Skip to main content

Foundational Trials

ALVEOLI

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome
  • Acronym: ALVEOLI
  • Year: 2004
  • Journal published in: New England Journal of Medicine
  • Citation: Brower RG, Lanken PN, MacIntyre N, Matthay MA, Morris A, Ancukiewicz M, et al; National Heart, Lung, and Blood Institute ARDS Clinical Trials Network. Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med. 2004;351(4):327-336.

Context & Rationale

  • Background
    Ventilator-induced lung injury (VILI) was increasingly conceptualised as arising not only from regional overdistension, but also from cyclic opening/closing (“atelectrauma”) when end-expiratory lung volume is low.
    Positive end-expiratory pressure (PEEP) can recruit lung units, raise end-expiratory volume, improve oxygenation, and theoretically reduce atelectrauma; however, excessive PEEP may worsen overdistension and haemodynamic compromise.
    By 2000, lung-protective ventilation with lower tidal volumes had demonstrated a mortality benefit in acute lung injury/ARDS, establishing a low-tidal-volume “backbone” on which to test adjunctive ventilator strategy components.1
    Prior “open lung” strategies often co-intervened (lower tidal volume, recruitment manoeuvres, and higher PEEP), making the independent effect of higher PEEP on patient-centred outcomes uncertain.2
  • Research Question/Hypothesis
    In mechanically ventilated patients with acute lung injury/ARDS managed with a low-tidal-volume, plateau-pressure-limited strategy, does a protocolised higher-PEEP approach reduce mortality (death before discharge home while breathing unassisted) compared with a protocolised lower-PEEP approach?
  • Why This Matters
    “How much PEEP?” is a high-frequency, high-stakes bedside decision with potential to help (recruitment, oxygenation, reduced FiO2 exposure) or harm (overdistension, barotrauma, reduced venous return).
    A pragmatic protocol (FiO2/PEEP tables) could be implemented widely if shown to improve patient-important outcomes without excess harms.

Design & Methods

  • Research Question: Whether a higher-PEEP strategy (vs a lower-PEEP strategy), delivered alongside low-tidal-volume ventilation, reduces mortality and improves clinical outcomes in acute lung injury/ARDS.
  • Study Type: Multicentre, randomised, parallel-group ARDSNet trial in adult ICUs (23 hospitals); open-label ventilator management; enrolment October 1999 to February 2002.
  • Population:
    • Setting: Mechanically ventilated ICU patients at ARDSNet centres.
    • Inclusion: Acute lung injury/ARDS with PaO2/FiO2 ≤300; bilateral infiltrates on chest radiograph; no clinical evidence of left atrial hypertension.
    • Timing: Enrolment required within 36 hours of meeting eligibility criteria.
    • Key exclusions: >36 hours since meeting criteria; age <13 years; pregnancy; increased intracranial pressure; severe neuromuscular disease; sickle cell disease; severe chronic respiratory disease; body weight >1 kg per cm of height; burns >40% body surface area; severe chronic liver disease; vasculitis with diffuse alveolar haemorrhage; coexisting condition with estimated 6-month mortality >50%; enrolment in another trial within 30 days.
  • Intervention:
    • Higher-PEEP strategy: PEEP titrated using a higher FiO2/PEEP table to achieve PaO2 55–80 mmHg or SpO2 88–95%.
    • Ventilator “backbone” (both groups): Volume assist-control; tidal volume targeted at 6 mL/kg predicted body weight (PBW), reduced to as low as 4 mL/kg PBW if needed to keep plateau pressure ≤30 cmH2O; respiratory rate adjusted (up to 35/min) for pH goals.
    • Protocol modification within higher-PEEP arm: After 171 patients, the higher-PEEP table was modified to eliminate combinations with PEEP <12 cmH2O (to increase separation).
    • Recruitment manoeuvres: Assessed in the first 80 higher-PEEP patients as single sustained inflations to higher airway pressures/volumes than tidal ventilation (exact pressure/duration not reported in the main manuscript); not continued thereafter.
    • Weaning readiness (both groups): Daily assessment when FiO2 ≤0.40 and PEEP ≤8 cmH2O; spontaneous breathing trials per protocolised approach.
  • Comparison:
    • Lower-PEEP strategy: PEEP titrated using a lower FiO2/PEEP table to achieve the same oxygenation targets (PaO2 55–80 mmHg or SpO2 88–95%).
    • Same low-tidal-volume / plateau-pressure-limited ventilation backbone as the intervention arm.
  • Blinding: Unblinded (ventilator settings not practically blindable); primary outcome largely objective (mortality/discharge status).
  • Statistics: Sample size 750 planned to detect a 10% absolute mortality reduction (from 28% to 18%) with 89% power at two-sided α=0.05; interim analyses after ~250 and ~500 patients with asymmetric stopping boundaries including a futility rule; primary comparisons used a 60-day cumulative mortality approach, with additional covariate-adjusted analyses (explicit “intention-to-treat” label not stated; outcomes reported for all randomised patients).
  • Follow-Up Period: Mortality and discharge status assessed using a 60-day cumulative approach; ventilator-free days, ICU-free days, and organ-failure–free days assessed to day 28.

