Non-Interactive Amblyopia Treatment Modalities in Children: A Systematic Review and Meta-Analysis

Background: Recently, a number of clinical trials are conducting on amblyopia treatments. Discovering an intervention that provides optimal outcome with minimal subsidiary effect is an essential issue in medical arena. Therefore, this review aimed to identify the most effective and safest amblyopia interventions in an integrative manner. Methods: Eligible studies were identied from Cochrane library (CENTRAL), PubMed Central, Google Scholar, ScienceDirect, and Scopus electronic database. Data pooling was performed for trials with a little statistical heterogeneity (P>0.05, I 2 <50%) using the xed-effect models. The mean difference (MD) and risk ratio (RR) at 95% condence interval (CI) for visual acuity improvement, the success of the treatment, reverse amblyopia, and adherence rate was pooled. Results : Eight trials consisted of 1253 participants were included . The pooled mean visual acuity improvement was similar for both six hours and full time patching (MD, 0.00; 95% CI, -0.54 to 0.55). In subgroup analysis, all spectrum of patching are equally effective in treating mild to moderate amblyopia but 6 hours to full-time patching regimen are effective for severe amblyopia. The therapeutic outcome of patching was statistically preferable to atropine (pooled MD, 0.25 lines; 95% CI, 0.01 to 0.48). Besides, the incidence of reverse amblyopia was lower in patching groups as compared to atropine (pooled RR, 0.19; 95% CI: 0.06 to 0.57). However, the treatment adherence for patching was lesser, (pooled RR, 0.9; 95% CI, 0.84 to 0.96). Conclusion: Patching, particulary six-hours patching as a rst line treatment provides optimal visual acuity improvement that balances the effect of reverse amblyopia and poor adhrence as compared to atropine and other spectrum of patching regimen but it should be under proactive monitoring to optimize the adherence issues. consisted of the journal name, publication year, aim of the study, unit of allocation, study design, inclusion and exclusion, age, participants, sex, follow up duration, type of amblyopia, the severity of amblyopia, intervention type, type of comparison, outcome measures, the effect of the intervention, adverse effects, adherence and conclusion of the studies. The extracted data were compared between the authors and any discrepancy was resolved together.

The target of amblyopia treatments is to equalize the visual function of the two eyes, which is commenced after treating the underlying cause and the visual acuity is stabilized with optical correction for a minimum of 4 months. 19 Recently, numerous clinical trials are being conducted on the effectiveness and safeness of passive amblyopia treatments, particularly PEDIG trials. Several studies reported that the success of amblyopia treatment varies from 30% -92%. [20][21][22][23] Factors behind the success were patient compliance, high refractive error, the magnitude of anisometropia, and severity of amblyopia, age at the time intervention, the dose of treatment, eccentric xation, binocularity and type of amblyopia. 20,[24][25][26] A review article that compared atropine with patching has recommended atropine penalization as a first line treatment 27 . Another study also compared that Part-time with full-time patching and concluded that both are equally effective 28 . The conclusion these reviews was focused on the therapeutic effect with little consideration of secondary outcomes such as reverse amblyopia and treatment adherence even though studies reported that the reverse amblyopia and compliance for treatment particularly patching were refractory to treatment. 29,30,31 Assuring overweighting of the bene t over the adverse effect is an essential step to choose a modality. Overall previous reviews are unidirectional and signi cant attention for visual acuity stabilization with optical correction was not given. [31][32][33] Hence, this review was aimed to generate evidence on therapeutic and adverse effects of amblyopia interventions in an integrative manner which assists the clinicians in managing amblyopic patients and researchers to develop new hypothesis.

Protocol
Page 3/15 The study was compiled following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses checklist format. 34 Study Eligibility Criteria For this review, articles with the characteristics of randomized controlled trials (RCTs), peer-reviewed articles published in English language were considered.
As eligibility criteria, studies conducted on children diagnosed with strabismic amblyopia, anisometropic amblyopia, or mixed amblyopia with no prior amblyopia treatment other than spectacles. Moreover, the spectacle correction should be followed for at least 18 weeks, through which visual acuity improvement would be stabilized before enrollment. Functional amblyopia is the most common and treatable type of amblyopia, which is caused by childhood strabismus, anisometropia, or mixed. Hence, children with organic amblyopia associated with ocular pathology or surgical intervention for other ocular disease were not included because such kind of conditions can mask of the effect of amblyopia intervention. However, pilot studies, ongoing trials, retrospective record reviews, review articles, trials for residual amblyopia (have a history of prior treatment of amblyopia other than spectacle) were excluded.
Limitation of language literacy and inaccessibility of the articles from the standard search engine were the reason to exclude articles published in a language other than English.

Search methods
The studies were identi ed from electronic databases searching such as Cochrane Central Register of Controlled Trials (CENTRAL)', PubMed Central, Google Scholar, ScienceDirect and Scopus between 08//2018 to 8/2019. Keywords such as 'amblyopia', 'patching', 'atropine', and "occlusion" were used to look for articles from the database. Additional trials were manually searched.

Selection of studies and data extraction
Initially, two authors independently assessed the titles and abstracts of the articles after duplicated records were ruled out. Disagreements between the two authors were alleviated through discussion in the presence of the third author. Each abstract was labeled as 'included' and 'excluded' after the abstract screening. Then full-text review was performed by all authors independently. For included studies, all relevant data were extracted using data extraction form, which was developed using Cochrane data collection form for intervention reviews(RCTs and non-RCTs) guideline. 35 The form consisted of the journal name, publication year, aim of the study, unit of allocation, study design, inclusion and exclusion, age, participants, sex, follow up duration, type of amblyopia, the severity of amblyopia, intervention type, type of comparison, outcome measures, the effect of the intervention, adverse effects, adherence and conclusion of the studies. The extracted data were compared between the authors and any discrepancy was resolved together.

Types of outcome measures and data item
Comparison of patching with atropine and part-time patching (Six hours per day) with full-time patching (more than10 hours per day and all waking hours per day) were considered. The primary outcomes were the mean visual acuity improvement from the baseline in the amblyopic eye (in log MAR line) and improving visual acuity of amblyopic eyes by ≥2 log MAR lines from the baseline (Success rate). Since the improvement of 2 or more lines of visual acuity had good statistical power to detect treatment difference and minimize chance as compared to other dichotomous criteria (better than 20/25, better than 20/32, equal vision in both eyes, and improved 3 or more lines), it was considered as a criterion to measure success. 36 The secondary outcomes were adherence rate and signi cant reverse amblyopia (reduction of visual acuity in the sound eye by ≥2 Log MAR lines from the baseline in the sound eye). 13 So many studies de ned treatment adherence rate as excellent (76%-100% of prescribed treatment completed), good (51%-75%), fair (26%-50%) and poor (≤25%) based on the author judgment. [37][38][39] Since the frequency was too small in this review, this classi cation was modi ed as good adherence (51%-100%) and Poor adherence (≤50%). Visual acuity reported in the form of letters was converted into a number of lines. A single letter of the Log MAR chart is valued as 0.02. The numbers of letters per row of a log MAR chart are ve letters. Thus, the value of a log MAR line is 0.1 log MAR unit (0.02x5 letters). Five log MAR letters improvement from the baseline is equivalent to one line improvement. The standard deviation of the mean improvement reported in letter form was changed into line by dividing the number of letters by ve.

Assessment of risk of bias within and across studies
Three authors assessed the risk of bias using the validated Cochrane collaboration tool. 40 The parameters were random sequence generation; allocation concealment; masking (blinding) of outcome assessors; blinding of the participants; incomplete outcome data; selective outcome reporting; and other sources of bias. Finally, each trial was labeled as 'low risk of bias,' unclear risk' and 'high risk of bias 'with supporting evidence. Publication bias was also evaluated using funnel plots and statistically by Egger's regression intercepts method.

