Speaking of weak peer-review: Holy shit that stinks!
When you read the paper’s description of the method for collecting data it is just a vague, buzz-word-heavy infomercial for Surgisphere. And the quotes from the guardian article about how they use AI and machine learning to gather data for the database makes no sense.
From my experience and pwrspective, extracting healthcare data in the way they describe would be impossible from the European Union (due to GDPR-laws you simply can’t export private information like individual EHR to a database outside of the EU, even if it is just snippets without immediately identifiable information). Maybe if they had a dedicated server within the EU for analyzing EU data, but even then GDPR is a legal nightmare with regards to sharing data from one place to another - it would take many months to get any legal department to sign off on such a thing. And that’s assuming the hospitals would have the legal paperwork in order to have consent to share the information in the first place (and would want to). All in all it just seems extremely unlikely for them to have managed to include the 50.000 patients from Europe which they claim.
I hadn’t read much into that part of the study’s methodology at first - mostly because I trusted that the peer-reviewers would have checked on such a fundamental thing.
Anyway, these observational studies are not the right study design to answer the question in the first place. The RCTs are beginning to come out. I’m not sure I completely agree with the conclusion (the study is underpowered), but it shows nothing convincing in favor of HCQ, even in the prophylaxis scenario: