Aims To compare amblyopic-eye visual acuity (VA) and binocularity improvement of a binocular game with part-time patching in the treatment of Chinese children with anisometropic amblyopia.
Methods 103 Chinese children aged 3–13 years with anisometropic amblyopia were recruited in a randomised clinical trial. Eligible participants were randomly assigned to the binocular, patching and combined groups. Primary outcome was amblyopic-eye VA improvement at 3 months. Secondary outcomes included reduction of suppression and change of stereoacuity.
Results Of 85 completed participants, 44 (52%) were women and mean (SD) age was 5.99 (2.33) years. At 3 months, mean (95% CI) amblyopic-eye VA improved 0.18 (0.10–0.26), 0.28 (0.19–0.36) and 0.30 (0.21–0.39) logarithm of the minimum angle of resolution in the binocular, patching and combined groups, respectively. After adjusting for baseline VA, the difference was statistically significant (F=6.29, p=0.003), favouring as follows: the combined group, the patching group and the binocular group. After treatment, Titmus (x2 binocular=9.75, p=0.007; x2 combined=9.35, p=0.009) and dynamic stereoacuity (x2 binocular=12.56, p=0.01; x2 combined=12.66, p=0.01) improved only in the binocular and combined groups. Among groups, only Titmus improvement differed significantly (F=49.55, p<0.001). Changes of other types of stereoacuity and interocular suppression were similar.
Conclusions The binocular game used in this study could improve amblyopic-eye VA and binocularity in Chinese children with anisometropic amblyopia, but it was less effective than patching in amblyopic-eye VA improvement and showed no superiority in binocularity over patching. It remains unclear whether the low treatment response of this binocular game was due to limitations of the study or its low treatment effect.
- treatment other
- optics and refraction
- clinical trial
Statistics from Altmetric.com
As amblyopia arises from binocular discordance, binocular treatments allowing children to overcome interocular suppression and experience binocular vision while playing the game are thought to likely yield better vision outcomes, not only visual acuity (VA) but also binocularity.1–3 Since 2006, a number of binocular treatments have been developed and initial studies demonstrated promising results.4–8 However, randomised clinical trials showed inconsistent results in amblyopic-eye VA improvement.9–12 The binocular game was shown more efficacious than patching in Kelly’s study,9 and combining the binocular game with patching was demonstrated more VA improvement than patching alone in Rajavi’s study,12 while the binocular game used in Pediatric Eye Disease Investigator Group (PEDIG)’s study was not found to be better than patching.10 11 Superior stereoacuity was not achieved with these binocular treatments.9–11
Of these trials, three compared anti-suppression therapy with patching and one compared Interactive Binocular Treatment (I-BiT) plus patching with patching alone. In anti-suppression therapy, images were presented dichoptically, with normal contrast images presented to the amblyopic eye and low-contrast images to the fellow eye.9–11 As for I-BiT, different parts of a two-dimensional image were shown to either eye, with moving targets shown to the amblyopic eye and stable targets to the fellow eye.12 Due to differences in the binocular game, type of amblyopia, and other factors in different trials, it is hard to compare these studies with each other. Whether binocular treatment is comparable to part-time patching in the treatment of amblyopia remains unclear. Whether binocular treatment could be acted as a complementary method of part-time patching is also undecided.
In this study, we tried a new binocular game based on push–pull method.13 During the training, an attention cue (flicker) presented to the amblyopic eye preceded the binocular competitive stimulation. Binocular competitive stimulation was then implemented in a virtual reality background with high-energy elements (contrast or signal to noise ratio) presented to the amblyopic eye and low-energy elements to the fellow eye. Reduced suppression and improved stereoacuity were reported in the initial study.2
The primary purpose of the present randomised clinical trial is to compare this binocular game with part-time patching in amblyopic-eye VA improvement in Chinese children with anisometropic amblyopia. Whether combining this binocular game with patching is better than binocular game alone or patching alone is also investigated. The secondary purpose of the trial is to investigate whether this binocular game is superior over part-time patching in binocularity improvement.
