Acute hypoxemia is common and deadly in resource variable settings. While studies in high income countries (HICs) have indicated a possible benefit to high flow oxygen as compared with standard flow oxygen, rigorous studies in low or lower middle income countries (LMICs) have not been performed. Studies in sepsis have demonstrated that interventions that improve outcomes in one context may actually be neutral or harmful in a different context.

The goal of this study is to test whether high flow oxygen results in better outcomes for hypoxemic adult patients, as compared with standard flow oxygen, in five LMIC hospitals. The main questions it aims to answer are:

1. For hypoxemic adults in these LMIC study settings, does high flow oxygen or standard flow oxygen result in lower mortality?
2. What are the facilitators and barriers to using high flow oxygen in these settings?
3. Does high flow or standard flow oxygen use more oxygen?

Participants will be randomized to receive either high flow oxygen through a large nasal cannula, or to receive standard flow oxygen, through nasal cannulas, face masks, or non-rebreather masks. Researchers will compare the outcomes for the two groups, to see if one group of patients has better outcomes than the other.

The study will also examine how much oxygen is used by the two patient groups, as well as other factors relevant to the feasibility of implementation of high flow oxygen in these sites.

Inclusion Criteria:

• age>=18 years AND
• admitted to a study site hospital within the 24 hours prior to screening AND
• SpO2<90% at time of first assessment OR
• receiving oxygen at time of first assessment

Exclusion Criteria:

• imminent death (high clinical suspicion of death within 24 hours of admission)
• patient or caregiver refusal of study participation
• history of chronic respiratory failure (SpO2<90% or oxygen dependence for at least three months)
• anatomical factors precluding the use of nasal cannula
• intubation or non-invasive ventilation by the clinical team prior to screening for the trial
• known hypoxemia at transferring facility for >48 hours
• lack of availability of either SFO or HFO devices or supplies at the time of randomization.

IPD will not be shared with other researchers for the trial overall. IPD is owned and managed by each of the sites for the trial, in accordance with the data protection requirements of each country where the sites are located. Individual sites may decide to share their own IPD with other researchers.