Too much of a good thing?

Recent research shows excess oxygen supplementation can harm.


New research highlighting the risks of excess oxygen is raising concerns about customary inpatient practices and pushing physicians toward change.

A meta-analysis of 16,037 acutely ill adults, published in the April 28 issue of The Lancet, found that liberal oxygen administration increased inpatient mortality rates by about 20% compared with conservative oxygen therapy. At the same time, liberal oxygen therapy did not improve other patient outcomes, such as disability, hospital-acquired pneumonia, or length of hospital stay.

Image by Getty Images
Image by Getty Images

“This means that on average, for every 71 patients treated, there's one excess death because of liberal oxygen therapy,” said study coauthor Lisa H-Y Kim, MD, BHSc, an adult respirology clinical fellow at McMaster University in Hamilton, Ontario. “This is quite substantial when you consider the ubiquitous use of supplemental oxygen in health care settings in general. [The study] is the first high-quality evidence showing excess harm with supplemental oxygen. This is distinct from the current notion that liberal oxygen is safe and, at worst, not beneficial.”

The new data add to existing concerns that hyperoxia could be harmful in ways not apparent during routine clinical practice, summarized by a data review in the June 2016 Respiratory Care. For example, hyperoxia following cardiac arrest, traumatic brain injury, and stroke has been shown to worsen outcomes. In addition, excess oxygen during paramedic and ED care of patients with chronic obstructive pulmonary disease (COPD) exacerbations has been associated with more frequent respiratory acidosis and need for mechanical ventilation as well as higher in-hospital mortality, according to the review.

The body of evidence about the harmful effects of excess oxygen continues to grow, said Richard Branson, MS, RRT, professor of surgery emeritus at the University of Cincinnati College of Medicine and a coauthor of the Respiratory Care paper. Mr. Branson said clinicians often provide too much oxygen to patients in part because they do not view oxygen as a drug and thus underestimate its risks.

“While FDA clearly defines medical-grade oxygen as a drug, doctors, nurses, [and] respiratory therapists deliver oxygen frequently without an order often because they believe it can't hurt and because the delivery is simple and noninvasive,” he said. “But the impact of excess oxygen—outside of COPD—is not obvious unless it is investigated.”

A hard practice to break

As early as the 1950s, reports about hyperoxic acute lung injury started to appear in the medical literature, followed by increasing concern in the 1960s over oxygen toxicity in adults and newborns in certain medical settings. Since then, multiple observational studies have examined the risks of excess oxygen, said Ronan O’Driscoll, MD, BsC, consultant physician in respiratory medicine for Salford Royal Foundation NHS Trust in the United Kingdom and a coauthor on British oxygen guidelines.

Despite data showing harmful effects from hyperoxia in a wide range of medical conditions, changes in the medical usage of oxygen have come slowly, Dr. O’Driscoll said. He noted that audits he led for the British Thoracic Society (BTS) found that, even among patients assigned appropriate oxygen saturation targets, hyperoxia is more than twice as common as hypoxia (8.8% of patients were above target by at least 2% vs. 3.0% below by at least 2%).

“Hypoxia is the only condition that I know of where clinicians routinely overcorrect a low reading,” said Dr. O’Driscoll. “However, giving excessive oxygen makes as much clinical sense as overcorrecting hypoglycemia, anemia, hypotension, hyponatremia, hypokalemia, etc. In all of these other instances, clinicians realize that overcorrection is actually harmful, but many clinicians behave differently with oxygen. They think if a little is good, more must be better,” he said.

The use of liberal oxygen in acutely ill patients is a doctrine passed down from one generation of physicians to the next, added Dr. Kim. Making matters more complicated is the broad spectrum of oxygen guidelines that exists across countries and within different medical professions, she said.

“Guidelines for supplemental [oxygen] overall have been inconsistent even across specialties, across different countries, [and] across different health care providers,” she said. “For example, [guidelines differ] from emergency response teams to respiratory therapists to physicians.”

Gauging optimal levels

In the Lancet study, investigators analyzed 25 randomized controlled trials comparing liberal and conservative oxygen therapy in acutely ill adults with sepsis, critical illness, stroke, trauma, myocardial infarction (MI), or cardiac arrest and in patients who had emergency surgery. Across the trials, the baseline median pulse oximetry (SpO2) in the liberal oxygen arm was 96%, with a range of 94% to 99%.

“The 94% to 96% span is where we start to see increased risk of mortality with liberal oxygen therapy,” Dr. Kim said.

However, further research is needed to determine the optimal upper oxygen saturation threshold, she added. In the interim, the Thoracic Society of Australia and New Zealand's 2015 guidelines recommend a target range of 92% to 96%. The BTS, meanwhile, recommends 94% to 98% for most patients and a target of 88% to 92% for patients at risk of hypercapnia.

