Skip to main content

Foundational Trials

ARMA 2000

Image: Shutterstock / Medical-R

Publication

Context & Rationale

  • Background
    • Traditional ventilation used tidal volumes of 10–15 mL/kg, prioritising carbon dioxide clearance and normal arterial pH. In acute lung injury, oedema and atelectasis reduce the aerated lung available to receive each breath, making otherwise conventional volumes potentially injurious.
    • Experimental work implicated excessive lung stretch in epithelial and endothelial injury, inflammation and systemic mediator release. Ventilation could therefore contribute to extrapulmonary organ failure as well as perpetuate pulmonary injury.
    • Earlier randomised trials produced conflicting results. A 53-patient trial demonstrated lower 28-day mortality with a protective strategy combining smaller tidal volumes, pressure limitation and higher positive end-expiratory pressure (PEEP), leaving uncertainty about the contribution of each component.1
  • Research Question/Hypothesis
    • Would reducing tidal volume and inspiratory plateau pressure improve survival and time free from ventilation in adults with acute lung injury or acute respiratory distress syndrome (ARDS), despite the potential for poorer carbon dioxide clearance and oxygenation?
  • Why This Matters
    • Mechanical ventilation was an unavoidable exposure for these patients. A reproducible adjustment to supportive care could influence survival across a broad population without requiring a new drug or device.
    • A survival endpoint was essential: better oxygenation, lower airway pressure or fewer air leaks alone would not establish that a ventilation strategy improved patient outcomes.

