To assess the trending ability of calibrated pulse contour cardiac index (CI_PC) monitoring during haemodynamic changes (passive leg raising [PLR] and fluid loading) compared with transpulmonary thermodilution CI (CI_TD).

Seventy-eight mechanically-ventilated patients admitted to intensive care with calibrated pulse contour following cardiac surgery were prospectively included and investigated during PLR, and after fluid loading. Fluid responsiveness was defined as a≥15% CI_TD increase after a 500ml bolus. Areas under the empiric receiver operating characteristic curves (ROC_AUC) for changes in CI_PC (ΔCI_PC) during PLR to predict fluid responsiveness and after fluid challenge to predict an increase at least 15% in CI_TD after fluid loading were calculated.

Fifty-five patients (71%) were classified as responders, 23 (29%) as non-responders. ROC_AUC for ΔCI_PC during PLR in predicting fluid responsiveness, its sensitivity, specificity, and percentage of patients within the inconclusive class of response were 0.67 (95% CI=0.55–0.77), 0.76 (95% CI=0.63–0.87), 0.57 (95% CI=0.34–0.77) and 68%, respectively. Bias, precision and limits of agreements and percentage error between CI_PC and CI_TD after fluid challenge were 0.14 (95% CI: 0.08–0.20), 0.26, –0.37 to 0.64 l min⁻¹m⁻², and 20%, respectively. The concordance rate was 97% and the polar concordance at 30° was 91%. ROC_AUC for ΔCI_PC in predicting an increase of at least 15% in CI_TD after fluid loading was 0.85 (95% CI: 0.76–0.92).

Although ΔCI_PC after fluid loading could track the direction of changes of CI_TD and was interchangeable with bolus transpulmonary thermodilution, PLR could not predict fluid responsiveness in cardiac surgery patients.