Résumé
Background: Predicting fluid responsiveness remains a difficult question in hemodynamically unstable patients. The author's objective was to test whether noninvasive assessment by transthoracic echocardiography of subaortic velocity time index (VTI) variation after a low volume of fluid infusion (100 ml hydroxyethyl starch) can predict fluid responsiveness. Methods: Thirty-nine critically ill ventilated and sedated patients with acute circulatory failure were prospectively studied. Subaortic VTI was measured by transthoracic echocardiography before fluid infusion (baseline), after 100 ml hydroxyethyl starch infusion over 1 min, and after an additional infusion of 400 ml hydroxyethyl starch over 14 min. The authors measured the variation of VTI after 100 ml fluid (Delta VTI100) for each patient. Receiver operating characteristic curves were generated for (Delta VTI100). When available, receiver operating characteristic curves also were generated for pulse pressure variation and central venous pressure. Results: After 500 ml volume expansion, VTI increased >= 15% in 21 patients (54%) defined as responders. Delta VTI100 >= 10% predicted fluid responsiveness with a sensitivity and specificity of 95% and 78%, respectively. The area under the receiver operating characteristic curves of Delta VTI100 was 0.92 (95% CI: 0.78-0.98). In 29 patients, pulse pressure variation and central venous pressure also were available. In this subgroup of patients, the area under the receiver operating characteristic curves for Delta VTI100, pulse pressure variation, and central venous pressure were 0.90 (95% CI: 0.74-0.98, P < 0.05), 0.55 ( 95% CI: 0.35-0.73, NS), and 0.61 ( 95% CI: 0.41-0.79, NS), respectively. Conclusion: In patients with low volume mechanical ventilation and acute circulatory failure, Delta VTI100 accurately predicts fluid responsiveness.