The accuracy of pulse oximetry in emergency department patients with severe sepsis and septic shock: a retrospective cohort study

摘要

Background Pulse oximetry is routinely used to continuously and noninvasively monitor arterial oxygen saturation (SaO₂) in critically ill patients. Although pulse oximeter oxygen saturation (SpO₂) has been studied in several patient populations, including the critically ill, its accuracy has never been studied in emergency department (ED) patients with severe sepsis and septic shock. Sepsis results in characteristic microcirculatory derangements that could theoretically affect pulse oximeter accuracy. The purposes of the present study were twofold: 1) to determine the accuracy of pulse oximetry relative to SaO2 obtained from ABG in ED patients with severe sepsis and septic shock, and 2) to assess the impact of specific physiologic factors on this accuracy. Methods This analysis consisted of a retrospective cohort of 88 consecutive ED patients with severe sepsis who had a simultaneous arterial blood gas and an SpO₂ value recorded. Adult ICU patients that were admitted from any Calgary Health Region adult ED with a pre-specified, sepsis-related admission diagnosis between October 1, 2005 and September 30, 2006, were identified. Accuracy (SpO₂ - SaO₂) was analyzed by the method of Bland and Altman. The effects of hypoxemia, acidosis, hyperlactatemia, anemia, and the use of vasoactive drugs on bias were determined. Results The cohort consisted of 88 subjects, with a mean age of 57 years (19 - 89). The mean difference (SpO₂ - SaO₂) was 2.75% and the standard deviation of the differences was 3.1%. Subgroup analysis demonstrated that hypoxemia (SaO₂ 2 was in the 90-93% range the SaO2 was Conclusions Pulse oximetry overestimates ABG-determined SaO₂ by a mean of 2.75% in emergency department patients with severe sepsis and septic shock. This overestimation is exacerbated by the presence of hypoxemia. When SaO₂ needs to be determined with a high degree of accuracy arterial blood gases are recommended.