Agreement and Correlation of Arterial and Venous Blood Gas Analysis in a Diverse Population

Abstract

Introduction: Agreement and correlation of arterial blood gases (ABGs) and venous blood gases (VBGs) at near normal values are well described but have not been well validated at extremes of values. We evaluated the agreement and correlation of ABG and VBG at extremes of values and assessed the utility of VBG as a screening tool. Methods: We performed a retrospective, observational study of ABG and VBG collected at the same time. Statistical analysis included Bland-Altman plot analysis, Pearson correlations, and sensitivity/specificity tests. For limits of agreement (LOA), we set a threshold of ±0.10 units for pH and ±2.4 mm Hg for partial pressure of carbon dioxide (pCO2). A threshold of 80% sensitivity was considered appropriate for VBG to be used as a screening tool for abnormal ABG values. Results: There were 1684, 1744, and 1769 paired pH, pCO2, and HCO3¯ results, respectively. Mean difference (MD) for arterial and venous pH was 0.017 (95% confidence interval [CI]: 0.014-0.020; 95% LOA: −0.11 to 0.15), and Pearson correlation was 0.78 ( P < .0001). At venous pH <7.35, MD was 0.025 (95% CI: 0.019-0.030; 95% LOA: −0.12 to 0.17). Sensitivity of venous pH <7.35 identifying arterial pH <7.35 was 85.0%. MD for arterial and venous pCO2 was −2.5 (95% CI: −2.9 to −2.0; 95% LOA: −21.7 to 16.8), and Pearson correlation was 0.70 ( P < .0001). At venous pCO2 >45 mm Hg, MD was −4.1 (95% CI: −4.9 to −3.3; 95% LOA: −25.7 to 17.5). Sensitivity of venous pCO2 >45 mm Hg identifying arterial pCO2 >45 mm Hg was 67.6%. MD for arterial and venous HCO3¯ was −0.3 (95% CI: −0.4 to −0.2; 95% LOA: −4.8 to 4.2), and Pearson correlation was 0.90 ( P < .0001). Conclusions: Venous blood gases cannot reliably replace ABGs due to poor agreement in acidemia and hypercarbia. Venous blood gases can be used as a screening tool for acidemia but are unsuitable for hypercarbia.