Varying dose of atropine in slowing myopia progression in children over different follow-up periods by meta-analysis

Author: Gan J, Li SM, Wu S, Cao K, Ma D, He X, et al.

Geographical coverage: Asia, Europe and the US

Sector: Biomedical

Sub-sector: Treatment

Equity focus: Children aged between 6 and 15 years old

Study population: School-aged children with myopia either receiving atropine or a placebo or non-atropine treatment.

Review type: Effectiveness review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis method: Not applicable

Background: Myopia is a growing global public health concern, particularly in some Asian regions. Topical atropine, in varying doses, has emerged as a potent treatment for myopia and is used in over 60% of myopic children in Taiwan. However, due to side effects associated with high doses (0.5-1%), it remains an off-label treatment in mainland China and most western countries, pending FDA approval. Consequently, moderate (0.01-0.5%) and low (0.01%) doses have been widely used in recent years. The authors previously found ethnic differences in atropine efficacy, with Asians showing greater effects than White children. This led to the first randomised controlled trials (RCTs) on low-dose atropine in mainland China, which resulted in a 34.2% reduction in myopia progression within a year. Despite these findings, uncertainties and controversies persist, necessitating further comparisons of different doses.

Objectives: To evaluate the overall efficacy and safety of different doses of atropine with more updated RCTs and cohort studies, to explore the dose-response relationship of atropine, and to investigate whether there was an efficacy difference across different treatment periods.

Main findings

Overall, authors found that both the efficacy and adverse effects of atropine are dose-dependent in slowing myopia progression in children. The efficacy of high-dose atropine was reduced after the first year of treatment, whereas low-dose atropine had better efficacy in a longer follow-up period.

Authors included 12 RCTs and 15 cohort studies in the review. The total sample size of participants included was 5,069, among which 3,024 received atropine treatment and 2,045 participants received placebo or non-atropine treatment, with a follow-up period from 12 to 144 months. Seven studies were conducted in mainland China, eight in Taiwan, four in the United States, two in Singapore, two in Hong Kong, two in Europe, one in Japan, and one in India, resulting in most participants being Asian. All RCTs were conducted in Asia. Studies involved 5,069 children aged 5 to 15 years.

In terms of quality, two of the 12 included RCTs were assessed as high risk of bias, and the cohort studies was generally high.

The average differences in myopia progression between the atropine and control groups, reported by the authors, were 0.73 dioptres (D), 0.67 D, and 0.35 D per year for high-dose, moderate-dose, and low-dose atropine, respectively (χ 2 = 13.76; p = 0.001, I² = 85.5%). After excluding studies with extreme results, according to the authors, atropine showed a significant dose-dependent effect on both refractive change and axial elongation, with higher atropine dosages leading to less myopia progression (r = 0.85; p = 0.004) and less axial elongation (r = -0.94; p = 0.005). Low-dose atropine resulted in less myopia progression (-0.23 D; p = 0.005) and less axial elongation (0.09 mm, p <0.001) in the second year compared to the first year, while high-dose atropine led to more axial elongation (-0.15 mm, p = 0.003). Higher atropine doses were linked to a higher occurrence of side effects, such as photophobia, with an odds ratio (OR) of 163.57, compared to an OR of 6.04 for low-dose atropine and 8.63 for moderate-dose atropine (p = 0.03).

Authors observed that conclusions remained unchanged after removing studies with significantly different characteristics. Meta-regression analysis identified ethnicity as the only significant moderator, with a greater effect on slowing myopia progression in Asian children compared to White children. A funnel plot for the publication bias test showed an asymmetric distribution, suggesting potential publication bias.

Methodology:

Inclusion criteria consisted of comparative studies (i.e., RCTs, and cohort studies) according to the following criteria: (1) a human study investigating the relationship between topical atropine and myopia in school-aged children (between 6 and 15 years); (2) using atropine in at least one intervention and placebo or non-atropine treatment in another as the control; and (3) reporting at least one outcome of interest, including the annual rate of myopia progression and any adverse effects. In addition, the dose of atropine was classified into three subgroups: low dose (0.01%), moderate dose (>0.01% to 0.5%) and high dose (0.5-1.0%).

Authors searched Medline, EMBASE and the Cochrane Library from their inception to May 2021 with language restriction in English. Authors also screened ClinicalTrials.gov and the reference lists of published reviews to identify additional relevant studies.

Two investigators independently screened the studies for inclusion and extracted data using standardised forms. All disagreements were reviewed by a third investigator. For any missing data, the authors of the trial reports were contacted or GetData GraphDigitizer 2.24 was used to read data from figures. Quality of the selected trials was assessed following the recommendations of Cochrane collaboration for RCTs. The Newcastle-Ottawa Quality Assessment Scale items with a “star system” were applied to assess the quality of cohort studies.

Data analyses were performed using Review Manager. The weighted mean difference and 95% confidence intervals (95% CIs) were calculated for refractive changes and axial elongation in different doses of atropine versus the control group, as well as the odds ratios (ORs) for adverse effects. Effect sizes were determined using the Cohen d formula. ORs with 95% CIs were calculated for proportions with fast (>1.0 dioptres (D) per year) and slow (<0.5 D per year) myopia progression. A fixed-effects model was used for low heterogeneity (p > 0.1, I² < 50.0%), otherwise a random-effects model was applied. Sensitivity analysis was conducted by excluding studies with significantly different characteristics. Subgroup analyses were pre-planned to compare treatment effects among children with different doses of atropine. A p-value <0.05 was considered statistically significant.

Applicability/external validity: The authors’ findings differ from existing literature due to their more detailed analysis. They highlight several limitations, including high heterogeneity, grouping of different atropine doses due to insufficient data, and combining RCTs and cohort studies. Ethnicity also influenced heterogeneity, with atropine showing greater effects in Asian children. Sensitivity analysis revealed stable outcomes, but potential publication bias warrants cautious interpretation. Over half the studies didn’t report adverse reactions, leading to incomplete data. The efficacy of atropine was reported during treatment, with varying follow-up periods. Most studies were conducted among Asians, with significant response differences between Asian and Caucasian individuals to myopia interventions.

Geographic focus: Most included studies and all RCTs were conducted in Asia. Data from one LMIC (China) was included.

Summary of quality assessment: The approaches used to identify, include and critically appraise studies were generally sound, with at least two authors undertaking all key tasks. However, the search was somewhat limited, in being restricted to material published in English and making no attempt to include published material. While the approaches to the analysis of the data were robust, and specified in detail, analysis was not undertaken based on the “risk of bias” status of included studies. In addition, it is unclear whether unit of analysis errors were considered in the analysis, which would have been appropriate, given the involvement of RCTs. We also note a high degree of heterogeneity and possible evidence of publication bias, although robust efforts are made to reduce and explain these features. For these reasons, there is medium confidence in the findings of this review.