Suprasternal ascending or descending aortic velocity peak variability assessment to predict fluid-responsiveness in...

Evaluating fluid responsiveness (FR) is crucial in managing critically ill patients. Measurement of respiratory variations of blood flow (V_peak) is physiologically sound, but blood flow sampling through the aortic valve (AV-V_peak) is not always feasible. We assessed the feasibility of suprasternal V_peak (SS-V_peak), at ascending or descending aorta level, as alternative to AV-V_peak.

Observational prospective study in spontaneously breathing healthy volunteers. We report the overall feasibility of AV- and SS-V_peak, and calculated their interchangeability, the mean bias with limits of agreement (LoA) and percentage error (PE). We defined FR as a 10% increase in cardiac output measured non-invasively with finger-cuff method after passive leg raising.

We enrolled 67 volunteers; SS-V_peak was feasible in 65 volunteers (97%), with sampling in the ascending and descending aorta in 22/65 (33.8%) and 43/65 (66.2%) volunteers, respectively. AV-V_peak was feasible in 64 volunteers (95.5%). When both V_peak were obtained (n = 62), interchangeability using a 12% cut-off was 67.7% (poor agreement with kappa coefficient 0.19 [-0.02;0.41]). Clinical concordance at ascending aorta level was non-significantly higher (16/22, 73% vs 26/40, 65%; p = 0.583). Prediction of FR with SS-V_peak using the 12% cut-off was poor: sensitivity 85%; specificity 9%; positive predictive value 82%; negative predictive value 11%. Bland-Altman's analysis revealed a mean bias -2.6% [-4.3%;-1.0%] with LoA ranging from -15.2% [- 18.1%;- 12.4%] to 10.0% [7.2%;12.8%]. The mean PE was 7.87%.

We report excellent feasibility for SS-V_peak, though with moderate interchangeability and accuracy; however, we found poor precision and poor performances in predicting FR in healthy volunteers.