To explain our rationale...
COVID-19 seems to present as 2 phenotypes: initially high compliance with poor V/Q match and low recruitability, ‘dry’ lungs; versus subsequent low compliance, ‘wet’ lungs, potentially recruitable.

Initial phenotype maybe due to pulmonary vascular pathology, some interstitial water on CXR, but no pulmonary oedema. Hypoxic but often comfortable on HFNO/CPAP because compliance normal.
Stretch receptors happy.

But hypoxia leads to high RR, and prolonged (hours? days?) high spontaneous MV causes lung injury, and likely hastens onset of later phenotype, true ARDS with wet lungs, low compliance, requiring recruitment, PEEP escalation, APRV, proning, ECMO...

First phenotype may also progress to second through natural course of disease.
But beware tolerating CPAP for days in patients with high MV.
Also, observe them closely during a CPAP trial, they deteriorate rapidly.

Proning may also work in first phenotype, by reducing V/Q mismatch. Also iNO seems effective, same reason, different mechanism.

Often first phenotype quite dry anyway, lowish PEEP required, common mis-step is to overPEEP dry lungs. Of course, be wary of net fluid balance, keep lungs dry (the SRF mantra…).
FiO2 seems to remain high for a week. Don’t worry so much, wait it out.

Second phenotype seems to behave classically, so treat as per severe ARDS. If compliance high (even without CT imaging), prone first; if low, try APRV. Not mutually exclusive. If refractory to both then refer ECMO.

Obvs key is to work out where patient is on trajectory on presentation to ICU.
Keep safe all, good luck.