Data synthesis and Summary measures
The treatment effect was estimated using mean difference (MD) for visual acuity change from the baseline and Risk ratio (RR) for incidence of reverse amblyopia, visual acuity improvement in the amblyopic eye by ≥2 Log MAR lines, and adherence at 95% con dence interval (CI). The directions of effect were computed from two hours to six hours patching, part-time patching to full-time patching, and from patching to atropine. Forest plots were constructed for each study and pooled data (The squares and horizontal lines correspond to the study-speci c risk ratio or mean difference at 95% con dence interval; the diamond represents the pooled risk ratio or mean difference at 95% Con dence interval; p-value indicates the level of signi cance; I 2 indicates the percentage of total statistical variation of effect sizes across the studies). Sources of variability or heterogeneity that results in distortion of the analysis and misleading were cautioned. The effect of methodological heterogeneity (variability related with trial design and quality) and clinical heterogeneity (Variability related with the participants' age, the types or timing of outcome measurements, and intervention) were assured by performing risk of bias assessment and considering subgroup analysis, moderator variables analysis and Meta-regression analysis, respectively. Moreover, statistical heterogeneity or variability in summary treatment effects between trials was tested during result pooling. Quantitative data pooling or Meta -analysis was considered for studies with low or moderate statistical heterogeneity (I 2 <50%, p>0.05) using the xed-effect models. 40,41 The qualitative synthesis was performed for pooled studies with signi cant statistical heterogeneity. The analysis was computed using review manager 5.3 and Comprehensive Meta-analysis version 2.

Literature search
Altogether, 1892 records were retrieved from an electronic database and manual searching. After the removal of duplicated records, 1592 records were exposed for abstract screening, and 48 articles appeared to be relevant for full-text review. After full-text review, 41 articles were excluded due to various reasons. The remaining 8 articles passed the eligibility criteria, and all of them were included in qualitative and quantitative synthesis (Figure1). 15,[37][38][39][42][43][44][45] Characteristics of included studies Eight randomized controlled trials with the total of 1253 participants were included. The mean age of the participants was ranged from 4.8years to 13.6years.
Majority of the studies used a logMAR chart even if some of them presented the visual acuity in Snellen's decimal and fraction (Table1). 39,42 Quality assessment (Risk of bias and publication bias) The risk of bias assessment was conducted based on the pre-speci ed criteria. The majority of the studies were free of allocation concealment bias (75%), random sequence generation (88%), attrition bias (88%), reporting bias (100%) and other potential bias (100%). However, a signi cant portion of the trials did not disclose blinding of participants (88%) and detection bias (38%) (Figure 2 and Figure 3). The supportive evidence for the author's judgment was documented ( Table 2). Publication bias was evaluated for trials that compared patching with atropine. It indicated that nearly all studies appeared on the top of the funnel plot and they were symmetrically distributed with respect to the weighted mean, which was also supported by Egger's regression intercept (Bo= -0.5; 95% CI,-2.12 to 1.09; p-value=0.3). Though publication bias is an unavoidable issue in a review article, it was not too foreboding for this review ( Figure 4)

Comparison of patching interventions
Totally four clinical trials compared different spectrum of patching modalities starting from two hours patching per day to full time patching 15,38,43,45 . A total of 518 children were enrolled whose pooled mean age was 6.25 years. All studies consisted of children with amblyopia associated with either anisometropia or strabismus or mixed. Two studies conducted on children with moderate amblyopia 38, 45 . The remaining two trails enrolled children with severe amblyopia 43 , and mixed amblyopia (children with moderate and severe degree of amblyopia) 15 . The pooled mean time that studies measured the outcome of intervention was 4 month of follow up 15, 38, 43, 45. Majority of the studies used Electronic-ETDRS chart to measure the visual acuity 38, 43,45 . Overall, it is noted that there are some extent of baseline variation of enrolled participants related to cause and degree of amblyopia.
Regarding the modality of intervention, one clinical trial compared the effectiveness of 2 hours patching and 6 hours patching for moderate mixed amblyopia, and the mean baseline visual acuity of amblyopic in both groups was 0.48 Log MAR unit 38 . This study reported that 79.3% of the participants from two hours patching and 76.4% of the participants from the six hours patching improved their visual acuity of the amblyopic eye by 2 ≥ log MAR lines from the baseline. The mean visual acuity improvement of amblyopic eye from the baseline was 2.4±1.34 lines in 2 hours patching group and 4±1.61 lines in 6hours patching group. The mean difference between the group is zero (MD) =0.00; 95% CI, -0.45 to 0.45). Hence, the author concluded both treatment modalities were equally effective for moderate amblyopia.
Stewart et al 45 also compared that the visual acuity improvement in the amblyopic eye in six hours and full-time patching in children with moderate amblyopia after 3.5 months of follow up. The baseline BCVA of both groups was 0.45log MAR unit. Post-treatment visual acuity improvements of amblyopic eye were 2.6±1.9 lines in six hours patching and 2.4±2.4 lines in full-time patching. The effectiveness difference was not statistically signi cant (MD, 0.2; 95% CI,-0.75 to 1.15). Similarly, Singh et al 15 compared the effectiveness of 2 hours patching, 4hours patching, and 6hours patching and full time patching and the baseline visual acuity of amblyopic eye were 0.67, 0.8, 0.68 and 0.76 Log MAR unit, respectively. The author reported that the four groups were matched for age, distribution of the three types of amblyopia and pre-treatment BCVA. Post-treatment visual acuity improvements of amblyopic eye from the baseline were 1.7±1.1lines in 2 hours patching, 2.6 ±1.7lines in 4hours patching, 3±1.6lines in 6 hours patching and 3.6±2.2lines, which was compared with pretreatment visual acuity. Hence, the study showed that signi cant visual improvement in all four groups has been achieved. Furthermore, regardless of the severity of amblyopia, a signi cantly better outcome was seen when visual improvement in full time patching, six hours patching and four hours patching group were compared with two hours patching. However, statistically signi cant effectiveness difference between four hours, six hours and full time patching groups was not evident. This study also compared the effectiveness difference for in mild-moderate and severe amblyopia amblyopia separately. The author concluded that all grades of patching are equally effective for mild to moderate amblyopia in children. However, signi cantly better visual outcome was achieved in six hours and full-time group as compared to four and two hours for severe amblyopia. Both Singh et al 15  Furthermore, PEDIG 2003a indicated that the incidence of reverse amblyopia was 6.5 % (6/92 cases) in 2hours and 8.9% (8/89cases) in 6 hours patching. The proportion of participants achieved good adherence was 83% for 2 hours group and 74% for 6 hours patching group 38 . PEDIG 2003b also reported higher reverse amblyopia in the full-time group (11%) as compared to six hours patching groups (4%). Singh et al 15  Overall, the nding of the four studies implies that the severity of amblyopia matters the effectiveness a certain of patching regimen. Qualitatively, the learntlesson is that all grades of patching, from 2 hours to full time patching, are equally effective in treating mild to moderate amblyopia. However, only 6 hours patching to full-time patching regimen are effective for severe amblyopia. Regardless of the severity of amblyopia, the effectiveness of three clinical trials that compared six-hours patching with full time patching was pooled 15,43,45 . In Singh et al 15 only participants enrolled in Six hours patching and full-time patching group was considered for meta-analysis. The weighted mean visual acuity improvement of amblyopic eye was similar for both groups. The effectiveness difference was nearly zero (MD, 0.00; 95% CI, -0.54 to 0.55), which indicates that the pooled synthesis is in agreement with the qualitative synthesis ( Figure 5). Furthermore, the incidence rate of reverse amblyopia is directly related with dose of patching despite less conclusive. Therefore, these ndings drive a new insight that six hours patching is an optimal choice that balances the primary and secondary outcomes of patching intervention for various degree of functional amblyopia management, particularly severe amblyopia.

Comparison of Patching with atropine
Four clinical trials compared the effectiveness of patching with atropine 37, 39 42, 44 . A total of 664 participants were enrolled whose pooled mean age was 9.3 years. Of them, two studies compared part-time patching (2hours patching) with atropine and they measured the visual acuity improvement at 4.25 month of follow up. 39,44 . The remaining two trials compared full-time patching with atropine and the outcome was measured at 6month of follow up. 37,42 .All trials measured the visual acuity improvement as an outcome using ETDRS vision chart.  19 lines in patching group and 2.34 ±1.14 lines in atropine from the baseline. Statistically signi cant difference between the modalities was not found (MD=0.04, at 95% CI -0.56 to 0.64).
Three trails reported incidence of developing reverse of amblyopia 37,39,44 . PEDIG 2002 37 reported that 1.4% of participants in patching and 8.7% in atropine groups have developed reverse amblyopia. The variation was statistically signi cant (RR, 0.16; 95% CI; 0.05 to 0.55). Likewise, PEDIG 2009 44 revealed a single case of reverse amblyopia from atropine group but no cases from patching (RR, 0.5; 95% CI; 0.02 to 11.42). PEDIG 2008 reported that there was no incidence of reverse amblyopia in both groups 39 . The combined incidence of developing reverse amblyopia was lower by 19% for patching groups as compared to atropine (RR, 0.19; 95% CI: 0.06, 0.57). With regard to adherence, 83% of the participants from patching group and 96% from atropine group achieved the level of good adherence 37 . Even though majority of the participants from both intervention were within the level of good adherence, it was noted that atropine is statistically favorable (RR, 0.87, 95% CI; 0.81 to 0.93). Similarly, PEDIG 2008 reported that 84 % of the participants from atropine group and 80% of participants from patching (RR 0.95, 95% CI; 0.82 to 1.09) 39 . However, Menon et al 2008 found that adherence was slightly better in patching (62%) than atropine group (57%) but it was not statistically signi cant variation (RR, 1.09, 95% CI; 0.71 to 1.67). 42 Therefore, regardless of the severity of amblyopia, the two hours patching and atropine produces similar visual acuity improvement. Even though the degree and signi cance of visual acuity improvement seems slightly better in full time patching as compared to atropine, it is inconclusive. Hence, the quantitative synthesis may resolve the controversy, in which the pooled estimate indicated that patching was statistically favorable to atropine (MD, 0.25 lines; 95% CI, 0.01 to 0.48) ( Figure 6). Besides, the weighted estimate of improving visual acuity in the amblyopic eye by ≥2 log MAR lines was better in patching group (74.1%%) than atropine (67.2% ) (RR, 1.08; 95% CI, 0.98 to 1.18). On the other hand, the incidence of developing reverse amblyopia was lower in patching groups as compared to atropine (RR, 0.19; 95% CI: 0.06, 0.57). Even if there was disagreement between the authors, the pooled estimate of adherence is better in atropine, (RR, 0.9; 95% CI, 0.84 to 0.96) (Figure 7).

Moderator variables and Subgroup analysis
Severity of amblyopia, the cause of amblyopia, a dose of patching and age of the participants were considered as moderator variables. The pooled mean visual acuity between part-time (2hours patching) patching and atropine was 0.21 lines (95% CI,-0.21 to 0.64), which was not signi cantly different. Similarly, for trials that compared full time patching with atropine, the mean post-treatment visual acuity different was 0.26 lines (95% CI, -0.02 to 0.54), which was not also signi cant( Figure 6). 37,42 Hence, the effectiveness difference between patching and atropine was not explained by the variation of the patching dose. The Meta-regression analysis showed that the effectiveness difference for patching and atropine did not associate with age (B0=-0.03; α=0.5; p=0.4) (Figure 8). Regardless of the cause of amblyopia, Singh et al found that part-time patching and full-time patching can produce a similar visual outcome for mild to moderate amblyopia. However, in severe amblyopia, six hours and full-time occlusion treatment were signi cantly more effective than two hours occlusion 15 .
It was also noted that there was no signi cant difference between 6 hours and full time patching for severe amblyopia. Furthermore, PEDIG 2003 a indicated that the mean visual acuity improvement in strabismus, anisometropic, and mixed amblyopia was similar for both modalities. 40

