Multivariable haemodynamic approach to predict the fluid challenge response: A multicentre cohort study

BACKGROUND: Beat-to-beat stroke volume (SV) results from the interplay between left ventricular function and arterial load. Fluid challenge induces time-dependent responses in cardiac performance and peripheral vascular and capillary characteristics. OBJECTIVE: To assess whether analysis of the determinants of the haemodynamic response during fluid challenge can predict the final response at 10 and 30 min. DESIGN: Observational multicentric cohort study. SETTING: Three university ICUs. PATIENTS: 85 ICU patients with acute circulatory failure diagnosed within the first 48 h of admission. INTERVENTION(S): The fluid challenge consisted of 500 ml of Ringer's solution infused over 10 min. A SV index increase at least 10% indicated fluid responsiveness. MAIN OUTCOME MEASURES: The SV, pulse pressure variation (PPV), arterial elastance, the systolic-dicrotic pressure difference (SAP-Pdic) and cardiac cycle efficiency (CCE) were measured at baseline, 1, 2, 3, 4, 5, 10, 15 and 30 min after the start of the fluid challenge. All haemodynamic data were submitted to a univariable logistic regression model and a multivariable analysis was then performed using the significant variables given by univariable analysis. RESULTS: The multivariable model including baseline PPV, and the changes of arterial elastance at 1 min and of the CCE and SAP-Pdic at 5 min when compared with their baseline values, correctly classified 80.5% of responders and 90.7% of nonresponders at 10 min. For the response 30 min after starting the fluid challenge, the model, including the changes of PPV, CCE, SAP-Pdic at 5 min and of arterial elastance at 10 min compared with their baseline values, correctly identified 93.3% of responders and 91.4% of nonresponders. CONCLUSION: In a selection of mixed ICU patients, a statistical model based on a multivariable analysis of the changes of PPV, CCE, arterial elastance and SAP-Pdic, with respect to baseline values, reliably predicts both the early and the late response to a standardised fluid challenge. TRIAL REGISTRATION: ACTRN12617000076370.