Oxygen supplementation in the inspired mixture is commonly used in critically ill patients and observational studies highlight that those patients remain hyperoxemic for substantial periods during Intensive Care Unit stay. However, exposure to inhaled oxygen-enriched mixtures is widely recognized as potentially harmful and cause of organ damage. Although, the specific level of arterial oxygen partial pressure (PaO2) considered harmful, or the dangerous duration of hyperoxia, is not determined yet as there are no clinical trials on humans that evaluate the appropriate percentage of oxygen considered safe to maintain an adequate tissue oxygen availability.

The study is designed as a multicentre, open-label, two parallel groups, randomized superiority clinical trial. The study will involve 10 European intensive care units and will recruit adult critically ill patients requiring mechanical ventilation with an expected length of stay of more than 72 hours admitted to the Intensive Care Unit. Within the conventional group, participants will receive an inspired oxygen fraction (FiO2) aiming to maintain an oxygen saturation by pulse oximetry (SpO2) equal or major than 98 percentage, accepting an upper limit of PaO2 of 150 mmHg and a lower limit of 60 mmHg. Patients in the conservative group will receive the lowest FiO2 to maintain SpO2 between 94 and 98 percentage, or when available a PaO2 between 60 mmHg and 100 mmHg. The primary objective of this study is to verify the hypothesis that strict maintenance of normoxia improves survival in a wide population of mechanically ventilated critically ill patients compared to the application of conventional more liberal strategies of oxygen administration. Survival will be measured at Intensive Care Unit discharge.

The confirmation of the efficacy of a conservative strategy for oxygen administration in reducing the mortality rate among critically ill patients will lead to a profound revision of the current clinical practice and a rationale revision of the current recommendations would be mandatory, maybe also in other clinical scenarios such as emergency departments.

Objectives: The primary objective of this study is to verify the hypothesis that strict maintenance of normoxia improves survival in a wide population of mechanically ventilated critically ill patients compared to the application of conventional more liberal strategies of oxygen administration. Survival will be measured at Intensive Care Unit discharge. Secondary objectives are to verify the hypothesis that, in a wide population of mechanically ventilated critically ill patients, the strict maintenance of normoxia compared to the application of conventional more liberal strategies of oxygen administration reduces the 90-day mortality, onset of new organ failures (respiratory, cardiovascular, renal and hepatic after 48 hours from Intensive Care Unit admission) during Intensive Care Unit stay, occurrence of nosocomial infections during Intensive Care Unit stay (only microbiologically documented bloodstream, respiratory and surgical site infections will be considered), the length of mechanical ventilation, vasopressor use and Intensive Care Unit stay, the occurrence of Intensive Care Unit acquired weakness and cognitive dysfunction.

Intervention: On the Intensive Care Unit admission, or later when they fulfil eligibility criteria, participants will be randomly assigned to a group of liberal conventional oxygen administration (CONVENTIONAL) or a group of conservative experimental oxygen administration (CONSERVATIVE). Within the conventional group, participants will receive a FiO2 aiming to maintain a SpO2 equal or major than 98 percentage, accepting an upper limit of PaO2 of 150 mmHg and a lower limit of 60 mmHg. Patients in the conservative group will receive the lowest FiO2 to maintain SpO2 between 94 and 98 percentage, or when available a PaO2 between 60 mmHg and 100 mmHg. The changes in oxygen therapy will be according to a nurse order-set. The indications to non-invasive ventilation or tracheal intubation and invasive mechanical ventilation (and to the most appropriate mode of ventilatory support) will be guided by principles of the Good Clinical Practice and by the clinical judgement of the attending physician. No other pharmacological therapy or treatment will be influenced by the study protocol. There are no restrictions to concomitant treatments provided to patients in this study.

Sample size and recruitment: Assuming a two-sided alpha level inferior than 0.05 and a power of 80 percentage, we calculated that 460 patients are needed per arm to detect a relative risk reduction of 40 percentage (absolute risk reduction of 6 percentage) compared to a mortality of 15 percentage observed in conventional group patients in previous studies. The global sample size is established to 1,000 patients. This includes an additional 80 participants to allow for drop-outs and an expected temporal trend of reducing mortality at ICU discharge. Every month, recruitment status will be evaluated, and a newsletter will be disseminated, including any issue arisen. The study will end after 12 months of enrollment plus 90 days for follow up.

