Mortality Risk Seen From Liberal Oxygen Delivery in MI, Stroke, Other Acute Illnesses

May 04, 2018

Short- and long-term mortality goes up after liberal use of supplemental oxygen, compared with a more conservative oxygen delivery approach, in patients with any of a range of acute illnesses, a meta-analysis of randomized trials shows.

The analysis, based on patient-level data from 25 trials, also saw the mortality risk from liberal oxygen as dose-dependent, in that in-hospital and long-term mortality rates climbed with oxygen saturations rising above the threshold of 94% to 96%.

No apparent gains were associated with the liberal approach; the two groups didn't differ in prevalence of hospital-acquired pneumonia, hospital length of stay, or subsequent patient disability.

The findings are "distinct" from "the current notion that at worst, liberal oxygen is not beneficial for acute illnesses," note the authors, led by Derek K. Chu, MD, and Lisa H.-Y. Kim, MD, both from McMaster University, Hamilton, Ontario, Canada.

Routine delivery of oxygen in many acute conditions has been dogma passed down over time across many medical subspecialties, Kim told | Medscape Cardiology. The approach had come to be seen as probably helpful "but at worst, of no benefit."

But the current analysis, she said, shows about a 20% increase in hospital mortality when oxygen is given to nonhypoxic critically ill patients, "which is quite significant considering how many patients get exposed to supplemental oxygen in the hospital setting."

Across the studies, she said, there was one death for about every 71 patients given liberal oxygen, a number needed to harm that rose to about 100 in the studies limited to patients with acute coronary syndromes.

"Oxygen shouldn't be treated as a harmless intervention," Kim said. "Now that we have high-quality data that too much oxygen is harmful, we should be really cautious in administering supplemental oxygen. We should really be, frankly, treating it as any prescribed drug, recognizing that it has both benefits and adverse effects."

The meta-analysis, published April 26 in the Lancet, encompassed 16,037 patients, most of whom participated in studies of cardiovascular disease or stroke. But it also included studies of patients with sepsis, traumatic injury, emergency surgery, and other critical conditions.

Its findings are consistent with those of a lot of prior research in different medical fields. They include the recent SO2S study, which saw no advantage or harm from administering low-dose oxygen to nonhypoxic patients with acute stroke; DETO2X-AMI, which produced similar findings in nonhypoxic patients suspected of having MI; and AVOID, in which nonhypoxic patients with acute MI who received oxygen experienced cardiac injury according to biomarkers and imaging.

That routine oxygen in critical care helps "is kind of an old wives' tale that no one ever tested. We just thought giving oxygen liberally would be the right thing to do," said Timothy D. Henry, MD, Cedars-Sinai Heart Institute, Los Angeles, California.

In an interview, Henry, who isn't associated with the meta-analysis, called it "a thoughtful advance in medicine." But he cautioned about its limitations, including variable definitions of "liberal" and "conservative" oxygen delivery across the studies, and that one of the study's admitted strengths, its heterogeneous population, also complicates how the analysis is interpreted.

Henry also took issue with the magnitude of the observed mortality risk from liberal oxygen. "I'm not 100% convinced that there's harm. The mortality difference is really small. But I am convinced that there's no benefit."

According to an accompanying editorial, even though the most conservative lower 95% confidence interval (CI) for the in-hospital mortality increase was low, at 1.03, "this could translate to a large number of deaths overall," given "the ubiquity of liberal oxygen administration for acute illness."

"Wider recognition that hyperoxaemia — similar to hypoxaemia — is a deleterious physiological insult should immediately affect clinical practice," writes John W. McEvoy, MD, Johns Hopkins University, Baltimore, Maryland.

The current meta-analysis, named Improving Oxygen Therapy in Acute-illness (IOTA), included only trials of adults admitted to a hospital with critical illness who were randomly assigned to liberal or conservative supplemental oxygen approaches.

The liberal-oxygen groups were defined as those with the higher objectively measured oxygen target; treatment groups using the lower target or allowing only room air were called conservative. Kim said there were no statistical interactions among the different approaches, so the variation likely didn't affect the conclusions.

The analysis excluded patients with chronic respiratory diseases or psychiatric disease, those on extracorporeal life support, and those receiving hyperbaric oxygen therapy or undergoing elective surgery.

Mortality data were available from 23 of the 25 trials included in the analysis, covering 15,754 patients.

Table. Relative Risks for Mortality, Liberal vs Conservative Oxygen Delivery Approaches

Mortality Period Relative Risk (95% CI) P Value
In-hospital 1.21 (1.03 - 1.43) .020
30 d 1.14 (1.01 - 1.29) .033
At longest follow-upa 1.10 (1.00 - 1.20) .044
aMedian, 3 months.


Mortality increases with the liberal approach were independent of whether oxygen was delivered in the intensive care unit or other settings or with invasive or noninvasive mechanical ventilation, the duration of oxygen delivery, and whether patients with hypoxemia at baseline had been excluded for the trial's primary endpoint, the group writes.

Henry said that regardless of the study's limitations, it shows that "there's no reason to use high-dose oxygen" and guidelines should be updated to reflect that.

"Personally, this should change care. For myocardial infarction, the guidelines should say that you should use conservative oxygen with a goal of 94% to 96% [oxygen saturation], which is their conclusion," Henry said.

Kim said that oxygen saturation cutpoint is indeed what the trials suggest, but "it needs clarification from further studies of just where the threshold lies."

Chu, Kim, their coauthors, McEvoy, and Henry report that they have no competing interests.

Lancet. 2018;391:1693-1705. Article, Editorial

Follow Steve Stiles on Twitter: @SteveStiles2. For more from | Medscape Cardiology, follow us on Twitter and Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: