Effect of Conservative vs Conventional Oxygen Therapy on Mortality Among Patients in an Intensive Care Unit, by Massimo Girardis, Stefano Busani et al. 
How easy is to break dogmas? Well, if you consider dogmas as a mere formality to prevent anarchy is not that hard. Life would be much easier if these organized critical care guerrillas didn’t exist. You could give a lecture about the quintessence of critical care in about 15 minutes. Hemoglobin target? 14g/dl, next? Traumatic brain injury has a lot to do with inflammation, give’em roids, next? Fever equal antibiotics, the more, the better! Next? CPR, stroke, oxygen…and the list goes on and on. I’m grateful for all those who stood up against common beliefs before it was cool!
So, here we are, discussing about mitochondria’s best lover. Mervyn Singer…I mean, oxygen! Oxygen! Probably the most prescribed drug in the world. Maybe behind normal saline and furosemide, because if the guy on the night shift gave NS the fist thing to do in the morning is a Lasix breakfast. Focus! Again, oxygen. Probably all of us in the past loved to see our patients with SaO2 of 100%, some still do. You will increase your oxygen delivery and all those neglected heart, liver, kidney and brain cells would be supplied with a huge amount of double-O and everybody will be happy. But, even my mama knew that oxygen also had detrimental effects, and every time I got bruised there she was with some hydrogen peroxide. But also other brilliant minds thought like her, and over the years we accumulated evidence about the harms of hyperoxia: atelectasis, vasoconstriction, reactive oxygen species and so on. Also good clinical evidence in both PROXY  and AVOID  trials, not only showing the safety in targeting normoxya but also that hyperoxya is harmful.
The stage was set for an ICU trial, and from 2010 to 2012 a single-center, open label, RCT was performed in Italy. Patients older than 18y/o who were expected to stay in ICU for 72h or more were eligible. The exclusion criteria was the usual plus acute de-compensation of COPD and ARDS with P/F ratio <150. Patients were randomized to a control and intervention group with specific targets for FiO2, PaO2 and SpO2:
The ICU team had full discretion of patients’ management with respect to good clinical practice.
Primary outcome was ICU mortality. They planned to enroll 660 patients to detect and absolute difference of 6% from a 23% ICU mortality rate. The study was stopped when 480 patients were enrolled after the results of an interim analysis not defined a priori. Interim analysis are tricky, for safety they are the best, but for efficacy, the alpha spending function ghost always come to my mind. But in this specific case they had a very noble purpose for stopping the trial. In 2012 huge earthquake damaged the Modena University Hospital (where the trial was held), both logistical and financial situation changed, recruitment ratio went down and after the appropriate consultations the trial was stopped.
Data and results
After few exclusions, 20 in intervention and 26 in control, the modified intention-to-treat population was 216 and 218 respectively. Although randomized, we can see few differences between groups: control group had slightly older patients, medical ICU admission, respiratory failure, shock, liver failure, infections and higher SAPS II. Of course these unbalances might happen just by chance and probably are not statistically relevant, but personally I don’t like the method of sealed envelopes for randomization in open label trials, which was what they used, specially if you are not doing random block sizes randomization, but maybe that’s just me. The patients’ main characteristics were:
The control group had higher PaO2 (102mmHg) and FiO2 (0,39) than the intervention group with PaO2 (87mmHg) and FiO2 (0,36), p<0.001. Regarding the primary outcome, 25 patients (11,6%) in intervention and 44 (20.2%) in control died during their ICU stay, p=0.01. Secondary outcomes: the intervention group had less in hospital mortality, new organ failure during ICU stay, bacteremia, and mechanical ventilation free-hours. The Kaplan-Meier curves are different from the start with higher probability of survival for the intervention group. Post-hoc analysis of ICU mortality showed decreased mortality in intervention subgroups: Respiratory failure at admission and mechanical ventilation at admission, interestingly the subgroups that makes everyone anxious to crank the O2.
I always thought that one study about oxygen target for critically ill patients were needed and the Oxygen-ICU trial somehow would fulfill my wishes. And it did, mostly because it’s consonant with my previous beliefs. But it wasn’t a flawless study. The earthquake sent by mother nature wasn’t investigators’ foul, but the study was terminated early; it was a single center open label; maybe some issues with randomization and maybe a Cox proportional hazards model would fit better than a Log-rank test…just maybe. The secondary endpoints, in a certain way are consonant with the possible pathophysiological process of hyperoxia leading to innate immune system disfunction. But the results for the primary outcome are there, the AVOID and PROXY trials are there and even some conservative doctor cannot help to see that there is something here, that 100% SpO2 doesn’t give full power to our patients. Enough with the crap that it won’t harm the patients. It will.
1. Girardis M, Busani S, Damiani E, et al. Effect of Conservative vs Conventional Oxygen Therapy on Mortality Among Patients in an Intensive Care Unit: The Oxygen-ICU Randomized Clinical Trial. JAMA. 2016;316(15):1583-1589.
2. Meyhoff CS, Wetterslev J, Jorgensen LN, et al. Effect of high perioperative oxygen fraction on surgical site infection and pulmonary complications after abdominal surgery: the PROXI randomized clinical trial. JAMA. 2009;302(14):1543-1550.
3. Stub D, Smith K, Bernard S, et al. A randomized controlled trial of oxygen therapy in acute myocardial infarction Air Verses Oxygen In myocarDial infarction study (AVOID Study). Am Heart J. 2012;163(3):339-345.e331.