A meta-analysis assessing change in pupillary diameter, accommodative amplitude, and efficacy of atropine for myopia control

Authors: Tran HD, Sankaridurg P, Naduvilath T, Ha TT, Tran TD, Jong M, Coroneo M, Tran YH.

Geographical coverage: China, USA, Singapore, Taiwan, Hong Kong and Italy.

Sector: Biomedical

Sub-sector: Treatment

Equity focus: Children

Study population: Children aged 4-16 years old

Review type: Effectiveness review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis method: Not applicable

Background:

Atropine is widely used for myopia control, but its use has concerns. Major issues include photophobia, difficulty with near-work at higher concentrations, and myopia “rebound” after treatment cessation. Unknown long-term effects on pupils, macula, and retina are also a concern. Some data suggest that accommodative amplitude and near vision recover to pre-treatment levels within months after stopping atropine. While previous studies have examined the effect of atropine concentrations on axial elongation and refractive error, its impact on pupillary diameter and amplitude of accommodation remains unexplored.

Objectives: To determine the effect of atropine on pupillary diameter, accommodative amplitude as well as myopia progression.

Main findings:

Overall findings of the review suggest that a nonlinear dose-response relationship exists between atropine and PD and AA.

A total of 13 eligible articles (six RCTs and seven observational studies) were included in the review. These comprised data from 2,907 participants aged 4 to 16 years old and with at least one year of follow-up. Nine different concentrations of atropine were used in these trials.

Authors found the relationship between atropine and change in PD and AA nonlinear; at <0.1+0.1 to +1.4) and AA (-1.6D; 95% CI: -3.9 to +0.7) was smaller, whereas at ≥0.10% atropine, the slope plateaued, but change in PD (+3.2 mm, 95% CI: +2.8 to +3.5) and AA (-10.7D; 95% CI: -12.2 to -9.2) was high. Reduction in myopia progression with atropine at <0.10% and ≥0.10% as compared to controls was 0.37D (95% CI: 0.16 to 0.58) versus 0.75D (95% CI: 0.17 to 1.33) for spherical equivalent and -0.10 mm (95% CI: -0.24 to 0.05) versus -0.23 mm (95% CI: -0.34 to -0.13) for axial length.

Authors note the need of further research to determine the concentration of atropine that provides maximal efficacy with tolerable side effects.

Methodology:

The review included studies with myopic participants aged 16 or younger, from randomized clinical trials or observational studies, with a control group, having myopia of 0.50D or worse at entry, followed for at least 12 months, and published in English. Excluded were studies with participants having refractive error of 0.25D or better at baseline, aged 17 or older, on multiple atropine concentrations, or treated with atropine and other methods. Control groups were treated with single-vision spectacles, 0.50% tropicamide, or placebo eye drops.

Authors searched medical databases like PubMed, EMBASE, Scopus, ProQuest, and Cochrane Library using keywords related to myopia and atropine. The search spanned articles from 1980 to June 2020, following PRISMA guidelines. The outcomes measured were changes in pupillary diameter and accommodative amplitude, and the annualised mean change in spherical equivalent and axial length with various atropine concentrations compared to control.

Two authors independently reviewed the data, initially assessing titles and abstracts, and removing ineligible articles and duplicates. Selected articles were evaluated using the Newcastle-Ottawa scale for observational studies and the Cochrane risk-of-bias tool for randomized clinical trials. Extracted data included authorship, publication year, research country, study design, participant age, baseline spherical equivalent and axial length, atropine concentration, and changes in pupillary diameter, accommodative amplitude, spherical equivalent, and axial length. If multiple articles reported on a single study, data from the earliest publication providing the annual change was prioritized.

Statistical analysis was performed using Review Manager, GraphPad Prism 8, and Microsoft Excel. The effect size was calculated by pooling the difference in accommodative amplitude (AA) and pupillary diameter (PD) with various atropine concentrations compared to control. The mean annual change in myopia progression was assessed. The Cohen d formula and a random effect model were applied. The I2 statistical test quantified heterogeneity and evaluated the risk of bias. Positive effect size indicated atropine’s superiority over control for spherical equivalent and pupil dilation, but the opposite for axial length and AA. Baseline data was used as control for side effect analysis, except in one trial where a historical control group was used. Further analysis determined the relationship between changes in PD and AA and atropine concentrations. The relation between weighted meta-analysis data and atropine concentrations was determined using various fits; the best fit was found using the logarithmic fit due to high R2 values and lower standardised residual errors.

Applicability/external validity: Limitations of studies included in the study noted by authors, reduce applicability of study. Included studies examined low doses of atropine only and all studies were undertaken on Asian eyes (with non-Asian eyes known to react differently).

Geographic focus: The review included two studies conducted in China. However, the authors did not attempt to investigate potential variations in findings across different geographical contexts. It’s worth noting that the physiological differences between Asian and non-Asian eyes could potentially influence the reaction to atropine.

Summary of quality assessment:

Important limitations were identified in the approaches used to identify, include and critically appraise studies. The search was limited to English language studies, and it’s unclear if reference lists from identified studies were reviewed or if relevant experts were consulted. The process for determining the inclusion and exclusion of articles could have been more transparent, particularly regarding whether it was a collaborative decision among multiple authors. Although the data analysis approach was generally sound, it would have been beneficial if the authors had explicitly addressed potential unit of analysis errors in the included RCTs. Due to these issues, our confidence in the review’s findings is somewhat limited.