[himinlomax]

There may be some subjectivity at play alright when deciding if borderline patients go to ICU or not, and that certainly lowers confidence in the numbers, but the effect size is quite large.

What would be interesting would be a follow up to see if patients have suffered lasting damages as has been reported; that could be done blind if the data is collected by imaging (for say heart damage) and analyzed by a different group. Though I have no idea how common those damages are.