Design & Methods

  • Research Question:
    • Superiority of a strategy beginning at 6 mL/kg predicted body weight (PBW), with plateau pressure ≤30 cm H₂O, over 12 mL/kg PBW with plateau pressure ≤50 cm H₂O, for mortality and ventilator-free days.
  • Study Type:
    • Investigator-led, NHLBI-funded, multicentre, parallel-group, randomised trial conducted in the intensive care units of 10 United States university centres and their participating hospitals.
    • Recruitment ran from March 1996 to March 1999. Central interactive voice randomisation allocated 861 patients: 432 to lower tidal volumes and 429 to traditional tidal volumes.
    • A factorial drug comparison accompanied ventilation allocation in the first 234 patients (ketoconazole versus placebo) and the last 194 (lisofylline versus placebo); 433 received no experimental study drug.
  • Population:
    • Inclusion: adults aged ≥18 years, intubated and mechanically ventilated, with an acute PaO₂/FiO₂ ratio ≤300, bilateral infiltrates consistent with pulmonary oedema, and no clinical evidence of left atrial hypertension or pulmonary-capillary wedge pressure ≤18 mmHg if measured.
    • Randomisation was required within 36 hours of meeting the respiratory eligibility criteria. These were contemporary acute lung injury/ARDS criteria, predating the Berlin definition.
    • Exclusions included pregnancy; increased intracranial pressure; neuromuscular disease impairing spontaneous breathing; sickle cell disease; severe chronic respiratory disease; weight exceeding 1 kg per centimetre of height; and burns affecting >30% of body surface area.
    • Other exclusions included bone marrow or lung transplantation, Child–Pugh class C liver disease, another condition carrying estimated six-month mortality >50%, participation in another trial within 30 days, or attending-physician refusal or unwillingness to provide full life support.
    • Consent was obtained from patients or surrogates except at one hospital where the requirement was waived.
  • Intervention:
    • Volume assist–control ventilation, reducing tidal volume to 6 mL/kg PBW within four hours of randomisation.
    • PBW was calculated from sex and height: men, 50 + 0.91 × (height in cm − 152.4) kg; women, 45.5 + 0.91 × (height in cm − 152.4) kg.
    • If plateau pressure exceeded 30 cm H₂O, tidal volume was reduced in 1 mL/kg steps to a minimum of 4 mL/kg PBW. If plateau pressure fell below 25 cm H₂O, volume could be increased towards 6 mL/kg PBW.
    • For severe dyspnoea, 7–8 mL/kg PBW was permitted if plateau pressure remained ≤30 cm H₂O. Pressure limits could be exceeded when tidal volume was already 4 mL/kg PBW or arterial pH was <7.15.
    • Plateau pressure was measured using a 0.5-second inspiratory hold every four hours and after changes in tidal volume or PEEP.
  • Comparison:
    • Volume assist–control ventilation beginning at 12 mL/kg PBW, reduced in 1 mL/kg steps if necessary to maintain plateau pressure ≤50 cm H₂O, down to a minimum of 4 mL/kg PBW.
    • If plateau pressure fell below 45 cm H₂O after volume reduction, tidal volume was increased until plateau pressure reached 45 cm H₂O or tidal volume reached 12 mL/kg PBW. The protocol did not require plateau pressure to reach 45–50 cm H₂O.
    • Both groups used the same oxygenation goals: PaO₂ 55–80 mmHg or SpO₂ 88–95%, with a common PEEP/FiO₂ table spanning PEEP 5–24 cm H₂O. Further PEEP increases to 34 cm H₂O were allowed but not required.
    • Both groups targeted arterial pH 7.30–7.45, using respiratory rates of 6–35 breaths/min. Rate increases were required for acidosis and bicarbonate was permitted; severe acidosis allowed relaxation of volume and pressure constraints.
    • Both groups used protocolised pressure-support weaning, required when FiO₂ was ≤0.4. Assigned ventilation continued until weaning or day 28 and resumed if ventilation was restarted within that period.
  • Blinding:
    • Ventilation assignment was unblinded to treating clinicians. Mortality was objective, although clinical decisions governing weaning, discharge and continued life support could not be blinded.
    • Blinding of the factorial drug interventions did not blind the ventilation comparison.
  • Statistics:
    • A maximum sample of 1,000 patients was planned to detect a reduction in mortality from 40% to 30%, with 85% power (β=0.15) at a two-sided α=0.05.2
    • Intention-to-treat analysis was specified. Mortality was compared using 180-day cumulative incidence, stratified by the five categories of concurrent drug allocation; discharge home breathing without assistance was a competing outcome.
    • The two primary outcomes were death before discharge home while breathing without assistance, and ventilator-free days during days 1–28. Ventilator-free days required unassisted breathing lasting at least 48 consecutive hours.
    • An independent data and safety monitoring board reviewed approximately every 200 patients, using O’Brien–Fleming efficacy and DeMets–Ware ineffectiveness boundaries. Skewed ventilator-free and organ-failure-free days were compared with Wilcoxon tests; P values were two-sided.
  • Follow-Up Period:
    • Clinical follow-up continued to day 180 or until the patient was breathing without assistance at home. Patients remaining in other healthcare facilities at day 180 were classified as discharged and breathing without assistance.
    • Ventilator-free days, nonpulmonary organ failure and barotrauma were assessed through day 28. Long-term functional recovery and quality of life were not assessed.

Key Results

This trial was stopped early after the fourth interim analysis because mortality was lower with the lower tidal-volume strategy. At stopping, 861 patients had been enrolled; the interim mortality comparison gave P=0.005, crossing the prespecified boundary of P=0.023. The final published comparison gave P=0.007.

