Glasses or ortho-k? What a non-inferiority trial means

Glasses or ortho-k? A head-to-head randomised trial, and what 'non-inferior' does and doesn't mean

Families in Hamilton who ask about myopia control usually arrive with the same question: are defocus spectacle lenses as good as overnight orthokeratology (ortho-k)? Until recently nobody had randomised children to one or the other and measured eye growth head-to-head. A Danish trial published in BMC Ophthalmology in August 2026 has now done that, and its design deserves as much attention as its result.

What the trial did

Andersen and colleagues at the University Hospital of Southern Denmark randomised 90 myopic children aged 6 to 12 years, 1:1, to either DIMS spectacle lenses (MiYOSMART, HOYA) or a four-zone reverse-geometry ortho-k lens (Dreamlite, CooperVision). Allocation was computer-generated and concealed. Mean age was 9.93 years, mean cycloplegic spherical equivalent −2.96 D, and 64.4% were girls. The primary outcome was axial length change over 18 months, measured with an IOLMaster 700.

This was a non-inferiority trial, not a superiority trial. The authors set out to test whether DIMS spectacles were no worse than ortho-k by more than a pre-specified margin of 0.13 mm of axial elongation. That margin was chosen because, in Danish myopic girls of this age, 0.13 mm of axial growth corresponds to roughly 0.25 D of refractive change, the smallest step a refraction can measure.

Why the ortho-k eye had to be measured differently

Ortho-k flattens and thins the central cornea, and biometers measure from the front corneal surface, so a thinner cornea makes the eye read shorter than it really is and can flatter the ortho-k result. The authors corrected for this by adding the change in central corneal thickness back onto each ortho-k reading. They also reported a secondary outcome, overall eye length: axial length plus subfoveal choroidal thickness on swept-source OCT. Because both treatments can thicken the choroid and push the retina forward, axial length alone may under-read true eye growth.

What they found

In the intention-to-treat analysis, axial length grew 0.22 mm (95% CI 0.15 to 0.28) in the DIMS group and 0.26 mm (95% CI 0.20 to 0.32) in the ortho-k group over 18 months. The adjusted between-group difference (DIMS minus ortho-k) was −0.05 mm (95% CI −0.13 to 0.04; P = 0.319). The unadjusted difference was −0.04 mm (95% CI −0.13 to 0.05; P = 0.332).

A 95% confidence interval is the range of values for the true difference that are reasonably compatible with the data; the P-value is the probability of seeing a difference at least this large if the two treatments were in fact identical. Here the interval runs from DIMS being 0.13 mm better to ortho-k being 0.04 mm better. Because the upper limit did not exceed the 0.13 mm margin, DIMS was declared non-inferior. Because it was not below zero, DIMS was not shown to be superior. Both statements are true at once, and neither means the two treatments are equal.

The per-protocol analysis of the 71 children with good adherence and no dropout gave a difference of −0.01 mm (95% CI −0.10 to 0.09; P = 0.908). For overall eye length, growth was 0.23 mm in the DIMS group and 0.27 mm in the ortho-k group, a difference of −0.04 mm (95% CI −0.12 to 0.04; P = 0.362), which is not statistically significant.

Dropout, adherence and safety

Dropout was 4 of 45 (8.89%) in the DIMS group and 15 of 45 (33.33%) in the ortho-k group (P = 0.004), with failed treatment initiation the main reason for leaving ortho-k. Among children who stayed, adherence was high in both arms: median no-wear was 0.64% of daytime hours for DIMS and 3.12% of days for ortho-k.

One adverse event occurred, in the ortho-k group: grade 2 corneal staining treated with preventive chloramphenicol drops. Grade 1 corneal staining was seen in 19 ortho-k wearers versus 6 spectacle wearers (right eye). Visual disturbance from the spectacle lenslets was not systematically assessed as an adverse event.

How we read this in Hamilton

Over 18 months, DIMS spectacles slowed axial growth to a degree that was not meaningfully worse than a well-fitted aspheric ortho-k lens, with fewer visits, fewer dropouts and no contact-lens-related infection risk. The trial does not say spectacles are better, and it does not make ortho-k redundant. Ortho-k still offers daytime freedom from correction, and a post hoc analysis hinted that the spectacle lens may do relatively better after age 10; that is hypothesis-generating only.

At Rose Optometry the optometrist team offers both options through our myopia control service, and the decision usually turns on age, motivation, sport, hygiene and family preference rather than efficacy alone. If you would like to talk through the evidence before committing, Jagrut offers a free Myopia Chat. Readers comparing optical options with pharmacological ones may also find our earlier analysis of low-dose atropine in the LAMP trial useful.

Limitations of this evidence

This is a single-centre trial of 90 children, and it was open label: children, families, clinicians and outcome assessors all knew the allocation. Only the statistical interpretation was blinded.
Follow-up was 18 months; the authors note that two years would align better with current recommendations.
Children with contraindications to ortho-k were excluded, so the trial says nothing about spectacle efficacy in the children most likely to need an alternative.
Only one ortho-k design was tested; results may not transfer to other geometries.
Dropout was much higher in the ortho-k arm, and dropouts had more myopic parents than completers. The mixed model and per-protocol analyses agreed, which is reassuring but not proof that attrition did not bias the result.
Ethnicity, outdoor time, near work and pre-treatment progression rate were not recorded. Four children were included without a valid baseline cycloplegic refraction owing to a technical error; a sensitivity analysis found no effect on the primary outcome.
Non-inferiority depends entirely on the margin chosen. A margin stricter than 0.13 mm would not have been met by the intention-to-treat interval.
Funding came from Danish charitable foundations and hospital PhD scholarships; EnsEyes Aps Kolding supplied the ortho-k lenses. The authors declare no competing interests and no manufacturer is listed as an author. Whether the spectacle lenses were donated is not stated in the text we retrieved.

The New Zealand Eye Research Centre is set up to run trials of this kind locally; read about our myopia clinical trial site capability.

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

Andersen LC, et al. Defocus incorporated multiple segment spectacle lenses versus orthokeratology lenses for myopia control: a non-inferiority randomised clinical trial. BMC Ophthalmol. 2026;26(1):488. PMID 42595991. https://doi.org/10.1186/s12886-026-05144-6