Plethysmography variability index for prediction of fluid responsiveness during graded haemorrhage and transfusion in sevoflurane-anaesthetized mechanically ventilated dogs

Objective To examine the accuracy of plethysmography variability index (PVI) as a noninvasive indicator of fluid responsiveness in hypovolaemic dogs. Study design Prospective experimental study. Animals Six adult healthy sevoflurane-anaesthetized Beagle dogs. Methods Dogs were anaesthetized with 1.3-fold their individual minimum alveolar concentration of sevoflurane. The lungs were mechanically ventilated after neuromuscular blockade with vecuronium bromide. Cardiopulmonary variables including mean arterial blood pressure (MAP), central venous pressure (CVP), transpulmonary thermodilution cardiac output (TPTDCO), stroke volume (SV), perfusion index (PI), pulse pressure variation (PPV), stroke volume variation (SVV) and PVI were determined during six stages of graded venous blood withdrawal (5 mL kg−1 increments) and six stages of graded blood infusion (5 mL kg−1 increments). The cardiopulmonary variables were analysed using paired t test or Wilcoxon signed rank test. Correlations between PPV and SVV or PVI were analysed by linear regression. The accuracy of PPV, SVV and PVI for predicting fluid responsiveness was examined by using receiver operating characteristic curve analysis. A value of p Results Blood withdrawal resulted in significant increases in PPV and PVI and decreases in MAP, CVP, TPTDCO, SV and PI. Blood infusion resulted in significant increases in MAP, CVP, TPTDCO, SV and PI and decreases in PPV and PVI. PPV and PVI showed a relevant correlation (p Conclusions and clinical relevance Noninvasive measurement of PVI predicted fluid responsiveness with moderate accuracy equal to PPV in sevoflurane-anaesthetized mechanically ventilated dogs. Provisional threshold values for identification of fluid responsiveness were PPV ≥ 16% and PVI ≥ 12%. Clinical trials are needed to confirm these threshold values in dogs.