Allocation and blinding: A block randomisation will be used with variable block sizes, stratified by centre using a web-based, randomisation interface.

Data collection and management: The study data will be collected along the entire Intensive Care Unit stay in a dedicated Case Report Form, that will be provided by the steering committee with proper options to minimize data entry errors: the data sheet will incorporate unmodifiable fixed intervals of values (for continuous variables) and pre-defined coding system (for binary or categorical variables). Data entry will be performed and double-checked from a dedicated researcher in each centre; in order to limit collection errors, 10 percent of all records will be randomly re-checked from the Principal Investigator in each participating centre. Data will be collected in an electronic Case report Form and transmitted by a dedicated platform to the Research and Innovation Office of the University Hospital of Modena protected by password to prevent unintentional modifications or deletion. Database management and quality control for this study will be under the responsibility of the Clinical Trial Quality Team of the University Hospital of Modena. In addition, each satellite centre will monthly communicate and report via e-mail with the coordinating centre about the number of recruited patients, eventual missing data or missing visit or any kind of problem correlated to data collection. Data related to the study will be stored for eventual further analysis or study purpose for 5 years after the end of the study. All the data about the included patients will be extrapolated from the clinical documentation and recorded in a Case Report Form from an adequately formed researcher. Demographic information (gender, age), co-morbidities, reason of Intensive Care Unit admission, type of admission will be registered at the inclusion; severity of critical illness (quantified by the Simplified Acute Physiology Score II) will be calculated by the data from the first 24 hours of Intensive Care Unit stay. During the entire stay, Sequential Organ Failure Assessment score will be calculated and registered daily. Every partial score will be registered separately (Nervous System, Respiratory, Cardiovascular, Liver, Renal and Coagulative). Blood Gas Analysis results will also be reported: FiO2, PaO2, PaCO2, hydrogen ion concentration, lactates, bicarbonates, Base Excess. Other daily parameters: duration of ventilatory support in hours, need and dose of vasoactive drugs (doses reported in µg/Kg/min), need of renal replacement therapy subsequent to the first 24 hours of Intensive Care Unit stay. The occurrence of Intensive Care Unit-acquired respiratory, blood and surgical site infections and the implicated microorganisms will be registered. Data from routine laboratory test will be reported: haemoglobin, platelets count, white blood cells count, coagulative parameters, parameters for liver and renal function; Central venous oxygen saturation and arterial lactates will be reported when available. The patient will be followed-up until 90 days and for patients who have been discharged from the hospital during this follow-up, the information will be collected with a phone interview. The vital status may be ascertained through public registries, in case of failure of all other ways of contact.

Statistical analysis: The intention to treat population will be considered for the primary analysis. A descriptive statistical analysis will be performed to describe every relevant variable. Kolmogorov-Smirnov normality test will be performed in order to verify the distribution of the variables. Results will be expressed in mean ± standard deviation or median and interquartile range as appropriate. Baseline and outcome variables will be compared between the two groups using the Mann-Whitney U test or t-test as appropriate. Categorical variables will be compared using Fisher's Exact test. The effect of conservative oxygen therapy on Intensive Care Unit and long-term mortality will be explored in the intention to treat population by a Kaplan-Meier analysis and Log-Rank for the hazard ratio. The primary and secondary outcomes will be also evaluated in pre-defined subgroups: quartile distribution of Simplified Acute Physiology Score II and Sequential Organ Failure Assessment score (total and for single organ) at admission, patients with respiratory Sequential Organ Failure Assessment score equal or major than 3 (respiratory dysfunction), patients with cardiovascular Sequential Organ Failure Assessment score equal or major than 3 (shock), surgical admissions compared to non-surgical admissions, documented infections at admission and distribution of length of stay in Intensive Care Unit. The relationship between oxygen exposure and Intensive Care Unit mortality will be evaluated according to the quartile distribution of the median value of the daily Intensive Care Unit time-weighted PaO2 values. Every test will be performed considering a two-sided p-value inferior of 0,05 for statistical significance.

