Central venous oxygen saturation is not predictive of early complications in cancer patients presenting to the emergency department.

Abstract

Central venous oxygen saturation (ScvO2) is easily observable in oncology patients with long-term central venous catheters (CVC), and has been studied as a prognostic factor in patients with sepsis. We sought to investigate the association between ScvO2 and early complications in cancer patients presenting to the ED. We prospectively enrolled adult cancer patients with pre-existing CVC who presented to the ED. ScvO2 was measured on their CVC. The outcome was admission to the intensive care unit (ICU) or mortality by day 7. ScvO2 was first studied as a continuous variable (%) with a ROC analysis and as a categorical variable (cut-off at < 70%) with a multivariate analysis. A total of 210 cancer patients were enrolled. At baseline, ScvO2 showed no significant difference between patients who were admitted to the ICU or died before day 7, and patients who did not (67%; IQR 62-68% vs. 71%; IQR 65-78% respectively, P = 0.3). The ROC analysis showed the absence of discrimination accuracy for ScvO2 to predict the outcome (AUC = 0.56). By multivariate analysis, ScvO2 < 70%was not associated with the outcome (OR 1.67; 95% CI 0.64-4.36). Variables that were associated with ICU admission or death by day 7 included a shock-index (heart rate/systolic blood pressure) > 1 and a performance status > 2 (OR 4.76; 95% CI 1.81-12.52 and OR 6.23, 95% CI 2.40-16.17, respectively). This study does not support the use of ScvO2 to risk stratify cancer patients presenting to the ED.