Fiberoptic bronchoscopy (FOB) is an important tool for the diagnosis of pulmonary diseases, more particularly in infectious pneumonia. In patients with severe acute hypoxemic respiratory failure, FOB may be contra-indicated until the patient is intubated and control of its oxygenation obtained. In the literature several authors showed that performing FOB under non invasive ventilation (NIV) preserved oxygenation of the patient; and the recent French Consensus on NIV recommends performing FOB under NIV in patients with acute hypoxemic respiratory failure.

Nevertheless this procedure remains uncomfortable in most patients with respiratory failure. In addition, patient's agitation may lead to desaturation, and compromise the realization of FOB.

In acute hypoxemic respiratory failure, FOB with BAL can be life threatening because of a deep oxygen desaturation. Performing FOB under NIV has been shown efficient to prevent episodes of desaturation. Several modes of ventilation have been tried ; Continuous Positive Airways Pressure and Pressure Support Ventilation both showed superiority when compared to oxygen supplementation alone. The recent French consensus on NIV recommends performing FOB under NIV in hypoxemic conditions.

Efficacy of NIV can be altered by patient agitation or asynchronism with the ventilator. Though, some authors have proposed the use of sedative agents like remifentanyl during NIV.

To our knowledge, none study has evaluated the beneficial effect of sedation during a FOB with BAL under NIV.

Propofol is an anesthetic agent which can be administered in TCI, to maintain a constant concentration in the target cerebral compartment. Propofol is widely used in different types of anesthesia. Using the same concept, authors have reported its use in sedation, with smaller target concentrations. TCI sedation with propofol preserves patient's spontaneous ventilation, and does not alter hemodynamic parameters.

Our hypothesis is that sédation by TCI with propofol will lead to best oxygen saturation than without this intervention in patients with acute hypoxemic respiratory failure and undergoing FOB with BAL under NIV.

Thus, we designed a randomized controlled study. Eligible patients will be randomized in two groups. In the treatment group, patients will receive sedation with propofol during the procedure of FOB under NIV. In the other group, patients will receive a placebo during the same procedure. Patients and fibroscopist won't be aware of the randomization group

Inclusion Criteria:

• Acute respiratory failure defined by clinical signs of respiratory failure (polypnea, use of accessory respiratory muscle) and a PaO2/FiO2 ratio < 250
• Need for a diagnosis FOB with BAL
• Informed consent signed

Exclusion Criteria:

• Contraindication of NIV
• FOB with bronchial biopsies
• Acute coronary syndrome
• Thrombopenia < 30.000 / mm3 despite platelets transfusion
• Coagulation disorders
• PaO2/FiO2 ratio < 80 under NIV
• Persistent respiratory acidosis under NIV (pH < 7,32)
• Propofol allergy
• Xylocaïne allergy
• Pregnancy
• Age < 18 years or > 90 years
• Weight > 150 kg or < 30 kg
• Inclusion in another clinical protocol