Binocular treatment for amblyopia: A meta-analysis of randomized clinical trial

Author: Roda M, Pellegrini M, Di Geronimo N, Vagge A, Fresina M, Schiavi C.

Geographical coverage: The US, China and Iran.

Sector: Biomedical

Sub-sector: Treatment

Equity focus: Children and adolescents aged 8-17.

Study population: Patients with amblyopia aged 3-17.

Review type: Effectiveness review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis method: Not applicable

Background: Patching or other means of deprivation of the non-amblyopic eye is the mainstay of amblyopia treatment. Recently, binocular approaches to treating functional amblyopia have been designed. The primary goal is to strengthen the amblyopic eye by improving fusion and stereopsis, in order to increase not only visual acuity (VA), but also binocular function. However, controversy remains regarding the efficacy of binocular amblyopia treatment and there is still no consensus regarding the effect of binocular treatment for amblyopia.

Objectives: To summarise the available evidence to determine whether binocular treatment is more effective than patching in children with amblyopia.

Main findings:

Overall, the meta-analysis found no convincing evidence supporting the efficacy of binocular treatment as an alternative to conventional patching.

A total of five studies were included in the meta-analysis. The selected trials were reported between 2016 and 2019. Three studies were conducted in the United States, one in Iran, and one in China. They examined the effect of binocular therapy for amblyopia compared to patching. Of 625 involved patients, all presented functional amblyopia, in particular: anisometropic amblyopia (n = 338), strabismic amblyopia (n = 89), combined anisometropic/strabismic amblyopia (n = 160), and unspecified (n = 38). The final visit was completed by 587 patients. The range of average ages was from 3 to 17 years. In all studies, binocular therapy consisted of dichoptic training played at home for a range from 20 minutes to 1 hour a day. Overall, according to the authors, all studies showed some concerns about risk of bias.

Authors found no significant difference in visual acuity between patients treated with binocular treatment, and patching was observed (SDM = -0.12; 95% CI: -0.45-0.20; p = 0.464). Furthermore, authors noted no significant difference in stereopsis between patients treated with binocular treatment, and patching was observed (SDM = -0.07; 95% CI: -0.61-0.48; p = 0.809). For both variables, the between-study heterogeneity was high (respectively, I² = 61% and I² = 57%).

Policy relevant findings suggest that binocular treatment cannot fully replace traditional treatment but, to date, it can be considered a valid complementary therapy in peculiar cases.Authors note that further studies are required to determine whether more engaging therapies and new treatment protocols are more effective.

Methodology:

Eligible studies for inclusion were English-language published randomized controlled trails (RCTs) involving patients aged 3-17 with anisometropic, strabismic, or combined amblyopia. These studies compared binocular treatment with patching and reported at least one outcome variable of visual acuity or stereoacuity. Exclusions were made for conference abstracts, letters, reviews, duplicate publications, studies without available raw data, studies with a paired-eye design, and studies combining binocular therapy with other therapies.

Authors searched PubMed, Scopus, Web of Science, and Cochrane Central Register of Controlled Trials databases was performed (last accessed 25 March 2020) to identify studies on binocular therapy for amblyopia. The key words “amblyopia” OR “lazy eye” combined with “binocular amblyopia treatment” OR “binocular amblyopia therapy” OR “binocular therapy” OR “binocular treatment” OR “binocular game” OR “dichoptic” were used in the literature search. No language restriction was applied. In addition, the references of selected retrieved articles were manually searched to identify additional relevant studies.

The full text of citations judged as potentially eligible were obtained and independently screened for eligibility by the same two reviewers. Disagreements were resolved by discussion with all authors.

Data was extracted independently from the included trials by two reviewers through a pilot-tested data extraction form. Discrepancies were mediated by a third reviewer. Study quality was assessed using the Revised Cochrane risk-of-bias tool for randomised trials.

Meta-analysis was conducted using the Meta package with R and RStudio software. Mean, standard deviation and sample size were used to calculate the standardised difference in means (SMD) with 95% confidence intervals (CI). A p <0.05 was considered statistically significant. The I² test was used to evaluate heterogeneity. The results were pooled using a random-effect model and the DerSimonian and Laird method. To explore individual study effects on overall pooling, sensitivity analyses were performed by the authors removing one study at a time from the analysis. Potential publication bias was assessed by visual evaluation of the funnel plots.

Applicability/external validity: The authors observe significant heterogeneity across the studies, which could potentially undermine the reliability of the findings. They attribute this variability to distinct differences in the design of the individual RCTs.

Geographic focus: The included RCTs were conducted in both high-income (USA) and low-to-middle-income countries (Iran and China). However, the potential variation of results across different geographical contexts was not taken into account.

Summary of quality assessment:

The approaches undertaken to identify, include and critically appraise studies were generally robust, with two reviewers independently conducting each of the key stages. However, there was no evidence that unpublished material was considered for inclusion, and it is unclear whether non-English studies were considered. While the approaches taken to the analysis of the data were generally robust, there was no discussion of unit of analysis issues and how these may have been dealt with in relation to the findings of different RCTs. For these reasons, we attributed medium confidence in the findings of this review.