Decompress on the physiology, not the X-ray. Then stratify the rib fractures you are tempted to send home.

The tension pneumothorax is a clinical diagnosis

The most dangerous sentence in the resus bay is “let’s get a film first.”

A tension pneumothorax is air that enters the pleural space on each breath and cannot get out. The pressure rises, the lung collapses, the mediastinum shifts, and the great veins returning blood to the heart are kinked and compressed. Cardiac output falls, and the patient arrests. This is a matter of minutes, not hours. Physiology

NICE is explicit that the diagnosis is made at the bedside, not on the film: “Use clinical assessment to diagnose pneumothorax for the purpose of triage or intervention.” NG39 1.3.1 And it sets the threshold for acting before any imaging in physiological terms: decompress before imaging “only if they have either haemodynamic instability or severe respiratory compromise.” NG39 1.4.1 The corollary is uncomfortable but correct: in the crashing patient you are allowed to be wrong. Decompressing a chest that turns out not to be under tension is a recoverable error; waiting for a radiograph in one that is, is often not.

The classic signs — tracheal deviation, distended neck veins, absent breath sounds, hyperresonance — are late and inconsistent, and tracheal deviation in particular is a pre-terminal sign. Do not wait to collect the full set. In the ventilated patient the presentation is different again: a sudden rise in airway pressures, a fall in saturations and blood pressure, and a chest that will not ventilate. Teaching

Needle or knife: why NICE asks for a thoracostomy

The needle you were taught to reach for fails in a large minority of adults.

For a generation the taught reflex was a wide-bore cannula in the second intercostal space in the mid-clavicular line. NICE now recommends something different: “use open thoracostomy instead of needle decompression if the expertise is available” NG39 1.3.5 pre-hospital, and in hospital, “perform chest decompression using open thoracostomy followed by a chest drain.” NG39 1.4.2 Why the change?

Because the needle often does not reach the pleura. Inaba and colleagues measured chest-wall thickness on the trauma CTs of 680 injured patients and found that a standard 5 cm decompression needle would be expected to fail in 42.5% of cases at the second space in the mid-clavicular line — the classic site — against 16.7% at the fifth space in the anterior axillary line, where the wall is on average 1.3 cm thinner. Inaba 2012 Even when it does reach, a narrow cannula kinks, blocks with blood or tissue, and dislodges. An open thoracostomy — a finger-sweep incision into the pleura — cannot fail for any of those reasons, which is exactly why it is preferred wherever the skill exists.

Inaba 2012, a CT-based anatomical study — it measures how often a needle could reach the pleura, not clinical outcomes. It is the reasoning behind the shift to a lateral approach and, where skill allows, a thoracostomy.

NICE gives no landmark and no tube size — those are your local guidance and the procedures tool. The teaching point is narrower and firmer: if you must needle, needle laterally, and treat a needle as a temporising measure, not the definitive treatment. The chest still needs a thoracostomy and a drain.

The open (“sucking”) chest wound

A dressing can save this patient or convert the injury into a tension pneumothorax — the difference is whether you watch.

An open chest wound large enough to draw air through the defect on inspiration impairs ventilation. NICE’s instruction is short: “cover the open pneumothorax with a simple occlusive dressing and observe for the development of a tension pneumothorax.” NG39 1.3.7 Two things are worth reading closely.

First, NICE says a simple occlusive dressing — not the older “three-sided” or “flutter-valve” teaching. The number of taped edges is not the point NICE makes. Any occlusive seal over a hole that is still letting air into the pleura can trap it, and a wound that was open (and therefore self-venting) can become a tension pneumothorax the moment you seal it. That is why the observation in the recommendation is the load-bearing word: whatever you put on, you watch for tension, and if it develops you decompress.

The military evidence has explored purpose-made vented chest seals, which are designed to let trapped air escape while sealing the wound. In a controlled swine model, a vented seal prevented tension pneumothorax across serial air injections up to 2 litres, whereas an unvented seal allowed tension to develop in every animal after about 1.4 litres of trapped air. Kheirabadi 2013 Both seals improved breathing immediately on application. Swine model This is animal data behind a device design, not a NICE recommendation — but it is the mechanism behind the warning: an occlusive dressing on the chest is something you commit to watching.

Blood in the chest: a haemorrhage problem wearing a breathing mask

The drain relieves the lung; it does nothing for the bleed.

A massive haemothorax presents as a shocked patient with a dull, quiet hemithorax. It is easy to treat it as a pneumothorax’s louder cousin — put in a drain and move on — but its danger is haemorrhagic. The hemithorax can hold several litres, and inserting a chest drain into it without replacing the blood and calling the surgeon can drop a marginally compensated patient into arrest.

So it runs down the haemorrhage pathway: activate the major haemorrhage protocol on physiology, resuscitate restrictively with blood in a fixed ratio, and give tranexamic acid as soon as possible if within 3 hours of injury. NG39 1.5.4–1.5.5, 1.5.13, 1.5.24 The full engine is in the major haemorrhage module; the anatomical decision here is only that a large or continuing drain output is a thoracic-surgical question, not an ED one.

Flail chest and pulmonary contusion: fear the lung, not the paradox

The moving segment is a marker; the injury that kills is underneath it.

