Who paid, and does it matter? Reading conflict-of-interest statements in myopia-control trials

Every myopia-control trial we read at the New Zealand Eye Research Centre ends with a short paragraph that most readers skip: the conflict-of-interest (COI) statement. Some declare no conflicts in a single line, some list grants, consultancies and patents in detail, and some say nothing at all. A new systematic review in the American Journal of Ophthalmology asked a simple question: does the way a trial declares its conflicts tell us anything about the quality of the trial, or about what it concluded?

This article is based on the published abstract; the full text is behind a paywall, so full-text details such as the exact criteria used to grade COI statements, the review's own funding, and the authors' stated limitations were not available to us. Every number below comes from the abstract.

What the review did

Kai and colleagues, a group led from the School of Public Health at Soochow University in China, searched PubMed, Embase, the Cochrane Library and Web of Science from inception to 6 October 2025. They included randomised controlled trials (RCTs) of myopia-control interventions in children or adolescents with at least six months of follow-up that reported cycloplegic spherical equivalent and/or axial length, the two outcomes we consider meaningful.

Each trial's COI declaration was graded into predefined transparency levels. Risk of bias was assessed with the Cochrane Risk of Bias 2 tool. Multiple meta-regression models then tested whether COI transparency was associated with the direction of a trial's conclusions, its statistical significance, and its effect size. The protocol was registered in PROSPERO (CRD420261291949).

What was found

A total of 134 RCTs were included. The distribution of COI reporting was striking:

• 16 trials (11.9%) had no COI statement at all;
• 85 trials (63.4%) gave a simple declaration of no conflicts;
• 14 trials (10.4%) gave an intermediate level of disclosure;
• 19 trials (14.2%) gave a detailed COI statement.

The first main finding concerned quality. The distribution of overall risk of bias, and specifically the domain covering selective outcome reporting, differed significantly across the COI transparency categories, with a higher proportion of low-risk judgements in trials with more detailed COI disclosures (both P < .01). In plain terms, trials that were more forthcoming about their conflicts also tended to be judged as better-conducted.

A P-value is the probability of seeing a difference at least this large if there were really no association at all; P < .01 means less than a one-in-a-hundred chance, which is conventionally treated as unlikely to be a fluke.

The second finding was that, across all 134 trials taken together, COI transparency was not significantly associated with study outcomes.

The third finding is the one that deserves careful reading. In stratified analyses, among trials of optical interventions (spectacle- and contact-lens-based approaches, as opposed to drugs or light), greater COI transparency was significantly associated with more favourable study conclusions, even after adjusting for industry funding and other covariates. The odds ratio was 2.66, with a 95% confidence interval of 1.19 to 5.92 and P = .02.

The 95% confidence interval is the range of values compatible with the data: if the study were repeated many times, about 95% of such intervals would contain the true value. Here it runs from a modest 1.19 to a large 5.92, so the association is statistically detectable but its true size is uncertain. Similar patterns appeared in trials with larger sample sizes and those in higher-impact journals, but the abstract reports that these associations were attenuated once industry funding was accounted for.

How we read this

There are two ways to interpret the optical-intervention result, and the abstract cannot settle which is right. The first is reassuring: detailed disclosure is a marker of well-resourced, well-run, industry-linked trials, and those trials tend to test products that work. The lower risk of bias among detailed disclosers supports this reading. The second is less comfortable: the trials most entangled with manufacturers are the ones most likely to conclude favourably, and a thorough COI paragraph does not neutralise that. The authors themselves land somewhere in between, concluding that COI transparency has "context-specific" relationships with conclusions and reporting quality, and that transparency should be interpreted alongside the funding context rather than instead of it.

We would add a point the abstract does not address. Favourable conclusions and large effects are not the same thing. Two trials can report similar axial-length differences and one call the result promising while the other calls it modest. Where a trial's headline reduction in progression is expressed as a percentage, that figure also depends heavily on how fast the control group progressed, a point we made when discussing the LAMP low-dose atropine trials and the repeated low-level red-light evidence.

The practical message is not "distrust sponsored trials". Most of the optical interventions used in the myopia control service at Rose Optometry rest on industry-supported RCTs, because that is who funds device trials. The message is to read the funding and COI sections as part of the methods, to prefer axial-length data over refraction alone, and to value independent replication. It is one reason we think a clinical trial site in Hamilton that can run investigator-initiated studies matters.

Limitations of this evidence

• We had access to the abstract only. The criteria that separated "simple", "intermediate" and "detailed" COI statements, the review's own funding and conflicts, and the authors' stated limitations could not be checked.
• This is a cross-sectional analysis across trials. It shows association, not cause. A more detailed COI statement does not make a trial better or worse; both may be driven by a third factor such as journal policy, year of publication or sponsor type.
• The headline odds ratio (2.66, 95% CI 1.19–5.92) comes from a stratified subgroup, optical interventions, while the overall analysis was null. Subgroup findings are more prone to chance and should be treated as hypothesis-generating.
• Only 19 trials fell in the "detailed" category and 14 in "intermediate", so the comparisons rest on small groups.
• The abstract reports P-values for the risk-of-bias comparison but not the actual proportions of low-risk trials in each category, so the size of that difference is unknown to us.

If you are weighing myopia-control options for a child in Hamilton and want to talk through which evidence applies, the free Myopia Chat with Jagrut is available to book online.

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

• Kai JY et al. Conflicts of Interest Transparency, Study Outcomes, and Risk of Bias in Myopia Controlled Trials: A Systematic Review and Meta-Analysis. Am J Ophthalmol. 2026;292:80-90. PMID 42600884. https://doi.org/10.1016/j.ajo.2026.08.019