Outcome Lower tidal volume (n=432) Traditional tidal volume (n=429) Effect p value / 95% CI Notes
Death before discharge home and breathing without assistance 31.0% 39.8% 22% relative reduction P=0.007; 95% CI for absolute mortality reduction 2.4 to 15.3 percentage points First primary outcome; cumulative-incidence analysis through 180 days, ending follow-up at recovery at home.
Ventilator-free days, days 1–28 12 ± 11 10 ± 11 Not reported P=0.007; 95% CI not reported Second primary outcome; mean ± SD. Incorporates survival and liberation from ventilation.
Breathing without assistance by day 28 65.7% 55.0% Not reported P<0.001; 95% CI not reported Unassisted breathing required ≥48 consecutive hours.
Days without nonpulmonary organ or system failure, days 1–28 15 ± 11 12 ± 11 Not reported P=0.006; 95% CI not reported Mean ± SD; also affected by survival.
Barotrauma, days 1–28 10% 11% Not reported P=0.43; 95% CI not reported New pneumothorax, pneumomediastinum, subcutaneous emphysema or pneumatocele >2 cm.
Delivered tidal volume, days 1–3 6.2 ± 0.8 mL/kg PBW 11.8 ± 0.8 mL/kg PBW Not reported P<0.001; 95% CI not reported Mean ± SD; achieved treatment exposure.
Plateau pressure, days 1–3 25 ± 6 cm H₂O 33 ± 8 cm H₂O Not reported P<0.001; 95% CI not reported Mean ± SD; inspiratory pressure changed alongside tidal volume.
PaCO₂, day 1 40 ± 10 mmHg 35 ± 8 mmHg Not reported P<0.05; exact P value and 95% CI not reported Mean ± SD; n=351 and 369 with measurements.
Arterial pH, day 1 7.38 ± 0.08 7.41 ± 0.07 Not reported P<0.05; exact P value and 95% CI not reported Mean ± SD; n=351 and 369. Mean values do not describe the frequency of severe acidosis.
  • Clinically important survival benefit: the mortality confidence interval excluded no difference. This endpoint should not be relabelled as 28-day mortality or complete all-cause mortality surveillance for 180 days after discharge.
  • Consistent clinical and biological signals: ventilator-free days and organ-failure-free days favoured lower volumes, although median ventilation duration among survivors was eight days in both groups. In 204 patients sampled on days 0 and 3, mean log-transformed interleukin-6 fell from 2.5 ± 0.7 in both groups to 2.0 ± 0.5 versus 2.3 ± 0.7; P=0.002 for the day-3 comparison and P<0.001 for the difference in decline. These findings support, but do not establish, an inflammatory mechanism.
  • Subgroups: baseline respiratory-system compliance was available in 517 patients, comprising 260 versus 257. The treatment-by-compliance-quartile interaction was not significant (P=0.49); subgroup relative effects and confidence intervals were not reported. Interaction with concurrent drug allocation was also not significant (P=0.16). Neither result establishes identical effects across subgroups.

Internal Validity

  • Randomisation and Allocation:
    • Central allocation reduced the opportunity for foreknowledge of the next assignment. The main article does not describe the sequence-generation algorithm or block sizes.
  • Dropout and Post-randomisation Exclusions:
    • The trial included 432 and 429 randomised patients in the assigned groups. The survival figure states that final status was unknown for nine patients; their data and those of 22 patients still hospitalised at the fourth interim analysis were censored.
    • The 22 patients represent incomplete follow-up at the analysis cutoff, distinct from the nine with unknown status. Arm-specific numbers and reasons for the unknown outcomes were not reported.
  • Performance and Detection Bias:
    • Unblinded care creates potential for differential co-intervention, but common oxygenation, acid–base and weaning procedures constrained major sources of performance bias.
    • Death is relatively resistant to observer interpretation. Ventilator liberation, discharge and organ-failure-free days depend more directly on clinical management.
  • Baseline Characteristics and Illness Severity:
    • Groups were broadly comparable: age 51 ± 17 versus 52 ± 18 years; APACHE III score 81 ± 28 versus 84 ± 28; PaO₂/FiO₂ 138 ± 64 versus 134 ± 58; and 1.8 ± 1.1 versus 1.8 ± 1.0 nonpulmonary organ failures.
    • Baseline minute ventilation was 13.4 ± 4.3 versus 12.7 ± 4.3 L/min (P=0.01). There was substantial illness severity and scope for a clinically meaningful reduction in ventilation-related injury.
  • Heterogeneity:
    • Pneumonia accounted for 33% versus 36%, sepsis 27% versus 26%, aspiration 15% versus 14%, and trauma 13% versus 9%. Randomisation supports an average causal effect across this mixture; it does not define the optimal strategy for every physiological phenotype.
  • Timing, Dose and Protocol Adherence:
    • The intervention addressed early established lung injury: enrolment within 36 hours and reduction to 6 mL/kg PBW within four hours of allocation. The exact distribution of time from eligibility to treatment was not reported.
    • Randomly selected ventilator and blood-gas measurements were audited, with quarterly centre reports. Mean delivered values demonstrate sustained separation; a complete patient-level protocol-violation or crossover count was not reported.
    • Permitted adjustments for pressure, dyspnoea and acidosis were components of the assigned strategies, rather than necessarily crossovers.
  • Separation of the Variable of Interest:
    • Mean tidal volume over days 1–3 was 6.2 ± 0.8 versus 11.8 ± 0.8 mL/kg PBW, with plateau pressure 25 ± 6 versus 33 ± 8 cm H₂O; both P<0.001.
    • On day 1, respiratory rate was 29 ± 7 versus 16 ± 6 breaths/min, while minute ventilation was 12.9 ± 3.6 versus 12.6 ± 4.5 L/min. PEEP was 9.4 ± 3.6 versus 8.6 ± 3.6 cm H₂O and FiO₂ 0.56 ± 0.19 versus 0.51 ± 0.17.
    • Thus, randomisation produced a strong contrast in the intended ventilation strategies, including their consequent physiological adjustments.
  • Adjunctive Therapy Use:
    • Other investigational rescue treatments were used in 15 versus 12 patients, including prone positioning in 14 versus nine. There was no indication of extensive compensatory rescue treatment in the traditional arm.
    • Among survivors discharged home breathing without assistance, sedatives were used on 65 ± 26% versus 65 ± 24% of days and neuromuscular blockers on 6 ± 14% versus 6 ± 15% of days; differences were not significant.
    • These are proportions of treatment days, not proportions of patients or cumulative drug doses. Similar treatment-day exposure does not establish equivalence of sedation depth or dose.
  • Outcome Assessment and Statistical Rigour:
    • Prespecified primary outcomes, an intention-to-treat plan, independent monitoring and an explicit stopping boundary support credibility. Control mortality was close to the 40% planning assumption.
    • The cumulative-incidence analysis appropriately recognised recovery at home as a competing event. Rank-based tests addressed the skewed distributions of ventilator-free and organ-failure-free days.
    • Mortality and ventilator-free days both favoured the intervention. The additional physiological, organ-specific and biomarker comparisons provide supporting evidence rather than independent confirmatory demonstrations of mechanism.

