Background: Routine use of central venous blood gases (VBGs) may reduce complications from prolonged arterial cannulation. We investigated the reliability of the VBG as a substitute for arterial blood gas (ABG) in multiple care settings.
Methods: We developed a VBG adjustment rule of ABG pH = VBG pH + 0.05, ABG CO(2) = VBG PCO(2) -5 mm Hg from prior studies and validated this relationship with simultaneous venous and arterial blood obtained from 187 medical/surgical intensive care, cardiac catheterization laboratory, and coronary care unit patients with central venous access.
Results: The overall accuracy of a normal adjusted VBG (aVBG) to predict a normal ABG was 90%. After adjustment, the mean systematic difference (bias) between ABG and VBG pH decreased from 0.035 +/- 0.02 to -0.015 +/- 0.02 and PCO(2) bias decreased from -4.5 +/- 3.5 to 0.5 +/- 3.5. Intraclass correlation coefficients for agreement improved after applying the adjustment rule to venous pH (from 0.84 to 0.93, P < .001) and PCO(2) (from 0.66 to 0.84, P < .001). Overall diagnostic accuracy of VBG improved from 45% to 74% after adjustment. Multiple logistic regression demonstrated that the factor independently associated with discrepancy between VBG and ABG diagnoses was an abnormal aVBG (OR 6.8, 95% CI 2.8-16.5).
Conclusions: Because of the high agreement between a normal aVBG with a normal ABG and the small bias between these tests, we recommend use of the adjusted central VBG.