To determine whether early endotracheal intubation compared to late endotracheal intubation affects clinical outcome (duration of invasive mechanical ventilation, mortality).

Non-invasive respiratory support, especially HFNO is usually well-tolerated, especially in COVID-19 patients and can be applied for a prolonged period of time. However, there is a heated debate whether later/delayed endotracheal intubation (due to HFNO / NIV) increases the risk of lung injury, so-called patient self-inflicted lung-injury (P-SILI). The concept of P-SILI assumes that high forces applied to the lung resulting from strenuous breathing effort generated by the patient exacerbates lung-injury and is associated with adverse clinical outcome. It is therefore important to investigate if a longer period of non-invasive respiratory support before invasive mechanical ventilation is associated with adverse clinical outcome. Thus, it is unknown if delaying invasive ventilation worsens clinical outcome. In this study, the research question is whether early compared to late endotracheal intubation improves clinical outcome (duration of invasive mechanical ventilation, mortality)? The aim is to identify the optimal S/F ratio and respiratory rate thresholds for initiating intubation to either improve survival or reduce the duration of invasive ventilation without compromising survival. This is achieved by comparing 25 dynamic treatment regimes defined by combinations of S/F thresholds (<any, 250, 200, 150, 100) and respiratory rate thresholds (>any, 16, 20, 25, 30). Thresholds indicating less severe illness (e.g., S/F < 250 and RR > 16) are considered 'early intubation,' while thresholds indicating more severe illness (e.g., S/F < 100 and RR > 30) are considered 'late intubation'.

Two sensitivity analyses will be conducted:

1. Restricting to patients without hypercapnia (pCO2 < 45), excluding those with missing pCO2 values or pCO2 > 45.
2. Including only patients with non-missing Glasgow Coma Scale (GCS) values and adjusting for GCS as a confounder.

Additionally, seven subgroup analyses will be conducted:

1. Patients diagnosed with COVID-19
2. Patients with a history of COPD
3. Patients with a history of CHF
4. Patients with BMI ≤ 30
5. Patients with BMI > 30
6. Patients on high-flow nasal oxygen (HFNO) at time zero
7. Patients on non-invasive ventilation (NIV) at time zero

Inclusion Criteria:

• Hospitalized (inpatient or ICU)
• 18 years or older
• Hypoxemic respiratory failure: S/F<250
• Not yet intubated

Exclusion Criteria:

• Do not intubate order
• Presence of tracheostomy

Patients from the United Stated for which data is registered in the National Covid Cohort Collaborative (N3C).