Thread by @walidgellad, We have a real problem now in science if preprints go directly [...]

We have a real problem now in science if preprints go directly to AP feeds in a matter of hours.
That just should not happen.
There could be big holes in this research and only scientists should pay attention to it now, not the public.
https://politi.co/3atgngC via @politico

More deaths, no benefit from malaria drug in VA virus study

With 368 patients, the study is the largest look so far of hydroxychloroquine with or without the antibiotic azithromycin for Covid-19.

https://politi.co/3atgngC

this reporting can have major impacts on recruitment in ongoing clinical trials. just like the remdesivir 'leak' can have impact on who participates in trials.

I will say a few words on why this study can't be as conclusive as this headline seems....

1. the VA is a tremendous resource for studying use of HCQ using observational data. Voluminous data. I'm speaking from experience.

2. but there is still confounding by indication. HCQ goes to those who are sicker. Statistics can't always fix that. Sicker people die.

3. There are funny things that reviewers will ask them about:

-how can you assign hcq exposure based on any use during hospitalization? what if it's started 1 day before they die?

-what about all the missing data

-how do you know they weren't on hcq before admission?

not saying finding won't end up being true, but there are a lot of questions that need to be answered in research like this.

i've said before that preprints and working papers should get 48hrs of public peer review before media picks them up.

one last piece. here is the first paragraph in the about page of MedRxiv

https://www.medrxiv.org/about-biorxiv