Since its first reported use in acute care in 1885, oxygen has become a mainstay of clinical practice. It’s the most used drug in emergency medicine today, where it’s typically prescribed to prevent hypoxemia by bringing the amount of oxygen in the blood to around 94–98%, towards normal levels.

This widespread use of oxygen has largely gone unquestioned: after all, it’s considered very safe and low levels can be dangerous. Yet, until very recently, there’s been a lack of experimental evidence supporting the use of oxygen therapy.

“We’ve never had any rigorous trials evaluating its impact on patient outcomes until the last five years,” says associate professor of medicine and critical care physician at Vanderbilt University Medical Centre, Matthew Semler. These trials represent a new wave of research that’s challenging conventional wisdom about oxygen: it’s a highly reactive molecule, so would very ill people be better off receiving a little less of it?

The results we have so far offer something of an answer, but there’s still a lot we don’t know. Adult studies have had conflicting findings and it’s not yet clear which patients might benefit most from being given conservative versus the standard liberal levels of oxygen. In children, however, lower targets do seem to be preferable.

“We now have a pretty good certainty that there’s not a huge treatment effect for conservative oxygen targets for everyone. But [oxygen therapy is] so common that it’s worth knowing: is there even a small treatment effect?” says Semler. This poses two follow-up questions he adds: can conservative oxygen therapy help most people in a small way? And can it help some patients in a big way?

“We now have a pretty good certainty that there’s not a huge treatment effect for conservative oxygen targets for everyone. But [oxygen therapy is] so common that it’s worth knowing: is there even a small treatment effect?”
Matthew Semler

A small improvement for most?

In theory, it makes sense that the bulk of critical care patients would see at least a small benefit from receiving less oxygen. “Oxygen is a horribly reactive thing that pulls electrons out of other molecules,” says Professor of Paediatric Intensive Care at University College London Great Ormond Street Institute of Child Health, Mark Peters. “If you’re exposed to a lot of supplemental oxygen for a long time, that may contribute to the severity of the biological injury you’re suffering.”

When too much oxygen is taken in, it can break down into free radicals – very reactive molecules that can cause cell damage. This can bring on inflammation that may be harmful to the body and brain.

The percentage of patients more likely to survive or spend less time on machines when receiving a conservative amount of oxygen.
The Lancet

Being more conservative with oxygen also follows the principle of minimum safe care: doing the least amount of intervention possible while ensuring patient safety. This can help minimise the risks that come with additional treatment while helping shortstay patients get out of hospital quicker, says Peters.

For instance, turning up oxygen may then mean that the patient needs more sedation – but if they are still on sedation, they can’t come off the ventilator. “A lot of trials have shown that the more aggressive [treatment] arm is harmful… nearly always, the less interventional group is better,” says Peters.

But in practice, evidence in favour of more conservative targets for adults isn’t decisive – although this approach wasn’t found to be harmful, either. In a review and meta-analysis of 25 trials including a range of critical care patients, there were more deaths in the hospital and at least 30 days later for those receiving liberal (94–99%) rather than conservative levels of oxygen. Yet there was no difference in illness between the groups at any of these points.

Other trials, including the ICU-ROX trial of 1,000 ICU patients, have found no difference in outcomes between conservative and liberal approaches. However, because there’s no standard definition for a conservative target, many studies report differing lower ranges of oxygen. Sometimes, these overlap with what would be given routinely. To really understand the treatment effect for entire groups of people, we need huge trials, says Semler. And a few are under way, such as the UK-ROX trial that will enrol 16,500 patients.

Evidence for lower targets is more convincing when it comes to children. A new, landmark study of 2,040 children – the Oxy-PICU trial – found that those receiving a conservative amount of oxygen (88–92%) versus a standard level (higher than 94%) were 6% more likely to either survive or spend less time on machines.

The researchers estimated that if this approach were rolled out across the country’s National Health Service, it could save 50 lives, 6,000 days spent in ICU beds, and save £20m per year in the UK alone.

Peters, who was the lead author on the study, says that the results are generalisable as the patients recruited were representative of the UK critical care population. They also align with other research, he adds. “There are three trials in kids, they all show a small signal in favour of the lower target… that adds to the credibility of our result.”

Bigger benefits for some

“There are a lot of conditions, especially for patients at the hospital, for which the administration of too much oxygen could cause that oxygen to break down into free radicals,” says Semler. And presently, we do have some idea of who those patients might be.

“Cardiac arrest is one of those groups where there’s real interest,” Semler explains. During cardiac arrest, the brain is starved of oxygen – but if that patient then receives a flood of oxygen, could that cause further damage to the brain? “The answer is that we’re still less certain than we would want to be,” he adds.

There have been some trials looking at conservative oxygen therapy in cardiac arrest patients, but again, the results are unclear. The BOX trial, which included 789 comatose patients following cardiac arrest, found a similar incidence of death and severe disability between conservative and liberal oxygen groups – yet, Semler notes, the estimates did favour the lower target.

Semler hopes that the ongoing international trial, Mega-ROX, should be able to provide some more definitive answers. With 40,000 total participants, the trial will compare the impacts of conservative versus liberal oxygen therapy on people with sepsis, brain injury, and following cardiac arrest. “It’ll be the biggest trial ever on oxygen targets,” he says.

Yet the question of who may benefit most from receiving less oxygen also raises another: who might be better off getting more of it? “Patients with severe infection, patients with sepsis or septic shock, and many of whom have acute respiratory distress syndrome. Those are the patients for whom there might be reasons why more oxygen would be better,” says Semler.

There are a few possible explanations why. One is that these patients, particularly in the case of sepsis, may not be getting enough oxygen in their cells – so they simply may benefit from having their levels topped up. Another could be that because oxygen is toxic to bacteria, it may help to slow or prevent an infection, Semler explains.

Some research has investigated oxygen targets for sepsis, but as with studies on other critical care populations there’s no conclusive answer as to which levels are best. Here, the Mega-ROX trial results may provide some insight.

Setting the right target

Randomised trials can help us uncover the treatment effect for patients within a group, but clinicians on the ground need to figure out which treatment is best for the patient in front of them. Someone’s age, sex, blood pressure, and more may all play a role here.

Ideally, this means setting personalised oxygen targets. We aren’t quite there yet, but Semler hopes that we’ll soon see large trials that consider a wide range of patient characteristics to help move the needle. This data could be used to create estimates for which type of oxygen target might be better for a given patient, he says.

“Oxygen is a horribly reactive thing that pulls electrons out of other molecules. If you’re exposed to a lot of supplemental oxygen for a long time, that may contribute to the severity of the biological injury you’re suffering.”
Mark Peters

Until then, it’s down to clinicians to consider the available evidence in deciding how they’ll administer oxygen. Conservative oxygen therapy is an existing treatment, so it can already be applied at a doctor’s discretion. “Most of my colleagues have already made the change to a lower target in critically ill kids who meet the criteria,” says Peters.

The number of days spent in ICU beds saved if conservative oxygen was rolled out across the NHS.
The Lancet

While the idea that we should use lower oxygen targets is gaining legs, the next hurdle is implementation: having this as a standard consideration in critical care decisions. “How to close that gap is a major challenge for health systems, even once we have very strong evidence,” says Semler. “It requires effort now, it requires personnel, and it requires focus and emphasis.”