Electronic supplementary material

NIV was administered with a face mask (Respironics inc., Murraysville, PA, USA or Koo Europe, Torino, Italy) or a helmet [12] (Castar, Starmed inc., Mirandola, Italy), depending on patient age and collaboration. Mask and helmet size were chosen to provide best fit and comfort to the patients. Helmet and face mask use is shown in Fig.1

Ventilator was initially set with positive end expiratory pressure (PEEP) at 5 cmH2O and pressure support (PS) at 10 cmH2O. PEEP and PS levels were progressively increased by step of 2-3 cmH2O, up to a maximum level of PEEP of 12 and a maximum PS level of 20 cmH2O, to obtain oxygen saturation >90%, decrease in oxygen need and the disappearance of accessory muscles activity. The rate of pressurization was set at the highest level tolerated by the patient and the inspiratory trigger was a flow-trigger set at the most sensitive value, without auto-triggering. When pressure control ventilation was needed, according to the evaluation of the attending physician, inspiratory positive airway pressure was applied at 6-8 cmH2O and gradually increased, if needed, up to a maximum of 20 cmH2O. The head of the bed was kept at a 45° angle.

During the first 24 h, NIV was continuously maintained for thelongest possible period, depending on patient tolerance. If the patientsrequired short periods without NIV for comfort reasons, they received 50% inspired oxigen (FiO2), through a Venturi mask for the shortest possible period.During the discontinuation, if peripheral oxygen saturationwas <88% with FiO2 =50%, NIV was immediately restarted.

After the first 24 hours, if oxygenation and clinical statusimproved, the patients were left to breath spontaneously with oxygensupplementation through a Venturi mask for short periods,gradually increasing the time between NIV sessions. Thepatients were considered successfully weaned from NIV ifthey maintained a PaO2/FiO2 ratio ≥ 200, without ventilatory support and in absence of dyspnoea,activation of accessory muscles or paradoxical abdominal movements for at least 24 hours. Our nurses have been trained to relieve the NIV-related discomfort by 1) using a colloid dressings on the different interface pressure point (Duoderm®, ConvaTec Italy, Roma, Italia); 2) unloading the skin-stripes contact points using weight traction during Helmet NIV; 3) by changing the NIV interfaces on a regular basis. If anxiety occurred, mild sedation was obtained with midazolam at very low doses continuous infusion.