Cavalcanti 2017.

Study characteristics
Methods	Randomised controlled trial conducted in 120 intensive care units (ICUs) from 9 countries (Brazil, Argentina, Colombia, Italy, Poland, Portugal, Malaysia, Spain, Uruguay) Time period of study: November 2011 through April 2017
Participants	1010 participants aged ≥ 18 (120 centres from 9 countries) Included: patients with ARDS (AECC) who receive MV for < 72 hours Excluded: use of vasoconstrictor drugs in increasing doses over past 2 hours or mean arterial pressure (MAP) < 65 mmHg, pneumothorax, subcutaneous emphysema, pneumomediastinum or pneumatocoele, contraindications to hypercapnia (such as intracranial hypertension or acute coronary syndrome), receiving palliative care only; or previously enrolled patients This event‐driven study was designed to continue until 520 events (28‐day deaths) had accrued and the number of events was estimated to provide 90% power, assuming a hazard ratio of 0.75 and type I error of 5%. This hazard ratio was estimated from 2 previous studies (Guérin 2013; Mercat 2008)
Interventions	Control (509) and intervention (501): MV ventilator mode: controlled volume; TV adjusted between 4 and 6 mL/kg PBW; plateau pressure ≤ 30 cmH₂O, respiratory rate (breaths/min) 6 to 35 to achieve arterial pH ≥ 7.30 Target ranges for oxygenation: PaO₂ between 55 and 80 mmHg; SpO₂ between 88% and 95% PEEP Control: PEEP/FIO₂ combination: mean PEEP 11.2 cmH₂O during first 72 hours Intervention: recruiting manoeuvres followed by decremental PEEP titration according to best static lung compliance. Mean PEEP 15.2 cmH₂O during first 72 hours
Outcomes	Primary: mortality at Day 28 Secondary: LOS in ICU, LOS in hospital, VFDs, pneumothorax requiring drainage within 7 days, barotrauma within 7 days, mortality in the ICU, before hospital discharge and at 6 months Exploratory outcomes: death with refractory hypoxaemia within 7 days, death with refractory acidosis within 7 days, death with barotrauma within 7 days, cardiorespiratory arrest on Day 1, need for commencement/increase in vasopressors or hypotension (MAP < 65 mmHg) within 1 hour after randomisation, refractory hypoxaemia (PaO₂ < 55 mmHg) within 1 hour after randomisation, severe acidosis (pH < 7.10) within 1 hour after randomisation
Notes	All study authors received grant support from Program to Support Institutional Development of Universal System (PROADI), from the Brazilian Ministry of Health, to conduct the study. Dr Amato also received grants from Timpel S.A. and Medtronic
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants randomly allocated because study authors used random allocation and a block randomisation scheme
Allocation concealment (selection bias)	Low risk	Allocation concealment ensured via central web‐based system
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Incomplete blinding (blinding of participants but not of personnel) but outcomes not influenced
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No blinding but outcomes not influenced by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Analysis on the basis of the intention‐to‐treat principle. Representatives of 3 participants assigned to the control group withdrew consent to use study data (not included in the final analysis). 23 participants were followed up and were censored between 2 and 6 months (partial data)
Selective reporting (reporting bias)	Unclear risk	LOS in ICU (secondary outcome) and all exploratory outcomes were not originally included in the protocol; they were included in the statistical analysis plan
Other bias	Low risk	Review authors believed the study to be free of other sources of bias