Thread by @houndcl

(1/) An update of analysis of Dr. Raoult’s #hydroxychloroquine #azithromycin trial on 1061 patients.
mediterranee-infection.com/pre-prints-ihu/
(warning: it’ more like conspiracy theory)

(2/) Actually the preprints have two versions.

“Version 2”
mediterranee-infection.com/wp-content/upl…

Final version
mediterranee-infection.com/wp-content/upl…

There're minor changes in the final version compared to v2. Most are Ct values and HCQ serum concentrations. Other baseline characteristics are the same.

(3/) Another difference is the way that NEWS scores were presented.

Version 2. Mean-Sd-median-range.

Final version. Low-medium-high brackets.

(4/) If we believe the “version 2” data is real, we can estimate the breakdown of each NEWS group based on mean, sd, median, range, number of low/medium/high.

This is what I got.

(5/) The actual numbers may be different, but it should be a generally good estimation, given so many constraints (mean, sd, median, range, num of low/medium/high).

Note: total statistics are slightly off because “5 patients belong to both poor clinical & viral outcome groups”.

(6/) This highlights that most patients had *VERY LOW* NEWS score upon admission.

VERY LOW means NEWS score = 0🤣

Based on ncbi.nlm.nih.gov/pmc/articles/P…, NEWS = 0 means everything LOOKS normal & age < 65.

(7/) this indicates ~ 80% patients are <65y & NO fever & NO short breath when admitted. Although the severity is (slightly) greater than the previous 80-pts cohort (only 15% fever), it is still significantly lower than the global stats.

80% NEWS = 0, what’s their symptoms?

(8/) what if this estimation is much off?

No way, because in order to achieve

948 low &
14 medium &
11 high &
mean = 0.4 in the good outcome group,

they will need *AT LEAST* 663 patients (62%) with NEWS = 0.

In fact, this number is ~800-ish from many simulations.

(9/) I am interested in how Raoult's team will compare this to a "control" group to demonstrate the combination actually works.

Is there such a control group on the earth (with <20% fever)?

(10/) conclusion

If one still believe there is no bias & no selection in this trial, take it.

otherwise, pls trust the RCTs by entities with greater level of responsibility and data integrity. RCT may not leads to magic cure, it doesn't fool you.

the end.

one good thing that Raoult confirmed is that,

there might be NO horrible synthetic lethality between #hydroxychloroqune and #metformin. This combination could lead to ~30% fatality in mice model.
biorxiv.org/content/10.110…

(if not all 5 death is from metformin comedication group).

https://twitter.com/chmn_victor/status/1248332569600540673?s=21

Total death (n=5) explained by age distribution.

Note: this estimation is rougher than NEWS scores, because only mean & sd were provided. Interpret with caution.

https://twitter.com/chmn_victor/status/1248332569600540673