A flail segment — two or more ribs each fractured in two places, so a portion of the chest wall moves paradoxically — looks dramatic. But the paradoxical movement itself is rarely the immediate problem. The danger is the pulmonary contusion beneath it and the respiratory failure that develops over the following hours as the bruised lung becomes oedematous and stiff. A patient who looks adequate on arrival can tire and fail late. Teaching

Two consequences follow. First, this is an admission with close respiratory monitoring, aggressive analgesia and early escalation to non-invasive or invasive ventilation if the patient tires — not a wire splint and a ward bed. Second, be careful with fluid: over-resuscitating a contused lung with crystalloid worsens the oedema, one more reason the whole pathway is built on blood and restraint rather than saline. Teaching

Where surgery fits is narrower than enthusiasm sometimes suggests. In a single-centre randomised trial, operative rib fixation for severe flail chest shortened intensive-care stay and cut the need for non-invasive ventilation after extubation, but did not improve lung function at three months or quality of life at six. Marasco 2013 Surgical stabilisation is a real option for selected patients, but it is a specialist decision made after admission — not an ED one, and not a reason to under-treat the pain and the lung in the meantime.

The rib-fracture patient who looks fine

The quiet danger of the department: a comfortable patient with a small number of fractures who deteriorates on day three.

NG39 runs out of advice for exactly the patient the department sees most: comfortable, breathing steadily, with a handful of rib fractures and no life threat. There is no NICE stratification for admit-or-discharge here — and it is a decision that is got wrong in both directions, because the patient looks well at the moment you are deciding and deteriorates later, when the pain stops them coughing, secretions pool, and pneumonia sets in.

The STUMBL score (Battle 2014) was built for this gap. It is a UK-derived, externally validated prognostic model that combines age, the number of rib fractures, chronic lung disease, pre-injury anticoagulation and oxygen saturation, and it discriminated complications well in both its derivation and validation cohorts. Battle 2014 It is not NICE guidance; use it alongside your local chest-wall-injury pathway, not instead of clinical judgement — the calculator is on the decision tool tab.

Three things the score encodes are worth saying in plain words:

• Age is a heavy variable. An older patient with three rib fractures can score into the admit range on age and anticoagulation alone. Three fractures in an 80-year-old is not the same injury as three in a 25-year-old.
• Analgesia is the treatment, not the comfort. NG39 asks for intravenous morphine first-line, titrated, with ketamine second-line. NG39 1.7.4–1.7.6 Regional techniques — a serratus anterior or erector spinae plane block, or an epidural for multiple fractures — sit beyond NG39 and are a decision for the acute-pain or anaesthetic team. Good analgesia is what lets the patient breathe deeply and cough; it is what prevents the late pneumonia the score is predicting.
• A small pneumothorax found on CT is not automatically a drain. In a randomised trial of ventilated trauma patients with occult pneumothoraces, observation carried a similar risk of respiratory distress to immediate drainage, and chest-drain complications were common. Kirkpatrick 2013 In the stable, spontaneously breathing patient a small occult pneumothorax is often observed with a repeat film — a decision for the trauma team, not a reflex.

The aorta and the pericardium

Two injuries the primary survey can miss, and where CT and echo come in.

Blunt thoracic aortic injury follows a rapid-deceleration mechanism — a high-speed frontal or side impact, a fall from height. The chest radiograph may show a widened mediastinum, but that is a prompt to image, not a diagnosis; the definitive test is CT angiography, which aligns with NG39’s immediate-CT recommendation for the stable or responding adult. NG39 1.4.5 For a confirmed injury, NICE is specific: “use an endovascular stent graft in patients with blunt thoracic aortic injury” NG39 1.5.43 — a vascular or cardiothoracic decision, with heart-rate and blood-pressure control directed by the receiving team in the meantime.

Cardiac tamponade follows penetrating praecordial injury, or severe blunt force. Blood in the pericardium compresses the heart; the eFAST or echo shows it. This is a surgical emergency whose definitive treatment is in theatre — or a resuscitative thoracotomy in the peri-arrest penetrating patient — not a decision this module makes. As with the chest, a negative eFAST does not exclude injury. NG39 1.3.3 Traumatic cardiac arrest and resuscitative thoracotomy are their own algorithm; this module deliberately stops at recognition.

What was dangerous about each plan

1. Case A — “a quick film first.” The unstable, severely compromised patient with a quiet hemithorax and a rising respiratory rate has a tension pneumothorax until proven otherwise, and NICE decompresses that patient before imaging. The film is the delay that kills. Decompress — ideally a thoracostomy, or a lateral needle as a bridge — then image. (Chapters 1–2)
2. Case B — “oral analgesia and discharge.” This is the quiet failure. She is 78, anticoagulated, with four fractures — a STUMBL profile that lands in the admit range on age and apixaban alone, before the fractures are counted. She is comfortable now and will be worse on day three, when the pain stops her breathing deeply. Score her, treat the pain properly, and admit rather than discharge on the day-one picture. (Chapter 6)

The pattern behind both: the loud injury tempts you to over-image, and the quiet one tempts you to under-admit. Neither the film nor the patient’s comfort at the moment of decision is the thing to trust.

What this is, and what it is not

The Decision tool implements NICE NG39 and quotes it verbatim, and reproduces the STUMBL score with attribution. The Teaching and Evidence tabs are the author’s synthesis of NG39 and the primary studies behind it, written to explain the tool, not to extend it. Nothing in them changes what the tool tells you to do. Where they go beyond NICE — the lateral needle site, the haemothorax surgical thresholds, the flail-chest physiology — the source is named and tagged.

Teaching and Evidence were added in September 2026 and have had no clinical review by anyone other than their author.