Transpulmonary thermodilution: A revised correction formula for global end-diastolic volume index derived after femoral indicator injection

Abstract

<abstract> <sec><title>Purpose</title><p>Transpulmonary thermodilution (TPTD) is usually performed by jugular indicator injection. In clinical practice, femoral venous access is often used instead, resulting in substantial overestimation of global end-diastolic volume index (GEDVI). A correction formula compensates for that. The objective of this study is to first evaluate the efficacy of the currently implemented correction function and then further improve this formula.</p> </sec> <sec><title>Methods</title><p>The performance of the established correction formula was investigated in our prospectively collected dataset of 98 TPTD measurements from 38 patients with both, jugular and femoral venous access. Subsequently, a new correction formula was developed: cross validation revealed the favourite covariate combination and a general estimating equation provided the final version, which was tested in a retrospective validation on an external dataset.</p> </sec> <sec><title>Results</title><p>Investigating the current correction function revealed a considerable reduction of bias compared to no correction. Concerning the objective of formula development, the covariate combination of GEDVI obtained after femoral indicator injection, age and body surface area is even favoured, when compared to the parameters of the previously published correction formula, as a further reduction of mean absolute error (68 vs. 61 ml/m²), a better correlation (0.90 vs. 0.91) and an increased adjusted R² (0.72 vs 0.78) is noticed in the cross validation results. Of particular clinical importance is, that more measurements were correctly assigned to the same GEDVI category (decreased / normal / increased) using the revised formula, compared with the gold standard of jugular indicator injection (72.4 vs. 74.5%). In a retrospective validation, the newly developed formula showed a greater reduction of bias (to 2 vs. 6 %) than the currently implemented formula.</p> </sec> <sec><title>Conclusions</title><p>The currently implemented correction function partly compensates for GEDVI overestimation. Applying the new correction formula on GEDVI measured after femoral indicator administration enhances the informative value and reliability of this preload parameter.</p> </sec> </abstract>