Publication

• Title: Effect of Lower vs Higher Oxygen Saturation Targets on Survival to Hospital Discharge Among Patients Resuscitated After Out-of-Hospital Cardiac Arrest: The EXACT Randomized Clinical Trial
• Acronym: EXACT
• Year: 2022
• Journal: JAMA
• Citation: Bernard SA, Bray JE, Smith K, et al; for the EXACT Investigators. Effect of lower vs higher oxygen saturation targets on survival to hospital discharge among patients resuscitated after out-of-hospital cardiac arrest: the EXACT randomized clinical trial. JAMA. 2022;328(18):1818-1826.

Context & Rationale

• Background

  • After return of spontaneous circulation (ROSC), oxygen is routinely delivered at high inspired fractions, making hyperoxaemia common early in the post-resuscitation period.
  • Observational work has linked hyperoxaemia to worse outcomes after cardiac arrest, but confounding and indication bias limit causal inference.
  • Conversely, even brief hypoxaemia may exacerbate secondary brain injury in the post-arrest phase; thus the optimal early oxygen strategy is uncertain.
  • Contemporary clinical uncertainty and the competing risks of hyperoxaemia vs hypoxaemia in early post-arrest care were highlighted alongside EXACT in an accompanying editorial.¹
• Research Question

  • In adults resuscitated after out-of-hospital cardiac arrest (OHCA), does targeting a lower oxygen saturation (SpO₂ 90–94%) from the prehospital phase through initial ICU assessment improve survival to hospital discharge compared with a higher target (SpO₂ 98–100%)?
• Why This Matters

  • Oxygen titration is ubiquitous, rapidly modifiable, and implementable at scale in EMS and ED workflows.
  • Defining a safe and effective early oxygen target could reduce mortality and neurologic disability, and standardise post-resuscitation care pathways.
  • Given the narrow therapeutic window, even modest shifts in target ranges could have large population impact.

Design & Methods

• Research Question

  Does a lower SpO₂ target (90–94%) vs a higher SpO₂ target (98–100%) after OHCA improve survival to hospital discharge?
• Study Type

  • Multicentre, parallel-group, randomised clinical trial with prehospital randomisation.
  • Enrolment: May 3, 2017 to March 10, 2020.
  • Setting: Two Australian states (Victoria and South Australia), involving two EMS systems and 15 receiving hospitals.
• Population

  • Core inclusion (prehospital): Adults (≥18 years) with OHCA of presumed cardiac cause, sustained ROSC, and coma (Glasgow Coma Scale ≤8).
  • Airway/monitoring requirement: Advanced airway in situ (endotracheal tube or supraglottic airway) and pulse oximetry available.
  • Oxygenation threshold at randomisation: SpO₂ ≥95% while receiving high inspired oxygen (eg, >10 L/min via reservoir bag or FiO₂ 1.0 on ventilator).
  • Key exclusions: Obvious non-cardiac aetiology (eg, respiratory, trauma, hanging, drowning), pregnancy, pre-arrest dependence in activities of daily living or known treatment limitations, and home oxygen therapy.
  • Screening yield (trial flow): 647 eligible; 428 randomised.
• Intervention

  • Target: SpO₂ 90–94% (“lower target”).
  • Duration: From prehospital randomisation through ED care until the first arterial blood gas measurement in ICU.
  • Safety provisions: 100% oxygen permitted for subsequent intubation and for hypoxic events.
• Comparison

  • Target: SpO₂ 98–100% (“higher target”).
  • Duration: Same intervention window (prehospital → ED → ICU to first arterial blood gas).
• Blinding

  • Prehospital and in-hospital clinicians were not blinded (oxygen target is visible at bedside).
  • Patients were unconscious at enrolment.
  • Outcome assessment at 12 months was performed by assessors blinded to treatment allocation.
  • The primary endpoint (survival to discharge) is objective, reducing detection bias despite open-label delivery.
• Statistics

  • Primary analysis: Compared groups as randomised (with exclusion of a small number of non-consented patients in South Australia).
  • Planned sample size: 1416 participants (trial stopped at 428).
  • Effect measures: Logistic regression for binary outcomes (reporting odds ratios), with additional models for other outcome types as appropriate.
• Follow-up

  • In-hospital follow-up to discharge for primary and most secondary outcomes.
  • Telephone follow-up at 12 months for survival and patient-centred outcomes (available for 80.3% of hospital survivors).

