Estimating venous admixture using a physiological simulator

Abstract

Estimation of venous admixture in patients with impaired gas exchange allows monitoring of disease progression, efficacy of interventions and assessment of the optimal inspired oxygen fraction. A pulmonary artery catheter allows accurate measurement, although the associated risks preclude its use solely for estimation of venous admixture. Non-invasive methods require assumed values for physiological variables. Many of the required data (e.g. haemoglobin concentration (Hb), base excess, inspired oxygen fraction, arterial oxygen (PaO2) and carbon dioxide (PaCO2) tensions, temperature) are available routinely in the intensive therapy unit. We have compared a typical iso-shunt-style estimation of venous admixture (assuming Hb, base excess, PaCO2 and temperature), and estimation using the Nottingham physiology simulator (NPS), with measured data. When the arteriovenous oxygen content difference (CaO2-CvO2) was assumed to be 50 ml litre-1, the 95% limits of agreement (LA95%) for venous admixture using the NPS were -3.9 +/- 8.5% and using an iso-shunt-style calculation, -6.4 +/- 10.6%. CaO2-CvO2 was 41.1 ml litre-1 in the patients studied, consistent with previous studies in the critically ill. When CaO2-CvO2 was assumed to be 40 ml litre-1, LA95% values were 0.5 +/- 8.2% and -2.1 +/- 10.1%, respectively.