Four eye emergencies where minutes decide the vision.

An eye injury severe enough to threaten sight rarely arrives on a quiet, isolated patient. In the TARN major-trauma registry for England and Wales, a concomitant head injury was present in 86.6% of patients who had an ocular injury TARN cohort Bashir 2024. The eye is a marker for force transmitted to the head and neck, and the reverse trap is just as real: the swollen, painful eye distracts you from the brain injury behind it. Assume a head and cervical-spine injury until each is cleared on its own terms — the CT head & C-spine module runs those decisions.

Two physical signs earn special mention. Bilateral periorbital haematoma — “panda” or “raccoon” eyes — points to a base-of-skull fracture, not to a local blow. And the mechanism itself carries safeguarding weight: RCEM Learning explicitly asks you to consider domestic violence and non-accidental injury in every ocular trauma. Most traumatic hyphaema is paediatric (about 70%, typically ball or projectile sport), so a hyphaema in a small child, or any eye injury whose story does not fit the findings, is a safeguarding event until proven otherwise.

The one examination that must happen before anything is done to the eye is visual acuity in both eyes, recorded — Snellen if you can, otherwise counts fingers, hand movements, perceives light, or no perception of light. It is the single most important prognostic and medicolegal number you will write, and the moment you start treating (irrigating, shielding, giving drops) you lose the chance to record the untreated baseline. Document the mechanism carefully for the same reason.

A chemical splash is the one presentation where irrigation comes before history, examination or investigation. RCEM Learning is unambiguous: chemical eye injuries are “true time-critical emergencies where immediate irrigation takes priority over history, examination or investigations,” and the British Burn Association guideline adds that you do “not delay immediate irrigation of the eye for detailed assessment of patient or acquiring a particular irrigation fluid, regardless of delay in presentation” BBA 2018. Anaesthetise, remove contact lenses and loose particulate matter, and run fluid — Hartmann’s, saline, an amphoteric solution, or plain water if nothing else is to hand — keeping the unaffected eye uppermost so you do not wash the injury across it.

Alkali is worse than acid, and this is chemistry, not opinion. Alkalis (cement, plaster, oven and drain cleaner, ammonia, lime) saponify cell membranes and penetrate deeply, going on injuring long after the splash. Acids tend to coagulate surface protein, which forms a partial barrier and limits the depth — with the notable exception of hydrofluoric acid, which behaves like an alkali. That is why the volume and duration of irrigation are measured against pH, not the clock: keep going, checking the inferior fornix, until the pH is stable in the 7.0–7.2 target range (RCEM Learning; mean normal tear pH is 7.6, per BBA), and resume if it drifts.

The white, comfortable eye is the dangerous one

The instinct that a red, weeping, painful eye is worse than a pale, quiet one is exactly inverted for chemical burns. Pain comes from an intact, innervated ocular surface. A white eye with perilimbal blanching is not a mild burn that has settled — it is a severe one that has killed the surface nerves and, more importantly, the limbal stem cells that live in the ring of vessels around the cornea. Roper-Hall grades the burn on exactly these two axes: corneal clarity, and how many clock-hours of limbus have been rendered ischaemic. The more blanched and painless the limbus, the worse the prognosis, because without limbal stem cells the cornea cannot re-epithelialise and scars over. The graded photographs are in the reference panel of the decision tool.

Two situations interrupt the “irrigate first” reflex, and both are covered in the next chapters: a suspected open globe (irrigate only gently, never with a Morgan lens, and get ophthalmology — decompressing or pressing on a ruptured globe extrudes its contents), and the presence of large solid particulate (lime, cement) which must be physically swept from the fornices or it keeps releasing alkali under the lid.

An open globe is a full-thickness wound of the cornea or sclera — in the Birmingham Eye Trauma Terminology, a rupture (blunt force bursting the wall from within) or a laceration (a sharp or high-velocity object cutting in). The clinical point is that once the wall is breached, the intraocular contents are held in only by their own position, and any rise in pressure on the globe — your finger, a tonometer, an ultrasound probe, a forced lid retraction, a Valsalva from vomiting — can push the iris, lens, vitreous or retina out through the wound. That extrusion is usually the moment vision is lost for good.

