Here is my first analysis of the new @IHU_Marseille preprint called 'Early Treatment with Hydroxychloroquine and Azithromycin in 10,429 COVID-19 Outpatients: A Monocentric Retrospective Cohort Study'
Found here: (not sure why in April 2020 folder).
https://mediterranee-infection.com/wp-content/uploads/2020/04/MS-IHU-Preprint.pdf …
-
Show this thread
-
Replying to @MicrobiomDigest @IHU_Marseille
Some additional points: they report adhering to STROBE guidelines, but they don't (to be fair, no one ever does) - no discussion of missing data/errors, statistical methods in the supplementaries, no demographics (even though this was almost certainly collected)
2 replies 2 retweets 25 likes -
The use of the term "IFR" throughout is boringly wrong. Infection fatality rate is the rate using a denominator of ALL INFECTIONS in a population, what they are reporting here is CASE fatality rate (for cases presenting to outpatients)
1 reply 2 retweets 24 likes -
Their interpretation of the odds ratio is wrong (again, very common error) - this would be true if they had calculated relative risks, but they are reporting the results of a logistic regression (to be fair, with such low numbers they are likely to be similar)pic.twitter.com/eDtcT4iWA9
1 reply 1 retweet 14 likes -
The controls in the study are insufficient for any meaningful conclusions. They clearly have access to comorbidity and other data (they calculated CCI/NEWS-2 scores for some patients) so it's weird they didn't uses these
2 replies 3 retweets 17 likes -
The exclusions are bizarre. They excluded patients who were "cured" - what does this mean? Asymptomatic at presentation? Negative PCR? The supplementary file also says those who refused use of data were excluded but there's no mention of this in the main methodspic.twitter.com/UjEZZMWY5q
1 reply 1 retweet 8 likes -
My mistake, this is mentioned in the methods section, just not with the other exclusions
1 reply 1 retweet 8 likes
Logistic regression is a bit problematic with such a tiny sample. You can see this in the ENORMOUS confidence intervals (i.e. male gender has between 29% and 1007% increase in odds)pic.twitter.com/SsZfbc9dbm
-
-
Very difficult to understand what value an unrandomised case series adds at this point in the pandemic with an extant evidence base that includes RCTs.
3 replies 1 retweet 10 likes -
None whatsoever, but that has never stopped the IHU crowd before
1 reply 1 retweet 13 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.