In the U.S., there are no general guidelines regarding hyperoxia in acutely ill patients, said John William McEvoy, MB, BCh, an assistant professor of cardiology at Johns Hopkins in Baltimore. Hospitals use varying oxygen strategies and there are few, if any, consistent recommendations for safe upper levels of saturation, said Dr. McEvoy, who wrote an editorial in The Lancet accompanying Dr. Kim's analysis.

In 2015, the American Heart Association for the first time encouraged clinicians to consider withholding supplementary oxygen in normoxic patients with suspected or confirmed acute coronary syndrome (ACS) as part of its updated guidelines for cardiopulmonary resuscitation and emergency cardiovascular care.

The association emphasized that the provision of supplementary oxygen to patients with suspected ACS and normal oxygen levels has not been shown to reduce mortality or hasten the resolution of chest pain. In addition, withholding supplementary oxygen in these patients has been shown to minimally reduce infarct size, according to the updated guidelines.

Experts have differing opinions on the recommended oxygen saturation cutoff. Dr. McEvoy said 95% is a reasonable threshold above which supplemental oxygen need not be given to acutely ill adult patients. Mr. Branson said a target range of 92% to 95% is preferable to avoid both hypoxia and hyperoxia. He noted that pulse oximeters are accurate for clinical use but have some variability and that most have a margin of error of 3%. Pulse oximeter accuracy may also be impacted by skin color and perfusion, he noted.

Dr. Kim recommends a lower oxygen saturation limit of 92%. “High-quality evidence, particularly in patients with stroke or MI, shows that patients with SpO2 greater than 90% to 92% should not receive supplemental oxygen,” she said. The evidence regarding patients with an initial SpO2 between 90% and 92% is less certain, she added.

“Although the evidence is the strongest in stroke and MI, I personally think that this is likely generalizable for all acutely ill patients in the hospital,” she said. “Again, the caveat for these recommendations is that they only apply to the patient without evidence of chronic hypercarbic respiratory failure. For this subset of patients, there is a well-established target range of 88 to 92%.”

“It makes sense for protocols to limit the administration of supplemental excess oxygen only to those patients who are actually hypoxic and to consider targeting both a lower and, importantly, an upper bound of safe oxygen levels,” Dr. McEvoy said.

Dr. O’Driscoll noted that a typical young adult has an oxygen saturation of 96% to 98% on room air, and older adults have a wider range of about 94% to 98%. Anxious patients breathing room air may have an oxygen saturation of 99% or even 100%, which is not harmful, he said.

“However, giving supplementary oxygen to a nonhypoxic patient involves delivering an irritant dry gas to the lungs that has many physiological effects and potential side effects,” he said. “There is no reason to aim for saturation above 98%—or even to aim this high—for patients who are on oxygen therapy. Most hypoxic risks to the patient are avoided if the oxygen saturation is maintained above 90%. An oximeter reading above 92 to 93% virtually guarantees that the blood oxygen saturation as measured by a blood gas analyzer will be above 90%.”

Implementing practice changes

Changing oxygen practices starts with wider recognition of hyperoxia at hospitals and vigilance in preventing the condition, Dr. Kim said. The shift requires a group effort by physicians, nurses, respiratory therapists, and emergency personnel.

Particularly for patients with COPD, traumatic brain injury, or chest pain and those experiencing cardiac arrest, physicians should rethink their impulse to provide supplemental oxygen, Mr. Branson said. He noted that conditions that do require high concentrations of oxygen are carbon monoxide poisoning, acute decompression sickness, and profound anemia.

“The first thing [physicians] should ask themselves is, ‘Does the patient need oxygen?’” he said. “The criteria for oxygen is a patient who is hypoxemic and we don't need to guess that. If you put a pulsometer on their finger and it says 86, give them some oxygen, but if it says 93, outside of those [specific conditions mentioned], then don't give the patient oxygen.”

Of course, be mindful that oxygen levels can suddenly fluctuate, Dr. McEvoy said. Experts also don't want clinicians to overreact to the recent research on hyperoxia and provide too little oxygen, he said, since the condition of acutely ill patients can change quickly.

“As such, an acutely ill patient who is short of breath may have an oxygen saturation of 98% one minute and very quickly deteriorate to 88% the next minute,” Dr. McEvoy said. “However, given we now know that excess oxygen can increase mortality, it behooves us to stop providing excess oxygen when not needed and, instead, to increase our vigilance of acutely ill patients so that those who are showing signs of deterioration are observed closely for falling oxygen saturations and receive the needed oxygen when the time is right.”