So the whole management is built around doing no further harm: “to avoid extrusion of intraocular contents, refrain from any unnecessary manipulation of the globe (including measuring intraocular pressure)” (RCEM Learning). Cover with a rigid shield, never a patch — a patch presses, a shield bridges. Keep the patient nil by mouth for theatre, give analgesia and antiemetics to head off the Valsalva of pain and nausea, leave any protruding foreign body exactly where it is, start prophylactic systemic antibiotics against endophthalmitis, and check tetanus. Ultrasound is contraindicated when an open globe is suspected; the imaging is CT in thin (1.5–2 mm) cuts if a foreign body is possible.

The signs, and the mechanism that outranks them

The signs are a full-thickness wound, a shallow or flat anterior chamber, iris (uveal) prolapse, a teardrop or peaked pupil pointing toward the wound, a positive Seidel test, and a deep lid laceration through the margin. But the most dangerous open globe is the one with no signs at all: a high-velocity mechanism — hammering metal on metal, grinding, glass, a power tool, a blast — means a retained intraocular foreign body until proven otherwise, and a small self-sealing entry wound can leave the eye comfortable with near-normal vision. “A missed intraocular foreign body with ferrous component will lead to late vision loss due to siderosis” (RCEM Learning). A comfortable eye and a good acuity do not exclude it; the history does the work the examination cannot.

The orbit is a cone of bone open only at the front, closed off at the rim by the tarsal plates and the canthal tendons. Bleeding behind the globe — most often after blunt facial trauma or orbital fracture — has nowhere to go, so pressure rises, the globe is pushed forward (proptosis) and the optic nerve is stretched and its blood supply compressed. That is orbital compartment syndrome (OCS), and retrobulbar haemorrhage is its commonest cause. Left alone, it produces optic-nerve ischaemia and irreversible blindness within a couple of hours.

The single most important sentence in this chapter: OCS is a clinical diagnosis, and the treatment is done at the bedside, not in the scanner. “Do not delay lateral canthotomy in known cases of retrobulbar hematoma” waiting for ophthalmology, and it should ideally be done within 2 hours of onset to avoid optic-nerve ischaemia (RCEM Learning). The picture that makes the diagnosis is a tense, proptosed globe with resistance to retropulsion, a relative afferent pupillary defect or falling acuity, painful restricted eye movements, and a high intraocular pressure. An IOP of ≥ 40 mmHg is a soft threshold for intervention StatPearls — but the whole picture decides, and you do not withhold decompression from a proptosed, blind, rock-hard eye because a pressure number is missing or because the CT has not happened.

Where the two-hour clock comes from

The “2 hours” is not a trial result; it is a rounded safety margin extrapolated from the retina’s tolerance to ischaemia. In Hayreh and Weingeist’s classic primate experiments, the retina suffered irreparable damage after 105 minutes of complete central-retinal-artery occlusion but recovered after 97–98 minutes experimental Hayreh 1980. That is the origin of the number, and two honest caveats follow from it. First, it is the retina’s tolerance in a clean arterial-occlusion model, not a guarantee about a compressed optic nerve in a bleeding orbit. Second, case reports describe good visual recovery after decompression well beyond two hours Ochi 2025 — so a patient presenting at three or four hours is not a lost cause you can deprioritise. Treat the clock as a reason to hurry, never as a reason to give up.

The decompression is lateral canthotomy with inferior cantholysis: release the lower limb of the lateral canthal tendon so the globe can move forward and the pressure falls. The full step-by-step, equipment and DOPS levels are in the lateral canthotomy procedure guide. It is a procedure a trained ED clinician can and should do without waiting — but it comes with one absolute contraindication, which is the whole of the next chapter.

Chapters 3 and 4 pull against each other, and the patient in the case sits exactly on the fault line. OCS says: cut now. The open globe says: touch nothing, and never decompress. Decompressing a ruptured globe with a canthotomy does precisely what you spend the open-globe chapter avoiding — it drops the resistance and extrudes the contents. A suspected open globe is therefore “a contraindication to the lateral orbital canthotomy” StatPearls.

