Authors: Zhao C, Cai C, Ding Q, Dai Q.
Geographical coverage: Taiwan, China, Singapore and Europe
Equity focus: Population aged 18 and under.
Study population: General population suffering from myopia.
Review type: Effectiveness review
Quantitative synthesis method: Meta-analysis
Qualitative synthesis method: Not applicable
Background: The effect and safety of atropine on delaying the progression of myopia has been extensively studied, but its optimal dose is still unclear.
Objectives: The purpose of this meta-analysis was to systematically evaluate the safety and effectiveness of atropine in controlling the progression of myopia, and to explore the relationship between the dose of atropine and the effectiveness of controlling the progression of myopia.
Overall, authors found the effectiveness of atropine in controlling the progression of myopia to be dose related. A 0.05% atropine was likely to be the optimal dose.
In total, 542 studies were identified with studies included in the meta-analysis. Combined, there were 809 patients in the atropine group and 814 patients in the control group. 0.05% atropine was used in one study, 0.5% atropine was used in five studies, and 1.0% atropine was used in the other four studies. Methodological evaluation of studies according to the Cochran Handbook are presented – most are low risk on most factors.
Myopia progression was better mitigated in the atropine treatment group than in the control group, with MD = -0.80 during the whole observation period. There was a statistical difference among 0.05, 0.5 and 1.0% atropine (p = 0.004). In addition, less axial elongation was shown, with MD = -0.26 during the whole observation period.
Sensitivity analysis was performed for the spherical equivalent refraction and the changes for the axis of the eyes. When Wang’s study (methodologically strongest) was excluded, the random-effect pooled estimate for the subgroup differences was least significant. The myopia progression was -0.79D, similar to that of all the included studies. Therefore, the original result was robust. There were no significant differences between the atropine group and the control group. Funnel plots did not suggest significant publication bias.
Several limitations were identified. Heterogeneity was still high after using the subgroup analysis. However, through the sensitivity analysis, the results of this meta-analysis were stable and consistent. There were no studies involving 0.01% atropine in this study. Some of the included studies did not report adverse reactions, and few studies reported the progression of myopia after atropine was discontinued.
Authors note that further determination and validation of the optimal dose required additional research, given the limitations to some of the studies included in this review.
Included studies needed to meet the following criteria: (1) a randomised placebo-controlled clinical trial; (2) spherical equivalent refraction more than -0.25D measured by cycloplegic autorefraction was diagnosed with myopia; (3) all patients were under 18 years old; (4) atropine was used for at least one year; and (5) the study reported at least the annual rate of myopia progression.
A purposive literature search was conducted in PubMed, Medline, EMBASE, and the Cochrane Central Register of Controlled Trials using Medical Subject Headings (MeSH) using free words closely related with myopia and atropine. Also searched were ClinicalTrials.gov and the reference lists of published reviews to find additional relevant studies. The final search date was 20 January 2020 (and the earliest included study was conducted in 1989). Only studies published in English were used.
Two investigators independently reviewed titles, abstracts and full-length articles to identify potentially eligible articles. Disagreements regarding eligibility were resolved through a discussion with a third investigator. When a study was reported more than once, only the latest study was included to avoid double inclusion of data. When a study contained different doses of atropine, only the dose recommended by the study was included. The list of excluded studies and reasons for exclusion are shown in a Supplementary Dataset.
Two investigators independently extracted data using pre-established extraction tables.
The quality of the included studies was assessed by the Cochrane Handbook, including six items: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other biases. Two reviewers determined the risk of bias which had three options (low, high and unclear). When necessary, we contacted the authors of the studies to obtain the full text or related information for an accurate assessment.
Review Manager (version 5.3; Cochrane Collaboration) was used for the data analysis. The statistical heterogeneity of included studies was tested by the Cochrane I2 test. If I2 was 50% or less, indicating a low-to-moderate heterogeneity, a fixed-effect model was used. If I2 was higher than 50%, indicating a high degree of heterogeneity, a random-effects model was applied. MD with a 95% confidence interval (CI) was used to estimate the effectiveness. A sensitivity analysis was performed by excluding the included studies one by one.
Applicability/external validity: Authors do not discuss the applicability or external validity of findings.
Geographic focus: The study encompasses 10 reviews, of which only three conducted in China were situated in middle-income countries, while the rest were in high-income countries.
Summary of quality assessment:
This study involved robust approaches to study inclusion and data extraction; however, it is unclear whether the literature search was sufficiently detailed – as it was limited to studies written in English and there was no evidence that relevant experts were consulted. The approach to data analysis was robust; however, it is unclear whether unit of analysis errors were accounted for the in the analysis, for any clustered RCTs that were included. For these reasons, we would have “medium confidence” in the results of this study.