Key Results

This trial was stopped early. Recruitment ceased after 428 of the planned 1416 participants due to the COVID-19 pandemic and associated disruption to research operations.

• Oxygen separation was achieved (eg, ED arrival oxygen flow median 2 vs 12 L/min; ED first PaO₂ median 95 vs 130 mm Hg; ICU first PaO₂ median 99 vs 114 mm Hg), but SpO₂ values in both groups were often high by ICU arrival.
• Prespecified subgroup finding (primary outcome): Evidence of heterogeneity by bystander CPR status (interaction P=0.01), with OR 0.48 (95% CI 0.32–0.74) in those receiving bystander CPR and OR 1.96 (95% CI 0.79–4.86) in those without bystander CPR (interpret cautiously given multiplicity and early stopping).
• Serious adverse events (reported in the manuscript text): Sustained hypoxic events (SpO₂ <90%) unresponsive to 100% oxygen occurred in 5 (2.3%) vs 3 (1.4%) participants; rearrest in the setting of hypoxia (SpO₂ <90%) occurred in 3 (1.4%) participants, all in the lower-target group.

Internal Validity

• Randomisation and allocation:
  • Prehospital randomisation enabled early intervention, but also introduces operational exclusions (eg, unscreened cases, lack of randomisation envelope) that can affect representativeness.
  • Baseline characteristics were broadly similar across groups (eg, age median 65 years in both; male ~77% in both; bystander CPR ~80% in both).
• Attrition and analysis population:
  • 428 randomised; 425 included in primary analysis (214 vs 211), with exclusions driven by consent procedures in one jurisdiction and withdrawal.
  • Follow-up to hospital discharge was complete; 12-month follow-up was incomplete (80.3% of hospital survivors), limiting certainty for longer-term patient-centred outcomes.
• Performance and detection bias:
  • Open-label oxygen targeting can influence co-interventions (eg, ventilator settings, decisions around escalation), though the primary endpoint is objective.
  • Neurologic status at discharge was assessed using Cerebral Performance Category (CPC), which can be influenced by documentation quality and discharge timing.
• Intervention fidelity and physiological separation:
  • Clear differences in oxygen delivery were observed in the prehospital/ED phase (eg, ED arrival oxygen flow and FiO₂ distribution), supporting protocol adherence.
  • The lower-target strategy substantially increased hypoxaemia events (SpO₂ <90%), indicating that the achieved “dose” included frequent excursions below the intended lower target range.
• Statistical considerations:
  • Early stopping markedly reduced power, increasing uncertainty around both benefit and harm estimates.
  • The primary outcome effect estimate was borderline (P=0.05), and interpretation should incorporate the under-recruitment and planned interim monitoring framework.
• Internal validity conclusion: The trial’s randomised design and objective primary endpoint support causal inference; however, open-label delivery, early termination, and frequent hypoxaemic excursions in the lower-target arm introduce meaningful uncertainty about whether harm is inherent to oxygen restriction per se vs difficulty safely maintaining the intended target window in routine care.

External Validity

• Population applicability:
  • Results apply most directly to adult OHCA patients with presumed cardiac cause, sustained ROSC, coma, and an advanced airway, who are already sufficiently oxygenated at randomisation (SpO₂ ≥95%).
  • Patients with primary respiratory arrest, severe early hypoxaemia, trauma, drowning, or other non-cardiac causes were excluded, limiting generalisability to those groups.
• System and workflow applicability:
  • Conducted in two Australian EMS systems with established post-arrest care pathways and receiving hospitals capable of early ABG measurement and ICU admission.
  • Implementation elsewhere may be constrained by pulse oximetry reliability, ability to blend oxygen/air, transport times, and local targets for ventilation and haemodynamics.
• External validity conclusion: Moderate—highly relevant to similar EMS/hospital systems treating adult cardiac-cause OHCA with early ROSC, but less generalisable to arrests of respiratory aetiology or settings without consistent monitoring and oxygen blending capability.