Conclusion on Internal Validity: Internal validity is strong for the comparison of these two ventilation strategies, supported by central randomisation, substantial treatment separation and concordant primary outcomes. Unblinded care, limited missing-outcome detail and stopping for benefit temper precision and mechanistic interpretation.

External Validity

  • Population Representativeness:
    • The trial enrolled mixed medical and surgical causes of acute lung injury, with PaO₂/FiO₂ ≤200 in 82% versus 85%. Its population was therefore predominantly substantially hypoxaemic, while still including less severe acute lung injury.
    • Applicability is strongest to adults receiving invasive ventilation early in an ARDS-like illness. The entry criteria cannot be mapped perfectly onto later definitions because their diagnostic and ventilator-setting requirements differ.
    • Children, patients with raised intracranial pressure, severe chronic respiratory disease, extreme obesity, important neuromuscular disease, major burns and several advanced comorbidities were excluded. Direct estimates of benefit and tolerability in these groups are unavailable.
  • Applicability Across Settings and Eras:
    • PBW-based tidal-volume adjustment and inspiratory holds are implementable with standard intensive care ventilators. Reliable height measurement, trained staff, blood-gas assessment and repeated review of respiratory mechanics are necessary for faithful delivery.
    • The trial was conducted in a well-resourced United States network. It did not directly test implementation where ventilator capabilities, staffing or monitoring were restricted.
    • Contemporary management incorporates different approaches to sedation, prone positioning and other supportive care. The observed mortality advantage should not be assumed to recur with the same magnitude against a modern comparator already using protective volumes and pressures.

Conclusion on External Validity: The principle of avoiding excessive tidal inflation is broadly applicable to invasively ventilated adults with ARDS. Generalisability is more limited for excluded populations, non-intubated disease, constrained monitoring environments and the exact benefit achievable over contemporary protective ventilation.

