Patient-ventilator asynchrony is known to frequently occur during lung protective ventilation in patients with ARDS. Previous clinical studies showed that patient-ventilator asynchrony was associated with worse outcome in ICU. Therefore, strategies to reduce patient-ventilator asynchrony need to be established promptly. Several asynchronies, e.g., breath stacking are caused by vigorous spontaneous breathing effort. Recently, the investigators' group found that higher positive end expiratory pressure (PEEP) reduced the intensity of spontaneous breathing effort of in severe ARDS model (rabbits, pigs) and patients with ARDS. Thus, the investigators conjectured that higher PEEP may reduce the intensity of spontaneous breathing effort and thereby reduce patient-ventilator asynchrony during protective ventilation strategy, compared with lower PEEP in patients with ARDS.

The cross-over study will enroll 10 participants fulfilled with the criteria of Berlin definition of moderate-to-severe ARDS and under mechanical ventilation in intensive care unit (ICU) at Osaka University Hospital. Informed consent will be obtained by legal representatives. After obtaining informed consent, participants will be randomly assigned to one of two groups: "higher PEEP - lower PEEP" or "lower PEEP - higher PEEP". PEEP, either higher or lower, will be set according to higher or lower PEEP/FIO2 table. The depth of sedation will be monitored quantitively and maintained to be the same throughout the protocol. The intensity of spontaneous breathing activity will be assessed with esophageal balloon manometry and electrical activity of diaphragm. At each PEEP setting, asynchrony index will be calculated.

Inclusion Criteria:

1. Patients ≧ 18 years old
2. Patients with moderate to severe ARDS under mechanical ventilation＊ ＊ Definition of moderate to severe ARDS is as per the Berlin definition (PaO2/FiO2 ≦ 200 mmHg with PEEP ≧ 5 cmH2O)

Exclusion Criteria:

1. Lack of informed consent
2. Continuous neuromuscular blockade at enrollment
3. DNR (do-not-resuscitate)
4. Moribund patient not expected to survive 24 hours
5. Massive hemoptysis
6. Increased intracranial pressure (> 18 mmHg)
7. Existence or high risk of pneumothorax
8. Known pregnancy
9. Actual body weight exceeding 1 kg/cm
10. Patient judged to be inappropriate for the trial by intensivist

Participants, their legal representatives, and outcomes assessor are blinded.