These propensity score based results show that even after adjustment, HCQ treatments do much worse in observed outcomes. https://www.thelancet.com/cms/10.1016/S0140-6736(20)31180-6/attachment/84423d57-4cf8-41d0-99ca-0e921f2c80ce/mmc1.pdf …
-
Show this thread
-
The usual, and reasonable, concern about propensity score based adjustment is that decision makers--here, the docs choosing whether to use HCQ treatments--know stuff that researchers can't measure. https://www.thelancet.com/cms/10.1016/S0140-6736(20)31180-6/attachment/84423d57-4cf8-41d0-99ca-0e921f2c80ce/mmc1.pdf …
1 reply 0 retweets 2 likesShow this thread -
If docs use HCQ as a Hail Mary pass for worst cases, & if that's hard to measure with observed data, then there's selection on unobservables,a nd it will look like HCQ makes thing worse b/c it is given to those who will have worse outcomes even w/out HCQ. https://www.thelancet.com/cms/10.1016/S0140-6736(20)31180-6/attachment/84423d57-4cf8-41d0-99ca-0e921f2c80ce/mmc1.pdf …
1 reply 0 retweets 1 likeShow this thread -
I (truly) wonder what additional variables docs have that these researchers couldn't/didn't use. From my quick read of the report's Table S7, those variables would have to matter an awful lot to explain away the results in this study. https://www.thelancet.com/cms/10.1016/S0140-6736(20)31180-6/attachment/84423d57-4cf8-41d0-99ca-0e921f2c80ce/mmc1.pdf …
1 reply 0 retweets 1 likeShow this thread -
To wrap up, this study alone makes clear that it's wrong to say there is no evidence that HCQ causes harm. If you accept the research approach used in this study, which is not uncommon, then there is actually powerful evidence. https://www.thelancet.com/cms/10.1016/S0140-6736(20)31180-6/attachment/84423d57-4cf8-41d0-99ca-0e921f2c80ce/mmc1.pdf …
1 reply 0 retweets 1 likeShow this thread -
I'm not a COVID expert, so I don't have an informed opinion as to whether p-matching approach used here is missing variables that could reverse these results. But imho someone who is would have to give some good reasons to disregard this study. https://www.thelancet.com/cms/10.1016/S0140-6736(20)31180-6/attachment/84423d57-4cf8-41d0-99ca-0e921f2c80ce/mmc1.pdf … /fin
1 reply 0 retweets 2 likesShow this thread -
Replying to @gelbach
zeynep tufekci Retweeted Robert M Califf
It’s not a “take”, go to the top of my thread and follow the RCT unblinding.https://twitter.com/califf001/status/1264636314504138752 …
zeynep tufekci added,
2 replies 0 retweets 0 likes -
Replying to @zeynep
I did read your thread! (In fact, I responded to it in at least one place; so did a lot of others, so I don't blame you for not responding). You claim that there is "no evidence", and you dismiss non-RCT evidence. Maybe that's not a "take", but imho it's mistaken.
1 reply 0 retweets 0 likes -
Replying to @gelbach
No don’t read my thread, read the RCT notice that I linked to.
1 reply 0 retweets 0 likes -
Replying to @zeynep
But I was responding to your claim! That RCT evidence is still being collected doesn't eliminate fact that non-RCT evidence exists. I wouldn't repeat this point but for the fact that your thread suggests there's something wrong w/people behaving as if the non-RCT evidence eixts.
1 reply 0 retweets 0 likes
No, no they unblinded and checked. Anyway, it’s in the link!
-
-
Replying to @zeynep
This doesn't respond to my point, which is about non-RCT evidence.
0 replies 0 retweets 0 likesThanks. Twitter will use this to make your timeline better. UndoUndo
-
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.