the treatments was excellent (>90%) with the majority of participants playing the game/watching the DVD for 30 min at each session
Holmes et al., 2016²⁵	• VA: 1.05 lines (binocular); 1.35 lines (patching) • VA (difference between groups): 0.3 lines^ns • VA (5 to <7 years without prior amblyopia treatment): 2.5 lines (binocular) and 2.8 lines (patching) • Stereo: did not differ significantly between treatment groups for the overall cohort or for participants with no history of strabismus at baseline	• Compliance data were monitored by the parent's report for the patching group and for the time connected to the game for the game group. In both cases there could be errors in compliance monitoring • Compliance with the use of the red-green glasses required to play the game was not monitorized	• 22% achieved greater than 75% compliance. See “limitations”
Kelly et al., 2016²⁶	• VA: 1.5 lines^* (binocular); 0.7 lines^ (patching) • VA (difference between groups): 0.8 lines^* • No differences between the binocular game vs patching treatments for change in stereo^ns, extent of suppression^ns and depth of suppression^ns	• Small sample size. • Short duration of treatment, it is necessary to analized the effects of treatment in the longer-term • Unassessed baseline factors (ie, BCVA, age) that may be important modifiers of treatment effect	• 85% (23 of 27) of children played at least 75% of prescribed treatment
Verdamurthy et al., 2016²⁷	• VA: just a trend for improved (p = .06) • Stereo: Significant change^* • Suppression: Significantly reduced^*. These effects were retained at follow-up^ • For the stereo-deficient group, no significant differences in noise vergence measured pre-training, post-training and at follow-up (p = .36). • Follow-up 2 month post-training: 5 out of 6 subjects maintained improvements in stereo^*	• Vergence measure is a subjective measure, and not be sufficiently sensitive to detect very small changes in oculomotor control. • Training was carried out in a virtual reality environment, trying to recreate a natural environment.	• Compliance data unespeficied
Barollo et al., 2017²⁸	• VA: 1.8 lines^** (after PL) and 0.9 lines^* (follow up 5−7 months post-training) • Vernier acuity: just a trend for improved (p = .06) • Reduction of crowding^* • CS: Significant change at intermediate spatial frequencies (7 cpd) after PL^and follow-up^	• A direct comparison can not be made with previous results	• Compliance data unespeficied
Bossi et al., 2017²⁹	• VA (overall): 2.7 lines^*** • VA (group 1): 2.6 lines • VA (group 2): 2.7 lines • VA (difference between groups)^ns • Stereo (group 1): Improvement was 165 ± 182″ ^* • Suppressión: no reduction^ns	• Treatment duration varied across children • Small sample size	• Compliance (calculated as the percentage of days when treatment was received) was 68.0 ± 12.2%. On average, 89.4 ± 24.2% of daily dose (54′/day; total: 75 h)
Singh et al., 2017³⁰	• VA: 1 line^* (VG + occlusion); 0.5 lines^* (occlusion) at 1 month • VA: 2.1 line^* (VG + occlusion); 1.7 lines^* (occlusion) at 3 months • VA (difference between groups): 0.5 lines^** (1 month); 0.4 lines^* (3 months) Stereo^ns, CS^ns	• Long-term studies with weaning of occlusion therapy may be required to assess for the recovery of stereo	• Patients who failed to follow-up at 1 month or follow the prescribed therapy or who had poor compliance were excluded. No further data specified
Žiak et al., 2017³¹	• VA: 1.5 lines^ • Stereo: 263.3 ± 135.1–176.7 ± 152.4 arcseg^	• Small sample size • Short follow-up • Absence of a control group • Stereo test used measures values of 400″ as max.	• Although no compliance data is provided, pre-post results of the 17 participants are presented, so it is understood that all of them completed the training
Gambacorta et al., 2018³²	• VA: 1.4 lines (≈38%) (DT group); 0.6 lines (≈15%) (MT group) • VA (difference between groups): 0.8 lines (3.14)^a • VA: 1.1 lines (A) and 0.7 (S) (Cohen's d = 1.33)^* • Stereo: improved ≈17% (DT) and ≈15% (MT) (0.1)^a • Follow-up 6−10 weeks pos-training: VA and stereo improvements were maintained in 57% of the participants at follow-up^***, and improvements did not differ between both groups^ns	• General dropout rate of 28% perhaps due to visits to the laboratory (2–3 times per week). • Small sample size	• 21 of 29 subjects (72%) completed 20 h of training under supervision with number of weeks to completion varying widely between participants (from 3 to 20 weeks) • Drop-out rate was about 31% for the MT group and 23% for DT group