So before you cut for a “compartment syndrome,” the one discriminator you must satisfy yourself on is: is the globe pushed out, or sunk in? Orbital compartment syndrome pushes the globe forward — proptosis, a tense globe resisting retropulsion. A ruptured globe more often sits back — enophthalmos — because the wall has lost volume, and it may show a full-thickness wound or a peaked pupil. Proptosis with a tense orbit is a canthotomy; enophthalmos, or any sign of rupture, is a shield-and-refer, not a cut.

There is a subtler version of the same trap even when there is no rupture. A high IOP on its own, without proptosis or a tense non-retropulsible globe, is more likely acute glaucoma or a hyphaema than OCS — and a canthotomy relieves the retrobulbar compartment, not the pressure inside the eye. Lower that pressure medically and involve ophthalmology; do not reach for scissors because a single number is high. And a coexisting hyphaema can keep the IOP, the RAPD and the acuity abnormal even after a technically perfect cantholysis, which you must not misread as a failed decompression.

A hyphaema is blood layering in the anterior chamber, usually from blunt trauma tearing iris or ciliary-body vessels. It looks dramatic and is often benign — but two things make it an emergency, and neither is the visible blood. The first is a rise in intraocular pressure as the blood blocks the drainage angle. The second, and the one that changes management, is the rebleed at day 2–5: a fresh haemorrhage into an eye whose vessels are already damaged, occurring in roughly 30% of cases and carrying the worse prognosis. Almost everything you do in the ED is aimed at that second bleed.

Sit the patient up, head of bed to at least 30°, so the blood settles inferiorly below the visual axis and you can grade the level; place a rigid shield. Grade by how much of the chamber fills (microhyphaema through grade IV, the total “eight-ball”); higher grades carry higher rates of raised IOP. Avoid NSAIDs, aspirin and anticoagulants until it resolves, and control pain and nausea aggressively — vomiting spikes the IOP and can restart the bleeding. Measure IOP only once an open globe is excluded, and put fluorescein on before the tonometer.

Sickle changes the whole risk calculation

Sickle cell disease or trait deserves specific attention, and it is easy to forget in a patient who has never had a crisis. The relatively hypoxic, acidotic anterior chamber makes red cells sickle; sickled cells jam the trabecular meshwork, so pressure rises fast and is harder to control — and the carbonic-anhydrase inhibitors (acetazolamide, dorzolamide) you would normally reach for to lower IOP worsen the sickling. So: test sickle status if it is unknown, treat the surgical thresholds as lower, and involve ophthalmology early. This is one of the clearest examples on the page of a treatment that is right for the general case and actively harmful for a specific patient.

Outpatient management is safe only with all of: grade ≤ II, normal IOP, no sickle disease or trait, no coagulopathy, and reliable daily ophthalmology review. Anyone failing any one of those is admitted, and every hyphaema gets an ophthalmology consultation.

This module screens for four sight-threatening emergencies. When none of them is present, the correct conclusion is “not a cleared eye,” not “discharge.” A normal-looking eye after trauma still needs a full, documented examination: visual acuity in both eyes, pupils and RAPD, lids and fornices everted, cornea with fluorescein and a Seidel test, the anterior chamber, the fundus and red reflex, and eye movements. The injuries this page deliberately leaves out — corneal abrasion, subtarsal foreign body, arc-eye keratitis, subconjunctival haemorrhage, and the definitive management of an orbital “blow-out” fracture — are worked up once the emergencies are excluded, not instead of them.

Three housekeeping items apply to any injured eye and are easy to drop under pressure: update tetanus, examine and protect the other eye, and record the mechanism in enough detail to satisfy a later medicolegal or safeguarding question. Where the mechanism was assault or the story does not fit, follow your domestic-violence and non-accidental-injury pathways, and consider photography with consent.

Finally, know the limits of this page. It does not give a canthotomy competence, it does not give drug doses, and it is not a substitute for your local ophthalmology and eye-casualty pathway, which takes precedence over anything written here. It is a map of where the traps are — the decision at the bedside is still yours.

Each of these is a real, recurring error — most of them a correct general reflex applied to the wrong eye.