⚠️Breaking: new trial find first drug to reduce #COVID19 mortality- For patients on ventilators the anti-inflammatory dexamethasone cut death risk from 40% to 28%. For patients on O2, death risk cut from 25% to 20%. Modest, but best of all: costs <$10! 🧵 bbc.com/news/health-53…
2) Results from the RECOVERY trial based at Oxford. Recovery has many parts with 11,500 people total- for this dexamethasone trial - 2104 patients given dexamethasone, and 4321 got just usual care. Rate ratio for mortality was RR=0.65 (meaning -35% drop) for ventilator patients. 
3) RR for mortality was 0.80 (-20% drop) among those needing oxygen. Hence drug works for both severe and moderate hospital cases, stronger for severe ventilator. But at least there is a clear mortality benefit, unlike the underpowered remdesivir trials.
4) “Based on these results, one death would be prevented by treatment of around eight ventilated patients or around 25 patients requiring oxygen alone”
5) And again, Dexamethasone is cheap, available from any pharmacy, and easily obtainable anywhere in the world. This is EXACTLY what we need, instead of a $1000 drug like remdesivir that is just marginally effective for shortening illness but not yet fully proven for mortality.
6) However there was no benefit to patients who did not require respiratory support (oxygen or ventilators), the trial found. So it seems this is a drug benefiting medium to later stage COVID illness once someone is in some respiratory distress. Ie. Don’t take unless really sick.
7) Full press release statement. The paper still not published yet. So awaiting further details. recoverytrial.net/files/recovery…
8) Found error by @BBCNews article top: The 5th sentence in the press release describes something called the ‘NUMBER NEEDED TO TREAT’ (NNT) - how many need to treated to save 1 life. But BBC figure suggest that drug prevents 1 in 8 deaths which isn’t true. Treat 8 ➡️ save 1 life!
9) Hooray, @BBCNews has fixed their figure finally! Notice the difference between new figure left and old figure right. NNT is certainly confusing to many. (FYI: statistically it is the inverse of the risk difference between trial arms, the RD being diff of death % in each group)