Strengths & Limitations

• Strengths:
  • Randomised, early (prehospital) intervention addresses a clinically important and common exposure window.
  • Multicentre design spanning two EMS systems and 15 hospitals enhances pragmatic relevance.
  • Objective primary outcome (survival to discharge) with complete in-hospital follow-up.
  • Detailed reporting of oxygen delivery and oxygenation measures demonstrates physiological separation.
• Limitations:
  • Stopped early due to COVID-19 (428/1416 planned), reducing power and increasing uncertainty around effect estimates.
  • Open-label design with potential for co-intervention and withdrawal-of-care influences.
  • Lower-target arm experienced frequent hypoxaemia (SpO₂ <90%), suggesting practical difficulty maintaining a narrow low target range in real-world care.
  • Selected population (already well oxygenated at randomisation; presumed cardiac aetiology; advanced airway) may not reflect all post-arrest patients.
  • Long-term outcomes were incompletely captured (12-month follow-up in 80.3% of hospital survivors) and not powered for between-group inference.

Interpretation & Why It Matters

• Clinical signal

  A lower oxygen saturation target (SpO₂ 90–94%) did not improve survival and was associated with more pre-ICU hypoxaemia and hypoxaemia with rearrest—suggesting that aggressive early oxygen restriction may be unsafe in routine post-OHCA care.
• Mechanistic inference

  If hyperoxaemia contributes to reperfusion injury, the benefit of avoiding it may be outweighed by the harm of even brief or intermittent hypoxaemia when implementing a narrow low SpO₂ target early after ROSC.
• Practice implication

  The trial supports a cautious “avoid extremes” strategy: titrate down from 100% oxygen when feasible, but prioritise preventing hypoxaemia—especially during prehospital transport and ED stabilisation.

Controversies & Subsequent Evidence

• Was the lower target actually “hypoxaemia”?
  • In correspondence, concerns were raised that SpO₂ 90–94% may represent clinically relevant hypoxaemia (and may sit below typical recommended targets), potentially reframing the comparison as “mild hypoxaemia vs normoxaemia” rather than “normoxaemia vs hyperoxaemia.”²
  • The trial investigators replied that the target choice was informed by prior phase 2 data and meta-analytic evidence, and emphasised the need for additional high-quality trials while acknowledging the safety signal around hypoxaemia.³
• Ventilation and PaCO₂ as potential confounders:
  • Median first PaCO₂ in the ED was 60 mm Hg in both groups, raising questions about hypoventilation and its interplay with oxygenation, perfusion, and neurological injury—an issue discussed in correspondence and reply.²³
• How does EXACT fit with other randomised evidence?
  • The BOX oxygenation component (PaO₂ targets in comatose OHCA survivors) did not show clear benefit of a lower vs higher oxygenation strategy within broadly normoxaemic ranges, suggesting that avoiding extremes may be more important than fine target optimisation in ICU-phase care.⁴
  • Patient-level meta-analysis of randomised data (pre-EXACT) suggested uncertainty regarding benefit from conservative oxygen strategies after cardiac arrest, with the key practical concern remaining avoidance of hypoxaemia during implementation.⁵
• Guideline direction (most recent published guidance):
  • Recent post-resuscitation care guidelines continue to recommend titrating oxygen after ROSC to avoid both hypoxaemia and sustained hyperoxaemia, with emphasis on reliable monitoring and arterial blood gas confirmation when available.⁶
• Interpretive bottom line from contemporaneous commentary:
  • The accompanying editorial argued that any potential benefits of reducing hyperoxaemia must be balanced against the real-world risk of inadvertently causing hypoxaemia, particularly in the prehospital setting where monitoring and titration are less controlled.¹

Summary

• EXACT tested early oxygen titration to SpO₂ 90–94% vs 98–100% in comatose adult OHCA survivors from prehospital care through initial ICU assessment.
• The trial was stopped early (428/1416 planned), limiting power; nevertheless, survival to hospital discharge was lower in the 90–94% group (38.3% vs 47.9%; OR 0.68; P=0.05).
• The lower-target strategy substantially increased hypoxaemia prior to ICU admission and hypoxaemia associated with rearrest.
• Discharge neurological outcomes and length-of-stay metrics did not show clear benefit for the lower-target arm.
• Overall, the findings caution against aggressive early oxygen restriction to SpO₂ 90–94% after OHCA and support an approach focused on preventing hypoxaemia while avoiding sustained unnecessary hyperoxaemia.

