Accuracy of different methods in predicting fluid responsiveness in septic patients: a comparison between ultrasound, PiCCO and traditional methods

Objective To compare the accuracy of ultrasound, pulse indicator continuous cardiac output monitoring (PiCCO) and traditional methods in predicting fluid responsiveness in septic patients. Methods Forty-six septic patients of both sexes, aged 18-72 yr, requiring mechanical ventilation treatment in the intensive care unit, were enrolled in the study.Venous blood samples were collected for determination of plasma B-type natriuretic peptide (BNP) concentrations by chemiluminescence assay, and central venous pressure (CVP) was recorded.Stroke volume variation (SVVTTE), distensibility index of inferior vena cava (dIVC) and velocity time integral changes of aortic blood flow (ΔVTI) were measured by ultrasound method.SVVPiCCO and global end-diastolic volume index (GEDVI) were measured by PiCCO method.The patients were divided into negative fluid responsiveness group and positive fluid responsiveness group according to the fluid responsiveness after volume expansion.The receiver operating characteristic curves of the parameters mentioned above in predicting fluid responsiveness were drawn.A consistency check for dIVC, ΔVTI and SVVPiCCO thresholds was conducted by using Kappa statistics.The agreement between SVVTTE and SVVPiCCO was analyzed by the Bland-Altman analysis. Results There were 24 patients in positive fluid responsiveness group and 22 patients in negative fluid responsiveness group.Compared with negative fluid responsiveness group, the plasma BNP concentration, CVP and GEDVI were significantly decreased, and SVVPiCCO, SVVTTE, dIVC and ΔVTI were increased before volume expansion in positive fluid responsiveness group (P<0.05). The area under the curve (95% confidence interval), sensitivity and specificity of the plasma BNP concentration were 0.894 (0.807-0.981), 81.8% and 79.2%, respectively, of CVP 0.859 (0.752-0.965), 81.8% and 79.2%, respectively, of GEDVI 0.772 (0.628-0.915), 72.7% and 75.0%, respectively, of SVVPiCCO 0.965 (0.922-1.008), 95.8% and 81.8%, respectively, of SVVTTE 0.940 (0.874-1.006), 91.7% and 86.4%, respectively, of dIVC 0.964 (0.920-1.008), 83.3% and 95.5%, respectively, and of ΔVTI 0.958 (0.909-1.008), 87.5% and 90.9%, respectively.The Kappa value for dIVC threshold and SVVPiCCO threshold was 0.826, and for ΔVTI threshold and SVVPiCCO threshold was 0.743 (P<0.01). The mean deviation of SVVTTE and SVVPiCCO was 0.209, 95% confidence interval (-2.967-3.385)%, and the limit of agreement (-2.46-2.62)%(P<0.05). Conclusion Ultrasound and PiCCO methods can accurately predict fluid responsiveness, have a good agreement and are superior to the traditional method in septic patients. Key words: Echocardiography; Monitoring, physiologic; Vascular capacitance; Sepsis