Predicting fluid responsiveness in ICU patients: from physiology to bedside

Fluid administration in patients with acute circulatory failure might increase cardiac output and restore oxygen supply for the different organs. However, in patients in which extra fluid does not increase cardiac output (the non-responders), this extra fluid is harmful. Predicting whether a patient will benefit from fluid administration therefore is a daily clinical challenge in the intensive care unit. For this reason, predicting fluid responsiveness has become one of the major topics of intensive care research during the last decades. ‘Static’ markers of volume status like blood pressure or central venous pressure have been studied for many years and were found to be unreliable in predicting fluid responsiveness. This thesis therefore focused on the use of ‘dynamic’ indices for predicting fluid responsiveness in mechanically ventilated patients, based on the swings in arterial blood pressure due to heart-lung interactions. First we found that the predictive value of dynamic indices is high in patients ventilated with high tidal volume and without cardiac arrhythmias. We also showed that the use of a lower tidal volume and the presence of cardiac arrhythmias reduced the area under the receiver operator curve to almost half. In order to improve our knowledge of the underlying physiology of heart-lung interaction, and to increase the usability of the dynamic indices in clinical practice with respect to different physiological conditions, we investigated the influence of tidal volume, lung compliance and chest wall compliance on the heart-lung interaction. We found that tidal volume linearly increases intrathoracic pressures and the value of the dynamic indices. We also found that a decrease in thoracic compliance increased intrathoracic pressures and dynamic indices. A change in lung compliance did not change these parameters. These findings were summarized in a physiologically based mathematical model of the interaction between the respiratory and cardiovascular systems, incorporating dynamic indices and fluid responsiveness. The model allows for the simulation of these complex relationships and may predict the effect of volume infusion in specific patients in the future. To further increase applicability in clinical practice, we also showed that the dynamic indices can be measured reliably non-invasively using an inflatable finger cuff.