Further Reading

Other Trials

• 2022 Kjaergaard J, Mikkelsen ME, Brown SM, et al. Oxygen targets in comatose survivors of cardiac arrest. N Engl J Med. 2022;387(16):1467-1476.
• 2018 Jakkula P, Reinikainen M, Hästbacka J, et al. Targeting low-normal or high-normal oxygenation and blood pressure after cardiac arrest: the COMACARE study. Intensive Care Med. 2018;44(12):2112-2121.
• 2018 Bray JE, Hein C, Smith K, et al; EXACT Investigators. Oxygen titration after resuscitation from out-of-hospital cardiac arrest: a multi-centre, randomised controlled pilot study (the EXACT pilot trial). Resuscitation. 2018;128:211-215.
• 2019 Young PJ, Bailey M, Bellomo R, et al; ICU-ROX Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group. Conservative oxygen therapy during mechanical ventilation in the intensive care unit. N Engl J Med. 2019;380(11):1029-1038.
• 2020 Young PJ, Bailey M, Bellomo R, et al. Conservative oxygen therapy in critically ill patients with suspected hypoxic-ischaemic encephalopathy. Intensive Care Med. 2020;46(12):2411-2422.

Systematic Review & Meta Analysis

• 2020 Young PJ, Bailey M, Bellomo R, et al; ICU-ROX Investigators. Conservative or liberal oxygen therapy after cardiac arrest: an individual-level patient data meta-analysis of randomised controlled trials. Resuscitation. 2020;157:15-22.
• 2015 Helmerhorst HJ, Roos-Blom MJ, van Westerloo DJ, de Jonge E. Association between arterial hyperoxia and outcome in subsets of critical illness: a systematic review, meta-analysis, and metaregression of cohort studies. Crit Care Med. 2015;43(7):1508-1519.
• 2025 Holmberg MJ, Ikeyama T, Garg R, Drennan IR, et al. Oxygen and carbon dioxide targets after cardiac arrest: a systematic review and meta-analysis. Resuscitation. 2025;211(8):110620.

Observational Studies

• 2010 Kilgannon JH, Jones AE, Shapiro NI, et al. Association between arterial hyperoxia following resuscitation from cardiac arrest and in-hospital mortality. JAMA. 2010;303(21):2165-2171.
• 2018 Elmer J, Scutella M, Pullalarevu R, et al. The association between hyperoxia and patient outcomes after cardiac arrest: analysis of a high-resolution database. JAMA Cardiol. 2018;3(4):314-321.
• 2021 McKenzie N, Finn J, Dobb G, et al. Non-linear association between arterial oxygen tension and survival after out-of-hospital cardiac arrest: multicentre observational study. Resuscitation. 2021;158:130-138.

Guidelines

• 2025 Nolan JP, Sandroni C, Böttiger BW, et al. European Resuscitation Council and European Society of Intensive Care Medicine guidelines 2025: post-resuscitation care. Resuscitation. 2025;215:110809.
• 2021 Nolan JP, Sandroni C, Böttiger BW, et al. European Resuscitation Council and European Society of Intensive Care Medicine guidelines 2021: post-resuscitation care. Intensive Care Med. 2021;47(4):369-421.

Notes

• The “lower target” strategy in EXACT aimed for SpO₂ 90–94% but was associated with frequent excursions below 90%, underscoring the practical challenge of maintaining a narrow lower target range during dynamic prehospital and ED care.

Bibliography

• 1 Elmer J, Guyette FX. Early oxygen supplementation after resuscitation from cardiac arrest. JAMA. 2022;328(18):1811-1813.
• 2 Wetsch WA, Böttiger BW. Lower vs higher oxygen saturation targets and survival to hospital discharge among patients resuscitated after out-of-hospital cardiac arrest. JAMA. 2023;329(9):766-767.
• 3 Bernard SA, Bray JE. Lower vs higher oxygen saturation targets and survival to hospital discharge among patients resuscitated after out-of-hospital cardiac arrest—In Reply. JAMA. 2023;329(9):767.
• 4 Kjaergaard J, Mikkelsen ME, Brown SM, et al. Oxygen targets in comatose survivors of cardiac arrest. N Engl J Med. 2022;387(16):1467-1476.
• 5 Young PJ, Bailey M, Bellomo R, et al; ICU-ROX Investigators. Conservative or liberal oxygen therapy after cardiac arrest: an individual-level patient data meta-analysis of randomised controlled trials. Resuscitation. 2020;157:15-22.
• 6 Nolan JP, Sandroni C, Böttiger BW, et al. European Resuscitation Council and European Society of Intensive Care Medicine guidelines 2025: post-resuscitation care. Resuscitation. 2025;215:110809.