The respiratory therapists at Mount Sinai Beth Israel, New York, know when Lina Miyakawa, MD, starts a week in the ICU, because she turns down the fraction of inspired oxygen (FiO2) levels if patients tolerate it.
Dr. Lina Miyakawa
“Hyperoxia in mechanical ventilation is a topic that’s near and dear to my heart,” Dr. Miyakawa, a pulmonary and critical care medicine specialist at Mount Sinai Beth Israel, said during SHM Converge, the annual conference of the Society of Hospital Medicine. “You can always find ‘wean down FiO2’ in my consult notes.”
While it is believed that humans have built up evolutionary defenses against hypoxia but not against hyperoxia, medical literature on the topic of hyperoxia with supplemental oxygen is fairly young. “In medical school we were taught to give oxygen for anybody with chest pain and concern about acute coronary syndrome,” she said. “This was until recent data suggested harm from liberal oxygen use.”
In a single-center trial of 434 critical care patients with an ICU length of stay of 72 hours or longer, Italian researchers examined the effects of a conservative protocol for oxygen therapy versus conventional therapy on ICU mortality (JAMA. 2016;316[15]:1583-9). The trial was stopped because the patients who were assigned to receive conservative therapy had a significantly lower mortality than the ones who received usual care (P = .01). “The study was not perfect, and the premature stoppage likely exaggerated the effect size,” said Dr. Miyakawa, who was not affiliated with the trial. “However, subsequent retrospective studies continue to support a benefit with conservative oxygen use, especially in different groups of patients. One of note is hyperoxia following cardiac arrest. There’s something called a two-hit model that speaks to worsening ischemia with reperfusion injury after the initial hypoxic event from the cardiac arrest itself” (See Intensive Care Med. 2015;41:534-6).
In a multicenter cohort study that drew from the Project IMPACT critical care database of ICUs at 120 U.S. hospitals between 2001 and 2005, researchers led by J. Hope Kilgannon, MD, tested the hypothesis that post-resuscitation hyperoxia is associated with increased in-hospital mortality (JAMA. 2010;303[21]:2165-71). The study population consisted of 6,326 patients who were divided into three groups: the hypoxic group (a PaO2 of less than 60 mm Hg); the normoxic group (a PaO2 of 60-299 mm Hg), and the hyperoxic group (a PaO2 of over 300 mm Hg). The mortality for the hyperoxic group was 63%, the hypoxic group at 57%, and the normoxic group at 45%.
More recently, the ICU-ROX Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group evaluated conservative versus liberal approaches in providing oxygen to 965 patients who were mechanically ventilated between 2015 and 2018 at 21 ICUs (N Eng J Med. 2020;382:989-98). Of the 965 patients, 484 were randomly assigned to the conservative oxygen group (defined as an SpO2 of 97% or lower) and 481 were assigned to the usual oxygen group (defined as having no specific measures limiting FiO2 or the SpO2). The primary outcome was the number of ventilator-free days from randomization until day 28, while the secondary outcome was mortality at 180 days. The researchers also performed a subgroup analysis of patients at risk for hypoxic-ischemic encephalopathy.
No significant differences were observed in the number of ventilator days between the two group (a median of 21 days in the conservative oxygen group versus 22 days in the usual oxygen group, respectively; P = .80) nor in mortality at 180 days (35.7% vs. 34.5%). However, in the subgroup analysis, patients with hypoxic-ischemic encephalopathy were noted to have more ventilator-free days (21 vs. 0 days), improved 180-day mortality (43% vs. 59%), and less functional impairment (55% vs. 68%) in the conservative-oxygen group.
“The results of this study suggest that conservative oxygen therapy has no additional advantage over standard oxygen therapy, but there may be benefits in those vulnerable to hyperoxia, which warrants further investigation,” Dr. Miyakawa said. “There are a few points to note on this topic. First, many of the previous studies had more liberal oxygen strategies than the ones used in this study, which could be the reason why we are seeing these results. In addition, O2 titration relies on imperfect approximations. PaO2 cannot be measured continuously; we really depend on the SpO2 on a minute-by-minute basis. Critically ill patients can also undergo episodes of hypoperfusion and shock state minute-by-minute. That’s when they’re at risk for hypoxemia. This would not be captured continuously with just O2 saturations.”
Dr. Miyakawa also highlighted the Liberal Oxygenation versus Conservative Oxygenation in Acute Respiratory Distress Syndrome trial (LOCO2) a prospective, multicenter, randomized, open-label trial involving patients with ARDS. It was carried out at 13 ICUs in France between June 2016 and September 2018 in an effort determine whether conservative oxygenation would reduce mortality at 28 days compared with the usual liberal-oxygen strategy (N Eng J Med. 2020;382:999-1008). The researchers detected a signal of increased mortality in the conservative oxygen group (34% vs. 27%), which led to a premature stoppage of the trial. “I’d like to postulate that the higher incidence of proning in the liberal oxygenation group compared to the conservative oxygen group (51% to 34%) may be the reason for the difference in mortality,” said Dr. Miyakawa, who was not affiliated with LOCO2. “This is supported from the 2013 PROSEVA Study Group, which reported that prone positioning in ARDS significantly decreases 28- and 90-day mortality” (see N Engl J Med. 2013; 368:2159-68).
She said that future trials on this topic “will have to address how a particular [oxygenation] target is both set and achieved in each group of patients, particularly those with specific organ injuries. In the meantime, in my opinion, avoiding excess oxygen seems sensible.”
Dr. Miyakawa reported having no financial disclosures.