Discussion
Overall, the pooled mean visual acuity improvement of amblyopic eye was similar for both groups (MD, 0.00; 95% CI, -0.54 to 0.55). All grades of patching are equally effective in treating mild to moderate amblyopia. However, only 6 hours patching to full-time patching regimen are effective for severe amblyopia. This implies that the severity of amblyopia affects the choice of patching regimen. Even if full time patching is effective for mild to moderate amblyopia, the incidence of reverse amblyopia should be under consideration. Therefore, these ndings drive a new insight that six hours patching is an optimal choice that balances the primary and secondary outcomes of patching intervention for various degree of functional amblyopia management under consideration of clinical heterogeneity related to some extent baseline variability. Evidence related to adherence and reverse amblyopia in patching treatment need furthers investigation.
Regarding atropine versus patching, majority of the trials concluded that both treatment modalities are equally effective independently. However, the quantitative synthesis estimated that patching was statistically favorable to atropine (MD, 0.25 lines; 95% CI, 0.01 to 0.48). Besides, the incidence of developing reverse amblyopia was lower in patching groups as compared to atropine (RR, 0.19; 95% CI: 0.06, 0.57). The pooled estimate of adherence is better in atropine, (RR, 0.9; 95% CI, 0.84 to 0.96) .This implies that patching could provide better visual acuity improvement with minimal incidence of reverse amblyopia. This nding is contrary to a study that found conventional occlusion and atropine produce comparable visual acuity improvement in the amblyopic eye and concluded atropine penalization as first line treatment for amblyopia. 27 As a subgroup analysis, this review found that effectiveness difference between patching and atropine is not explained by the variation of the patching dose, severity of amblyopia and age at the time enrollment.
The effort invested to evaluate the bene t and harms of each intervention is the good stand of this review despite data limitation. Moreover, serious caution was taken while selecting the studies; particularly those studies conducted on residual amblyopia, outcome measurement methods and time, and visual acuity stabilization with optical correction. Though publication bias and risk of bias are an unavoidable issue in a review article, it was not too foreboding. The majority of the studies were RCT trial, which is free of allocation concealment bias, random sequence generation, attrition bias and reporting bias, which minimizes methodological heterogeneity and validates the appropriateness of the synthesis. Even if the majority of included trials free from the risk of bias, blinding of the participants was not clearly disclosed in some studies. Since cost-effectiveness synthesis is not considered, economically e cient treatment modality is not implicated. This review also did not include active or interactive amblyopia treatments which require the interaction of patients with the treatment like a video game, vision therapy, and near activities even though there are substantial research revolution on such amblyopia treatment modlaities. Hence, this area would be potential for future researches who interest to generate evidence. Moreover, this study did not aim to nd which interventions provide rapid improvement.
Overall, this systematic review implicated that the choice of amblyopia treatment should rely on both primary and secondary outcomes. Selection of the modality depends on the severity of amblyopia. This review also indicates that atropine and full-time patching should be given under close follow up to prevent reverse amblyopia. Treatment adherence is related to patching dose. As a result, strong physician-parents integration might be crucial for maximizing the adherence rate and adverse effects.

Conclusions
Patching, paticullary six-hours patching,provides optimal visual acuity improvement for severe amblyopia that balances the effect of reverse amblyopia and poor adhrence as compared to atropine and other spectrum of patching regimen. All spectrum of patching less than six hour patching are equally effective for mild to moderate amblyopia and minimize the side effects related to reverse amblyopia and poor adherence as compared to atropine and six hours to full-time Availability of data and materials: The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request Competing interests: None of the above authors have any proprietary interests or con icts of interest related to this submission.
Funding: This work is not nanced by any organization.
Author's Contributions: All authors conceived, designed the study and analyzed the data. NF. and GT. retrieved the articles, screened the articles, extracted relevant data ,assessed the risk of bias for included articles . MS and AA wrote the rst draft of the article. Finally, All authors read and approved the nal manuscript.    The visual acuity improvement in the amblyopic eye after part-time (6 hours) versus full-time patching (The squares and horizontal lines correspond to the study-speci c mean difference and 95% con dence interval; the diamond represents the pooled mean difference; p-value indicates the level of signi cance; I2

Acknowledgement: Not Applicable
indicates the percentage of total variation across studies) Figure 6 The visual acuity improvement in the amblyopic eye after patching versus atropine Figure 7 The success, adherence and incidence rate of reverse amblyopia in patching versus atropine