A brief 🧵 about RECOVERY: In addition to lots of people on the internet, the authors themselves make this claim in the preprint... but it is not true—and it's not a matter of opinion, it's quite clearly false. 1/
https://twitter.com/nickmmark/status/1359969730069602304
If we were concerned with whether toci's effect were neutral or beneficial it might be a fair claim; but we are concerned with whether toci helps *or harms*. 2/
We don't have data about the 269 patients who were randomized to toci and didn't receive it. Remember RECOVERY treatment teams were not blinded—they knew they were deciding whether to give toci to their patient. 3/
For example: objective evidence of secondary infection was part of the exclusion criteria for randomization. But what if you merely suspected your patient might have an infection? Maybe you wouldn't give the immunosuppressive agent they were just randomized to receive. 4/
If we acknowledge that there's any element of clinical judgment involved—and there's no evidence I've seen to suggest otherwise—then this doesn't mean the effect size is underestimated, it means the study's confidence is overestimated. 5/5
• • •
Missing some Tweet in this thread? You can try to
force a refresh