Key Results

This trial was stopped early. Stopped for futility at the second interim analysis after 549 patients had been enrolled.

Outcome Higher PEEP Lower PEEP Effect p value / 95% CI Notes
Death before discharge home while breathing without assistance (primary endpoint; unadjusted) 27.5% 24.9% Difference (Lower − Higher): −2.6 percentage points 95% CI −10.0 to 4.7; P=0.48 Absolute difference reported as Lower PEEP minus Higher PEEP (as published).
Death before discharge home while breathing without assistance (covariate-adjusted analysis) 25.1% 27.5% Not reported (modelled rates shown) 95% CI for difference −3.6 to 8.4; P=0.47 Adjusted for baseline imbalances (notably age and baseline PaO2/FiO2).
Breathing without assistance by day 28 72.3% 72.8% Not reported P=0.89 Secondary outcome; similar between groups.
Ventilator-free days (day 1–28) 13.8 ± 10.6 14.5 ± 10.4 Not reported P=0.50 Higher PEEP did not increase ventilator-free days.
ICU-free days (day 1–28) 12.3 ± 10.3 12.2 ± 10.4 Not reported P=0.83 No signal for shorter ICU stay.
Barotrauma 11% 10% Not reported P=0.51 No statistically significant increase in barotrauma despite higher mean airway/plateau pressures.
Days without circulatory, coagulation, hepatic, or renal failure (day 1–28) 16 ± 11 16 ± 11 Not reported P=0.82 No difference in non-pulmonary organ-failure–free days.
Protocol separation (day 1): PEEP (cmH2O) 14.7 ± 3.5 8.9 ± 3.5 Not reported P<0.01 Demonstrated meaningful separation in the variable of interest.
Protocol separation (day 1): FiO2 0.44 ± 0.17 0.54 ± 0.18 Not reported P<0.05 Higher PEEP achieved oxygenation targets with lower FiO2.
Physiology (day 1): PaO2/FiO2 220 ± 89 168 ± 66 Not reported P<0.01 Oxygenation improved early with higher PEEP (physiologic efficacy without patient-centred benefit).
  • Despite improved oxygenation and lower FiO2 exposure (day 1 PaO2/FiO2 220 ± 89 vs 168 ± 66; FiO2 0.44 ± 0.17 vs 0.54 ± 0.18), higher PEEP did not improve mortality or ventilator-free days.
  • Higher PEEP increased airway pressures (e.g., day 1 plateau pressure 27 ± 6 vs 24 ± 7 cmH2O; not shown in table), yet without a detected increase in barotrauma (11% vs 10%).
  • Stopping early for futility makes the trial most informative about the absence of a large mortality benefit in an unselected acute lung injury/ARDS population managed with low tidal volumes.

