The Blurry Side of the Chamber: What Hyperbaric Oxygen Does to the Lens

$Stylised clinical illustration of a human eye in cross-section: the crystalline lens drawn as concentric refractive-index bands with light rays from a distant point converging in front of the retina to show a myopic focus error, beside faint silhouette outlines of a hyperbaric hood and an oronasal mask face-piece on a deep slate-navy background$

Thirty sessions into a course of hyperbaric oxygen treatment for a non-healing diabetic foot ulcer, many patients report the same thing before anyone asks them about it. Road signs have gone soft at the edges. The television has drifted closer. The glasses that were fine three months ago now feel half a step weak.

Chamber staff know this symptom well. It is a myopic shift — a change in the eye’s focusing power towards nearsightedness — and it has appeared in the literature since the 1970s, in complication reviews and in the last paragraph of many consent conversations. What has been missing is a sense of scale, and of what the shift depends on. A systematic review published in Undersea and Hyperbaric Medicine supplies both, and its most interesting result is not that the side effect happens but that its size is partly a function of the equipment.

A Side Effect With a Signature

Knut Evanger and colleagues at Haukeland University Hospital in Bergen screened 153 reports and found eight studies that met their criteria: published in English, full text, with the oxygen delivery method, pressure, session count and a conventional refraction measurement all specified. Patients with artificial lenses were excluded, as were single case reports. The eight studies together covered 263 patients.

Within that small pool the pattern is orderly, and the numbers move together in both directions — how far the refraction shifts, and what proportion of patients cross the half-dioptre threshold that counts as a clinically detectable change.

After twenty treatments in a multiplace chamber using an oronasal mask, the mean shift was 0.4 dioptres and 53 per cent of patients reached at least half a dioptre. Eighteen treatments in a monoplace chamber produced 0.6 dioptres and 70 per cent. Thirty monoplace treatments produced 0.95 dioptres and 91 per cent. The top of the range came after forty treatments using a hood in a multiplace chamber: 1.6 dioptres, and every patient in that group crossed the threshold.

The reviewers’ conclusion is deliberately flat. Both the magnitude and the incidence of myopic shift increase with the number of sessions, and they rank in the order multiplace-mask, then monoplace, then multiplace-hood.

One caveat belongs beside those numbers: means hide the spread. In one of the studies the median change among affected eyes was around two-thirds of a dioptre, while the worst single eye moved by more than two.

The Lens Is the Target, Not the Cornea

Why should breathing oxygen at pressure change where light focuses? The same group answered that question with imaging rather than speculation. In a 2015 study in Optometry and Vision Science, patients underwent Scheimpflug imaging of the crystalline lens after the first day of treatment and again after nineteen days.

The optical density of the lens nucleus fell — by 2.8 units in the circular measurement area and 2.2 units in the oval one — and backward-scattered light decreased. The mean myopic shift was 0.58 dioptres. Everything else stayed put: lens curvature, lens thickness, anterior chamber depth, corneal power, intraocular pressure and the axial length of the eye. The eye’s focusing power had changed without any of the eye’s dimensions changing, which points to the refractive index of the lens substance itself, altered by oxygen-driven oxidation of lens proteins.

A companion study published four years earlier sharpened the same point by comparing two kinds of eyes. Sixteen patients with their natural crystalline lenses and six who had already had cataract surgery with an intraocular implant were treated at 240 kPa for 90 minutes a day, five days a week, for twenty days. A shift of at least half a dioptre appeared in 81 per cent of the phakic patients’ eyes, with a median of about 0.6 dioptres and a maximum beyond two. Among the pseudophakic patients there was no shift at all — the median changes were a fraction of a dioptre in the opposite direction.

Patients with lens implants are, in other words, effectively exempt — a useful sentence to have ready in a clinic, and one that makes the mechanism hard to argue with: remove the crystalline lens and the phenomenon disappears.

Hoods, Masks and a Randomised Difference

The ordering the review reports is not a statistical artefact of mixing studies, because one of the included trials tested it directly. Michael Bennett and colleagues randomised 120 patients — mean age 57.6, four-fifths of them men — to receive their oxygen by hood or by oronasal mask across a course of twenty and then thirty treatments.

After twenty sessions the hood group had shifted by 0.92 dioptres and the mask group by 0.52, a difference of 0.40 dioptres (95 per cent confidence interval 0.22 to 0.57). After thirty sessions the gap had widened: 1.25 dioptres against 0.63, a difference of 0.62 dioptres (0.39 to 0.84). Both comparisons were significant at p<0.0001. Recovery was also slower and less complete among hood users at four and twelve weeks. The authors’ recommendation follows from their own data: use an oronasal mask, in an air environment, when possible.

The physical reason is not mysterious once the hardware is pictured. In a multiplace chamber the chamber atmosphere is compressed air and the oxygen arrives through a delivery device: a hood floods the whole head, eyes included, with a continuous stream of near-pure oxygen, while an oronasal mask confines the gas to the airway and lets the eyes sit in chamber air. Two patients on identical prescriptions, breathing the same gas at the same pressure for the same ninety minutes, do not give their lenses the same exposure.

Monoplace chambers land between the two, and the review offers no mechanism for why. That is honest of it. It also means the practical takeaway is narrower than “masks are better”: the eye’s oxygen dose is a design variable, in the same way that inspired partial pressure and time are.

