The primary evalution criteria had a so few occurrence that it has been a fundamental obstacle to randomize patient. The use of mecanical aide to cough is more widly employ, creating a second limitation.

After the admission in ICU, most patients have to be intubated in order to control haematosis in case of acute respiratory failure, to reduce the metabolic crisis during severe haemodynamic shock, or to protect upper airways in case of impairment of consciousness. After the initial phase of etiological treatment, as soon as the patients no more required the intubation, weanibility has to be checked (thanks to a weaning trial with or without pressure support) before the separation attempt is decided on an evaluation of the overall extubability, based in particular on a subjective assessment of cough strength. The cases of re-intubation can be related to several factors such as: i) a ventilatory insufficiency indicating an imbalance between the muscular pomp function and the mechanical constraint of the chest; ii) an acute cardiogenic oedema; iii) an obstruction of the superior airways, possibly due to an imbalance between the bronchial overload and the cough efficacy.

Preventing extubation failure should avoid exposing such patients to an over-risk of morbidity and mortality due to the consequences of a prolonged invasive ventilation. This prevention can be implemented thanks to an early detection of the patients the most at risk and then a coherent intervention to manage of each risk factor involved. For example, the inspiratory insufficiency can be fixed by the use of Non-Invasive Ventilation (NIV), as proposed for patients developing a final hypercapnia at the end of the weanibility test.

Several studies have been conducted to improve the prediction of extubation failure. As this is notably influenced by the cough efficacy before extubation, it has been proposed to assess the peak flow expiratory during voluntary cough. The Cough Peak Flow could for example replaced some semi-quantitative measures of cough strength associated to a low reproducibility. The most validated threshold for a weak cough is < 60 L/min and it has recently been demonstrated that it remained valuable when directly assessed using the built-in ventilator flow-meter.

In the meantime, new devices of mechanical cough assistance have been developed and are frequently used for patients presenting a chronic neuro-muscular disease affecting their ability to spontaneously clear their airways from an inappropriate bronchial overload. However, the interest of such devices for a systematic use after extubation has not been validated with a sufficient level of evidence to be recommended, in particular because of the bias of the single randomised monocentric study.

The main objective of the study consists in demonstrating in an open multicentre randomised study (focused on the patients with an objective low cough strength) the superiority of a systematic strategy combining mechanical cough assistance and non-invasive ventilation on standard care (manual post-extubation physiotherapy and NIV for restrictive indications) to reduce the re-intubation rate at 48h.

Inclusion Criteria:

1. Patients older than 18 years
2. Patients intubated for more than 24h before the extubation attempt
3. Patients able to follow simple commands (M6 response at the GCS) and to provide a signed agreement (or, in case of neurological impairment, to confirm to a third person their will to be included in the study by their appropriate code).
4. Successful weaning trial with a low pressure support (7 cmH2O) during 30-60 min or longer in case of neuromuscular disorders and with PEEP=0 cmH2O
5. Patients included for their first extubation attempt after a successful weaning trial based on classic weaning criteria (with the exception of: self-extubation, accidental extubation, extubation after less than 24 hours of ventilation).
6. No decision of care limitation on respiratory failure (reintubation considered).
7. Informed consent form signed by the patient (if cognitive capacities are preserved despite intubation) or by relatives (if cognitive and judgment capacities are impaired but with a capacity to respond to instructions sufficient to carry out spirometric measurements).
8. Randomisation if the cough is objectively assessed as weak (CPF < 60 L/min et VT < 0.55 L). In other case, the patient with a strong cough are only included in an observational study with an outcome assessment.

Exclusion Criteria:

1. Patient with tracheostomy
2. Weaning trial conducted without a pressure support technique during the last test
3. Failure of a weaning trial using a pressure support technique
4. Decision not to extubate after a successful weaning trial
5. Extubation of a patient already included after a previous extubation
6. Final extubation
7. Patients not affiliated to French health care system
8. Patients in poor medical condition (hemodynamic, respiratory instability)
9. Patients moribund or with previous decision of care limitation
10. Absence of informed consent document or patient under legal protection
11. Any clinical argument for a laryngeal oedema
12. No possibility of CPF assessment: ventilator non-adapted for assessment, no comprehension of the cough order despite a M6 response at the GCS (no coughing effort initiated).
13. Patient previously using a mechanical cough assistance
14. Deglutition disorders objectified or supposed (severe lesion of the brainstem), with a theoretical risk of tracheostomy
15. Contra-indication to mechanical cough assistance: pneumothorax or any lesion at risk of pneumothorax (such as costal fracture, emphysema, previous pneumothorax or other barotraumatism)
16. Contra-indication to Non-Invasive Ventilation: agitation, non-treated pneumothorax, thoracic wound, severe vomiting, active digestive haemorrhage with hematemesis, severe cranio-facial traumatism with pneumo-encephaly