If anything, the internet has made me cynical and despairing about "research" which is the reporting of an interesting hint some real research needs to be done on.
The biggest factor I think is poorly controlled (other than skewed/incomplete reporting/data) is the fact that the more desperately sick someone is, the more measures get thrown at them; since their outcomes are going to be worse, the measures get a skewed reportage of poor outcomes. A case can be made (and has been in the lay press) that being admitted to the ICU increases your chances of not leaving the hospital alive....OF COURSE.
I think what is causing you so much concern, as it does me, is making sweeping and simplified conclusions on clinical data.
Of all the studies I deal with, clinical studies are probably the most dfficult to really make definate conclusions out of.
For example, this topic.
We know from labratory models that clinical conditions like ARDS and pulmonary fibrosis are multifactorial. Attributing a specific outcome in an emergency setting to a specific intervention in my mind is a fools errand. The study will always be flawed and the conclusions will always highly suspect.
The argument is that clinical studies are required in an emergency setting. But I am not sure I agree with that. With so many variables unaccounted for in a dynamic system, often with multiple simultaneous treatments, it amounts to little more than guessing or the inability to correlate. I am of the mind that any useful studies are going to have to be performed in lab or highly controlled clinical environments and extrapolated to the emergency field. I acknowledge that it is an extremely slow process and at times exceedingly boring.
Otherwise when you look at an outcome, and then claim in a limited population, what you are saying is "we guess right sometimes."
As KB points out in his critique of the oxygen causing major harm study, and I fully agree with, there is no way that high flow oxygen has such a profound effect in such a short period. Particularly when I saw nothing of the severity of the disease process, comorbidities, simultaneous treatments or pathology reports.
The most damning flaw for the conclusion was that oxygen administered by EMS caused death in the ICU, but the ED was permitted any treatment. That is a big unaccounted for variable.
Like I siad though, this is going to be par for the course when you try to answer "what" without "why" or "how."