skip_previous play_arrow skip_next
00:00 00:00
playlist_play chevron_left



Dr Swapnil Pawar November 10, 2019 312

share close

Conservative Oxygen Therapy during Mechanical Ventilation in the ICU

The ICU-ROX Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group

Oxygen is a vital element in medical practice and among the most universally used agents for the treatment of critically ill patients. Guidelines for the provision of oxygen give recommendations for target oxygen saturations and for the weaning of oxygen therapy. Despite this, the titration of supplemental oxygen in mechanically ventilated patients is infrequent with resultant hyperoxia being common in the intensive care unit (ICU).

Hyperoxia can be defined as a state in which supraphysiological levels of oxygen are inspired and/or reach the arterial circulation. A formal definition for arterial hyperoxia does not exist, but a partial pressure of arterial oxygen (PaO2 ) higher than 120 mmHg (>16 kPa) has previously been characterised as mild hyperoxia and PaO2 >200 mmHg (26.7 kPa) as severe hyperoxia.

Physiologic Rationale –

Conservative oxygen therapy may contribute to acclimatisation and cellular adaptation to the lower ranges of normoxia, which may result in improved efficiency of ATP production and protection of mitochondria. From a physiological point of view, this makes perfect sense as the human body has evolutionarily adapted to successfully maintain aerobic metabolism with a fraction of inspired oxygen (FiO2 ) of 21% in ambient air at sea level. It provides the ideal environment for the eukaryotic cell that is perfectly capable of efficient oxygen consumption at partial pressures of arterial oxygen of 75-100 mmHg (10-13.3 kPa) and corresponding mitochondrial oxygen concentrations of approximately 11 mmHg (1.5 kPa).
The side effects of supraphysiological oxygenation can be roughly subdivided in cell damage, inflammation, pulmonary complications, neurological symptoms and vascular effects.

As we know, oxygen delivery and tissue oxygen tension not only depend on the arterial oxygen tension but also on perfusion and oxygen consumption.

  1. higher oxygen concentrations have been shown to increase peripheral vascular resistance and decreased cardiac output. As a result, oxygen delivery in the organs at risk may actually be impaired through peripheral vasoconstriction and bradycardia.
  2. hyperoxygenation is associated with a disturbing healthy balance between oxidants and antioxidants. Oxygen-free radicals, which are commonly referred to as reactive oxygen species (ROS), are versatile molecules with an important role in cell signalling and homeostasis.
  3. Pulmonary complications are more frequently encountered due to hyperoxia and include absorption atelectasis, inflammation and pulmonary oedema.
  4. Neurological symptoms include DCI

What’s Known –

  1. Excess oxygen delivery was reported to be a very common phenomenon, in which about 50% of the patients showed hyperoxemia and 4% in severe hyperoxemia. Helmerhorst 2016
  2. in a study of 36,307 patients admitted to the intensive care units (ICU), no difference in mortality was noted between the patients exposed to hyperoxia and those who did not. de Jonge E et al Crit Care 2008
  3. Page et al. (2018 Crit Care) found that there was an association between hyperxia and increased mortality (adjusted odds ratio[aOR] 1.95, 95% confidence interval[CI] 1.34–2.85).
  4. In a meta-analysis of randomized trials involving adults with acute illnesses, the use of oxygen without limitation according to achieve arterial oxygen saturation was associated with a higher rate of death than more restrictive approaches.
  5. In a single-center ICU trial published in JAMA in 2016 in which approximately two thirds of the patients were receiving invasive mechanical ventilation at the time of randomization, the use of conservative oxygen therapy, a therapeutic regimen designed to minimize exposure to high levels of oxygen, was associated with a lower rate of death and a higher number of ventilator-free days than usual oxygen therapy.
  6. The study by Helmerhost published in 2017 in Critical Care collected Data from all admissions with more than one PaO2 measurement were supplemented with anonymous demographic and admission and discharge data from the Dutch National Intensive Care Evaluation registry. Severe hyperoxia defined as PaO2 > 200 mm Hg was associated with higher mortality rates and fewer ventilator-free days in comparison to both mild hyperoxia and normoxia for all metrics except for the worst PaO2. Adjusted effect estimates for conditional mortality were larger for severe hyperoxia than for mild hyperoxia. This association was found both within and beyond the first 24 hours of admission and was consistent for large subgroups. The largest point estimates were found for the exposure identified by the average PaO2, closely followed by the median PaO2, and these estimates differed substantially between subsets. Time spent in hyperoxia showed a linear and positive relationship with hospital mortality.
  7. Conflicting results were also found in subgroups – patients with cardiac arrest, in which Elmer et al. showed a decreased survival (aOR 0.83, 95% 0.69–0.99, P = 0.04), [14] while Ihle et al. did not find any difference; [15] and 2) patients after traumatic brain injury(TBI), in which Asher et al. found a reduced mortality by hyperoxia while a contrary result was showed by Davis et al.
  8. The relationship between arterial hyperoxia and hospital mortality is quite consistent over several subgroups, but associations with delayed cerebral ischaemia, poor cerebral performance and disability have also been shown in patients after cardiac arrest, traumatic brain injury and stroke.[23,24] Also, a decline in ventilation-free days (28 days after admission) have been observed in patients where the average PaO2 over the ICU admission time was larger than 200 mmHg.
  9. a controversial recommendation by the World Health Organisation (WHO) stating that ‘adult patients undergoing general anaesthesia with endotracheal intubation for surgical procedures should receive 80% fraction of inspired oxygen intraoperatively and, if feasible, in the immediate postoperative period for 2-6 hrs’. The rationale for such measures is to improve oxidative killing of bacteria by neutrophils. However, not only was there insufficient evidence from methodologically limited studies with heterogeneous outcomes, but other effects of hyperoxic ventilation were completely ignored in this international recommendation.