This prospective study was conducted at Eye & ENT Hospital of Fudan University and approved by the Institutional Review Board of Eye & ENT Hospital of Fudan University. Written informed consent was obtained from parent or legal guardian, and either written or verbal assent was obtained from each participant as required. The study was listed on www.chictr.org.cn under identifier ChiCTR1800015907.
Anisometropic amblyopia is the most common form of amblyopia in China. To exclude the confounding of type of amblyopia, we included only anisometropic amblyopia in this study. Children aged 3–13 years with unilateral anisometropic amblyopia were recruited. The diagnosis of amblyopia was according to amblyopia preferred practice pattern.14 The best-corrected VA of the amblyopic eye was worse than the normal value of corresponding age (0.30 logarithm of the minimum angle of resolution (logMAR) for 3-year to 5-year old, 0.20 logMAR for more than 6-year old) and with an interocular difference of at least two lines.15 The best-corrected VA of the fellow eye was within the normal value of corresponding age. Anisometropia was defined as ≥1.50 D difference between eyes in spherical power or ≥1.00 D difference between eyes in cylindrical power in any meridian. Criteria for refractive correction were based on cycloplegic refraction within 3 months: hypermetropia ≥3.00 D, astigmatism ≥1.50 D and myopia ≥0.5 D. Participants were required to demonstrate stable VA (five or less letters change over 4 weeks) after appropriate refractive correction within the inclusion range before randomisation.
Children receiving any amblyopic treatment other than optical treatment in the past 4 weeks were excluded. Except amblyopia and refractive error, children with other ocular diseases or ocular misalignment more than 10 prism diopters using a prism and alternate cover test (PACT) at near were excluded. Children who were born at less than 32 weeks of gestational age or had systemic diseases, congenital infections or malformations, or developmental delay were excluded. Children who could not understand the examination or play the game were excluded.
Randomisation and treatment
Randomisation was performed using a random number table. The numbers were assigned sequentially as each subject entered the study. After conforming eligibility and obtaining written informed consent, children were assigned to the appropriate treatment.
Treatment was prescribed for 3 months and performed at home. The binocular group was prescribed 40 min of a binocular game a day and was divided into two training sessions. Each session had two items and each item lasted for 10 min with 10 min interval. The patching group was prescribed 2–6 hours of patching of the fellow eye a day depending on the severity of amblyopia. Mild (<0.3 logMAR), moderate (0.3~0.7 logMAR) and severe (>0.7 logMAR) amblyopia were prescribed 2, 4 and 6 hours of patching a day, respectively. The total hours could be divided into several sessions, but every session was at least 2 hours. The combined group was prescribed both the binocular game and patching of the fellow eye.
The binocular game was played on a computer at the participant’s habitual reading distance while wearing polarised anaglyphic glasses (over the appropriate refractive correction spectacles, if applicable). Using the improved push–pull model personalisation training system provided by National Engineering Research Centre for Healthcare Devices (Guangzhou, China), the detailed parameters of all participants’ eyes imbalance (eg, contrast ratio, signal to noise ratio) were obtained through a series of visual sensory examinations of binocular integration abilities and binocular stereoscopic vision energy maps, and a short period of about 5–10 min plasticity check. The suppression was tested using dichoptic motion coherence test.5 9 The difference between the eyes ≤15% was regarded as normal contrast ratio. The signal to noise ratio was divided into eight levels and level eight was regarded as normal. As for fine stereoacuity, an “E” mark (3°×3°) composed of random dots was in the centre of the distribution map (54 cd/m2) with a grey background (44 cd/m2, size 5°×5°). The parallax was 400, 300, 200 and 100 seconds of arc, respectively. As for dynamic stereoacuity, an “E” image (6°×6°) composed of random dots was in the centre of the distribution map (36 cd/m2) with a grey background (44 cd/m2, size 8°×8°). The parallax changed from zero to 800 seconds of arc with a low speed (50 seconds of arc/millisecond) or a high speed (600 seconds of arc/millisecond). The period was 1.2 seconds. The dynamic random dot density remained the same. Participants wore polarised glasses to judge the opening direction of “E”. To examine fine stereoacuity, parallax was tested from high to low and every parallax was tested three times. If all three judgements were correct, next parallax was tested, or the current parallax was the participant’s fine stereoacuity. 100 seconds of arc was regarded as normal. Dynamic stereoacuity was tested from high speed to low speed and every speed was tested three times. Passing all three low speed tests was regarded as normal.
