Gideon is being modest here: this is MASSIVE. A single RCT, which has caused 2 meta-analyses to show benefit of ivermectin, is almost certainly FRAUD. Remove that study, and both meta-analyses (Lawrie and Hill) show NO benefit. Astonishing story.https://twitter.com/GidMK/status/1415764372362649601 …
-
Show this thread
-
Replying to @AlastairMcA30
@AlastairMcA30, you're making unsupported claims again. the Hill review included leave-one-out sensitivity analysis, and still found significance with this study's omission.3 replies 0 retweets 4 likes -
Replying to @NeilTStacey
Alastair McAlpine, MD Retweeted Health Nerd
Not really: all benefit rests with essentially 2 studies: Elgazzar and Niaee. One of which is fraud and the other is dodgy. Dr Hill has responded to me that he’s already re-running his meta-analysis. So… very supported.https://twitter.com/gidmk/status/1415906647185125377?s=21 …
Alastair McAlpine, MD added,
Health NerdVerified account @GidMKReplying to @AlastairMcA30 @pash22 @DrAndrewHillYep. It's also worth noting that once you exclude Elgazzar, any potential benefit entirely rests on another very high-risk study Niaee et al, which has some extremely worrying aspects H/T@K_Sheldrick pic.twitter.com/nu7qwQh3wl2 replies 1 retweet 9 likes -
Replying to @AlastairMcA30 @NeilTStacey
Also worth noting that the Hill meta-analysis didn't entirely exclude Elgazzar - the leave-one-out approach excluded either of the estimates (mild/severe) from the paper, but not the entire studypic.twitter.com/aI3LPHnKBW
1 reply 2 retweets 9 likes -
Here's what I get when I run the standard inverse variance random-effects model in Stata 15 using the metan command. Left including, right excluding Elgazzarpic.twitter.com/mWKmIgI1Z5
3 replies 4 retweets 10 likes -
-
Replying to @GidMK @NeilTStacey
Fancy that… it’s like, when you exclude the dodgy trials, there doesn’t seem to be much benefit…
3 replies 1 retweet 5 likes -
And if you include the last one...https://bmcinfectdis.biomedcentral.com/articles/10.1186/s12879-021-06348-5 …
2 replies 0 retweets 0 likes -
That study has lower hospitalization in the IVM group (~0.67 RR) but can't conclude anything either way because of small sample size (35 incidences). Adding it into a meta-analysis would have little effect.
1 reply 0 retweets 0 likes -
According to Gid, it actually does have an effect. It tips the scale to ‘no benefit’ even if you leave the dodgy Niaee study in…
1 reply 0 retweets 2 likes
Yep. The number of deaths is small, but once you remove Elgazzar the benefit is marginal no matter how you run the analysis, so even a small change results in a CI crossing 1
-
-
I'd need a bit of convincing that a study with 35 hospitalizations total significantly affects a meta-analysis of mortality. What were the number of deaths in each group? I couldn't find that number in a quick scan of the paper.
1 reply 0 retweets 0 likes -
4/250 deaths IVM, 3/251 deaths control. It's not an enormous effect, but as I said once you remove the potentially fraudulent research the confidence interval is already close to or crossing 1 anyway depending on how you run the analysis
0 replies 0 retweets 1 like
End of conversation
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.