Despite this evidence, there is a lack of good clinically directive data regarding strategies for oxygen administration in adults undergoing mechanical ventilation.

Hypothesis –

conservative oxygen therapy would result in more ventilator free days than usual oxygen therapy in adults who were expected to undergo mechanical ventilation in the ICU beyond the day after recruitment.
Design: Randomized clinical trial across 21 ICUs in Australia and New Zealand conducted by the ICU-ROX investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group. Randomization using computer-generated random numbers; block randomization in a 1:1 ratio and stratified by trial center.

Study population: 

Eligibility: ≥18 years of age who were expected to receive mechanical ventilation in the ICU beyond the day after recruitment. Less than 2 h of invasive or non-invasive ventilation. Those not enrolled within the 2-hour time window were categorized as missed. 

Exclusion criteria: Hyperoxia appropriate in the judgment of the physician (e.g., CO poisoning) 

Hyperoxia needs to be avoided (e.g., COPD, paraquat poisoning, etc)

Death imminent, no active measures of treatment 

Life expectancy less than 90 d due to the underlying medical condition 

Drug overdose, including alcohol intoxication 

Long term NIV dependence 

Neurological conditions: Guillain–Barré syndrome, cervical cord injury above C5, muscular dystrophy, or motor neuron disease

In the “best interests” of the patient, non-enrollment was appropriate


Enrolled in previous studies  

Intervention: “Conservative group”: The alarm was set to sound when the Spo2 was 97%. Exposure to a Spo2 of 97% or was higher minimized.

Control: “Usual care”: No specific methods to limit the Fio2 or the Spo2. No upper alarm limits for Spo2. Fio2 of less than 0.3 during invasive ventilation was discouraged.

Care common to both groups: Minimum SaO2 of 90%; minimum PO2 of 60 mm Hg (regardless of SaO2). FiO2 titrated accordingly. Other aspects of care, including ventilator weaning and extubation practices, were at the discretion of the treating clinician. No restriction on oxygen when transported outside the ICU; an increase in FiO2 was considered standard practice for procedures including as bronchoscopy, suctioning, or tracheostomy. Usual oxygen therapy while on ECMO. Assigned oxygen-therapy strategy until discharge from the ICU or 28 days after randomization, whichever was earlier.

Statistical analysis: Ventilator-free days assumed to be16.4±11.3 days. The absolute difference of 2.6 ventilator-free days was assumed at day 28. 1000 patients for 90% power, an alpha level of 0.05. Analysis- the intention to treat basis. 

