Determinants and practice variability of oxygen administration during surgery in the U.S., a retrospective cohort study.

Overview

abstract

BACKGROUND: The best approaches to supplemental oxygen administration during surgery remain unclear, which may contribute to variation in practice. We aimed to assess determinants of oxygen administration and its variability during surgery. METHODS: Using multivariable linear mixed-effects regression, we measured the associations between intraoperative fraction of inspired oxygen and patient, procedure, medical center, anesthesiologist, and in-room anesthesia provider factors in surgical cases of 120 minutes or longer in adult patients who received general anesthesia with tracheal intubation and were admitted to the hospital after surgery between January 2016 and January 2019 at 42 medical centers across the U.S. participating in the Multicenter Perioperative Outcomes Group data registry. RESULTS: The sample included 367,841 cases (median [25 th, 75 th] age, 59 [47, 69] years; 51.1% women; 26.1% treated with nitrous oxide) managed by 3,836 anesthesiologists and 15,381 in-room anesthesia providers. Median (25 th, 75 th) fraction of inspired oxygen was 0.55 (0.48, 0.61), with 6.9% of cases <0.40 and 8.7% >0.90. Numerous patient and procedure factors were statistically associated with increased inspired oxygen, notably advanced ASA classification, heart disease, emergency surgery, and cardiac surgery, but most factors had little clinical significance (<1% inspired oxygen change). Overall, patient factors only explained 3.5% (95% CI, 3.5 to 3.5) of the variability in oxygen administration and procedure factors 4.4% (4.2 to 4.6). Anesthesiologist explained 7.7% (7.2 to 8.2) of the variability in oxygen administration, in-room anesthesia provider 8.1% (7.8 to 8.4), medical center 23.3% (22.4 to 24.2), and 53.0% (95% CI, 52.4 to 53.6) was unexplained. CONCLUSIONS: Among adults undergoing surgery with anesthesia and tracheal intubation, supplemental oxygen administration was variable and appeared arbitrary. Most patient and procedure factors had statistical but minor clinical associations with oxygen administration. Medical center and anesthesia provider explained significantly more variability in oxygen administration than patient or procedure factors.