@richardursomd not had time to look through this thoroughly but it doesn't pass the sniff test.

(SHORT THREAD)

Immediately publication in the corrupted @NEJM raises red flags.

A previously hardly published (first) author suddenly has a 55 hospital collaboration? Unlikely

The author is an intensivist in Brazil, so is seeing patients on ICU. Hardly "mild" cases.

Yes, perhaps he managed to get the 55 hospitals to recruit to his study in the mild phase but they are still hospitalised patients. In Brazil.

This is a public hospital in Brazil.

Now, don't get me wrong. Brazil has some great doctors and hospitals, but it requires a lot of organisation to get 55 hospitals to collaborate to a high quality study. A lot.

Anyway, back to the study.

Patients were meant to be excluded when they had taken the drugs assessed in the study.

Except they weren't. 50% had taken them.

And, wait, what? Over 50% of the patients were on ceftriaxone?

Ceftriaxone is a 3rd generation cephalosporin. We're not talking about your average amoxil here. This is serious stuff. That means those patients are seriously ill.

If you're on ceftriaxone you're not going home any time soon - it's given intravenously.

And another reason that these patients are not "mild" cases.

The median time from onset of symptoms to treatment was 8 days. You can die in 8 days. If you're in hospital after 8 days you're in trouble.

And another reason - the mean hospital stay was 10 days. 10 days is a hell of a long time to stay in a hospital with "mild" disease.

And then there was another weird thing...

The age distribution. Even though we had 10 day hospital stays the majority of the patients were under 60, which is a low risk group for mortality (death rate less than 4%).

75% of this study population - that stayed in hospital 10 days - were under 60.

#thesepeoplearesick

Now here's another weird thing about this study.
NOBODY was lot to follow-up. This is almost unheard of, particular with a telephone follow-up study. The study is saying that everybody - all of 504 people - answered the phone.

I don't believe that. Not for a minute.

The other odd thing is the death rate.
2.7% overall. That's amazingly good for this group of patients.

Remember Brazil's death rate peaked at 7% - for all cases. This group would be heading for well above that.

So, what was in the "standard care" arm?
Who knows....

So, how do we interpret this data?

We can't. It's not possible without the dataset where we can work out whether suitable patients did actually benefit from treatment.

The @NEJM have agreed to mandate data publication but they don't enforce it....

According to the data sharing agreement anybody can access the full data set by contacting the authors on this email address

[email protected]

Good luck. Please share your responses.

In the meantime, don't believe this study negates #hydroxychloroquine. At all.

#lancetgate