Estimation of the inferior caval vein saturation using high-fidelity non-invasive haemodynamic values and validation of modelled estimates.

Abstract

Monitoring venous saturation allows identification of inadequate systemic oxygen delivery. The aim was to develop a model using non-invasive haemodynamic variables to estimate the inferior caval vein saturation and to determine its prognostic utility. This is a single-centre, retrospective study. A Bayesian Pearson's correlation was conducted to model the inferior caval vein saturation. Next, a Bayesian linear regression was conducted for data from all the patients and from only those with parallel circulation. Venous saturation estimations were developed. The correlation of these estimates to the actual inferior caval vein saturation was assessed. The resulting models were then applied to two validation cohorts: biventricular circulation (arterial switch operation) and parallel circulation (Norwood operation). One hundred and thirteen datasets were collected across 15 patients. Of which, 65% had parallel circulation. In all patients, the measured and estimated inferior caval vein saturations had a moderate and significant correlation with a coefficient of 0.64. In patients with parallel circulation, the measured and estimated inferior caval vein saturation had a moderate and significant correlation with a coefficient of 0.61. In the biventricular circulation cohort, the estimated inferior caval vein saturation had an area under the curve of 0.71 with an optimal cut-off of 49. In the parallel circulation cohort, the estimated interior caval vein saturation had an area under the curve of 0.83 with an optimal cut-off of 24%. The inferior caval vein saturation can be estimated utilising non-invasive haemodynamic data. This estimate has correlation with measured inferior caval vein saturations and offers prognostic utility.