Where atropine's benefit levels off: the dose-response plateau in a 33-trial meta-analysis

When a child's myopia keeps progressing on low-dose atropine, the usual instinct is to step up the concentration. But does each step up buy a proportionate amount of extra effect? A systematic review and dose-response meta-analysis by Zhang and colleagues, published in Survey of Ophthalmology in 2026, set out to answer that question by pooling the placebo-controlled trials and modelling efficacy against concentration.

One caveat before the numbers. We could not access the full paper, so this summary is based on the published abstract only. Everything quoted below comes from that abstract, and several things a careful reader would want to check are listed in the limitations section.

What the review did

The authors searched PubMed, Embase, Web of Science, the Cochrane Library and two trial registries (the World Health Organization platform and ClinicalTrials.gov). They included randomised clinical trials comparing any dose of atropine against placebo in myopic children. From 3566 studies assessed, 33 trials were eligible, involving 6301 children aged 4 to 18 years. The trials covered 10 different concentrations, and mean follow-up was 19.5 ± 12.3 months.

A dose-response meta-analysis fits a curve through all the available doses, to test whether more drug gives proportionately more effect or whether the relationship bends.

What it found

The relationship was nonlinear (P < 0.001). Compared with placebo, the mean reduction in annual spherical equivalent refraction progression at each concentration was:

• 0.01%: 0.21 D (95% CI, 0.13 to 0.28)
• 0.02%: 0.35 D (95% CI, 0.23 to 0.46)
• 0.03%: 0.42 D (95% CI, 0.28 to 0.56)
• 0.04%: 0.45 D (95% CI, 0.30 to 0.60)
• 0.05%: 0.46 D (95% CI, 0.32 to 0.61)
• 0.1%: 0.49 D (95% CI, 0.34 to 0.63)
• 1%: 0.99 D (95% CI, 0.66 to 1.31)

These are between-group differences against placebo, not within-group change.

A 95% confidence interval (CI) is the range within which the true average plausibly lies; narrower is more precise. A P-value is the probability of seeing a result this strong if there were really no effect; below 0.05 is the usual threshold for "unlikely to be chance".

Read down the list and the shape is clear. The modelled effect rises steeply between 0.01% and 0.03%, then flattens: the estimates for 0.04%, 0.05% and 0.1% sit close together, with confidence intervals that overlap almost entirely. The authors conclude that the gain in efficacy "may plateau beyond a certain range", and that stepping up the concentration for children who respond inadequately should be confined to a specific concentration range. The abstract does not say exactly where that range ends.

The cost side of the curve

Side effects did not plateau in the same way, at least as far as the abstract reports. Higher doses were associated with decreased amplitude of accommodation (P = 0.02), larger pupil diameters (P = 0.01) and a higher frequency of photophobia (P = 0.02). The abstract gives no figures for how large these effects were at each concentration, so the trade-off cannot be quantified from what we could read.

Refraction versus axial length

Every efficacy number in the abstract is spherical equivalent refraction. None is axial length. That matters for two reasons.

First, the long-term harms of myopia are tied to how far the eye has elongated, so axial length is the measurement closest to the outcome we care about. Second, atropine acts on the focusing muscle. Refraction measured in a child using atropine can be influenced by that effect on accommodation as well as by any change in eye growth, and the influence could plausibly differ by concentration. A dose-response curve built on refraction is therefore a curve built on a proxy. A plateau in refraction is not automatically a plateau in eye growth.

We have discussed the same proxy problem in our earlier summary of the evidence for low-dose atropine.

How solid is the top of the curve?

The 1% estimate looks like it breaks the plateau: 0.99 D against 0.49 D for 0.1%. The authors themselves caution that the estimates for 0.1% and 1% were based on fewer and smaller trials, and that results for doses exceeding 0.1% should be interpreted with caution.

There is also a large gap in the data. Between 0.1% and 1% the abstract reports nothing, so the curve across that stretch is drawn by the model, not by trials. And any benefit at high concentrations has to be weighed against the side effects above and against what happens when treatment ends, a question covered in our post on rebound after stopping myopia treatment.

Limitations of this evidence

• Abstract only. We could not read the full text. We could not check funding or conflicts of interest, the risk-of-bias assessment, heterogeneity statistics, how many trials and children sat behind each concentration, the results for the three concentrations not listed in the abstract, drop-out rates, or whether axial length was analysed.
• Thin data above 0.1%. The authors name the number, design heterogeneity and sample sizes of the higher-concentration trials as a limitation.
• Missing baseline history. Included studies frequently lacked pre-intervention refractive history.
• Comparisons across trials. A dose-response model compares concentrations largely across different trials, which is weaker than one trial randomising children to each dose.
• Averages, not individuals. A pooled mean difference says nothing direct about children who have already progressed on a lower dose, the group escalation is aimed at.
• Follow-up averaged 19.5 months, short against the many years over which childhood myopia progresses.

What we do with this in practice

This review does not settle which concentration any child should use. If the benefit of going higher flattens while pupil and focusing side effects keep rising, then automatically escalating the dose is harder to justify, and the alternatives deserve a look: checking adherence, reviewing time outdoors and near work, or adding or switching to an optical treatment. A sibling post this week looks at a study of stepwise atropine escalation, which tests that strategy more directly.

Atropine is a prescription medicine, and its suitability, concentration and availability are things to discuss with your optometrist. The optometrist team at Rose Optometry in Hamilton can talk through how this evidence applies to your child. The team also offers a free Myopia Chat. The optometrists at Rose Optometry are Anjali Hira, Emilie Lawson, Jacqueline Rowe, Jagrut Lallu, Jason Shen, Jessica Wood and Stella Wong, and myopia management is part of their clinical practice.

Jagrut Lallu — BOptom (Hons), MSc Specialty Lenses (Hons), FIAOMC. Therapeutic Optometrist & Contact Lens Specialist. Founder, New Zealand Eye Research Centre; Partner, Rose Optometry; WCO Asia-Pacific Myopia Ambassador; International Myopia Institute NZ Ambassador; Clinical Senior Lecturer, Deakin School of Optometry; Honorary Teaching Fellow, University of Auckland.

Reference

• Zhang XJ, Liu M, Yu M, Ho ECF, Wong RTC, Zaabaar E, Chen W, Huang H, Guo PY, Sun J, Lam AK, Cheng CY, Yam JC, Shih KC, Leung CKS. Efficacy, tolerability, and threshold effect of atropine eye drops for myopia control: A systematic review and dose-response meta-analysis. Survey of Ophthalmology. 2026;71(6):1732-1745. PMID: 42070753. https://doi.org/10.1016/j.survophthal.2026.04.002