Then participants in the binocular and combined groups carried out a personalised push–pull model training with an attention cue (flicker) presented to the amblyopic eye followed by high-energy elements (contrast or signal to noise ratio) presented to the amblyopic eye, low-energy elements presented to the fellow eye and binocular integrated elements presented to the both eyes on the basis of the balance between the eyes. Low-energy elements were reduced by algorithm of contrast attenuation or Gaussian low-pass filtering. All participants were asked to align cross/ring before the start of each game session so that the image components of the game could be aligned, to compensate for small angles of ocular misalignment (no more than 10 prism diopters with PACT at near), thereby allowing binocular game play. The points were scored in the game by determining whether the bird was in front of the box (online supplementary eFigure 1).
Study visits and testing procedures
After randomisation, follow-up visits were planned at 1 month and 3 months. At baseline and each follow-up visit, VA in each eye was measured without cycloplegia in optimal refractive correction and converted to the logMAR scale. Tumbling E Early Treatment Diabetic Retinopathy Study chart was used throughout the study for each participant. Ocular alignment was tested with PACT. Stereoacuity was examined using Titmus stereoacuity test (Stereo Optical Co, Chicago, Illinois, USA). VA and Titmus were conducted by study-certified examiners masked to the treatment. Additionally, suppression (contrast ratio and signal to noise ratio) and stereoacuity (fine stereoacuity and dynamic stereoacuity) were examined using the improved push–pull model personalisation training system.
Our primary hypothesis was that the binocular treatment was comparable to the conventional patching in amblyopic-eye VA improvement, and the combined treatment was better than either of them. Based on previous studies9 10 12 and our pilot study, the binocular treatment and the conventional patching resulted in a mean (SD) amblyopic-eye VA improvement of 0.10 (0.10) logMAR, and the combined treatment resulted in a mean (SD) improvement of 0.17 (0.09) logMAR. The required sample size of 103 was computed to have 80% power with a type I error of 0.05 and no more than 10% loss to follow-up.
All analyses included only participants completing the 3-month follow-up with an intent-to-treat approach. Titmus and fine stereoacuity were converted to log seconds of arc for analyses, and nil stereoacuity was arbitrarily assigned a value of 4 log seconds of arc. The primary outcome was amblyopic-eye VA improvement from baseline to 3 months. An analysis of covariance model, adjusted for baseline VA, was conducted to determine whether amblyopic-eye VA improvement differed among groups. Kruskal-Wallis test was conducted to determine whether amblyopic-eye VA of every group improved with time. Linear regression model was further used to analyse the associated factors to amblyopic-eye VA improvement. The differences among different time points per group and group differences in contrast ratio and dynamic stereoacuity were analysed using Pearson χ2 test. As for binocular difference of signal to noise ratio, amblyopic-eye signal to noise ratio, Titmus and fine stereoacuity, Kruskal-Wallis test was conducted to determine the changes per group from baseline to 1-month visit and 3-month visit, and analysis of variance was performed to determine the differences among groups. Analyses were performed using SAS analysis of V.9.4 (SAS, Cary, North Carolina, USA). All p values were two-sided, and significance was set at p<0.05.
Between 5 January 2016 and 26 July 2017, a total of 103 participants were enrolled in this study and randomly assigned to the binocular group (n=36), the patching group (n=38) or the combined group (n=29). Baseline characteristics were similar in these three groups (table 1).
The 3-month visit was completed by 33 participants (92%) in the binocular group, 30 participants (79%) in the patching group and 22 participants (76%) in the combined group (figure 1). Baseline characteristics of the completed participants were similar (online supplementary eTable 1). Masking of VA and Titmus testers was maintained at 100% of the visits for all these three groups.