Strengths & Limitations

  • Strengths:
    • Large multicentre randomised comparison with public funding and independent safety monitoring.
    • Clear, reproducible intervention with substantial achieved separation in both tidal volume and plateau pressure.
    • Two clinically relevant primary outcomes, supported by organ-failure and biomarker findings.
    • Common oxygenation and weaning procedures, formal compliance monitoring and broad inclusion of acute lung injury aetiologies.
  • Limitations:
    • Stopping for benefit introduces uncertainty about the exact effect magnitude.
    • A 6-versus-12 mL/kg PBW strategy comparison cannot locate the optimal tidal volume within a narrower clinically relevant range.
    • Simultaneous changes in volume, pressure and consequent respiratory management prevent attribution to one isolated component.
    • Unblinded treatment, incomplete reporting of patient-level deviations and limited subgroup precision.
    • A recovery-dependent mortality endpoint and absence of long-term functional outcomes limit assessment of the full patient-centred effect.

Interpretation & Why It Matters

  • A change in the purpose of ventilation
    • ARMA established that how ventilation is delivered can alter survival. Avoiding excessive inflation became a central therapeutic objective, alongside maintaining adequate gas exchange.
    • On day 1, PaO₂/FiO₂ was lower with protective ventilation, 158 ± 73 versus 176 ± 76, despite better survival. Improvement in oxygenation is therefore an unreliable surrogate for the net benefit of a ventilation strategy.
  • The practical intervention
    • The durable lesson is to combine a PBW-based tidal-volume starting point with pressure monitoring, oxygenation support and active management of acidosis and synchrony.
    • Current ATS guidance retains a strong recommendation to limit tidal volume to 4–8 mL/kg PBW and plateau pressure to <30 cm H₂O. A starting value around 6 mL/kg PBW fits that strategy; ongoing physiological assessment remains necessary.3
  • What the causal result establishes
    • The trial establishes superiority of the tested lower-volume, lower-pressure strategy over the tested traditional strategy.
    • It does not identify a universally optimal volume, isolate the mortality effect of plateau pressure, or demonstrate that pressure-controlled ventilation is superior to volume assist–control ventilation.

