Passive leg raising can predict fluid responsiveness in patients placed on venovenous extracorporeal membrane oxygenation

INTRODUCTION In ICUs, fluid administration is frequently used to treat hypovolaemia. Because volume expansion (VE) can worsen acute respiratory distress syndrome (ARDS) and volume overload must be avoided, predictive indicators of fluid responsiveness are needed. The purpose of this study was to determine whether passive leg raising (PLR) can be used to predict fluid responsiveness in patients with ARDS treated with venovenous extracorporeal membrane oxygenation (ECMO). METHODS We carried out a prospective study in a university hospital surgical ICU. All patients with ARDS treated with venovenous ECMO and exhibiting clinical and laboratory signs of hypovolaemia were enrolled. We measured PLR-induced changes in stroke volume (ΔPLRSV) and cardiac output (ΔPLRCO) using transthoracic echocardiography. We also assessed PLR-induced changes in ECMO pump flow (ΔPLRPO) and PLR-induced changes in ECMO pulse pressure (ΔPLRPP) as predictors of fluid responsiveness. Responders were defined by an increase in stroke volume (SV) > 15% after VE. RESULTS Twenty-five measurements were obtained from seventeen patients. In 52% of the measurements (n = 13), SV increased by > 15% after VE (responders). The patients' clinical characteristics appeared to be similar between responders and nonresponders. In the responder group, PLR significantly increased SV, cardiac output and pump flow (P < 0.001). ΔPLRSV values were correlated with VE-induced SV variations (r² = 0.72, P = 0.0001). A 10% increased ΔPLRSV predicted fluid responsiveness with an area under the receiver operating characteristic curve (AUC) of 0.88 ± 0.07 (95% confidence interval (CI95): 0.69 to 0.97; P < 0.0001), 62% sensitivity and 92% specificity. On the basis of AUCs of 0.62 ± 0.11 (CI₉₅: 0.4 to 0.8; P = 0.31) and 0.53 ± 0.12 (CI₉₅: 0.32 to 0.73, P = 0.79), respectively, ΔPLRPP and ΔPLRPO did not predict fluid responsiveness. CONCLUSIONS In patients treated with venovenous ECMO, a > 10% ΔPLRSV may predict fluid responsiveness. ΔPLRPP and ΔPLRPO cannot predict fluid responsiveness.