ARDS

in review 6 min read Updated 2026-10-29
ARDS

Acute (≤1 week) hypoxaemic failure, bilateral opacities, not explained by heart failure or fluid alone, with a predisposing insult. Treat the cause, then: tidal volume 6 mL/kg predicted body weight, plateau ≤30 cmH₂O (strong), higher PEEP in moderate–severe ARDS (conditional; ESICM no recommendation), prone ≥12 h/day in severe ARDS (ATS strong); ESICM: ≥16 h once P/F stays <150 after PEEP is optimised (SSC conditional). Conditional: corticosteroids, NMBA only for early severe ARDS (already deeply sedated, or dyssynchrony despite sedation), VV-ECMO referral for P/F <80 despite all this. Avoid prolonged recruitment manoeuvres and HFOV.


definition (global, 2024)

criteria
allrisk factor (pneumonia, non-pulmonary sepsis, aspiration, trauma, transfusion, [[acute pancreatitis
oedemanot primarily cardiogenic or fluid overload, but ARDS can coexist with either if a risk factor is present
non-intubatedP/F <300 or S/F <315 (SpO₂ <97%) on HFNC >30 L/min or NIV/CPAP ≥5 cmH₂O
intubatedmild P/F 200–300 · moderate 100–200 · severe <100 (S/F equivalents 315 / 235 / 148)
  • what’s new since Berlin: ARDS can be diagnosed before intubation (on HFNC), from SpO₂/FiO₂ when no arterial gas, and from ultrasound in place of imaging
  • S/F only works when SpO₂ <97%: on the flat part of the curve, saturation stops reflecting PaO₂
  • severity grades guide treatment: proning, NMBA and ECMO apply to moderate–severe or severe only

Workup that changes management: echo (LV function, filling pressures → cardiogenic contribution), cultures and respiratory viral/atypical panel, BAL if immunocompromised or no clear cause. Bilateral infiltrates without a risk factor → think of mimics that need a different treatment: diffuse alveolar haemorrhage, acute eosinophilic pneumonia, acute interstitial pneumonia, organising pneumonia, drug toxicity.


ventilation

The lung is small, not stiff (“baby lung”): only the aerated part takes each breath, so a normal tidal volume overdistends it. Ventilator-induced injury comes from volume (overdistension) and from opening and closing (atelectrauma). Hence: small breaths, enough PEEP to keep the lung open, and repeated recruitment manoeuvres avoided.

settingtarget
modevolume control is usual
tidal volume6 mL/kg predicted body weight (range 4–8) — from height and sex, never actual weight
plateau pressure≤30 cmH₂O — drop Vt toward 4 mL/kg if above
driving pressure (Pplat − PEEP)aim <15 cmH₂O — observational association with survival, not a tested target
PEEPhigher PEEP/FiO₂ strategy for moderate–severe (conditional); not in mild (possible harm)
oxygenationSpO₂ 88–95% / PaO₂ 55–80 mmHg
CO₂permissive hypercapnia, pH ≥7.25 (lower tolerated if needed); raise rate before Vt (watch auto-PEEP)
  • ARMA (2000): 6 vs 12 mL/kg cut mortality from 40% to 31% (NNT ~11). The founding trial of the whole approach
  • driving pressure was the variable most strongly tied to survival in a pooled reanalysis (Amato, NEJM. 2015); useful for judging a PEEP change (if driving pressure falls, the PEEP recruited lung)
  • higher PEEP: IPD meta-analysis showed a survival gain in moderate–severe ARDS and possible harm in mild. ESICM found no effect overall and makes no recommendation; ATS 2024 and SSC suggest it. Default to the higher PEEP/FiO₂ table in moderate–severe ARDS, watching BP and plateau pressure. FiO₂ ≥0.8 on PEEP 8 is under-PEEPed on either table

ARDSNet PEEP/FiO₂ tables (step along the row to keep SpO₂ 88–95%; PEEP ≥5)

FiO₂0.30.40.50.60.70.80.91.0
lower PEEP55–88–101010–141414–1818–24
higher PEEP5–1414–1616–20202020–222222–24
  • no prolonged recruitment manoeuvres (strong, all three guidelines): ART (2017) had higher mortality with recruitment + incremental PEEP titration (barotrauma, haemodynamic collapse). Brief manoeuvres are not routine either
  • no HFOV (strong): OSCILLATE (2013) stopped early for harm
  • sepsis-associated hypoxaemic failure without ARDS: 6–8 mL/kg is fine (lower adds sedation burden for no benefit)
  • conservative fluid once out of shock: more ventilator-free days, no mortality difference (FACTT (2006))