Internal Validity

  • Randomisation and Allocation: Central interactive voice system; permuted blocks of 4 and 6; stratified by hospital (supports allocation concealment and balance across centres).
  • Dropout / Exclusions: No post-randomisation exclusions were reported; outcomes are presented for all randomised patients (Higher PEEP n=276; Lower PEEP n=273).
  • Performance / Detection Bias: Unblinded ventilator management introduces potential performance bias (co-interventions, clinician behaviour), but the primary outcome (mortality/discharge status) is relatively objective.
  • Baseline Characteristics: Groups were broadly similar; notable imbalances included older age in Higher PEEP (54 ± 17 vs 49 ± 17 years) and lower baseline PaO2/FiO2 (151 ± 67 vs 165 ± 77), addressed via adjusted analyses (no change in inference).
  • Timing: Enrolment within 36 hours of meeting criteria supports testing early PEEP strategy, when recruitability and potential atelectrauma mitigation might be most relevant.
  • Dose / Separation of the Variable of Interest: Clear protocol separation achieved:
    PEEP (cmH2O) day 1: 14.7 ± 3.5 (Higher) vs 8.9 ± 3.5 (Lower); day 3: 12.9 ± 4.5 vs 8.5 ± 3.7; day 7: 12.9 ± 4.0 vs 8.4 ± 4.3.
    FiO2 day 1: 0.44 ± 0.17 vs 0.54 ± 0.18; day 3: 0.40 ± 0.14 vs 0.52 ± 0.18; day 7: 0.40 ± 0.11 vs 0.52 ± 0.20.
    PaO2/FiO2 day 1: 220 ± 89 vs 168 ± 66; day 3: 206 ± 76 vs 169 ± 69; day 7: 218 ± 85 vs 181 ± 115.
    Plateau pressure (cmH2O) day 1: 27 ± 6 vs 24 ± 7; day 3: 26 ± 7 vs 24 ± 6; day 7: 26 ± 6 vs 26 ± 8.
  • Protocol Adherence: Both arms were ventilated at low tidal volumes (approximately 6 mL/kg PBW), supporting that the trial tested “PEEP on top of lung-protective ventilation”.
  • Heterogeneity: Broad inclusion (PaO2/FiO2 ≤300) likely introduced clinical heterogeneity (from mild to severe ARDS physiology), which can dilute a treatment effect if benefit is severity-dependent.
  • Protocol Changes: Higher-PEEP table was modified after 171 patients and recruitment manoeuvres were evaluated only in the first 80 higher-PEEP patients (factual design features that may increase within-arm heterogeneity over time).
  • Outcome Assessment: Ventilator-free and ICU-free days were protocol-defined; mortality/discharge endpoint is robust to measurement bias, though discharge practices can vary across sites.
  • Statistical Rigor: Pre-specified interim analyses with a futility rule; early stopping reduces power to detect smaller (but potentially meaningful) benefits than the large effect size used in planning.

Conclusion on Internal Validity: Overall, internal validity is moderate to strong: randomisation/allocation processes were robust and protocol separation was clear, but lack of blinding, protocol evolution within the higher-PEEP arm, and early stopping for futility limit certainty about smaller subgroup-specific effects.

External Validity

  • Population Representativeness: Typical adult ICU acute lung injury/ARDS population in a multicentre network, enrolled early (≤36 hours), increasing relevance to routine critical care.
  • Important Exclusions: Significant chronic lung disease, severe chronic liver disease, extreme obesity (weight >1 kg/cm height), pregnancy, and conditions with high short-term mortality were excluded, limiting generalisability to these subgroups.
  • Applicability of the Intervention: Uses pragmatic FiO2/PEEP tables (implementable at bedside), but does not incorporate modern personalised PEEP approaches (e.g., recruitability- or transpulmonary pressure-guided titration).
  • Contemporary Context: Adjuncts now commonly used for moderate–severe ARDS (prone positioning, neuromuscular blockade, ECMO) were not protocolised and may modify the net effect of higher PEEP strategies in present-day practice.

Conclusion on External Validity: Findings are reasonably generalisable to adult ICU patients with early acute lung injury/ARDS managed with lung-protective ventilation, but applicability is more limited for extreme phenotypes, resource-limited settings, and modern care bundles where co-interventions may interact with PEEP strategy.

Strengths & Limitations

  • Strengths: Large multicentre randomised trial; pragmatic, protocolised PEEP delivery; clear separation in PEEP/FiO2; low-tidal-volume ventilation in both groups; objective primary outcome; pre-specified interim analyses and adjusted analyses addressing baseline imbalances.
  • Limitations: Unblinded intervention; higher-PEEP protocol modified mid-trial; recruitment manoeuvres applied only early in the higher-PEEP arm; early stopping for futility (reduced power for smaller effects); broad eligibility (PaO2/FiO2 ≤300) may obscure severity-dependent benefit or harm; limited reporting of co-interventions (e.g., prone positioning) that might influence outcomes.

