Agreement between Pleth Variability Index and oesophageal Doppler to predict fluid responsiveness

Optimisation of stroke volume using oesophageal Doppler is an established technique to guide intraoperative fluid therapy. The method has practical limitations and therefore alternative indices of fluid responsiveness, such as ventilator-induced variation in the pulse oximetric signal (Pleth Variability Index (PVI)) could be considered. We hypothesised that both methods predict fluid responsiveness in a similar way.

Seventy-five patients scheduled for open major abdominal surgery were randomised to fluid optimisation using fluid bolus algorithms based on either PVI (n = 35) or Doppler (n = 39). All patients were monitored with both methods; the non-guiding method was blind. Primary endpoint was the concordance between the methods to predict fluid responsiveness. We also analysed the ability of each method to predict a stroke volume increase ≥ 10% after a fluid bolus, as well as the accumulated intraoperative bolus fluid volume.

PVI indicated a need for fluid in one-third of the situations when Doppler did so, Cohen's kappa = 0.03. A fluid bolus indicated by the PVI algorithm increased stroke volume by ≥ 10% in half the situations. The same was found for the Doppler algorithm. The mean total bolus volume given was 878 ml when the fluid management was governed by PVI compared to 826 ml with Doppler (P = 0.71).

PVI- and Doppler-based stroke volume optimisations agreed poorly, which did not affect the amount of fluid administered. None of the algorithms showed a good ability to predict fluid responsiveness. Our results do not support the fluid responsiveness concept.