In this study, the investigators aim to validate a non-invasive marker of fluid-responsiveness in preterm infants (term below 37 gestational week) with acute circulatory failure based on standardized abdominal compression. This would allow physicians to identify which patient could benefit from a fluid expansion, thus avoiding a potentially useless or even dangerous fluid expansion, leading to fluid overload. To this end, the investigators will evaluate the diagnostic accuracy (sensitivity and specificity) of stroke volume variation induced by standardized abdominal compression for the diagnosis of fluid responsiveness (based on the gold-standard test: significant increase in cardiac index after fluid expansion).

Fluid expansion is the cornerstone of acute circulatory failure treatment in children and preterm infants with acute circulatory failure. Although this therapy drastically reduced mortality, several studies in recent years have highlighted the adverse effects of excessive fluid expansion especially for preterm infants such as PDA, chronic lung disease and mortality. Currently, the search for indicators to predict fluid responsiveness is a major issue in neonatal intensive care. These indicators are based on Franck-Starling's law: if small changes in preload lead to an increase in cardiac output, then fluid responsiveness can be expected. However, in children, the only validated indicator (respiratory variability of peak aortic velocity) can only be used in the absence of any spontaneous respiratory movement, a rare situation in practice. Recently, two pediatric studies investigated a simple clinical test: hepatic or abdominal compression. This clinical maneuver, by increasing venous return via mobilization of the hepato-splanchnic reserve, induces a transient and reversible preload increase. The evaluation of the hemodynamic effects of this "endogenous fluid expansion" allowed, according to the authors, to accurately predict fluid responsiveness. However, several factors reduce the applicability of these results: the small number of studies on this subject, the smaller volume of fluid used than in clinical practice, and the population studied, composed almost exclusively of children in postoperative of cardiac surgery.

Thus, there is a need of reliable preloads markers to guide intravenous volume rescusitation in neonates.

In this study, the investigators will evaluate the diagnostic accuracy of abdominal compression for the diagnosis of fluid responsiveness in preterm infants (<37WA) with acute circulatory failure, hospitalized in neonatal intensive care unit for a medical or a non-cardiac surgical condition, for whom a fluid expansion was prescribed by the physician in charge. The index test will be the stroke volume variation following a standardized abdominal compression (before fluid expansion). The gold standard test will be the variation of cardiac output between baseline and after fluid expansion, a variation > 15% defining fluid responsiveness.

In this non interventional study of diagnostic accuracy, patients will undergo an extra echocardiographic assessment, but no supplemental blood test or invasive parameters will be collected. Simple clinical parameters will be collected within 4 hours after the fluid expansion.

Inclusion Criteria:

• Premature newborn under 37 weeks of amenorrhea hospitalized in neonatal intensive care
• Prescription of a 10 to 20ml/kg fluid expansion (crystalloid, blood products or albumin) by the physician in charge, whose goal would be to increase cardiac output (based on clinical, biological and/or echographic criteria).

Exclusion Criteria:

• Acute cardiogenic pulmonary edema
• Hemodynamic instability making the delay necessary for abdominal compression and ultrasonography dangerous for the patient
• Non corrected congenital cardiopathy, or inferior to 15 days postoperative. Intra-abdominal hypertension or painful abdominal palpation
• Abdominal surgery in the last 15 days
• Supine position contraindicated or deleterious
• No investigator available to assess ultrasonographic measures
• Impairment of echocardiographic acoustic window or restless patient making ultrasonography impossible

As this is a pilot study in a population of premature infants (term bellow 37 weeks of amenorrhea), the projected number of patients included will be 20, based on the department's active patient file, in order to have a minimum of 40 fluid expansion required for the study.