Ventilator modes, pressures and troubleshooting: mechanical ventilation.


escalation in moderate–severe ARDS

flowchart TD
  A[Intubated ARDS<br/>Vt 6 mL/kg PBW · Pplat ≤30] --> B[Optimise PEEP<br/>higher PEEP/FiO₂ table if moderate–severe]
  B --> C{P/F still &lt;150<br/>after 12–24 h stabilisation?}
  C -->|yes| D[Prone ≥16 h/day<br/>repeat daily until P/F ≥150 supine]
  C -->|no| E[Continue LPV · treat cause · conservative fluid]
  D --> F{Dyssynchrony despite<br/>sedation + vent changes?}
  F -->|early severe, &lt;48 h| G[NMBA bolus or ≤48 h infusion]
  D --> H{P/F &lt;80 or pH &lt;7.25 with PaCO₂ ≥60<br/>despite all, ventilated &lt;7 days?}
  H -->|yes| I[Call ECMO centre early]
  • prone (ATS strong for severe ARDS; ESICM strong for P/F <150 with PEEP ≥5 after stabilisation; PROSEVA also required FiO₂ ≥0.6; SSC conditional): PROSEVA (2013) cut 28-day mortality from 33% to 16% with sessions of ~17 h/day. Proning works by making ventilation and stress more even across the lung, not by the oxygenation gain. Optimise PEEP first, but don’t let PEEP titration delay proning. Contraindications are relative: unstable spine, open abdomen, raised ICP, recent sternotomy
  • neuromuscular blockade (ATS conditional for early severe ARDS; ESICM strong against routine continuous infusion): ACURASYS (2010) showed a benefit against deep sedation; ROSE (2019) showed none against light sedation. Use for early (<48 h) severe ARDS with dyssynchrony that sedation and vent changes don’t fix, as boluses or ≤48 h infusion
  • VV-ECMO (ATS conditional, ESICM strong for patients meeting EOLIA criteria): EOLIA (2018) showed 35% vs 46% mortality, not significant, but with 28% crossover; meta-analysis favours ECMO. Referral triggers (EOLIA): P/F <50 for >3 h, P/F <80 for >6 h, or pH <7.25 with PaCO₂ ≥60 for >6 h despite optimal ventilation and proning, ventilated <7 days, reversible cause, no futility markers. Call early, since transfer takes time
  • no ECCO₂R outside trials (ESICM strong against)
  • inhaled pulmonary vasodilators improve oxygenation without improving survival: rescue or bridge only

corticosteroids

  • suggested for ARDS (ATS 2024 and SCCM 2024, conditional, moderate certainty): mortality RR ~0.84, fewer ventilator days. ESICM 2023 didn’t address steroids
  • no single regimen. Use the cause-specific regimen where one exists: severe CAP → hydrocortisone (community-acquired pneumonia), COVID-19 → dexamethasone (influenza and covid-19), HIV-PJP → prednisone (hiv opportunistic infections). Otherwise DEXA-ARDS (2020): dexamethasone 20 mg IV daily × 5 days, then 10 mg daily × 5 days or until extubation, started early (within 24 h). Courses >7 days were associated with better survival than shorter ones (SCCM 2024 subgroup)
  • don’t start after ~2 weeks of ARDS: late initiation was associated with harm
  • watch hyperglycaemia, infection (immunocompromised, TB/Strongyloides risk), weakness. Many trials stopped steroids at extubation

key points

  • treat the cause (septic shock, community-acquired pneumonia): source control and antibiotics for sepsis or pneumonia, and look for the mimics above when no risk factor is obvious
  • HFNC is the first non-invasive step in non-intubated ARDS; prolonged NIV that is failing delays intubation and worsens outcome → noninvasive respiratory support
  • survivors: weakness, cognitive impairment, PTSD and reduced exercise capacity are common at 1 year → extubation (ICU-acquired weakness)

Prepared 2026-10-29. Not yet verified — check every number against the sources.

Key references

+12 more sources

Suggest a correction ARDS