A clinician has suggested a small desktop device. The child looks into it for three minutes, twice a day, and over a year their short-sightedness is expected to get worse more slowly. There is a waiting list. A decision is wanted this week.
Searching for the answer produces two piles. One consists of clinics and manufacturers describing a breakthrough. The other consists of headlines about children's retinas. Neither pile contains the thing a parent needs, which is what has actually been measured, by whom, and what remains unknown.
Why anyone is treating short-sightedness at all
Spectacles correct myopia, so the instinct is to treat it as an inconvenience with a solved answer. The reason clinicians work to slow it is that the eye keeps growing longer, and the risk attached to that growth is not about focus.
Holden and colleagues pooled 145 studies covering 2.1 million participants in 2016 and estimated 163 million people with high myopia in 2000, at 2.7% of the world population. Their projection for 2050 was 938 million, at 9.8%. That is close to a sixfold increase in the group whose eyes are long enough to carry raised risk of retinal detachment, myopic maculopathy and glaucoma.
Which is the frame for everything below. The target is not a prescription. It is axial length, in millimetres, accumulated across childhood.
What the randomised trial measured
Jiang and colleagues published the multicentre trial in Ophthalmology in 2022. Two hundred and sixty-four children aged 8 to 13 with myopia between -1.00 and -5.00 dioptres were randomised to red light therapy plus single-vision spectacles, or to spectacles alone.
The device was a desktop unit emitting 650 nm red light at roughly 1600 lux, described in the paper as 0.29 mW for a 4 mm pupil and as class I. Three minutes per session, twice daily, at least four hours apart, five days a week, at home under a parent's supervision.
| Red light + spectacles | Spectacles only | |
|---|---|---|
| Axial elongation | 0.13 mm (95% CI 0.09-0.17) | 0.38 mm (95% CI 0.34-0.42) |
| Refraction change | -0.20 D (95% CI -0.29 to -0.11) | -0.79 D (95% CI -0.88 to -0.69) |
| Difference | 0.26 mm and -0.59 D | reference |
Jiang Y, Zhu Z, Tan X, et al. Effect of repeated low-level red-light therapy for myopia control in children: a multicenter randomized controlled trial. Ophthalmology. 2022;129(5):509-519. Of 264 randomised, 246 (93.2%) were analysed.
A quarter of a millimetre of axial growth prevented in a year is a large effect by the standards of myopia control. The authors also recorded what their monitoring found on safety, and it matters that it is quoted rather than skipped.
No severe adverse events (sudden vision loss of two or more lines or scotoma), functional visual loss indicated by best-corrected visual acuity, or structural damage seen on OCT scans were observed.
That is a twelve-month trial in 246 children with OCT imaging. It is not nothing, and anyone presenting this treatment as untested is misrepresenting the literature.
Then somebody put the devices on a radiometer
Trials watch for harm appearing. A different question is whether the light being delivered sits inside the exposure limits that exist for looking into a light source at all. Those are two separate exercises, and the second one was not done on commercial devices until recently.
2024: two devices, and a title that says it plainly
Ostrin and Schill published measurements of two devices in Ophthalmic and Physiological Optics in 2024, under the title Red light instruments for myopia exceed safety limits. Both were confirmed to be Class 1 laser products.
For the Sky-n1201a they reported 654 nm, 0.2 mW through a 7 mm aperture at 10 cm, and a calculated maximum permissible exposure for photochemical damage of 0.55 to 7.0 seconds across 2 to 7 mm pupils. For the Future Vision device they reported 652 nm, 0.06 mW, and a photochemical maximum permissible exposure of 50 to 625 seconds. Their discussion stated that for both devices, three minutes of continuous viewing approached or surpassed the maximum permissible exposure.
2024: a manufacturer replies, in the same journal
Battersby published a response to that paper in Ophthalmic and Physiological Optics in 2024, disputing the analysis. It appears in the literature alongside the original.
This article does not adjudicate between them, and no article written by someone who has not repeated the measurements should. What matters for a parent is that the disagreement is real, documented, and between people who are qualified to have it.
2026: four devices, and the number that separates them
The same two authors published an expanded evaluation in JAMA Ophthalmology in 2026. Four commercially available devices, measured with an integrating sphere radiometer at 1 and 10 cm through a 7 mm aperture, with retinal irradiance calculated for pupil diameters from 2 to 7 mm, and classified against ANSI Z80.36-2021 and ANSI Z136.1-2022.
| Device | Type | Time to group 1 limit |
|---|---|---|
| EyeRising | laser (Class 2M) | 1.4 seconds |
| Sky-n1201 | laser (Class 1) | 2.8 seconds |
| Future Vision | laser (Class 1) | 253 seconds or longer |
| AirDoc | LED, diffuse illumination | 22,761 seconds |
Ostrin LA, Schill AW. Safety evaluation of 4 red light therapy devices for myopia. JAMA Ophthalmol. 2026;144(3):255-258. Laboratory evaluation performed November 2023 to April 2024. The recommended treatment time across these devices is 180 seconds.