Two participants were excluded post-randomisation. One was subsequently found to be ineligible for fundus disease and the other was unable to understand the examination. Of the 16 participants who withdrew before 3 months, nine participants cited reasons unrelated to the treatment and the others did not give a reason. No participants in the binocular and patching groups received non-protocol alternative treatment.
The patching adherence relied on parental report. Three (10%) participants in the patching group and two (9%) participants in the combined group were reported completing less than the prescribed treatment. The adherence of the binocular game was recorded electronically, but only the number of login was recorded, not the time how long the game was played. Five (15%) participants in the binocular group and six (27%) participants in the combined group were recorded completing <75% of login. Further analysis showed that the more the number of login, the greater the amblyopic-eye VA improvement in the binocular group at 3 months (r=0.53, p<0.001).
After treatment, amblyopic-eye VA improved significantly in all three groups (binocular group: x2=7.45, p=0.02; patching group: x2=29.19, p<0.001; combined group: x2=19.29, p<0.001) (table 2）(figure 2), implying that all these treatments were effective for anisometropic amblyopia in children aged 3–13 years. Moreover, amblyopic-eye VA improvement was associated with baseline amblyopic-eye VA (figure 3). The more severe the amblyopia, the greater the amblyopic-eye VA improvement. Linear regression model confirmed that baseline amblyopic-eye VA (t1m=9.09, p<0.001; t3m=10.49, p<0.001) and the treatment (t1m=3.43, p=0.001; t3m=2.29, p=0.02) had a statistically significant association with amblyopic-eye VA improvement.
To compare the effectiveness of different treatments, mean amblyopic-eye VA improvement in different groups was calculated at 3 months (table 2). After adjusting for baseline VA, the difference reached statistically significant (F=6.29, p=0.003), favouring as follows: the combined group, the patching group and the binocular group. In a post hoc analysis, the patching and combined groups had greater amblyopic-eye VA improvement than the binocular group (tpatching-binocular=3.41, p=0.001; tcombined-binocular=2.36, p=0.02), but the difference between the patching group and combined group was not statistically significant (t=0.74, p=0.46).
At 3-month visit, amblyopic-eye VA improved by at least 0.2 logMAR from baseline for 13 (39%), 18 (60%) and 14 (64%) participants in the binocular, patching and combined groups, respectively. The rate of amblyopic-eye VA improvement did not differ significantly among groups (x2=4.04, p=0.13). Two participants in the binocular group and one participant in the combined group were found 0.10 logMAR worse from baseline.
In this study, contrast ratio and signal to noise ratio were used to demonstrate the interocular suppression (table 2). After treatment, more participants had normal contrast ratio than baseline (binocular group: x2=10.24, p=0.01; patching group: x2=13.53, p=0.001; combined group: x2=8.88, p=0.01). Reduced binocular difference of signal to noise ratio was only found in the binocular group (binocular group: x2=6.05, p=0.05; patching group: x2=3.97, p=0.14; combined group: x2=3.79, p=0.15). Increased amblyopic-eye signal to noise ratio was detected in all three groups (binocular group: x2=21.37, p<0.001; patching group: x2=9.90, p=0.01; combined group: x2=16.05, p<0.001) (online supplementary eFigure 2).
At 3-month visit, no difference was found in contrast ratio improvement (t=2.20, p=0.33). The reduction of binocular difference of signal to noise ratio (F=1.46, p=0.24) and the increase in amblyopic-eye signal to noise ratio (F=0.99, p=0.38) were similar among groups (table 2).
In addition to Titmus, fine stereoacuity and dynamic stereoacuity were tested using the system-based binocular stereoscopic vision energy maps in this study (table 2). After treatment, the binocular group and the combined group had better Titmus (binocular group: x2=9.75, p=0.007; combined group: x2=9.35, p=0.009) and dynamic stereoacuity (binocular group: x2=12.56, p=0.01; combined group: x2=12.66, p=0.01), but Titmus (x2=4.04, p=0.13) and dynamic stereoacuity (x2=3.60, p=0.46) did not change in the patching group. No improvement of fine stereoacuity was detected after treatment (binocular group: x2=3.74, p=0.17; patching group: x2=4.05, p=0.13; combined group: x2=3.91, p=0.14) (online supplementary eFigure 3).
