Pulmonology Topic

ARDS: the injured lung is small, not stiff.

The injured lung isn't stiff — it's small. A full breath lands in a lung the size of a child's, and that strain is what kills. Five cards: what counts as ARDS, the settings that protect it, the rescue ladder, what it costs the right heart, and how to get back off.

Reviewed September 2026 · verify against current guidelines

ARDS · 1 of 5

Is this ARDS?

Acute respiratory distress syndrome is a definition, not a test. Four criteria have to hold at once, and the fourth is the one that decides — the edema must not be explained by a failing heart.

All four, or it is not ARDS — Berlin definition

Within one week of a known insult, or new and worsening respiratory symptoms.
Bilateral opacities on chest radiograph or CT, not fully explained by effusion, collapse or nodules.
Not fully explained by cardiac failure or fluid overload. Objective assessment — usually echocardiography — if no risk factor is present.
Impaired oxygenation with PEEP or CPAP of at least 5 cmH₂O. The ratio then grades it.

Grading — PaO₂ / FiO₂ in mmHg, on PEEP ≥ 5; severity sets the mortality, not the diagnosis

P/F	Grade	Mortality
201–300	Mild	Around 27%. Includes patients on non-invasive support.
101–200	Moderate	Around 32%. Below 150 is where proning and paralysis were trialed.
≤ 100	Severe	Around 45%. The band where rescue therapy and ECMO get considered.

Bilateral infiltrates that are not ARDS

Cardiogenic edema. The one the definition explicitly excludes. Look for a dilated left atrium, a raised filling pressure, and improvement with diuresis alone.
Diffuse alveolar hemorrhage. Falling hemoglobin with no external bleeding; the tell is progressively bloodier lavage aliquots.
Acute eosinophilic pneumonia. Fever and infiltrates that steroids reverse in days. Lavage eosinophils above 25%.

Two traps — the ratio moves with the ventilator, not only with the lung: raising PEEP or FiO₂ changes it within minutes, which is why severity is graded after a period of standardized settings. And "not fully explained by" does not mean "absent" — a patient can have heart failure and ARDS at once, and treating only the heart leaves the other half untreated.

ARDS · 2 of 5

The settings that keep the lung.

The injured lung is small, not stiff. Only a fraction of it is open to ventilate, so a normal tidal volume is delivered into a baby-sized lung — and the harm is the strain, not the volume.

Driving pressure

ΔP = Vt / C_rs = P_plat − PEEP. Driving pressure is tidal volume normalized to the lung you actually have. It tracks mortality better than either term alone.

Set it in this order — each step has a number that tells you to stop

1 · Predicted body weight	From height and sex, never actual weight. A 1.6 m woman gets the same lung as a 1.6 m woman twice her mass.
2 · Tidal volume 6 mL/kg	Start at 6, range 4–8. ARMA cut mortality from 40% to 31% against 12 mL/kg.
3 · Plateau ≤ 30 cmH₂O	Measured on a 2-second inspiratory hold. Over 30, drop Vt by 1 mL/kg at a time, to a floor of 4.
4 · Driving pressure ≤ 15	If ΔP stays high at a legal plateau, the lung is smaller than the numbers suggest.
5 · PEEP from the table	Paired to FiO₂, targeting PaO₂ 55–80 mmHg or SpO₂ 88–95%.

The two PEEP tables — pick one and stay on it; the higher table suits moderate to severe disease

FiO₂	.30	.40	.50	.60	.70	.80	.90	1.0
Lower	5	5–8	8–10	10	10–14	14	14–18	18–24
Higher	5–14	14–16	16–20	20	20	20–22	22	22–24

What you accept to keep the settings

Permissive hypercapnia. A rising CO₂ is the price of a small tidal volume. Tolerate a pH down to about 7.20 before you trade lung protection for it; raise the rate first, up to 35.
Oxygenation that looks wrong. A saturation of 88–95% is the target, not a failure to correct. Chasing a normal number with FiO₂ and volume is how the ventilator does the injuring.

Two traps — a plateau under 30 does not make the settings safe: with a stiff chest wall the lung may be under-stretched, and with a compliant one it can be over-stretched at 28. And the plateau is only valid on a passive patient — a spontaneous effort during the hold makes the number meaningless.

ARDS · 3 of 5

When oxygenation fails anyway.

Two of these save lives, one does not, and one buys time. The order is the order of the evidence.

Escalate in this order — P/F < 150 triggers the first two

Intervention	Evidence	Verdict
Prone positioning	At least 16 consecutive hours. PROSEVA halved 28-day mortality, 33% to 16%.	Do it first
Neuromuscular blockade	ACURASYS found benefit at 48 hours; ROSE, with lighter sedation in controls, found none.	If needed
Inhaled vasodilator	Nitric oxide improves the P/F ratio without improving survival. A bridge, not a treatment.	Buys time
Recruitment maneuver	ART found higher 28-day mortality with stepwise recruitment, 55% against 49%.	Harm signal

Why turning the patient over works

Supine, the ventral lung is overdistended while the dorsal lung collapses with its perfusion preserved — ventilation and blood flow are mismatched. Prone, the dorsal lung recruits and inflation becomes more uniform, so ventilation reaches well-perfused dorsal lung.

Proning is a mortality drug, not an oxygenation one. Perfusion stays dorsal whichever way the patient lies; ventilation is what redistributes. The survival benefit comes from spreading strain across more open lung, so an oxygenation non-responder is not a treatment failure — judge the session by the 16 hours completed.

Two traps — late proning is a different intervention from early: PROSEVA enrolled within 36 hours, and the benefit has never been shown for a lung injured for a week. And the ART harm signal is against aggressive stepwise recruitment, not against raising PEEP off a table.

