A useful article! At least it shows people struggling to clarify concepts and terminology. A similar effort may be helpful for working out various effect sizes here as well.
As for ‘arbitrary’, aiming for total liberation from arbitrariness may be as futile as the search for an ‘assumption-free analysis’ (which I once heard advocated at a JSM audience mic). In any scientific or technological undertaking, there is always a large body of “unproblematic background knowledge” that we take for granted as we examine the questions at hand.
From Karl Popper’s The Logic of Scientific Discovery Ch.5, ‘The Problem of the Empirical Basis’
“Uncertain treatment effect” has some value but I’m looking for a phrase that has a good ring to it in the context of Pr(phrase). Don’t think that Pr(more than uncertain benefit) is quite there, but maybe I’m wrong. Maybe we should not call it anything but just say Pr(effect > MCID/k)?
I am not sure where you are on this now or if it fits what you had in mind, but “subclinical benefit” could be something that readers might relate to. As for uncertain, I can think of “potential” or “plausible”. But yeah sticking with the math is usually cleaner I guess.
1 Like
Subclinical benefit has a nice ring to it. We would need to have a modifier so as to exclude a more trivial or zero benefit. Something like non-trivial but subclinical benefit?
1 Like
hmmm… I’ve instinctively (is that allowed) reacted against “subclinical” before. The context was in measurement of a reduction in urine output that was below the change that was used to define acute kidney injury. This was termed “subclinical”, even though we demonstrated a mortality association with the change. i.e. it was a “clinical” effect because there was a measurable value and outcome of clinical relevance to the patient - nothing “sub” about it. I get the impression that for many busy clinicians if it can be classed as subclinical it can be ignored. If “subclinical benefit” were used, how would that be interpreted?
3 Likes