Skip to main content

Foundational Trials

OSCAR

Image: Shutterstock / ChaNaWiT

Publication

  • Title: High-frequency oscillation for acute respiratory distress syndrome
  • Acronym: OSCAR (Oscillation in ARDS)
  • Year: 2013
  • Journal published in: New England Journal of Medicine
  • Citation: Young D, Lamb SE, Shah S, et al. High-frequency oscillation for acute respiratory distress syndrome. N Engl J Med. 2013;368(9):806-813.

Context & Rationale

  • Background
    • ARDS carries high short-term mortality despite supportive care and lung-protective conventional ventilation.
    • High-frequency oscillatory ventilation (HFOV) delivers very small oscillatory tidal volumes at high frequency with sustained mean airway pressure, aiming to minimise volutrauma/atelectrauma while maintaining recruitment.
    • Before OSCAR, adult HFOV evidence comprised small heterogeneous trials and physiology-driven protocols; oxygenation often improved, but effects on patient-centred outcomes and safety in routine practice remained uncertain.
  • Research Question/Hypothesis
    • In invasively ventilated adults with ARDS, does early HFOV (delivered with a pragmatic protocol) reduce 30-day all-cause mortality compared with conventional ventilation?
  • Why This Matters
    • HFOV is resource-intensive (specialised ventilator, training, frequent blood gases) and frequently necessitates deep sedation and neuromuscular blockade.
    • Potential physiological trade-offs (hypercapnia, haemodynamic effects from higher mean airway pressures) make robust efficacy and safety data essential before widespread adoption.
    • A pragmatic multicentre trial could resolve whether HFOV improves outcomes beyond “real-world” conventional ventilation in UK ICUs.

Design & Methods

  • Research Question: Whether early HFOV, compared with conventional ventilation, reduces 30-day mortality in ventilated adults with ARDS.
  • Study Type: Pragmatic, multicentre, parallel-group, randomised controlled trial in adult ICUs (29 UK centres); open-label with objective primary outcome.
  • Population:
    • Setting: Adult intensive care units across England, Wales, and Northern Ireland.
    • Inclusion criteria: Mechanically ventilated adults with ARDS; PaO2/FiO2 ≤200 mmHg while receiving PEEP ≥5 cm H2O; acute onset with bilateral pulmonary infiltrates and no evidence of left atrial hypertension.
    • Key exclusions: Age <16 years; weight <35 kg; ventilation ≥7 days before randomisation; other interventional trial; airway narrowing/air-trapping conditions; recent lung surgery (contraindicating HFOV).
  • Intervention:
    • Delivery: HFOV initiated as soon as possible after randomisation (median 6.1 hours) using a dedicated oscillator (Novalung/Metran R100).
    • Initial settings: FiO2 1.0; bias flow 20 L/min; cycle volume 100 mL; frequency 10 Hz; mean airway pressure set 5 cm H2O above the plateau pressure at enrolment.
    • Targets: Oxygenation guided by SaO2 88–92% (if pH >7.25) or PaO2 60–75 mmHg; ventilation guided by pH 7.25–7.35 with permissive hypercapnia.
    • Weaning strategy: Reduce FiO2 first, then mean airway pressure; switch to pressure-controlled conventional ventilation once mean airway pressure ≤24 cm H2O and FiO2 ≤0.40 with PaO2 ≥60 mmHg for ≥12 hours.
    • Implementation support: Structured training programme (2306 staff trained across 198 sessions) and 24/7 telephone support; protocol modified early after hypercapnia-related serious adverse events (including allowance of endotracheal tube cuff leak; used in 30 patients).
  • Comparison:
    • Conventional ventilation: Usual care at each site; sites were encouraged to use pressure-controlled ventilation, tidal volume 6–8 mL/kg ideal body weight, plateau pressure ≤30 cm H2O, and an ARDSNet lower PEEP/FiO2 table.
    • Co-interventions/rescue: Permitted at clinicians’ discretion; crossover to HFOV was not prohibited.
  • Blinding: Open-label (ventilator strategy not practically blindable); primary outcome (mortality) is objective, but co-interventions could be influenced by treatment assignment.
  • Statistics: Original target 1006 patients (503/group) to detect a 9% absolute mortality reduction (45% to 36%) with 80% power at the 5% significance level; revised to 802 patients (401/group) to detect a 10% absolute reduction with 80% power at the 5% significance level after interim assessment of recruitment and control mortality; intention-to-treat analysis; adjusted survival analysis via logistic regression; three interim analyses (after 100, 340, and 640 patients) without formal stopping rules.
  • Follow-Up Period: Primary endpoint at 30 days post-randomisation; ICU and hospital outcomes to discharge; longer-term follow-up planned in protocol (not reported in the index publication).

