Point-of-care ultrasound to estimate central venous pressure: a comparison of three techniques.

To determine the most accurate predictor of central venous pressure among three point-of-care venous ultrasound techniques. Cross-sectional study. Medical ICU in an academic medical center. Convenience sample of 67 spontaneously breathing patients who had an intrathoracic central venous catheter to allow measurement of central venous pressure. Measurement of the internal jugular vein height to width ratio (aspect ratio), the inferior vena cava diameter, and the percent collapse of the inferior vena cava with inspiration (collapsibility index) by ultrasound. Complete data for analysis were available in 65 patients, as the inferior vena cava could not be visualized in two patients. A central venous pressure of 10 mm Hg was chosen a priori as a clinically significant cutoff. The range of central venous pressure values was 1-23 mm Hg with a median value of 7 mm Hg. The maximal inferior vena cava diameter correlated moderately with central venous pressure (R = 0.58), whereas the inferior vena cava collapsibility index and the internal jugular vein aspect ratio showed poor correlation (R = 0.16 and 0.21, respectively). The area under the receiver operating characteristics curve (area under the curve) to discriminate a low central venous pressure (< 10 mm Hg) was 0.91 for inferior vena cava diameter (95% confidence interval 0.84-0.98), which was significantly higher than the internal jugular vein aspect ratio (area under the curve 0.76; 95% confidence interval 0.65-0.89) or the inferior vena cava collapsibility index (area under the curve 0.66; 95% confidence interval 0.51-0.80) (p = 0.0001). An inferior vena cava diameter < 2 cm predicted a central venous pressure < 10 mm Hg with a sensitivity of 85% (95% confidence interval 69% to 94%), specificity of 81% (95% confidence interval 60% to 93%), and positive predictive value of 87% (95% confidence interval 71% to 95%). Inferior vena cava collapsibility index was not an independent predictor of central venous pressure after adjusting for inferior vena cava diameter in a multiple linear regression model. Among spontaneously breathing patients largely without vasopressor support, the maximal inferior vena cava diameter is a more robust estimate of central venous pressure than the inferior vena cava collapsibility index or the internal jugular vein aspect ratio.