Data monitoring: An independent Data Safety Monitoring Board, consisting of 2 experts in clinical research in intensive care and 1 bio-statistic will be established before patient enrolment. The Data Safety Monitoring Board Charter will be prepared by the steering committee and signed by the members of the Data Safety Monitoring Board before the trial commences. The Data Safety Monitoring Board will receive a study report every 3 months starting from the first enrolled patient and will have access to all results and make the appropriate considerations about the appropriateness of the sample size, the efficiency and quality of data collection system. No stopping rules are foreseen, either for utility or for efficacy. Moreover, the Data Safety Monitoring has the right to stop the trial for safety reasons.

Safety: Throughout the course of the study, every effort must be made to remain alert to possible Adverse Events. If this occurs, the first concern should be for the safety of the subject. The Investigator's responsibilities include the following: (a) Daily monitoring of all Adverse Events and laboratory abnormalities, regardless of the severity or relationship to study treatment. Only Serious Adverse events will be reported to the Steering Committee and, then, to Data Safety Monitoring Board and Research and Innovation Office of the University Hospital of Modena; (b) Determine the seriousness, relationship, and severity of each Adverse Event; (c) Determine the onset and resolution dates of each event; (d) Report each serious event by fax or email to the Steering Committee and, then, to Data Safety Monitoring Board and Research and Innovation Office of the University Hospital of Modena within 24 hours of the study site staff becoming aware of the event; ( e) Pursue Serious Adverse Events follow-up information actively and persistently. Follow-up information must be reported to the Steering Committee within 24 hours of the study site staff becoming aware of new information and entered in the electronic Case Report Form; (f) Ensure all Serious Adverse Event reports in the electronic Case Report Form are supported by documentation in the subjects' medical records.

Ethics approval: The entire study protocol, including informative material for the patients and modules for the informed consent, will be evaluated from the Local Ethics Committee from the coordinating centre and from all the collaborating centres. Every intention to modify any element of the original protocol after the first approval will be promptly notified to the Ethics Committee and will be applicated only after its written authorization.

Consent and confidentiality: A written informed consent will be asked to each eligible patient before enrolment. If the patient will be unable to comprehend or to give his consent (because of compromised neurological status), the following consent options are acceptable: (a) A priori consent by a substitute decision-maker; (b) delayed consent from a substitute decision-maker; (c) Delayed consent from the patient; (d) waiver of consent; (e) consent provided by an ethics committee or other legal authority. Which options are available at individual participating sites will be determined by the relevant ethics committee and subject to applicable laws. Every patient is free to leave the study protocol at any stage of the study and can request to retire his consent and, consequently, to ask the elimination of all his data from the database. Data about personal and private information, included sensible data, will be treated following current legislation on data protection; patients will be identified with a coding system and data registered in an anonymous form.

Inclusion Criteria:

• Critically ill patients admitted to participant Intensive Care Units
• Age major than 18 years without regards about sex and ethnicity
• Expected length of Intensive Care Unit stay of more than 72 hours
• Need of any respiratory support (invasive or non invasive mechanical ventilation) at admission and with an expected length of respiratory support major than 6 hours
• Acquisition of informed consent

Exclusion Criteria:

• Pregnancy
• Admission to Intensive Care Unit after elective surgery
• Intensive Care Unit readmission (after a first discharge) in the study period
• Invasive or non invasive mechanical ventilation greater than 12 hours in the 28 days before study inclusion
• Clinical decision to withhold life-sustaining treatment or "too sick to benefit" or patients with a life expectancy of less than 28 days due to a chronic or underlying medical condition
• Previous enrolment in other interventional studies of targeted oxygen therapy
• Acute respiratory failure on chronic obstructive pulmonary disease
• Acute respiratory distress syndrome with a PaO2/FiO2 ratio less than 150
• Long-term supplemental oxygen therapy
• Patients candidate to hyperoxia treatment for reasons including (but not limited to) carbon monoxide poisoning or to hyperbaric oxygen therapy

The study is conceived as single blinded: only the patients will be not aware of the group allocation.