To investigate whether the binocular treatment was superior in stereoacuity improvement, changes of these three types of stereoacuity from baseline to 3-month visit were compared among groups. Only the difference of Titmus improvement was statistically significant (F=49.55, p<0.001). The binocular and combined groups had greater Titmus improvement than the patching group, but the change did not reach statistically significant in a post hoc analysis (tbinocular-patching=0.38, p=0.71; tcombined-patching=1.34, p=0.19). Fine stereoacuity improvement (F=1.38, p=0.25) and dynamic stereoacuity improvement (F=3.34, p=0.19) did not differ significantly among groups.
One participant in the patching group reported moderate skin irritation during follow-up which was resolved by the 3-month visit. No other adverse events were reported.
In Chinese anisometropic children aged 3–13 years, the binocular treatment used in this study was found to be less effective than patching in amblyopic-eye VA improvement. Combining this particular binocular treatment with patching did not show further amblyopic-eye VA improvement. Moreover, this particular binocular treatment did not demonstrate superiority in reduction of interocular suppression and improvement of stereoacuity over patching.
The binocular game in this study based on push–pull method which excited the amblyopic eye while inhibiting the fellow eye to re-balance excitatory and inhibitory interactions.13 With the advantage of addressing the unequal input to binocular cortical neurons, it was regarded as a promising treatment for amblyopia. Our study demonstrated that this binocular game could improve amblyopic-eye VA by 0.18 logMAR with only 40 min training a day for 3 months. Consistent with the initial study,2 this binocular game could reduce interocular suppression, and improve Titmus and dynamic stereoacuity. Based on these results, this binocular game could be useful for anisometropic amblyopes who may not be amenable to patching.
In our study, mean amblyopic-eye VA improvement with the binocular treatment was lower than Kelly’s study9 at 1-month visit, but higher than Holmes’s study10 and Gao’s study16 at both 1 month and 3 months. In addition to different age and amblyopia type in these studies, the differences were partly due to the methods of the binocular treatment and the compliance of children. The greater amblyopic-eye VA improvement in our binocular group may be owing to the virtual reality background, which may enhance the stimulation to the amblyopic eye. Moreover, the contrast starting point was set individually based on the measurement of interocular suppression rather than the arbitrary 20% used in other studies.9–11
It remains unclear whether the low treatment response to the binocular treatment was owing to poor treatment adherence. Li et al 17 reported that a supervised, in-office treatment with movies resulted in a mean VA gain of 0.2 logMAR in just 2 weeks with 9 hours of treatment. Kelly et al 9 demonstrated a mean VA improvement of 0.15 logMAR in 2 weeks and 0.17 logMAR in 4 weeks for 1 hour a day 5 days a week using a binocular treatment with an adventure game. These studies suggested that better compliance with more appealing game and more frequently supervision may lead to larger VA gains. Greater treatment effect was also observed in younger children10 with better treatment adherence than the older ones.11 Consistent with previous studies,10 11 16 children in our study often lost interest in the game after a number of days or weeks, although electronical records showed only 15% participants in the binocular group and 27% participants in the combined group completing <75% of login. Moreover, our binocular treatment was divided into four 10 min items. Multiple short sessions can produce the same cumulative game play time as a long daily session, it is unknown whether continuous stimulation is important for the effectiveness of the binocular treatment.
Mean amblyopic-eye VA improvement with patching in our study was greater than previous studies.9–12 It may be attributed to the patching dosage: 2–6 hours a day in our study while 2 hours a day in others. As the mechanisms of the binocular treatment and patching were different, the binocular treatment was supposed to be acted as a complementary method of patching and combining these two treatments may enhance the treatment response. To test this hypothesis, we included the combined group in this study. However, the difference of amblyopic-eye VA improvement between the combined group and the patching group did not achieve statistically significant. In view of amblyopic-eye VA improvement, it remains undecided that this particular binocular game could be acted as a complementary method of patching in the treatment of children with anisometropic amblyopia.