Controversies & Other Evidence

  • Was the comparator representative, and what did the trial establish?
    • The accompanying editorial questioned whether the benefit primarily reflected avoiding excessive inspiratory pressure. The correspondence challenged the permissive control pressure limit against earlier recommendations to constrain plateau pressure.45
    • The important correction is that 50 cm H₂O was a ceiling, not a treatment target. Mean traditional-arm plateau pressures were 33 and 34 cm H₂O on days 1 and 3. The 37 cm H₂O mean on day 7 described patients still ventilated and measured at that time, a selected subset rather than the initial randomised population.5
    • Nevertheless, the control strategy deliberately pursued 12 mL/kg PBW whenever its pressure ceiling allowed. Without an intermediate-volume or clinician-directed usual-care arm, ARMA cannot quantify superiority over 8–10 mL/kg PBW with stricter pressure limitation.
    • Eichacker and colleagues’ five-trial meta-analysis attributed the disparate trial results to differences in control-arm ventilation. This is a legitimate comparator question, but associations across a few heterogeneous trials cannot establish the outcome of a randomised comparison that was never performed. Differences in body-weight conventions further complicate comparisons of nominal mL/kg settings.6
    • Avoiding injury caused by a treatment is itself a clinically important benefit. The comparator debate limits the scope and magnitude of extrapolation; it does not invalidate the randomised contrast.
  • Tidal volume, plateau pressure and driving pressure:
    • Tobin proposed that the pressure achieved might be more informative than the prescribed volume, noting that the early trials showing benefit had higher control-arm plateau pressures. Airway plateau pressure also reflects chest-wall mechanics and cannot fully describe regional pulmonary stress.4
    • Hager and colleagues subsequently analysed ARMA data, including day-1 plateau pressure in 787 patients, and found no defensible plateau-pressure threshold below which tidal-volume reduction could be assumed unnecessary. This argues against treating 30–35 cm H₂O as a proven safety boundary. Analyses conditioned on achieved, post-randomisation pressure do not constitute a new randomised test within pressure strata.7
    • A pooled secondary analysis of 3,562 patients from nine trials associated each approximately 7 cm H₂O increase in driving pressure with higher mortality: RR 1.41; 95% CI 1.31 to 1.51; P<0.001. This supports attention to tidal volume relative to functional respiratory-system size, but mediation modelling does not prove that prescribing a particular driving-pressure target improves survival.8
  • The intervention was a strategy with linked physiological consequences:
    • PEEP and FiO₂ were managed by the same algorithm, yet the lower-volume group required higher early settings. The correspondence correctly challenged the inference that benefit occurred “despite” higher PEEP: PEEP could contribute to protection, reflect derecruitment, or both. Its independent contribution was not randomised.5
    • The trial also compared different respiratory rates and modestly different carbon dioxide and pH values. It tested lower volumes with active management of acid–base consequences, rather than unrestricted acceptance of severe hypercapnic acidosis.
    • The editorial proposed better acidosis management as one explanation for differences from earlier trials; the correspondence highlighted experimental reasons why buffering might not be beneficial. Neither bicarbonate use nor the degree of hypercapnia was separately randomised, so ARMA establishes neither buffering benefit nor therapeutic benefit from acidosis.45
    • Higher respiratory frequency, inspiratory flow and patient effort may alter the mechanical burden of ventilation. The trial did not quantify their independent contribution or establish the best strategy for preventing dyssynchrony.
  • Mechanism and the meaning of apparently concordant outcomes:
    • Lower interleukin-6 and more organ-failure-free days are consistent with reduced inflammatory injury. The correspondence also proposed improved cardiac output and microcirculation; the investigators acknowledged insufficient data to assess these pathways. A specific causal chain from less lung stretch to reduced systemic inflammation to survival was not demonstrated.5
    • Similar barotrauma rates do not refute reduced ventilator-induced lung injury: clinically apparent air leaks are neither a sensitive nor a specific measure of microscopic overdistension injury.
    • Ventilator-free days and organ-failure-free days are influenced by death, so their agreement with mortality is not fully independent corroboration. Conversely, equal median ventilation duration among survivors does not prove absence of a liberation benefit, because the intervention can change which patients survive.
  • Early stopping and the mortality estimand:
    • Stopping followed a prespecified monitoring plan rather than an unplanned inspection. Nonetheless, trials stopped for benefit can overestimate effect magnitude; the confidence interval is more informative than treating the published 22% relative reduction as a fixed expectation.2
    • The endpoint captured death before recovery at home, with follow-up capped at 180 days. Survival after discharge, later functional dependence and quality of life were outside this endpoint; patients in other facilities at day 180 received a favourable administrative classification.
    • The nine unknown outcomes leave some uncertainty about informative missingness. The 22 additional censored observations at the interim cutoff should not be conflated with loss to follow-up.
  • Does every patient need the same tidal volume?
    • ARMA’s non-significant compliance interaction was based on 517 patients and broad quartiles. It cannot establish a uniform treatment effect, particularly when respiratory-system compliance combines lung and chest-wall properties.
    • A 2021 Bayesian secondary analysis of 1,096 patients from five trials found a 93% posterior probability that the mortality effect varied with respiratory-system elastance. The median interaction odds ratio was 0.80 per cm H₂O/(mL/kg), with a 90% credible interval of 0.63 to 1.02; greater benefit was more probable at higher elastance. This is suggestive effect-modification evidence, not prospective validation of an individualised prescription.9
    • A 2025 critical reappraisal renewed the objection that ARMA did not establish 6 mL/kg PBW as the optimum against intermediate volumes. That distinction is sound; observational comparisons lacking a mortality difference cannot establish equivalence or justify routine higher volumes in ARDS. Even ARMA’s lower-volume arm allowed adjustments between 4 and 8 mL/kg PBW.10
  • Subsequent trials: PEEP and recruitment:
    • ALVEOLI randomised 549 patients already receiving low tidal volumes to higher or lower PEEP. Hospital mortality was 27.5% versus 24.9%; P=0.48. It did not demonstrate an overall survival benefit from the tested higher-PEEP strategy and cannot retrospectively isolate the small early PEEP difference in ARMA.11
    • ART randomised 1,010 patients to recruitment with compliance-titrated PEEP or low PEEP. Mortality at 28 days was 55.3% versus 49.3%; HR 1.20; 95% CI 1.01 to 1.42; P=0.041. A strategy intended to improve recruitment can therefore cause harm; its effects should not be inferred from improved mechanics or oxygenation alone.12
  • Subsequent trials: how far should tidal volume be reduced?
    • REST enrolled 412 patients with acute hypoxaemic respiratory failure. Further tidal-volume reduction facilitated by extracorporeal CO₂ removal produced 90-day mortality of 41.5% versus 39.5%; RR 1.05; 95% CI 0.83 to 1.33; P=0.68. Serious adverse events occurred in 30.7% versus 8.6%. This tests a device-supported strategy, so its result cannot be attributed solely to the lower volume.13
    • VT4COVID randomised 215 patients with COVID-19 ARDS to targets of 4 versus 6 mL/kg PBW without extracorporeal support. Its hierarchical mortality/ventilator-free-day outcome did not differ significantly: win ratio 0.85; 95% CI 0.60 to 1.19; P=0.38. Severe respiratory acidosis occurred in 33% versus 13%; absolute difference 20 percentage points; 95% CI 9 to 31; P=0.0004. These results limit extrapolation from ARMA to routine further volume reduction.14
  • Subsequent trials: extension beyond ARDS:
    • PReVENT randomised 961 ventilated adults without ARDS to lower or intermediate tidal volumes. Mean ventilator-free days were 15.2 versus 15.5; mean difference −0.27 days; 95% CI −1.74 to 1.19; P=0.71. This distinct population and comparison did not reproduce an incremental benefit and does not overturn ARMA’s result in established lung injury.15
  • Contemporary synthesis and guidelines:
    • The 2023 ESICM synthesis illustrates that the total evidence is less uniform than a single pivotal trial suggests: the primary three-trial mortality analysis gave RR 0.96; 95% CI 0.72 to 1.28; P=0.768, and the analysis including all eligible studies gave RR 0.82; 95% CI 0.66 to 1.02; P=0.069. The guideline nevertheless strongly recommends 4–8 mL/kg PBW, integrating the physiological rationale, heterogeneous trial comparisons and the overall balance of benefit and harm.16
    • The 2024 ATS update retains volume and pressure limitation as foundational care and conditionally supports higher PEEP without prolonged recruitment manoeuvres in moderate-to-severe ARDS. These recommendations concern an integrated protective strategy and do not establish a universally optimal fixed tidal volume.3
    • A 2026 international Delphi consensus addresses implementation in resource-limited settings, including infrastructure, diagnostics, respiratory support, sedation and weaning. It complements existing guidance where trial evidence is sparse; it does not provide a new randomised estimate of ARMA’s effect in those settings.17