Gao et al., 2018³³	• VA: 0.6 lines^ns (active group); 0.7 lines^ns (placebo group). • VA (difference between groups)^ns • No difference for changes of any secondary outcomes were found between both groups ^ns	• Unable to monitor participants’ attention to the video game at home or whether they wore anaglyphic glasses correctly.	• Compliance with more than 25% (≥10.5 h) was achieved in 64% (active group) and 83% (placebo group) of participants
Jia et al., 2018³⁴	• VA: 1.7 lines^ • CS: an improvement of 278.4% at the trained spatial frequency^ • Stereo: 929.11″ to 80.42″ ^* • Suppression^ns • Dominance duration ratio (amblyopic eye): 9%–15%^*	• Training time too short	• Compliance data unespeficied
Kelly et al., 2018³⁵	• VA:1.4 lines^*** (VG group) • Stereo: Significant change^* (VG group) • Suppression: Extent/depth of suppression were reduced^ (VG group) • Depth of suppression was reduced more in children aged <8 years than in those aged ≥8 years^	• Pooled data from two ongoing studies of binocular treatment and was limited in that there was no control group	• Game group: 87% prescribed treatment time Movie group: 100% prescribed treatment time
Liu & Zhang, 2018³⁶	• VA^ns • Stereo: 146.9″ to 103.1″ ^ • Maximal tolerable noise contrast^* • Improvements persist for 10 months after DT in 54% cases	• Results based on 70% of cases with type A amblyopia • Results may be specific to the training used • Small sample size, and not run a control group	• Although no compliance data is provided, pre-post results of the 13 participants are presented, so it is understood that all of them completed the training
Manh et al., 2018³⁷	• VA: 0.74 lines (binocular group); 1.26 lines (patching group) • VA (difference between groups): 0.52 lines^ns	• Poor treatment adherence	• Binocular group (data from the iPad device for 97% of participants): 13% of participants completed >75% of the prescribed treatment
Mezad-Koursh et al., 2018³⁸	• VA (training group, 12 weeks): 2.6 lines^ • VA (control group, 4 weeks)^ns • VA (difference between groups)^ • VA (training group, follow-up 24 weeks): remained stable^** • Improvement is greater the longer the treatment time	• Nonrandomized study • Small sample size • Most subjects were previously treated and had residual amblyopia • No compare the effectiveness on different types of amblyopia	• All 19 patients in the treatment group completed 8 weeks of treatment; 16 (84%) completed 12 weeks of treatment.
Moret et al., 2018³⁹	• VA (training group): 1.9 lines^* • VA (control group)^ns • VA (difference between groups) • VA (training group, follow-up 6 months): remained stable^ • CS: significantly improved in both groups^*** • CS (difference between groups)^ns	• unespeficied	• Although no compliance data is provided, pre-post results of the 20 participants are presented, so it is understood that all of them completed the training.
Portela et al., 2018⁴⁰	• Stereo: Improvement with RPST was 50% (RPST) and 46.42% (Wirt Circles) and it was statistically different when success was considered a gain of two levels on Wirt Circles and stereoacuity 140″ or less^* • Stereo remained stable after 6 months when measured with RPST	• Computer and software settings: stimulation category be set manually, but it should be automatic according to the patient's evolution ^{Martín et al., 2020} • Study design: Too small sample of subjects with type A amblyopia; age range was restricted to 7−14 years; the stereoacuity range was between 800"-200" ^{Martín et al., 2020}	• 100% compliance was considered if patients finished the training in less than 12 weeks (5 sessions/week). Compliance was excellent, with a median percentage value of 88.36%
Holmes et al., 2019⁴¹	• VA (4 weeks): 0.26 lines (binocular group); 0.34 lines (control group) • VA (difference between groups): 0.08 lines^ns • No difference for changes of any secondary outcomes were found between both groups ^ns • Analyzing possible differential treatment effect by baseline characteristics, no factors were found to be statistically significant, including basal stereo acuity	• Possible biases due to the lack of correct monitoring (not playing for the entire time that the handheld device recorded, not wear the red- green glasses)	• Spectacle wear (across 8 weeks, >75% of time): 90% (binocular) 98% (control) • PAVG (across 8 weeks, >75% of time): parent report: 75%; log file data: 56% • Median total hours of game play was 31 h of the intended 40 h at 8 weeks