3366 screened
848 excluded 
8 did not consent
1510 missed the window 
1000 randomized
Con: 484; Lib: 481 (after refusal for evaluation of primary endpoint)
Baseline characteristics: well matched
Clear separation of FiO2 between groups:
Time on a FiO2 of 0.21: Median 29 hours (5 to 78) vs. 1 hour (interquartile range, 0 to 17)

Time with SaO2 97% or less: Median 27 hours (11 to 63.5) and 49 hours (interquartile range, 22 to 112)

The mean FiO2 during the first 10 days of mechanical ventilation in the ICU and the lowest and highest FiO2 values until day 28

Time-weighted mean Pao2 values during the first 10 days of mechanical ventilation

Both were lower in the conservative group. 

Primary outcome: Ventilator-free days from randomization till day 28. Patients who died: 0 ventilator-free days. 

Median: Conservative vs. liberal: 21.3 (0.0 to 26.3) vs. 22.1 (0.0 to 26.2)

Mean: 15.5±11.8 vs. 16.0±11.5 

No. of days of ventilation among survivors: 2.95 (2.61 to 3.33)

3.11 (2.76 to 3.51)

Secondary outcomes

All-cause mortality at 90 d: 166/479 (34.7) vs. 156/480 (32.5); odds ratio, 95% CI: 1.10 (0.84 to 1.44)

All-cause mortality at 180 d: 170/476 (35.7) vs. 164/475 (34.5); 1.05 (0.81 to 1.37)

Rate of unemployed patients at 180 days, among those who were employed prior to the illness: No difference between groups

Cognitive function at 180 d, assessed using the Telephone Interview for Cognitive Status (TICS) questionnaire. Higher scores, better cognitive status. No difference between groups 

Health-related quality of life at 180 d assessed using the five-level EuroQol five dimensions (EQ-5D-5L). In the mobility and personal-care domains of the quality-of-life assessment, greater frequency of moderate problems and a lower frequency of severe problems than with usual-oxygen; no difference in other domains. 

The Glasgow outcome score was used in patients with a primary neurological problem. 

Ventilator-free days at 28 d in patients with HIE: 87 (18.0%) vs. 79 (16.4%) patients. 21.1 (0 to 26.1) days vs. 0 (0 to 26) (p=0.007)

Post-hoc analysis
Post-hoc analysis of HIE: 37 of 86 patients (43%) in the conservative-oxygen group vs. 46 of 78 (59%) in the usual-oxygen group (relative risk, 0.73; 95% CI, 0.54 to 0.99; hazard ratio, 0.67; 95% CI, 0.43 to 1.03);

Unfavorable outcome on the Extended Glasgow Outcome Scale:  43 of 78 patients (55%) vs. 49 of 72 (68%), respectively (relative risk, 0.81; 95% CI, 0.63 to 1.05).

Patients who had acute brain pathologies other than acute ischemic hypoxic encephalopathy appeared to do better with usual care oxygen 

Patients with sepsis also did better in the usual care oxygen group (Day 180 death 25.9% vs 20.1%)

Strengths –

  1. RCT ( Multicenter)
  2. Included the sickest cohort ( 33 % mortality)
  3. Tested a wide variety of outcomes – thus adds significant collateral information on this topic and in general about critically ill patients.
  4. Also tested process outcomes. 
  5. Highlights the point – Oxygen is the drug and need to be used carefully.
  6. The signal in Post cardiac arrest population – Hyperoxemia is harmful.

Limitations –

  1. Unblinded
  2. Did not evaluate mortality as the primary outcome
  3. Some outcome variables (e.g., employment status) were compared only among survivors.
  4. Some data, particularly related to the quality of life and cognition, were missing.
  5. Eligible patients who were not enrolled in the trial had less severe illness and lower rates of death than those who were enrolled. Accordingly, study findings may not apply to patients with less severe illness.
  6. did not include mandates regarding weaning or extubation in the protocol, changes in the Fio2, Spo2, and Pao2 that occurred because of treatment assignment may have affected clinicians’ decisions to wean and extubate particular patients. Authors allowed clinicians to increase oxygen in the two groups in some specific circumstances thus allowed hyperoxia.

In many of the previous trials, liberal oxygen interventions were considerably more liberal than the oxygen regimen used in the usual-care group, and relatively few of the patients were critically ill. Different results may also be found with different regimens for conservative oxygen therapy.

Summary –

Much needed RCT

Challenged the dogma of minimum 30-40% FiO2 in all ICU patients on mechanical ventilation

Interesting Subgroup analysis which will need further exploration.

Rate it
Previous episode
Similar episodes