Despite the widespread use of pulse oximetry for intraoperative estimation of arterial oxygen saturation, there is growing evidence that certain patient populations may be vulnerable to inaccurate pulse oximetry measurements and that unrecognized hypoxemia is associated with end-organ damage and adverse outcomes. In this single-center retrospective cohort study, we sought to better elucidate the relationship between intraoperative occult hypoxemia and postoperative mortality among patients undergoing anesthesia and surgery. Data were collected from our departmental data warehouse for adult patients (≥18 years) undergoing anesthesia between 2008 and 2019 with at least 1 intraoperative arterial blood gas recorded. The number of occult hypoxemic events, defined as arterial oxygen saturation (Sao2) of <88% despite oxygen saturation measured by pulse oximetry (Spo2) >92%, were determined. Mortality data were obtained from the Social Security Death Master File and used to determine 30-day and 1-year postoperative mortality. Propensity score overlap-weighted Firth logistic regression and Cox proportional-hazard modeling were performed to analyze whether at least 1 occult hypoxemic event was predictive of 30-day and 1-year mortality. There were 25,234 patients and 62,707 paired readings included in the final analysis. There were 351 patients (1.4%) with at least 1 occult hypoxemic reading. The overall 30-day mortality rate was 3.3% and 1-year mortality rate was 10.2%. In the overlap-weighted models, patients who experienced at least 1 occult hypoxemic event had significantly higher odds of both 30-day mortality (odds ratio [OR] = 2.89, 95% confidence interval [CI], 1.46-5.72, P = .002) and 1-year mortality (hazard ratio [HR] = 1.90, CI, 1.48-2.43, P < .001). There was no significant interaction between occult hypoxemia and self-reported race/ethnicity for predicting mortality. Intraoperative occult hypoxemic events are associated with significantly higher odds of 30-day and 1-year mortality, independent of self-reported race/ethnicity.
Read full abstract