ARDS · 4 of 5

The right heart pays for it.

Every lever that protects the lung raises the load on the right ventricle. Acute cor pulmonale appears in roughly a fifth of moderate to severe ARDS, and the ventilator is part of the cause.

How lung protection becomes right heart failure

High PEEP, hypercapnia and hypoxic vasoconstriction → pulmonary vascular resistance rises → right ventricular afterload rises → the RV dilates and the septum shifts left → LV filling falls, and so does the output → a protected lung and a failing pump.

Four risk factors — none present, about 4% have cor pulmonale; all four, about three quarters

Threshold	Factor
≥ 18	Driving pressure, in cmH₂O. The one you can change tonight.
≥ 48	PaCO₂, in mmHg. Permissive hypercapnia has a cost, and this is it.
< 150	The P/F ratio. The same threshold that triggers proning.
Pneumonia	As the cause of the ARDS. Fixed, but it tells you to go looking.

When the lung cannot be protected at all — the EOLIA thresholds

P/F < 50	for more than 3 hours
P/F < 80	for more than 6 hours
pH < 7.25	with PaCO₂ ≥ 60 for more than 6 hours

EOLIA missed its endpoint — 35% against 46% at 60 days, not significant with 28% crossover to rescue ECMO. Read it as a threshold to telephone a center, not proof of benefit.

Two traps — a dilated right ventricle inverts the fluid logic: the next bolus becomes septal shift and less left-sided filling, so the treatment is unloading, not filling. And the transfer window closes early — a patient who has been on high settings for a week is a worse candidate than the same patient on day two.

ARDS · 5 of 5

Getting back off.

Liberation is not a single decision at the end. It is five things you start on day one and undo in order — and the longest delays are sedation and accumulated fluid, not the lung.

Undo it in this order — each step is checked daily, not once

1 · Lighten the sedation	Target RASS 0 to −2 with a daily interruption. Deep sedation on day one predicts longer ventilation and higher mortality.
2 · Settle the dyssynchrony	Watch the flow and pressure traces, not the number. Double triggering and reverse triggering are the two that injure.
3 · Take the fluid back	Once the shock has resolved. FACTT bought 2.5 more ventilator-free days with a conservative strategy, without costing survival.
4 · Screen, then trial	A spontaneous breathing trial of 30 to 120 minutes, on pressure support or a T-piece, once the screen below is passed.
5 · Extubate, then support	High-flow or non-invasive support straight after, in the patients at risk of failing.

The screen before the trial — a general guide; individual patients vary

Threshold	Criterion
≤ 0.40	FiO₂, with PEEP of 8 or less and adequate oxygenation.
< 105	Rapid shallow breathing index — rate over tidal volume in liters.
Awake	Following commands, with an effective cough and a manageable secretion load.
Off pressors	Or on a low, stable dose. Shock is the commonest reason the trial fails.

Why sedation is the step that has to come first. A deeply sedated patient cannot be screened, cannot cough, and cannot be assessed for delirium — so every step below it stalls. Depth is the rate-limiting variable, not the lung. It also hides the diaphragm. Under deep sedation the muscle wastes within days, and the effort you eventually need for the trial is no longer there to find.

Two traps — passing a trial is not the same as being ready to extubate — the airway, the cough and the secretions are a separate judgment from the lung. And an ARDS survivor leaves with problems the ventilator chart does not show: weakness, cognitive impairment and low mood are common a year out, so the handover matters as much as the extubation.

ARDS acute respiratory distress syndromeP/F PaO₂ / FiO₂ ratioPEEP positive end-expiratory pressureΔP driving pressurePplat plateau pressureRASS Richmond Agitation-Sedation ScaleECMO extracorporeal membrane oxygenation

Educational reference only — verify against local protocol.

Sources

Verify against current guidelines and local protocol before acting.

ARDS Definition Task Force. Acute Respiratory Distress Syndrome: The Berlin Definition. JAMA 2012;307:2526–2533.
ARDS Network (ARMA). Ventilation with lower tidal volumes as compared with traditional tidal volumes for ALI and ARDS. NEJM 2000;342:1301–1308.
Amato MBP, et al. Driving pressure and survival in the acute respiratory distress syndrome. NEJM 2015;372:747–755.
Guérin C, et al. (PROSEVA). Prone positioning in severe acute respiratory distress syndrome. NEJM 2013;368:2159–2168.
Papazian L, et al. (ACURASYS). Neuromuscular blockers in early ARDS. NEJM 2010;363:1107–1116.
National Heart, Lung, and Blood Institute PETAL Network (ROSE). Early neuromuscular blockade in ARDS. NEJM 2019;380:1997–2008.
Writing Group for the ART Investigators. Effect of lung recruitment and titrated PEEP vs low PEEP on mortality in ARDS. JAMA 2017;318:1335–1345.
Combes A, et al. (EOLIA). ECMO for severe acute respiratory distress syndrome. NEJM 2018;378:1965–1975.
Mekontso Dessap A, et al. Acute cor pulmonale in ARDS: prevalence, predictors, and clinical impact. Intensive Care Med 2016;42:862–870.
NHLBI ARDS Clinical Trials Network (FACTT). Comparison of two fluid-management strategies in acute lung injury. NEJM 2006;354:2564–2575.
Grasselli G, et al. ESICM guidelines on ARDS: definition, phenotyping and respiratory support strategies. Intensive Care Med 2023.

Downloads

Every card for this topic — carousels and tables, print-ready for the wards or for sharing.

ARDS: the injured lung is small, not stiff.

Carousel5 slides