Hypovolemia is one of major factor of haemodynamic instability. Fluid administration is not totally riskless. Indeed, it can create or inflate pulmonary oedema, alter gaz exchanges and increase post operative respiratory complications. Furthermore, fluid administration is not always followed by a cardiac output increase.

Predicting preload responsiveness before administering fluid by reliable and reproductible methods is necessary in critically ill patients.

Dynamic indicators are approved at the bedside such as passive raising leg test, pulse pressure variation, respiratory variation of the diameter of the superior vena cava. However, all these tests cannot be used for all patients. For example in the cases of spine or pelvis injury, or traumatic brain injury, patients with difficult condition for transthoracic echography.

The investigators hypothesize that EtCO2 (end tidal carbon dioxide) variation after an 15 seconds end-expiratory occlusion test could predict fluid responsiveness in mechanically ventilated patients in the intensive care units.

EtCO2 is a parameter which can be easy to collect, reproductible, and totally non invasive. This method could be especially appropriate for patients for whom the classical test of fluid responsiveness cannot be used

For each patient under mechanical ventilation, answering inclusions and non inclusions criteria, and eligible to a fluid perfusion, the physician in charge collect vital parameters such as cardiac frequency, blood pressure, cardiac output and end tidal carbon dioxide.

The cardiac output is measured by transthoracic echography, or invasive devices such as transpulmonary thermodilution or pulmonary arterial catheter.

The physician achieve a 15 seconds interruption of mechanical ventilation at end expiration, and collect the end tidal carbon dioxide variation.

Fluid perfusion of 500 ml of crystalloid is performed. Then the physician collect the same vital parameters, including a new cardiac output measure.

The patients for whom the cardiac output increased about more than 15 percent are considered as responders, the others are considered as non responders.

Furthermore, socio demographic parameter, reason for admission, parameters of mechanical ventilation, use of vasopressor drugs and water balance are collected.

Inclusion Criteria:

• older than 18 years old
• critically ill patients
• under mechanical ventilation
• whose cardiac output is measurable by transthoracic echography, or monitored by a transpulmonary thermodilution catheter or pulmonary arterial catheter
• eligible to a fluid perfusion, by the physician in charge appreciation

Exclusion Criteria:

• younger than 18 years old
• pregnant patients

Patients in intensive care unit, under mechanical ventilation and eligible to a fluid administration, who are capable to hold a 15 seconds end expiratory occlusion test.