Variations of pulse pressure and central venous pressure may predict fluid responsiveness in mechanically ventilated patients during lung recruitment manoeuvre: an ancillary study

Abstract

Abstract Background Maintaining a constant driving pressure during a prolonged sigh breath lung recruitment manoeuvre (LRM) from 20 to 45 cmH20 peak inspiratory pressure in mechanically ventilated patients has been shown to be a functional test to predict fluid responsiveness (FR) when using a linear regression model of hemodynamic parameters, such as central venous pressure (CVP) and pulse pressure (PP). However, two important limitations have been raised, the use of high ventilation pressures and a regression slope calculation that is difficult to apply at bedside. This ancillary study aimed to reanalyse absolute variations of CVP (ΔCVP) and PP (ΔPP) values at lower stages of the LRM, (40, 35, and 30 cm H20 of peak inspiratory pressure) for their ability to predict fluid responsiveness. Methods Retrospective analysis of a prospective study data set in 18 mechanically ventilated patients, in an intensive care unit. CVP, systemic arterial pressure parameters and stroke volume (SV) were recorded during prolonged LRM followed by a 500 mL crystalloid volume expansion. Patients were considered as fluid responders if SV increased more than 10%. Receiver-operating curves (ROC) analysis with the corresponding grey zone approach were performed. Results Areas under the ROC to predict fluid responsiveness for ΔCVP and ΔPP were not different between the successive stepwise increase of inspiratory pressures [0.88 and 0.89 for ΔCVP at 45 and 30 cm H20 (P = 0.89), respectively, and 0.92 and 0.95 for ΔPP at 45 and 30 cm H20, respectively (P = 0.51)]. Using a maximum of 30 cmH2O inspiratory pressure during the LRM, ΔCVP and ΔPP had a threshold value to predict fluid responsiveness of 2 mmHg and 4 mmHg, with sensitivities of 89% and 89% and specificities of 67% and 89%, respectively. Combining ΔPP and ΔCVP decreased the proportion of the patients in the grey zone from 28 to 11% and showed a sensitivity of 88% and a specificity of 83%. Conclusions A stepwise PEEP elevation recruitment manoeuvre of up to 30 cm H20 may predict fluid responsiveness as well as 45 cm H20. The combination of ΔPP and ΔCVP optimizes the categorization of responder and non-responder patients.