Interpretation & Why It Matters

  • Clinical signal
    In an unselected acute lung injury/ARDS cohort ventilated with low tidal volumes, “routine” protocolised higher PEEP improved oxygenation but did not translate into improved survival or faster liberation from ventilation.
  • Mechanistic implication
    Physiologic improvements (higher PaO2/FiO2, lower FiO2) are not sufficient surrogates for outcome benefit; the balance between recruitment and overdistension likely varies by lung recruitability and ARDS severity.
  • Bedside consequence
    ALVEOLI supports a cautious, individualised approach: higher PEEP is not automatically “better” when low tidal volumes and plateau pressure limitation are already in place, and the clinical justification for higher PEEP should extend beyond oxygenation alone.

Controversies & Subsequent Evidence

  • Powering and early stopping: The trial was powered for a large absolute mortality reduction (10%); stopping for futility strengthens inference against a large benefit, but cannot exclude smaller benefits (or harms) in clinically important subgroups.
  • Protocol evolution within the intervention arm: The higher-PEEP table was modified after 171 patients to increase separation; this pragmatic adaptation improves “dose” separation but complicates interpretation as a single uniform intervention.
  • Subsequent large RCTs with higher-PEEP/open-lung approaches: Two major later trials (LOVS and EXPRESS) similarly did not demonstrate a clear mortality benefit for higher-PEEP strategies when applied broadly, reinforcing the probability that any benefit is modest and/or concentrated in specific phenotypes.34
  • Meta-analytic synthesis and heterogeneity of treatment effect: An individual patient data meta-analysis of the three major “higher vs lower PEEP” trials (including ALVEOLI, LOVS, and EXPRESS) reported that overall benefit was not uniform and suggested effect modification by ARDS severity (numerical subgroup estimates not reproduced here).5
  • Updated evidence synthesis: A Cochrane review (2021 update) concluded that higher vs lower PEEP likely results in little to no difference in mortality (RR 0.97; 95% CI 0.90 to 1.04) and little to no difference in barotrauma (RR 1.00; 95% CI 0.64 to 1.57), consistent with ALVEOLI’s neutral clinical outcomes despite physiologic improvements.6
  • Guideline trajectory: More recent consensus guidance incorporates ALVEOLI and later trials/meta-analyses, generally favouring lung-protective ventilation with consideration of higher PEEP in selected (often moderate–severe) ARDS, balancing oxygenation gains against overdistension and haemodynamic compromise.78

Summary

  • ALVEOLI randomised 549 patients with acute lung injury/ARDS to protocolised higher vs lower PEEP, with low-tidal-volume ventilation in both arms.
  • The trial stopped early for futility at the second interim analysis.
  • Higher PEEP improved oxygenation and reduced FiO2 requirements (day 1 PaO2/FiO2 220 ± 89 vs 168 ± 66; FiO2 0.44 ± 0.17 vs 0.54 ± 0.18), confirming physiologic efficacy.
  • There was no improvement in the primary endpoint (mortality before discharge home while breathing unassisted: 27.5% vs 24.9%; difference (Lower − Higher) −2.6 percentage points; 95% CI −10.0 to 4.7; P=0.48) and no difference in ventilator-free days (13.8 ± 10.6 vs 14.5 ± 10.4; P=0.50).
  • Neutral outcomes, combined with later RCTs and syntheses, support selective (rather than routine) escalation of PEEP, especially when low tidal volumes and plateau-pressure limitation are already implemented.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • ALVEOLI tested a pragmatic FiO2/PEEP table strategy (with low tidal volume in both arms) rather than physiology-guided “best PEEP” titration; later evidence increasingly frames PEEP as a severity- and phenotype-dependent intervention rather than a universal default.

Overall Takeaway

ALVEOLI is a landmark ARDSNet trial because it isolated the effect of a higher-PEEP strategy on top of low-tidal-volume ventilation and showed that physiologic improvements in oxygenation do not necessarily produce patient-centred benefit. Its neutral results (and early futility stopping) anchored subsequent syntheses and guideline recommendations toward selective rather than routine higher PEEP, emphasising careful balancing of recruitment against overdistension and haemodynamic cost.

Overall Summary

  • Higher PEEP improved oxygenation and reduced FiO2 requirements but did not improve mortality, ventilator-free days, or ICU-free days in an unselected ALI/ARDS cohort ventilated with low tidal volumes.
  • Stopping early for futility supports the absence of a large mortality benefit, while leaving open the possibility of smaller or subgroup-specific effects addressed in later trials and meta-analyses.

Bibliography