Conservative Vs Conventional Oxygen therapy : How much or how low ?

Massimo Girardis, Stefano Busani and Mervin Singer et el  published a single center study in October, 2016 issue of JAMA comparing conservative oxygen therapy with conventional one in critically ill patients. 



All adult patients admitted to an ICU in Italian hospital from 2010 to 2012. Following were excluded pregnancy, ICU readmission, a decision to withhold life-sustaining treatment, immunosuppression or neutropenia, and enrollment in another study.


Patients were randomized to control (PaO2 values up to 150 mm Hg and an SpO2
between 97% and 100%, TOTAL 244 patients) or treatment group (lowest possible FiO2 to maintain the PaO2 between 70 and 100 mm Hg or SpO2 values between 94% and 98% TOTAL 236 patients)


There was significantly lower mortality in ICU in the treatment group (11.6%)  compared
with the conventional group (20.2%) (absolute risk reduction, 0.086 [95% CI, 0.017-0.150]; relative
risk, 0.57 [95% CI, 0.37-0.90]; P = .01). Hospital mortality was also lower in treatment group (not a predefined outcome). There was low incidence of shock, and duration of mechanical ventilation.

The trial was terminated early due to low inclusion rate.


Single-center open-label study, albeit of reasonable size, conducted in the ICU of a university hospital and stopped early for low inclusion rate.

The sample size did not allow a detailed analysis of the effects of hyperoxia in different population subsets.



Life can not sustain without oxygen.  Both hypoxia and hyperoxia have been associated with adverse outcomes. de Jonge et al demonstrated U-shaped relationship between PaO2 and in-hospital mortality, the lowest of the mortality being at PaO2 values of 110–150 mmHg; mortality increased both at PaO2 values <67 and >225 mmHg.

Damiani E showed association between arterial hyperoxia and increased mortality in critically ill patients.  Possible explanation for the effects of hyperoxia are detrimental effects of hyperoxia on the innate immune  system,  attenuation of cytokine production by human leukocytes , structural changes within alveolar macrophages, with a significant impairment of their antimicrobial activity and a marked reduction in the production of inflammatory cytokines in response to stimulation.

Hyperoxia has also shown to be harmful in the post-cardiac arrest, traumatic brain injury, and ischemic
stroke patients. In small animals, 100% oxygen can be lethal in a matter of hours.

This study clearly points to the dangers of hyperoxia, though a large study is warranted to confirm the findings.  Another recent trial by Panwar et el, demonstrated no difference in the outcome in conservative(SpO2) of 88-92% [n = 52]) vs liberal oxygen treatment(SpO2 of greater than or equal to 96%[n=51]) in ICU patients.

How about hypoxia

The original low tidal volume study(ARDSnet) used PaO2 of 55 or above as their target to treat ARDS.  At very high
altitudes, humans can survuve with PO2 values of 50 mm Hg or less. Similarly, the developing fetus near delivery has PaO2 values often below 30 mm Hg.  However, the arterial oxygen is not equivalent to tissue oxygenation.  Tissue oxygenation also depends on cardiac output as well as hemoglobin’s oxygen carrying capacity. Furthermore, many compensatory mechanisms by which humans adapt to hypoxia may be blunted in the disease states. Systemic inflammation and organ dysfunction may impair ventilation, create poor ventilation-perfusion matching, limit cardiac function, disrupt proper control of systemic perfusion distribution, impair tissue oxygen extraction, and limit the hemoglobin response.

Taking all this in to account, traditionally,  arterial PaO2 of  55 mm Hg, hemoglobin levels of at least 7 g/dL (but perhaps as high as 9–10 g/dL in high O2 demand states), and cardiac indices above 2 L/min/m2 are the target for the tissue oxygenation.

One recent retrospective study looked in to effect of hypoxia in survivors of ARDS who received ECMO found no deleterious effect. However, there are no randomized trials which compared hypoxia vs normoxemia.  Recently, the UK and Australian Benefits of Oxygen Saturation Targeting (BOOST) II trials showed an oxygen saturation target of 85% to 89%, rather than 91% to 95%, may increase the risk for death or disability at 2 years corrected age in infants born before age 28 weeks.  However, results from neonates may not be applicable to adults. 

We need future trials to compare optimal lower threshold of PaO2 along with ideal monitoring mechanisms for tissue oxygenation.









Please follow and like us: