If point estimate for drug x for covid is 20% relative reduction in mortality, but 95% CI crosses 1, and we decline to prescribed. Then with more studies the CI tightens and point estimate doesn't change. Now clear it works 100 million patients not treated. 2/
-
-
Replying to @K_Sheldrick @GidMK and
Do we have moral culpability for those that died that would have survived? I'm not sure there is such a huge difference between harm caused and harm not prevented. Waiting for more info is still an active decision not to treat. 3/3
2 replies 0 retweets 9 likes -
But the converse is also extremely problematic - HCQ we justified based on similar arguments and it appears to cause modest increases in death rates from covid. If the end result is a treatment that does cause harm, what is the moral culpability?
5 replies 0 retweets 3 likes -
High. Treatment decisions aren't no-risk. There's the argument about harms from drink driving vs punishment (moral culpability) too. 1/
1 reply 0 retweets 2 likes -
Replying to @K_Sheldrick @GidMK and
Like in Australia if you speed drunk through a red light and there's nobody crossing the road you'll get your licence suspended. If a pedestrian happened to be there you get 10 years jail. I think this is wrong, there's no difference in the risks taken, just luck 2/
1 reply 0 retweets 0 likes -
Replying to @K_Sheldrick @GidMK and
I think if the decision made was the best decision that could be made with the best info available at the time then the doctor is guiltless regardless of what comes out. I accept that that's slightly different to just chalking up deaths though. 3/
1 reply 0 retweets 1 like -
Replying to @K_Sheldrick @GidMK and
I think that's different to saying declining to prescribe a drug that's more likely than not to have net benefit, because evidence doesn't meet an arbitrary certainty threshold, is ok. 4/
1 reply 0 retweets 1 like -
Replying to @K_Sheldrick @GidMK and
I guess what I'd say is that I don't see any deaths caused by prescribing HCQ as more or less important than deaths caused by not prescribing dex. But the orthodox model would see the first as more important because we assign more weight to harms of action than inaction. 5/
1 reply 0 retweets 1 like -
Replying to @K_Sheldrick @GidMK and
I recognise though that this is a minority view. Most of my colleagues DO see harms caused by taking weakly evidenced action as more important than harms caused by not taking weakly evidenced actions. 6/6
2 replies 0 retweets 0 likes -
I think in this context you could argue that the harms of not taking action are often better understood than the harms of taking action. Won't always change the calculus, but I think it's an important point
1 reply 0 retweets 2 likes
Moreover, the justification for using low-certainty evidence can apply to a massive net of treatments. I mean, rare cancers are one thing, but the evidence for any particular low-certainty drug for covid is not that different to a dozen or more other treatments
-
-
Yep, and I think we have to have an appropriately comprehensive understanding that isn't just point estimate but considers: - pre-test probability -potential harm + safety of drugs - sensitivity to single results being false - availability of alternative, evidenced therapies
2 replies 0 retweets 0 likes -
Replying to @K_Sheldrick @GidMK and
I guess the other thing is this is actually two separate problems to solve. This can't just be solved at the coalface even by optimal decision making. We also have to fix lack of trial coordination and poor quality that so much research effort adds so little evidence.
2 replies 0 retweets 3 likes - Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.