Key Results

This trial was not stopped early. Recruitment target was revised after interim assessment (1006 → 802 planned); 795 patients were randomised and analysed (HFOV 398; conventional 397).

Outcome HFOV Conventional ventilation Effect p value / 95% CI Notes
30-day all-cause mortality (primary) 166/398 (41.7%) 163/397 (41.1%) Absolute risk difference 0.6 percentage points 95% CI −6.1 to 7.5; P=0.85 Adjusted OR for survival (conventional vs HFOV) 1.03; 95% CI 0.75 to 1.40; P=0.87
ICU mortality at first ICU discharge 42.1% 44.1% Absolute difference 2.0 percentage points 95% CI −4.8 to 8.8; P=0.57 No evidence of between-group difference
Hospital mortality at first hospital discharge 48.4% 50.1% Absolute difference 1.7 percentage points 95% CI −5.3 to 8.8; P=0.62 No evidence of between-group difference
Ventilator-free days (to day 30) 17.1 ± 8.6 17.6 ± 8.8 Not reported P=0.42 Similar duration of invasive support
Duration of mechanical ventilation (days) 14.9 ± 13.3 14.1 ± 13.4 Not reported P=0.41 Includes time on HFOV and conventional ventilation
ICU length of stay (days) 17.6 ± 16.6 16.1 ± 15.2 Not reported P=0.18 No evidence of between-group difference
Neuromuscular-blocking agent use (days) 2.5 ± 3.5 2.0 ± 3.4 Not reported P=0.02 Greater exposure in HFOV group
Serious adverse events related to HFOV (hypercapnia) 2 events Not reported Not reported Not reported One event required alkalinisation and resolved; one event involved pulseless electrical activity arrest where hypercapnia was a contributing factor among three identified factors; protocol modified and additional training implemented
  • Mortality at 30 days was virtually identical (41.7% vs 41.1%), and the confidence interval excluded the pre-specified large absolute mortality reduction sought in the sample-size calculation.
  • HFOV improved oxygenation (e.g., day 1 PaO2/FiO2 192 ± 77 vs 154 ± 61 mmHg) but was associated with higher PaCO2 and lower pH (day 1 PaCO2 55 ± 17 vs 50 ± 19 mmHg; pH 7.30 ± 0.10 vs 7.35 ± 0.10), and greater neuromuscular blockade exposure.
  • Important subgroup effects were not reported in the index publication.

