skip_previous play_arrow skip_next
00:00 00:00
playlist_play chevron_left

evidence based medicine

BOX Trial – Oxygenation targets in OHCA Comatose survivors

Dr Swapnil Pawar January 12, 2023 184

share close
  • cover play_arrow

    BOX Trial – Oxygenation targets in OHCA Comatose survivors
    Dr Swapnil Pawar


Oxygen Targets in Comatose Survivors of Cardiac Arrest 

Schmidt H, Kjaergaard J, Hassager C, et al. Oxygen Targets in Comatose Survivors of Cardiac Arrest. N Engl J Med. 2022 Oct 20;387(16):1467-1476. doi: 10.1056/NEJMoa2208686. Epub 2022 Aug 27. PMID: 36027567.

Blog written by – Dr Jose Chacko



Following successful resuscitation from cardiac arrest, reperfusion injury may exacerbate damage to the brain. The appropriate oxygenation target during mechanical ventilation in comatose survivors of out-of-hospital cardiac arrest is unknown. Findings from observational studies and animal models suggest worsening of brain injury with hyperoxia. In the ICU-ROX study, subgroup analysis suggested improved outcomes with lower oxygen targets in patients with hypoxic-ischemic encephalopathy. The Blood Pressure and Oxygenation Targets in Postresuscitation Care (BOX) trial was aimed to evaluate outcomes with a restrictive compared with a liberal oxygenation strategy among comatose survivors of out-of-hospital cardiac arrest.

Population and design

The study was conducted in two tertiary care centers in Denmark between March 2017 and December 2021. The study included patients who attained return of spontaneous circulation (ROSC) within 20 minutes and remained comatose after out-of-hospital cardiac arrest from a presumed cardiac cause. The study excluded patients who suffered cardiac arrest due to a presumed non-cardiac cause if >4 hours had elapsed from ROSC to screening and if the systolic BP remained <80 despite supportive treatment. The BOX trial was a 2-by-2 factorial study with oxygen and blood pressure targets. The study used block randomization, stratified by site.

Restrictive group

In the restrictive group, PaO2 of 68 to 75 mm Hg (9 to 10 kPa) was targeted.


In the liberal group, the target PaO2 was 98 to 105 mm Hg (13 to 14 kPa)

Management in both groups

Patients in both groups were sedated and mechanically ventilated with temperature control to 36°C for at least 24 hours. After 24 hours, the core temperature was gradually allowed to rise, and sedation was weaned off. Arterial blood-gas analysis was performed at regular intervals for 120 hours (5 days) after randomization or until removal of the arterial catheter. The FiO2 was initially set at 0.3 in the restrictive and 0.6 in the liberal group and adjusted to the target PaO2 in each arm. The FiO2 was increased if the SaO2 dropped below 93%. Ventilation management was at the clinician’s discretion.

Sample size

The authors calculated a sample size of 732 patients to detect a 10% absolute difference in the primary outcome (death or severe disability at 90 d) with 80% power, assuming a mortality of 28% at a two-sided alpha level of 0.05. The study aimed to randomize 800 patients, with 400 in each group.


On the intention to treat analysis, 789 patients were included; 394 were in the restrictive group and 395 in the liberal group. Baseline characteristics were well-matched between the two groups. Restrictive vs. liberal: shockable rhythm 85 vs. 84.5%; witnessed arrest: 84.5 vs. 86%; bystander CPR: 89.2 vs. 85.8%; time to ROSC: mean duration of 21 min in both groups. Patients in both groups had similar P/F ratios on admission to the ICU. Separation of oxygenation levels was evident within 2–4 hours and maintained for the first 48 hours. The median duration of ventilation was 57 vs. 61 hours.


Outcome Restrictive (394 patients) Liberal (395 patients) Treatment effect
Primary outcome Death or severe disability (dependent for ADLs, coma, vegetative, brain dead) whichever occurred first at 90 d post-randomization 126 (32.0%) 134 (33.9%) 0.95 (0.75–1.21)   (not significant)
Death at 90 d 113 (28.7%) 123 (31.1%) 0.93 (0.72–1.20)  (not significant)
AKI requiring RRT 34 (8.6%) 47 (11.9%) 0.85 (0.69–1.03)  (not significant)
CPC at 90 d* 1 (1–5) 1 (1–5)
Modified Rankin Score at 90 d# 2 (0–6) 1 (0–6)
Montreal Cognitive Assessment score at 90 d 27 (24–29) 27 (24–28)
Neuron‐specific enolase at 48 h 17 (11–36) 18 (11–34)


*CPC 1: Good cerebral performance: conscious, alert, able to work, might have mild neurologic or

psychological deficit

#Modified Rankin Score 1: No significant disability despite symptoms; able to carry out all usual duties and activities

Modified Rankin Score 2: Slight disability; unable to carry out all previous activities, but able to look after own affairs without assistance

Adverse events

There was no difference in pre-specified adverse events. The most frequent adverse events included infection, bleeding, and seizures. No difference was observed in the primary outcome on pre-specified subgroup analysis of gender, age, presence of hypertension, renal impairment, COPD, STEMI, and shockable vs. non-shockable rhythm,


  • Adequately powered
  • Groups well matched at baseline
  • The largest RCT so far that has addressed the question of oxygenation targets following cardiac arrest.


  • The target PaO2 levels were attained after reaching the hospital, usually after admission to ICU. Separation in oxygen levels was achieved 2 h after ICU admission. This may be too late, especially because reperfusion injury and hyperoxia damage occur early.
  • The liberal group also had what may be considered to be within the “normal” oxygenation range (PaO2: 68–75 vs 98–105 mm Hg). There may be no appreciable difference in clinical outcomes within these two ranges. In the restrictive group, the PO2 levels were on the upper limit of the target range. The P/F ratios were generally high, indicating relatively normal gas exchange in both groups.
  • The study included only two centres from Denmark.
  • Nearly 90% of arrests were witnessed, with bystander CPR in almost 90% – which may be challenging to attain in other settings.
  • Lower mortality (28.7 vs. 31.1%) compared to other studies (TTM trials: 50%)
  • The 10% difference in the primary outcome may have been too optimistic, resulting in a less-than-ideal sample size.
  • Possible bias related to an unblinded study – especially in regard to end-of-life care
  • The results may not be extrapolatable to non-cardiac causes
  • Physical follow-up was limited due to the Covid pandemic
Rate it
Previous post
eCritCare Podcast
  • 380


Double Sequential External Defibrillation – A game-changer

Dr Swapnil Pawar January 1, 2023

play_arrow Double Sequential External Defibrillation – A game-changer Dr Swapnil Pawar   Alternate defibrillation strategies for refractory ventricular fibrillation Cheskes S, Verbeek PR, Drennan IR, et al.  N Engl J […]

Read more trending_flat