Thread by @ashishkjha

Dexamethasone RecoveryTrial data out

Not yet peer-reviewed but thank goodness for preprint

A few reflections

1. The randomization worked. The groups were pretty balanced

2. They pre-specified their subgroups. That's important

So the results?

Thread

medrxiv.org/content/10.110…

1. Large over-all benefit. Absolute mortality 3% lower for Dexamethasone

2. Subgroups. Now it gets interesting! Tells us a lot about COVID19:

--No 02 support: Mortality looks higher for Dex

--02 only: Mortality 3.5% lower for Dex

--Ventilator: Mortality 11.7% lower for Dex!!

The benefit in 02 only and particularly for ventilated patients are very large

I can't find much wrong with approach or analysis but will see if others do

But this study hints at a bigger point about this disease

.@jeremyfaust and others have raised the question -- why is this different? There was no similar benefit for SARS or MERS. Why for COVID?

Key line from paper:

Beneficial effect...is dependent on using the right dose, at the right time, in the right patient

4/5

Early in disease course: key is controlling virus and steroids likely harmful

Later in disease course: Immune response may be the problem and drugs like Dexamethasone helpful

That means that if you take the drug too early, it may end up doing more harm than good

5/6

Finally:

This is great. A fabulous study done under the most difficult of circumstances by @UniofOxford researchers within @NHSuk

Huge bow to the Recovery Trial investigators

One more thing

Because we can't edit in twitter, a minor clarification.

The groups in this trial are pretty balanced.

But even if they weren't -- doesn't mean that randomization was flawed.