Internal Validity

  • Randomisation and Allocation:
    • Central 24/7 telephone randomisation; 1:1 allocation using minimisation stratified by centre, age, and baseline PaO2/FiO2 category.
    • Allocation concealment is likely robust up to assignment (central system), but post-allocation masking was not feasible.
  • Drop out or exclusions (post-randomisation):
    • Randomised: 795 (HFOV 398; conventional 397).
    • Primary endpoint: vital status at 30 days reported for all randomised participants (no loss to follow-up reported).
    • Non-adherence: 10/398 (2.5%) allocated to HFOV did not receive HFOV (3 died before initiation; 4 ventilator malfunction; 1 recovered; 2 clinician non-compliance).
  • Performance/Detection Bias:
    • Open-label design introduces potential performance bias for co-interventions (sedation strategies, rescue therapies), but mortality is objective and less susceptible to detection bias.
  • Protocol Adherence:
    • Time to HFOV initiation: median 6.1 hours after randomisation.
    • HFOV exposure among those treated: median 3 days (IQR 2–5); maximum initial HFOV run 24 days.
    • Control contamination: 10/397 (2.5%) allocated to conventional ventilation received HFOV at some point after randomisation.
  • Baseline Characteristics:
    • Groups were closely matched for age (55.2 ± 15.8 vs 54.3 ± 15.0 years), severity (APACHE II 19.8 ± 6.3 vs 20.1 ± 6.1), and hypoxaemia (PaO2/FiO2 113.9 ± 33.6 vs 113.5 ± 34.6 mmHg).
    • Duration of ventilation before randomisation was short and similar (2.1 ± 1.7 vs 2.0 ± 1.7 days), supporting “early” application.
  • Heterogeneity:
    • Pragmatic delivery across 29 ICUs increases clinical heterogeneity (local ventilation and co-intervention practices) but reflects real-world care.
    • Structured training programme (2306 staff trained across 198 sessions) aimed to minimise centre-to-centre variability in HFOV delivery.
  • Timing:
    • ARDS patients were enrolled relatively early (≈2 days of ventilation before randomisation) and HFOV commenced promptly after assignment.
  • Dose:
    • HFOV “dose” (mean airway pressure) on day 1 was 26.9 ± 6.2 cm H2O (with ongoing titration); conventional ventilation day 1 plateau pressure was 30.9 ± 11.0 cm H2O (not directly comparable to mean airway pressure).
    • Frequency and oscillatory parameters were applied consistently (day 1 HFOV frequency 7.8 ± 1.8 Hz; cycle volume 213 ± 72 mL).
  • Separation of the Variable of Interest:
    • Oxygenation separation: day 1 PaO2/FiO2 192 ± 77 (HFOV) vs 154 ± 61 mmHg (conventional).
    • Ventilation/acid–base separation: day 1 PaCO2 55 ± 17 (HFOV) vs 50 ± 19 mmHg (conventional); day 1 pH 7.30 ± 0.10 (HFOV) vs 7.35 ± 0.10 (conventional).
    • Co-intervention separation: neuromuscular-blocking agent use day 1 was 52.5% (HFOV) vs 41.6% (conventional).
  • Key Delivery Aspects:
    • Control group delivered tidal volume at baseline was at the upper end of the encouraged 6–8 mL/kg ideal body weight (baseline 8.6 ± 1.8 vs 8.3 ± 2.3 mL/kg; day 1 8.3 ± 2.9 mL/kg in conventional group).
    • HFOV protocol required transition back to conventional ventilation for weaning, reducing the duration of separation late in the course of illness.
  • Crossover:
    • 10/397 (2.5%) in the conventional group received HFOV after randomisation, potentially diluting any treatment effect.
  • Outcome Assessment:
    • Primary outcome (30-day mortality) is objective and time-bound, limiting measurement bias.
  • Statistical Rigor:
    • Revised sample size after interim feasibility/mortality assessment was prespecified as part of trial oversight; intention-to-treat analysis preserved randomisation benefits for primary comparison.
    • Primary outcome analysed with chi-square; adjusted logistic regression for survival reported with 95% confidence interval.

Conclusion on Internal Validity: Overall, internal validity appears moderate to strong for the primary mortality endpoint given central randomisation, near-complete outcome ascertainment, and clear between-group physiological separation, but is tempered by the open-label design, pragmatic (less standardised) conventional ventilation, and limited crossover/non-adherence.

External Validity

  • Population Representativeness:
    • Enrolled patients had typical ICU ARDS severity for the era (APACHE II ≈20; mean PaO2/FiO2 ≈114 mmHg), and were recruited across a wide range of UK ICUs.
    • Exclusions were largely safety- or feasibility-driven (e.g., airway obstruction/air trapping, prolonged ventilation), supporting broad applicability to adult ventilated ARDS populations.
  • Applicability:
    • Findings generalise well to high-income ICUs where conventional ventilation is clinician-directed and HFOV would require a training programme and dedicated ventilators.
    • Generalisation to centres using highly protocolised low tidal volume/high PEEP conventional ventilation (or to settings with extensive HFOV expertise) is less certain.
    • Resource-limited settings may not be able to reproduce the training/support infrastructure used in OSCAR.

Conclusion on External Validity: External validity is good for broadly similar adult ICU ARDS populations in high-income settings, particularly where “usual care” ventilation is pragmatic, but less certain for contexts with highly protocolised conventional ventilation or specialised, high-expertise HFOV programmes.

Strengths & Limitations

  • Strengths:
    • Large, multicentre pragmatic randomised trial with objective primary endpoint and complete short-term follow-up.
    • Early enrolment and prompt initiation of assigned strategy (median 6.1 hours), targeting the window where ventilator strategy might plausibly modify injury trajectory.
    • Explicit HFOV protocol with substantial implementation support (training and 24/7 troubleshooting).
  • Limitations:
    • Open-label design with potential co-intervention differences (sedation, paralysis, rescue therapies).
    • Conventional ventilation was encouraged but not standardised; delivered tidal volumes were commonly around 8 mL/kg ideal body weight early, limiting inference versus strictly protocolised lung-protective ventilation.
    • Small but present non-adherence (HFOV not initiated in 2.5%) and crossover from conventional ventilation to HFOV (2.5%).
    • HFOV protocol required transition to conventional ventilation for weaning, reducing sustained exposure to the experimental strategy later in illness.