Law & Backus 2019⁴²	• 44% of participants (mixed-contrast group) showed improvement in stereodepth individually, none (fixed-contrast group) showed improvement individually^ns • Stereodepth (difference between groups)^	• Training relatively sparse and only 1–2 sessions per week for a total of 10 sessions. It is possible that learning would be greater with additional or more frequent training	• Although no compliance data is provided, first four and last four sessions results of the 19 participants are presented, so it is understood that all of them completed the training
Liu & Zhang, 2019⁴³	• VA: 1.2 lines^** (clinical E-Chart and single-E computerized); 0.8 lines^* • Stereo: 60.2 ± 4.9%^* • CS: mainly at higher spatial frequencies^ • Additional MT did not produce further AV and stereoacuity gains	• Results based on >70% of cases with type A amblyopia • Results may be specific to the training used • Small sample size • No follow-up	• Although no compliance data is provided, pre-post results of the 11 participants are presented, so it is understood that all of them completed the training
Sauvan et al., 2019⁴⁴	• VA: 0.8 lines^ (nonpatched group); 1.9 lines^ (patched group) • VA (difference between groups): 1.1 lines^ns • VA (follow-up 1 month): Improvement VA is maintained only in the patched group^** • CS, stereo, interocular suppression^ns	• Non-homogeneous groups • Non-randomization • Short period of training • Better measure of stereopsis • More sensitive test of binocular balance • Extend the periods of occlusion to see if its benefits for DT can be enhanced	• Although no compliance data is provided, pre-post results of the 17 participants are presented, so it is understood that all of them completed the training
Birch et al., 2020⁴⁵	• VA:1.5 lines (game group); 0.7 lines (patching group) • VA (difference between groups):0.8 lines^** • 35% of children of binocular game had recovered normal VA. Only 8% of the children in the patching group recovered normal VA for age • Only in the game group, baseline VA and ocular alignment were associated with response to treatment (p < .001). Age, etiology, prior treatment and baseline stereo were not associated with response to treatment • Dose-response: Only in the game group, moderate linear relationship between hours of game with VA improvement. 5 h (50% adherence) improved 0.1 logMAR; 10 h (100%), 0.18 logMAR; 15 h (150%) 0.26 logMAR	• Single-site, small-cohort randomized clinical trial • The inclusion/exclusioncriteria may limit generalization to other groups of amblyopic children • VA testing was not masked • Results may be dependent on the criteria chosen to dichotomize each variable • The brief duration may limit improvement of VA • Possible biases due to the lack of correct monitoring ( number of hours of patching objectively, not wear the red- green glasses)	• Parents overestimated time spent playing the game by 13% • Game group completed 10.3 ± 3.0 h (103% prescribed treatment time • Patching group completed 27.7 ± 2.6 h (99% prescribed treatment time)
Gu et al., 2020⁴⁶	• VA^*, CS^, stereopsis^ and interocular diference^** improved through behavioral measurements and SSVEP in MT group • (Due to the length of the results, it is recommended to see the "results" section in the original article)	• Control group had only five subjects • Training effects may be due to the influences of both training and patching • Effects of patching were not entirely ruled out in this study. Further investigations with more subject and only training (no patching) are necessary	• Although no compliance data is provided, pre-post results of the 32 (out of 46) participants are presented, so it is understood that not everyone completed the training

Note: A: anisometropic amblyopia; BCVA: Best corrected visual acuity; BiT: binocular treatment; cpd: cycles per degree; CS: contrast sensitivity; DT: dichoptic therapy; h: hour/ hours; IOR: interocular ratio; min: minutes; MT: monocular training; ns: no significance; PL: perceptual learning; S: strabismic amblyopia; SSVEPs: Steady-state visually evoked potentials; VA: visual acuity; VG: video game.

Cohen's “d” effect size has been calculated when the p-value is not shown in the study, but means and SD are shown (d≥0.80 is considered a large effect size)."Both" refers applied equally to both groups. The average improvement in VA lines is always referred from the baseline.

p < 0.05;

^**

p < .01;

^***

p < .001.