Summary

  • ARMA randomised 861 adults with early acute lung injury/ARDS to initial tidal volumes of 6 or 12 mL/kg PBW, with plateau-pressure ceilings of 30 or 50 cm H₂O and common supportive protocols.
  • The trial stopped after its fourth interim analysis for efficacy. Mortality before recovery at home was 31.0% versus 39.8%; P=0.007, with a 95% CI for absolute reduction of 2.4 to 15.3 percentage points.
  • Ventilator-free days were 12 ± 11 versus 10 ± 11; P=0.007. Barotrauma was 10% versus 11%; P=0.43. Survival benefit therefore did not require a demonstrable reduction in clinical air leaks.
  • Strong treatment separation supports the causal comparison, but linked changes in volume, pressure and respiratory management prevent attribution to tidal volume alone. The study did not determine the optimal volume for every patient.
  • The lasting standard is PBW-based volume limitation with pressure monitoring and physiological reassessment. Subsequent evidence supports careful individualisation and does not justify assuming that progressively smaller volumes or more aggressive recruitment will improve outcomes.

Overall Takeaway

ARMA is a landmark trial because it demonstrated that a strategy designed to reduce excessive lung inflation can improve survival in acute lung injury and ARDS. Its enduring contribution is the clinical priority given to preventing ventilation-related injury, using a reproducible combination of volume limitation, pressure monitoring and supportive adjustments. It establishes a foundation for protective ventilation while leaving the optimal prescription for an individual patient unresolved.

Overall Summary

  • Lower-volume, lower-pressure ventilation improved survival compared with ARMA’s traditional strategy; apply that result through measured PBW, pressure monitoring and repeated physiological assessment, while recognising that the trial did not establish one optimal tidal volume for every patient.

Bibliography

Last updated September 8th, 2026