The authors' stated conclusion was that laser-based red light therapy instruments deliver irradiance levels that reach ANSI safety limits within exposure times below the recommended 180-second treatment time, and that this, combined with emerging clinical reports of retinal damage and the recent reclassification of red laser devices as Class III in China, highlights the need for rigorous independent safety validation before widespread paediatric use.
The distinction that does the most work
Look at the last two rows of that table again. 253 seconds and 22,761 seconds. Both are described in the literature as red light therapy for myopia, and one of them reaches the same limit sixteen thousand times faster than the other.
The LED device produced diffuse illumination. The laser devices deliver light as a point or near-point source, which concentrates it on a small patch of retina. That is the physical reason the numbers separate, and it is why treating these as one category is the single most misleading thing anyone can do with this subject.
Where the two sets of numbers actually meet
The efficacy trial and the safety analysis are not measuring different devices so much as making different assumptions about the eye.
Jiang and colleagues described their device as 0.29 mW for a 4 mm pupil, and class I on that basis. Ostrin and Schill calculated retinal irradiance across pupil diameters from 2 to 7 mm, and their most-quoted figures are for 7 mm.
A pupil is not a fixed aperture. It widens in dim surroundings and it is larger in children than in adults. A treatment delivered at home, in the evening, by a child leaning into an eyepiece is not obviously a 4 mm pupil situation.
This article is not in a position to say which assumption is correct, and stating that plainly is more useful than picking one. It is, though, the specific thing the disagreement turns on, and a parent who knows that can ask a much better question than one who does not.
What regulators have done so far
China's regulator reclassified red laser devices of this kind as Class III, which Ostrin and Schill cite in their 2026 conclusion as part of the case for independent validation. Class III is the tier reserved for devices carrying the highest level of regulatory control.
In the United States the treatment has not received FDA approval, and a reader there cannot simply buy the device on the strength of results published elsewhere. In several other jurisdictions equivalent devices are available under local approvals. Where you live changes what is in front of you, and it does not change what has been measured.
The part of this that is not disputed by anyone
There is an intervention for childhood myopia with a cluster randomised trial behind it, no device, no exposure limit and no cost. It is daylight.
He and colleagues randomised twelve primary schools in Guangzhou, published in JAMA in 2015. Six schools added one 40-minute outdoor class to each school day and encouraged outdoor activity at home. Six continued as normal. Nineteen hundred children, followed three years.
| Extra outdoor time | Usual schedule | |
|---|---|---|
| Children who became myopic | 30.4% | 39.5% |
| Difference | -9.1 percentage points (95% CI -14.1 to -4.1), P<0.001 | reference |
| Refraction change over 3 years | -1.42 D | -1.59 D (P=0.04) |
| Axial elongation | 0.95 mm | 0.98 mm (P=0.07, not significant) |
He M, Xiang F, Zeng Y, et al. Effect of time spent outdoors at school on the development of myopia among children in China: a randomized clinical trial. JAMA. 2015;314(11):1142-1148.
Read the last row rather than skipping it. Outdoor time reduced how many children became short-sighted. It did not significantly slow the elongation of eyes already growing.
That is the honest relationship between the two halves of this article. Daylight is a preventive lever, aimed at onset. Red light therapy is aimed at progression in children who are already myopic. They are not competing for the same job, and a parent told to choose between them has been given a false choice.
And the dose turns out to have a shape
Chen and colleagues put smartwatches on 2,976 children in Shanghai, published in JAMA Network Open in 2024, and measured both time outdoors and the light intensity during it. Mean time outdoors was 90 minutes a day at a mean 2,345 lux.
Only one pattern was associated with less myopic shift: continuous episodes of at least fifteen minutes at 2,000 lux or above. Shorter scattered bursts, and time outdoors in dimmer conditions, were not.
Which is a more useful instruction than a daily minute count. Fifteen unbroken minutes in real daylight counts for something that three separate five-minute trips to the car does not.
What to take into the appointment
- Is it laser or LED, and what is the ANSI classification. The 2026 evaluation put four devices between 1.4 and 22,761 seconds to the same limit. The category name does not tell you where a given device sits in that range.
- What pupil diameter is the safety classification based on. This is the point the published disagreement turns on, and it is a fair question to ask of anyone recommending the treatment.
- What is my child's current axial length, and what is the plan for measuring it. Axial length in millimetres is the outcome that matters. A treatment aimed at it should come with a schedule for measuring it.
- What are the alternatives here, and what does each one do. Low-concentration atropine and specific spectacle and contact lens designs are also used for progression. They have their own evidence and their own drawbacks.
- How much daylight is my child actually getting. It is free, the evidence for onset is a cluster randomised trial, and it is the one thing on this page nobody disputes.