Interpretation & Why It Matters

  • Clinical implication
    • In a pragmatic UK ICU setting, early HFOV did not reduce mortality or shorten ventilation/ICU stay compared with conventional ventilation, despite improving oxygenation.
    • Routine HFOV use to improve patient-centred outcomes in adult ARDS is not supported by OSCAR’s results.
  • Mechanistic signal
    • HFOV produced better PaO2/FiO2 but higher PaCO2 and lower pH, consistent with trade-offs between recruitment/oxygenation and effective CO2 clearance.
    • Greater neuromuscular blockade exposure suggests clinically relevant burdens and potential downstream risks (e.g., weakness, delirium pathways), even when mortality is unchanged.
  • Systems and policy relevance
    • The trial tested HFOV as it might be implemented at scale (training plus pragmatic comparator), making its neutral results particularly influential for de-implementation decisions.

Controversies & Subsequent Evidence

  • Discordant contemporaneous trial evidence (OSCAR vs OSCILLATE):
    • OSCILLATE (published in the same NEJM issue) stopped early for harm with higher mortality in the HFOV strategy compared with a protocolised conventional ventilation strategy.3
    • The accompanying editorial highlighted the mismatch between physiological plausibility (improved oxygenation) and patient-centred outcomes, and emphasised potential haemodynamic cost and sedation/paralysis requirements of HFOV strategies.1
  • Comparator quality and pragmatic design:
    • Correspondence argued that OSCAR’s conventional ventilation arm achieved tidal volumes just above 8 mL/kg (ideal body weight), whereas OSCILLATE’s comparator was more tightly protocolised to contemporary lung-protective targets; this difference could influence apparent treatment effects and contributes to between-trial interpretability challenges.2
    • Commentary emphasised that “HFOV versus conventional ventilation” is not a single intervention comparison: outcomes plausibly depend on the HFOV strategy (e.g., mean airway pressure targets) and on how optimised conventional ventilation is in the control arm.4
  • Haemodynamic and right-ventricular concerns:
    • Physiological studies demonstrate that higher mean airway pressures during HFOV can impair right-ventricular function and haemodynamics, providing a mechanistic pathway by which “open lung” HFOV strategies could worsen outcomes in some patients.8
  • Evidence syntheses after OSCAR/OSCILLATE:
    • Cochrane review concluded HFOV does not reduce mortality compared with conventional ventilation and may increase harm; conclusions were influenced by strategy heterogeneity and evolving conventional care over time.5
    • Later meta-analysis similarly found no mortality benefit and highlighted concerns regarding higher mean airway pressure strategies and adverse haemodynamic consequences.6
  • Subgroup hypothesis – severity of hypoxaemia:
    • An individual patient-data meta-analysis evaluated effect modification by baseline hypoxaemia severity, sustaining interest in whether any narrowly defined subgroup might benefit, but without establishing HFOV as routine therapy.7
  • Guideline impact (current practice):
    • Updated ATS clinical practice guideline recommends against routine HFOV in moderate or severe ARDS (strong recommendation, high certainty of evidence).9

Summary

  • In 795 adults with ARDS across 29 UK ICUs, early HFOV did not reduce 30-day mortality compared with conventional ventilation (41.7% vs 41.1%).
  • HFOV improved oxygenation but increased PaCO2 and lowered pH early after randomisation, with greater neuromuscular blockade exposure.
  • Clinical outcomes beyond mortality (ventilator-free days, duration of ventilation, ICU/hospital length of stay) were similar between groups.
  • The comparator arm reflected pragmatic conventional ventilation; delivered tidal volumes were typically around 8 mL/kg ideal body weight early in the trial.
  • Subsequent evidence (including OSCILLATE and later syntheses) shifted practice away from routine HFOV in adult ARDS and informed guideline recommendations against its routine use.

Overall Takeaway

OSCAR is a landmark pragmatic trial demonstrating that early HFOV, as deliverable across a broad network of UK ICUs with substantial training support, improves oxygenation but does not reduce mortality or shorten recovery in adult ARDS. Alongside contemporaneous and subsequent evidence, it helped drive de-implementation of routine HFOV in adult ARDS and reinforced the primacy of optimised conventional lung-protective ventilation and evidence-based adjuncts.

Overall Summary

  • Early HFOV in adult ARDS did not improve survival versus conventional ventilation in a pragmatic multicentre trial, despite better oxygenation and greater neuromuscular blockade exposure.

Bibliography