As the binocular treatment addressed root problem of amblyopia: the unequal input to binocular cortical neurons, it may improve not only VA but also binocularity. Consistent with previous studies,9 18 reduced interocular suppression measured by contrast ratio and signal to noise ratio was observed in the binocular group in our study. However, the differences of these changes were almost similar among groups, suggesting that the binocular treatment had no superiority over patching in reduction of interocular suppression. Previous studies10 11 demonstrated that some participants in the binocular group reached a point with no contrast difference between the two eyes by 4 weeks while amblyopic-eye VA was still under normal value. In our study, four participants in the binocular group and six participants in the combined group had similar contrast in both eyes at baseline. For these participants, the different input of the binocular treatment depended on signal to noise ratio instead. How to set the different input for those without contrast difference between the two eyes needs more studies.
To test stereoacuity comprehensively, we included dynamic stereoacuity in this study. Static stereoacuity was processed by ventral pathway (P-stream), whereas dynamic stereoacuity was processed by dorsal pathway (M-stream).19 Due to these independent pathways, patients who failed static stereoacuity may still pass dynamic stereoacuity.19 20 Our results showed that more participants had normal dynamic stereoacuity than normal static stereoacuity at baseline (normal dynamic stereoacuity: 22; normal Titmus: 2; normal fine stereoacuity: 4). After treatment, dynamic stereoacuity in the binocular group and the combined group improved and the difference of dynamic stereoacuity improvement between the combined group and the patching group was marginally statistically significant (t=3.18, p=0.07). Our study also showed improved Titmus in the binocular and combined groups after treatment and significant difference of Titmus improvement among groups. However, no improvement of fine stereoacuity was detected, consistent with previous randomised controlled trial studies using Random dot stereoacuity.9–11 These differences may be attributed to the different type of stereoacuity test, as monocular cue may exist in Titmus, thereby resulting in better stereoacuity improvement. Whether combining this binocular game with patching would be benefit for stereoacuity improvement needs to be further investigated.
First, inaccurate assessment of adherence was a major limitation of our study. For the binocular treatment, only the number of login was recorded which was unable to tell the real time how long the game was played. For patching, adherence was reported by parents and bias may exist. Second, the percentage of loss to follow-up was relatively high in this study. Although the main reason was attributed to the heavy study burden in China, high loss to follow-up may interfere with the results. Third, the binocular treatment dosage was similar for different severity of amblyopia and was divided into multiple short sessions while the occlusion dosage was different for different severity of amblyopia. The great difference of dosage (40 min vs 2–6 hours) may underestimate the efficacy of binocular treatment and increase the efficacy of patching. To keep consistency with other studies and reduce the dosage difference, 2 hours of occlusion was suggested in future studies. Fourth, previous studies5 10 18 showed that age, prior amblyopic treatment and severity of amblyopia were potential confounders. Due to the small sample size, further stratification analysis was not done in this study.
The binocular game used in this study could improve amblyopic-eye VA and binocularity in Chinese children with anisometropic amblyopia, suggesting that it could be useful for those who may not be amenable to patching. However, this binocular game was less effective than part-time patching in amblyopic-eye VA improvement. Moreover, it did not demonstrate superiority in reduction of interocular suppression and improvement of stereoacuity over patching. It remains unclear whether the low treatment response of this binocular game was due to limitations of the study or its low treatment effect. Whether it could be acted as a complementary method of patching in amblyopia therapy is also undecided. A more appealing game with frequently supervision and accurate assessment of adherence is needed to tell the actual efficacy.
JY and H-WM contributed equally.
Contributors JY: substantial contribution to conception and design, analysis and interpretation of data, drafting the article, viewing it critically for important intellectual content and final approval of the article to be published. HWM: substantial contribution to acquisition of data, analysis and interpretation of data, drafting the article and final approval of the article to be published. XQ: substantial contribution to conception and design, viewing it critically for important intellectual content and final approval of the article to be published.
Funding This research was supported by the Foundation of Shanghai Municipal Health and Family Planning Commission (grant number 201540366).
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The Institutional Review Board of Eye & ENT Hospital of Fudan University 201540366.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data are available in a public, open access repository.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.