One further wrinkle complicates the tidy story that oxygen alone is responsible. In 2002, a Japanese group examined twelve men who had worked in a pneumatic caisson — compressed air, with no oxygen enrichment — for up to eleven weeks. Seven of the nine who were myopic at that first examination had become so after starting the job, and six months after the work ended the mean refractive change had swung significantly back towards hyperopia. The authors called for refraction testing as part of industrial health control, and a letter in The Lancet the following year flagged the same occupational pattern. Pressure by itself appears able to move the lens, so “hyperoxic myopia” may not be a complete explanation.

Recovery, and the Cataract Question

Most of the shift is transient, and the best longitudinal data on recovery come from a Swedish cohort of 29 patients treated with a standard forty-session protocol. A shift of at least half a dioptre occurred in 45 of 58 eyes — 77.6 per cent. Median refraction changed by 0.75 dioptres in the right eye and 0.66 in the left by the end of treatment. Twelve weeks after the course, refraction had returned essentially to baseline, with one small residue: a median persisting change of about a quarter of a dioptre in the left eye that was statistically significant and clinically trivial.

The same study found something less dismissible. Grading the lens with the LOCS III system showed modest increases in nuclear colour and opalescence during the course, alongside small reductions in corneal thickness and intraocular pressure. The authors’ summary was cautious but pointed: the transient shift was probably a refractive index effect, and reduced lens transparency was a potential consequence of treatment.

That is the thread that leads to cataract. In 2007, two American authors reported a 49-year-old woman who received 48 treatments at 2.5 ATA for ninety minutes for chronic refractory osteomyelitis of the sacrum and a repeatedly failing sacral myocutaneous flap. She was myopic before treatment and had no cataract. Eight weeks after finishing, her myopia had worsened and nuclear cataract changes were visible in both eyes; diabetes, corticosteroid use and ultraviolet exposure were all excluded. The report notes that cataract had previously been described mainly after prolonged courses of 150 or more treatments, which is precisely why a 48-session case was worth publishing. One case proves nothing about causation. It does show that the question is open.

For context on how the field normally treats this, consider a large complication audit: 782 patients and 11,376 sessions in a multiplace chamber at 240 to 250 kPa, with myopia and cataract growth listed among the recognised ocular effects next to middle-ear trouble — more than 17 per cent of patients reported ear discomfort and 3.8 per cent had verified barotraumatic lesions — and four generalised seizures from central nervous system oxygen toxicity. The 2026 side-effects chapter in the same journal treats refraction change the same way: expected, quantifiable, worth explaining.

Against that, the negative evidence deserves its place. A 2026 expedition study followed twelve closed-circuit rebreather divers through 173 dives over 31 days — mean maximum depth 38 metres, mean dive time 101 minutes. Weekly visual-acuity checks fluctuated without any overall trend, and the authors concluded that reasonable oxygen exposure within recognised limits does not appear to significantly affect acuity, though a Snellen chart is a blunter instrument than refraction and a month of working dives is not forty clinical sessions. The same team’s 2024 case report describes a 54-year-old diver whose refraction moved from -3.75 and -5.75 to -5.50 and -7.75 dioptres after 75 mostly-enriched-air dives on a 22-day liveaboard; the PRISMA review published with it screened 22 articles and found the data too heterogeneous to derive any exposure limit that would prevent the effect.

What to Do With a Number Like This

None of this changes who should be treated. For an unconscious carbon monoxide patient or a failing graft, a reversible half-dioptre that resolves within three months is not a reason to withhold therapy.

What it does change is the monitoring conversation. Refraction is cheap, non-invasive and repeatable, and for any course longer than a couple of weeks it arguably belongs next to the ear examination on the checklist rather than in a footnote. Patients should be told not to buy new glasses mid-course and to expect things to settle afterwards; optometrists should be told their patient is spending ninety minutes a day in a chamber, because a sudden myopic change in a 60-year-old has a different differential in that context. For commercial divers, pilots, caisson and chamber workers the stakes are sharper still, since a forced half-dioptre of myopia is a fitness-to-work question rather than an inconvenience.

Where the side effect sits in the paperwork is worth stating precisely. The FDA’s letter to health care providers of 25 August 2025 is a device-safety document: it describes reports of fires causing serious injuries and deaths, and asks for adherence to manufacturer instructions, staff training, patient monitoring, cotton clothing and the exclusion of prohibited electrical and static devices. It classifies hyperbaric devices as Class II, cleared through the 510(k) pathway under Product Code CBF. It says nothing about refraction, and it does not need to.

Medicare’s national coverage determination is narrower still. “Program reimbursement for HBO therapy will be limited to that which is administered in a chamber (including the one man unit) and is limited to the following conditions,” it reads, then names fifteen — decompression illness, gas embolism, carbon monoxide intoxication, crush injuries, chronic refractory osteomyelitis, the “preparation and preservation of compromised skin grafts (not for primary management of wounds)”, and others. No ocular diagnosis appears on the list, and the UHMS indications list in its fifteenth edition contains no ocular entry either. A myopic shift is a hazard that accompanies a covered treatment; it is not an indication for one.

The final gap is the one the wellness market occupies. The review’s evidence covers 18 to 40 sessions at clinical pressures. Nobody has published a refraction dataset for the 1.3 ATA lounges where mild hyperbaric sessions are sold, so the claim that such sessions carry no risk of this effect rests on nobody having looked rather than on anyone having checked.

Put together, the picture is not alarming, and it is not tidy either. Nearsightedness after hyperbaric oxygen is common, usually temporary, occasionally persisting at the margins, and — the genuinely new part — sensitive to how the oxygen is delivered. Behind it sit eight studies, 263 patients, at least one case report of a cataract nobody predicted, and a small research community in which the authors of the review are also among the authors of the evidence it reviews. The eye, it turns out, keeps a dose record. Someone should be reading it.

Sources

← Back to blog